Aspects of the historical phonology of Manx

Christopher  Lewin

This thesis elucidates some of the hitherto poorly understood aspects of the diachronic development of Manx phonology. By tracing phonological changes from earlier varieties of Gaelic, and within the attested period of written and recorded Manx, it frames these developments within the wider contexts of Gaelic dialectology and historical linguistics. Manx provides an important source for understanding the linguistic development of the Gaelic languages. A lack of systematic treatments and reliable datasets for the language, however, has obscured this fact and led to its neglect within Gaelic studies. The thesis focuses, in particular, on the development of the language’s prosody, suprasegmental features, vowel system and sonorants, the latter having a particular bearing on vowels. Five principal methodologies are deployed to investigate these topics: • Re-evaluation of existing descriptions and datasets provided by previous scholarship, especially those collected by Rhŷs in the 1880s and 1890s, and material from the last generation of speakers presented by Broderick in his Handbook of Late Spoken Manx. • Interpretation of the evidence of the two main Manx orthographies and non-standard variations thereof. • Analyses based, as far as possible, on the whole attested lexis of the language, making use of Cregeen’s and Kelly’s dictionaries. • Quantitative approaches to all of these sources of data where appropriate. • Instrumental phonetic analysis of recordings of the terminal speakers of Manx. Chapter one places Manx in its historical and dialectological context, reviews previous scholarship, discusses the primary sources, and introduces the interpretative difficulties of the orthographies. Chapter two examines developments in the short and long vowels, and the impact of the consonant system on vowel changes. Chapter three examines the development of the vowels ao(i) /əː/ and ua(i) /uə̯/ in Manx. The written evidence, description and recorded data are complex, and some scholars have claimed that these vowels fell together with one another and with other vowels. It will be shown that these vowels in fact remained contrastive for the most part in Late Manx. Chapter four investigates developments in the sonorant consonants, especially the R, L and N phones. Changes in vowels preceding historically tense sonorants are also examined, as well as the origins and spread of the phenomenon of preocclusion. Chapter five examines suprasegmental and prosodic features including stress shift, unstressed long vowel shortening, and the conditioning factors for these. Chapter six provides concluding remarks assessing the thesis’ contribution to current scholarship, and the prospects for future research. [This doctoral thesis was completed in the University of Edinburgh with supervision from Dr William Lamb (Celtic and Scottish Studies) and Dr Pavel Iosad (Linguistics and English Language). I am grateful to the AHRC Centre for Doctoral Training in the Celtic Languages for funding the research.]

Aspects of the historical phonology of Manx Christopher Lewin Tràchdas airson ceum Dotair Feallsanachd Oilthigh Dhùn Èideann Thesis presented for the degree of Doctor of Philosophy The University of Edinburgh 2019 ii Declaration Tha mi a’ dearbhadh gur mise a-mhàin ùghdar an tràchdais seo, agus nach deach an obair a tha na bhroinn fhoillseachadh roimhe no a chur a-steach airson ceum eile. I confirm that this thesis has been composed solely by myself, and that the work contained within it has neither previously been published nor submitted for another degree. Christopher Lewin iii iv Geàrr-chunntas ’S e a tha fa-near don tràchdas seo soilleireachadh a thoirt seachad air grunn chuspairean ann an cinneachadh eachdraidheil fòn-eòlas Gàidhlig Mhanainn nach robhas a’ tuigsinn gu math roimhe seo. Le bhith a’ leantainn atharrachaidhean fòneòlach bho àm na Seann Ghàidhlig is na Meadhan-Ghàidhlig air adhart, agus taobh astaigh ùine na fianais sgrìobhte agus clàraichte a th’ againn airson Gàidhlig Mhanainn fhèin, tha e a’ suidheachadh nan cinneachaidhean seo ann an co-theacsa farsaing a thaobh rannsachadh dhualchainntean is cànanachas eachdraidheil na Gàidhlig. ’S e tobar cudromach de dh’fhiosrachadh airson cinneachadh nan cànanan Gàidhealach a thuigsinn a tha ann am fianais Gàidhlig Mhanainn. Tha cion sgrùdaidhean siostamach is sheataichean-dàta earbsach airson na cànain air seo a chumail am falach ge-tà, agus air fàgail gun do rinneadh dearmad oirre am broinn raon rannsachaidh na Gàidhlig. Tha cuideam air leth air a thoirt anns an tràchdas air atharrachaidhean ann am feartan prosaideach agus os-mhìreach, air siostam nam fuaimreagan, agus cuideachd air na consain shonarach, a tha dlùth-cheangailte ris na fuaimreagan. Tha na modhanrannsachaidh a’ tarraing air còig prìomh thùsan:  Ath-mheasadh air tuairisgeulean is seataichean-dàta a tha air an toirt seachad le sgoileirean mu thràth, gu h-àraid na chaidh a thrusadh le Rhŷs anns na 1880an is 90an, agus clàraidhean den ghinealaich mu dheireadh de luchdlabhairt a tha air an toirt am follais le Broderick anns an Handbook of Late Spoken Manx.  Eadar-mhìneachadh na fianais a tha an lùib an dà phrìomh dhòigh-litreachaidh a tha air an cleachdadh gu Gàidhlig Mhanainn a sgrìobhadh, agus tionndaidhean neo-ghnàthach dhiubh.  Sgrùdadh stèidhte cho fad ’s a ghabhas air stòr-fhaclan na cànain air fad, a’ cur gu feum faclairean Cregeen agus Kelly bhon naoidheamh linn deug.  Measadh uimhireil air na tùsan dàta seo gu lèir far a bheil sin iomchaidh. v  Sgrùdadh innealach fogharach air clàraidhean den luchd-labhairt mu dheireadh. Tha Caibideil 1 a’ toirt seachad cunntas goirid air eachdraidh Ghàidhlig Mhanainn is a co-theacsa cànanach an taca ri dualchainntean Gàidhlig eile, a’ measadh obair sgoileirean an ama a dh’fhalbh, a’ toirt sùil air na bun-thùsan, agus a’ mìneachadh nan duilgheadasan is nan cothroman a tha an lùib nan dòighean-litreachaidh dhan sgoileir. Tha Caibideil 2 a’ sgrùdadh chinneachaidhean anns na fuaimreagan goirid is fada, agus a’ bhuaidh a th’ air a bhith aig siostam nan consan air na cinneachaidhean sin. Tha Caibideil 3 a’ coimhead gu mionaideach air cinneachadh nam fuaimreagan ao(i) /әː/ agus ua(i) /uә̯/ ann an Gàidhlig Mhanainn. Tha an fhianais sgrìobhte, na tuairisgeulan agus an dàta clàraichte gu math toinnte, agus tha cuid de sgoileirean a’ cumail a-mach gu robh an dà fhuaimreig seo air tuiteam còmhla ri chèile agus ri fuaimreagan eile. Leigear fhaicinn gu robhas fhathast a’ cumail suas eadardhealachadh eadar na fuaimreagan seo airson a’ chuid as motha anns a’ Ghàidhlig Mhanannaich Anmoich. Tha Caibideil 4 a’ coimhead air cinneachaidhean ann an siostam nan consan sonarach, gu h-àraid na fuaimean R, L is N. Thathar cuideachd a’ sgrùdadh nan atharrachaidhean ann am fuaimreagan ro na sonaraich theanna eachdraidheil, agus a’ toirt sùil às ùr air pròiseas an ro-dhùnaidh. Tha Caibideil 5 a’ gabhail beachd air feartan os-mhìreach agus prosaideach, a’ gabhail a-steach gluasad a’ bheum agus giorrachadh fhuaimreagan fada gun bheum, agus na factairean cumhachaidh a thug buaidh orra seo. Tha Caibideil 6 a’ toirt seachad cho-dhùnaidhean far am measar na tha an tràchdas a’ cur ri sgoileireachd an latha an-diugh, agus dè tha fhathast ri dhèanamh a thaobh rannsachaidh san àm ri teachd. vi Abstract This thesis elucidates some of the hitherto poorly understood aspects of the diachronic development of Manx phonology. By tracing phonological changes from earlier varieties of Gaelic, and within the attested period of written and recorded Manx, it frames these developments within the wider contexts of Gaelic dialectology and historical linguistics. Manx provides an important source for understanding the linguistic development of the Gaelic languages. A lack of systematic treatments and reliable datasets for the language, however, has obscured this fact and led to its neglect within Gaelic studies. The thesis focuses, in particular, on the development of the language’s prosody, suprasegmental features, vowel system and sonorants, the latter having a particular bearing on vowels. Five principal methodologies are deployed to investigate these topics:  Re-evaluation of existing descriptions and datasets provided by previous scholarship, especially those collected by Rhŷs in the 1880s and 1890s, and material from the last generation of speakers presented by Broderick in his Handbook of Late Spoken Manx.  Interpretation of the evidence of the two main Manx orthographies and nonstandard variations thereof.  Analyses based, as far as possible, on the whole attested lexis of the language, making use of Cregeen’s and Kelly’s dictionaries.  Quantitative approaches to all of these sources of data where appropriate.  Instrumental phonetic analysis of recordings of the terminal speakers of Manx. Chapter one places Manx in its historical and dialectological context, reviews previous scholarship, discusses the primary sources, and introduces the interpretative difficulties of the orthographies. Chapter two examines developments in the short and long vowels, and the impact of the consonant system on vowel changes. Chapter three examines the development of the vowels ao(i) /әː/ and ua(i) /uә̯/ in Manx. The written evidence, description and recorded data are complex, and some scholars have claimed that these vowels fell together with one another and with other vowels. It will be shown that these vowels in fact remained contrastive for the most part in Late Manx. Chapter four investigates developments in the sonorant consonants, especially the R, L and N phones. Changes in vowels preceding historically tense sonorants are also examined, as well as the origins and spread of the phenomenon of preocclusion. Chapter five examines suprasegmental and prosodic features including stress shift, unstressed long vowel shortening, and the conditioning factors for these. Chapter six provides concluding remarks assessing the thesis’ contribution to current scholarship, and the prospects for future research. viii Lay summary This thesis is concerned with the pronunciation of Manx Gaelic, the historical Celtic language of the Isle of Man which was the community language of the bulk of the island’s population until the mid nineteenth century, with a few elderly speakers remaining into the second half of the twentieth century. It examines changes in the pronunciation during the period for which we have extensive written, and latterly sound-recorded, evidence (the seventeenth century onwards), and also the changes which separate Manx from the other Gaelic languages past and present. It does not tackle the revived language spoken today mostly by adult learners, which is quite a distinct topic, and deserving of separate attention in its own right. Serious academic study of the spoken language did not begin until the 1880s, and the first audio recordings date from the first decade of the twentieth century, with the bulk being made in the mid twentieth century from some of the very last speakers. This material is extremely valuable, but inevitably limited, and by necessity we must rely extensively on earlier written material and on deductions from variation and changes in the orthography (spelling system). Fortunately, Manx has a fairly extensive corpus of written material, albeit mostly religious translations, including a complete Bible translation (finished in 1772). This material is written in two largely independent orthographies, both of which are in turn largely independent of the conservative literary standards of Ireland and Scotland. This makes Manx unique among Gaelic dialects in having an independent and vigorous orthographic tradition during this period, and provides extensive evidence of changes in pronunciation which would be obscured if the Irish-Scottish system had been in use. The fact that the Manx orthographies are English-based (although with significant innovations and adaptations to represent non-English sounds) has led to neglect and even derision on the part of scholars, but this thesis shows that the orthographies are considerably more systematic, and therefore useful as evidence for linguistic changes, than has previously been assumed. In contrast to most previous linguistic scholarship on Manx, which is largely impressionistic in its use of data, quantitative approaches form a major focus of this thesis, and often reveal details, and even major trends, which challenge other scholars’ claims. These methodologies include taking an exhaustive approach to the vocabulary of the language, drawing on four major dictionaries or glossaries, and analysis of large amounts of variant spellings from the earliest Manx manuscript and other sources. There is also computerized analysis of some of the audio recordings of the last native speakers using the phonetics software package Praat, which is a first for Manx studies. Re-evaluation of the data and analyses of previous scholarship is also a major part of the project, including especially the important early fieldwork and descriptions by Sir John Rhŷs, the first Professor of Celtic at Oxford University. In terms of the topics covered, it was not possible to deal with every area of Manx historical phonology in depth. The decision was taken to focus on vowels, sonorant consonants (L, N, M and R sounds), and stress patterns. The first chapter is an introduction to the history of the language, previous scholarship, and the sources and methods used in the rest of the thesis. Chapter 2 covers most of the vowel sounds, while chapter 3 goes into further depth about a particularly complex area of vowel developments. Chapter 4 concerns the sonorant consonants, and related developments including vowel lengthening and preocclusion — the insertion of a stop consonant [b], [d], [ɡ] before some of the sonorants. Chapter 5 deals with certain suprasegmental or prosodic phenomena. This means properties of the phonology above the level of individual vowels and consonants, especially stress patterns, which show particularly intriguing and complex changes in Manx. Chapter 6 provides a brief conclusion, bringing different threads from the thesis together. x Aspects of the historical phonology of Manx Taing / Acknowledgements Tha mi fada an comain mo phàrantan, mo chèile Edit, agus mo mhic, Torcall — a nochd faisg air deireadh na saothrach — agus mo charaidean is mo cho-oileanaich, airson an cuid foighidinn agus brosnachaidh ann an dòighean thar cunntais is tomhais. Tha mi air leth taingeil dom luchd-stiùiridh a threòraich air an turas mi, an Dr Uilleam Lamb agus an Dr Pavel Iosad, agus don luchd-obrach air fad ann an Roinn na Ceiltis agus Eòlas na h-Alba aig Oilthigh Dhùn Èideann. Tha mi cuideachd fo fhiachaibh don fheadhainn air fad a bhrosnaich is a theagaisg mi agus mi ag ionnsachadh nan cànanan Gàidhealach. Mu dheireadh, tha mi fada an comain luchd-labhairt agus sgrìobhaidh Gàidhlig Mhanainn fad nan linntean, a tha an cuid fhaclan a’ nochdadh air feadh na h-obrach seo. S’mooar ta my wooise as my eeaghyn daue ooilley. ’S ann leam-sa a-mhàin a tha gach mearachd is mì-thuigse a th’ air fhàgail. Er jerrey ooilley, ta mee chymney yn obbyr shoh da’n sleih Gailckagh ooilley ta as vees ayns Mannin. Christopher Lewin Samhainn 2019, Dùn Èideann (tionndadh na deuchainn) An t-Iuchar 2020, Dùn Èideann (an tionndadh deireannach). 1 Aspects of the historical phonology of Manx That a language so venerable for its antiquity and so estimable on many accounts should be so generally neglected, is much to be lamented. The consequence of this neglect has been, that numerous corruptions have crept into the dialect in general use, and so many anglicisms been adopted, that the Manks is now seldom spoken or written in its original purity. Despised and neglected, however, as the language appears to be at present, it is susceptible of high improvement, and justly entitled to the attention of the scholar. The sublime strains of Ossian mark the capabilities of the language, and commend it to the regard of the philologist as a subject of curious enquiry, and deserving accurate investigation. (Archibald Cregeen, A Dictionary of the Manks Language, 1835: iii) [I]t is always a source of delight to me to be able to trace the phonetics of a language from the earliest dawn of its documentary existence down to the most curtailed pronunciations of its vocables in the mouths of one’s contemporaries. In the Manx of the present day we have one of the lineal descendants of the Goidelic attested by the earliest Ogmic monuments of Great Britain and Ireland. Besides, the study of Manx phonology is by no means a bad corrective of the effect of seeing Irish written in an orthography which is more historical than phonetic. Manx, it is true has no vast stores of literature; but from the point of view of the phonologist even poverty of that kind has its consolation. For it leaves the natural tendencies of the language less trammelled, and keeps a freer sphere of evolution for its sounds. The result in Manx, as it would be found also in the other Goidelic dialects, is, that the changes of sound to which it testifies, work out with a precision falling not hopelessly short of mathematical accuracy. To suppose that modern Goidelic, because not blessed with a vigorous literature, must be a lawless jargon — lawless like the savages that speak it, as it is sometimes put — is not only not true, but is almost the exact contrary of the truth, so far at least as concerns the phonology. The mere spelling is a different matter, though even that has its interest, a wider interest, in fact, than has hitherto been usually supposed in the case of Manx. (John Rhŷs, The Outlines of the Phonology of Manx Gaelic, 1894: x) No apology is needed for the considerable amount of space devoted to Scottish and Manx. So closely are the three Gaelic languages allied that it would be futile to investigate the history of any one of them without taking full account of the other two. It is, perhaps, too much to expect that Manx, that Cinderella of Gaelic tongues, should ever attract many students […] (T. F. O’Rahilly, Irish Dialects Past and Present, 1932: viii–ix) 2 Aspects of the historical phonology of Manx Eisht dooyrt ad rish, Abbyr nish SHIBBOLETH; as dooyrt eh SIBBOLETH: son cha daink e hengey lesh dy ockley eh dy kiart. Eisht ghow ad, as varr ad eh ec aaghyn Yordan; as huitt ec y traa shen jeh ny Ephraimiteyn, daa housane as da-eed. Then said they unto him, Say now SHIBBOLETH: and he said SIBBOLETH: for he could not frame to pronounce it right. Then they took him, and slew him at the passages of Jordan: and there fell at that time of the Ephraimites forty and two thousand. (Briwnyn 12:6) Son eisht ver-yms da’n pobble glare ghlen, dy vod ad ooilley geamagh er ennym y Chiarn, dy hirveish eh lesh un aigney. For then will I turn to the people a pure language, that they may all call upon the name of the Lord, to serve him with one consent. (Zephaniah 3:9) Shen-y-fa te enmyssit Babel, er-yn-oyr dy nee ayns shen hug y Chiarn shaghrynys er glare ooilley yn seihll: as veih shen ren y Chiarn ad y skeayley harrish slane eaghtyr y thallooin. Therefore is the name of it called Babel; because the Lord did there confound the language of all the earth: and from thence did the Lord scatter them abroad upon the face of all the earth. (Genesis 11:9) As dob eh reesht. As tammylt ny lurg shen, dooyrt paart jeh’n cheshaght reesht rish Peddyr, Son shickyrys t’ou uss fer jeu: son she Galilean oo, as ta dty ghlare dy hoilshaghey eh. And he denied it again. And a little after, they that stood by said again to Peter, Surely thou art one of them: for thou art a Galilaean, and thy speech agreeth thereto. (Mark 14:70) Ta mee coyrt booise da my Yee, dy vel mee loayrt ny s’lhee glare na shiu ooilley I thank my God, I speak with tongues more than ye all (1 Corinthianee 14:18) 3 Aspects of the historical phonology of Manx 4 Aspects of the historical phonology of Manx Contents Declaration ................................................................................................................. iii Geàrr-chunntas .......................................................................................................... v Abstract ..................................................................................................................... vii Lay summary ............................................................................................................. ix Taing / Acknowledgements ....................................................................................... 1 List of tables .............................................................................................................. 17 List of charts ............................................................................................................. 20 List of figures ............................................................................................................ 21 Prefatory notes ......................................................................................................... 23 0.1 Transcription practices ................................................................................ 23 0.1.1 Phonetic and phonological transcriptions ................................................. 23 0.1.2 Transcriptions cited from other authors .................................................... 24 0.1.3 Examples cited in original orthography .................................................... 24 0.2 Citation of Gaelic forms .............................................................................. 25 0.3 Names of Gaelic varieties ............................................................................ 26 0.4 Linguistic terminology ................................................................................ 27 0.5 Citation of reference works ......................................................................... 27 0.6 General abbreviations and symbols ............................................................. 28 0.7 Abbreviations of names of speakers and fieldworkers ................................ 30 0.8 Map of the Isle of Man ................................................................................ 31 Chapter 1 Introduction ........................................................................................ 33 1.1 Purpose and scope of the thesis ................................................................... 33 1.2 Descriptive and theoretical concerns ........................................................... 35 1.3 Historical background of Manx ................................................................... 36 1.3.1 1.4 Periodization of Manx .......................................................................... 38 1.3.1.1 Early, Classical and Late Manx .................................................... 38 1.3.1.2 The language of the terminal speakers ......................................... 39 1.3.1.3 Revived Manx ............................................................................... 40 The place of Manx within the Gaelic dialect continuum ............................ 40 5 Aspects of the historical phonology of Manx 1.5 Review of previous scholarship ................................................................... 43 1.5.1 Material collected for Edward Lhuyd .................................................. 44 1.5.2 Rhŷs (1894) .......................................................................................... 45 1.5.3 Strachan (1897) .................................................................................... 49 1.5.4 Marstrander (1932) ............................................................................... 49 1.5.5 Kneen (1931, 1938) .............................................................................. 50 1.5.6 Carmody (1947, 1954) ......................................................................... 52 1.5.7 Jackson (1955)...................................................................................... 52 1.5.8 Wagner (1969)...................................................................................... 54 1.5.9 Broderick (1984–86) ............................................................................ 55 1.5.10 Toponymic and onomastic evidence .................................................... 56 1.5.11 Summary .............................................................................................. 57 1.5.12 Recent linguistic literature ................................................................... 58 1.6 Primary sources ........................................................................................... 58 1.6.1 The written corpus and the orthographies ............................................ 58 1.6.2 Scholarly views on the Manx orthographies ........................................ 59 1.6.3 Phillips’ Prayer Book translation and orthography .............................. 60 1.6.4 The Classical Manx orthography and variants ..................................... 68 1.6.4.1 Characteristics of the Classical Manx orthography ...................... 68 1.6.4.2 Homophones differentiated by spelling ........................................ 69 1.6.4.3 One sound, several spellings ......................................................... 69 1.6.4.4 One spelling, several sounds ......................................................... 70 1.6.4.5 English v. ‘continental’ vowel values ........................................... 71 1.6.4.6 Representation of palatalization .................................................... 72 1.6.4.7 Redundant symbols ....................................................................... 76 1.6.5 6 Late spellings (especially Cregeen)...................................................... 77 Aspects of the historical phonology of Manx 1.6.6 Interpreting the Manx orthographies: summary of difficulties ............ 79 1.6.7 Non-standard orthographies ................................................................. 80 1.6.8 Dictionaries .......................................................................................... 80 1.6.8.1 Kelly (1866) .................................................................................. 81 1.6.8.2 Cregeen (1835) ............................................................................. 82 1.6.8.3 Revival era English–Manx dictionaries ........................................ 83 1.6.8.4 Pitfalls in the use of Manx dictionaries ........................................ 84 1.6.9 1.7 Native speech: recordings and transcriptions....................................... 85 1.6.9.1 The Manx of the ‘last native speakers’ ......................................... 85 1.6.9.2 Features of Terminal Manx phonology ........................................ 89 Outline of synchronic phonology of Classical Manx .................................. 90 1.7.1 Stressed vowels .................................................................................... 91 1.7.2 Unstressed vowels in pretonic position ................................................ 92 1.7.3 Unstressed vowels in post-tonic preconsonantal position.................... 92 1.7.4 Unstressed vowels in post-tonic final position .................................... 93 1.7.5 Diphthongs and triphthongs ................................................................. 93 1.7.6 Consonants ........................................................................................... 95 1.7.7 Stress .................................................................................................... 96 1.7.8 Vowel nasalization ............................................................................... 97 1.7.9 Presentation of data .............................................................................. 98 Chapter 2 2.1 The Manx vowel system ..................................................................... 99 Short vowels ................................................................................................ 99 2.1.1 a /a/ > /a/, /e/, /o/ .................................................................................. 99 2.1.1.1 a /a/ > /e/ ....................................................................................... 99 2.1.1.2 a /a/ > /o(ː)/, /u(ː)/ ....................................................................... 101 2.1.2 ai /a/ > /a/, /e/ ..................................................................................... 105 2.1.2.1 ai /a/ > /a/ .................................................................................... 105 7 Aspects of the historical phonology of Manx 2.1.2.2 ai /a/ > /e/ .................................................................................... 107 2.1.2.3 ai /a/ > /o/ .................................................................................... 110 2.1.3 ea /e/ > /e/, /a/, /o/............................................................................... 110 2.1.3.1 Analysis....................................................................................... 117 2.1.3.2 ea /e/ > /o/ ................................................................................... 121 2.1.3.3 ea /e/ > /i/ .................................................................................... 122 2.1.3.4 Lexical diffusion ......................................................................... 123 2.1.3.5 Semantic splits between /e/ and /a/ variants ............................... 123 2.1.4 ei /e/ > /e/, (/i/) .................................................................................... 124 2.1.5 o /o/ > /o/, /a/, (/u/) ............................................................................. 125 2.1.5.1 Conditioning environments for /o/ > /a/...................................... 131 2.1.5.2 Diachronic development of /o/ > /a/ and Gaelic dialectological context 132 2.1.5.3 foddey ‘far, long’, G. fada, OIr. fota ........................................... 133 2.1.5.4 /o/ > /u/ ........................................................................................ 134 2.1.6 oi /o/ > /o/, /a/ ..................................................................................... 135 2.1.6.1 2.1.7 i, io /i/ ................................................................................................. 137 2.1.7.1 2.1.8 io /i/ > /u/..................................................................................... 138 u, ui /u/................................................................................................ 139 2.1.8.1 ui /u/ > /wa/ ................................................................................. 142 2.1.9 Morphophonological alternations /a/, /o/ > /i/, /u/ ............................. 142 2.1.10 OIr. air-, aur- ..................................................................................... 145 2.2 8 kemmyrk ‘refuge’, G. coimirce etc. ............................................. 137 Long vowels .............................................................................................. 147 2.2.1 Stressed final vowels in monosyllables .............................................. 147 2.2.2 Fronting of /aː, oː/ > /ɛː/ ..................................................................... 148 Aspects of the historical phonology of Manx 2.2.3 /ɛː/ and /eː/ .......................................................................................... 153 2.2.4 a(i)th /ah/ > /aː/ .................................................................................. 154 2.2.5 /oː/ and */ɔː/ ....................................................................................... 156 2.2.6 Monophthongization of /iә̯, ɨә̯, uә̯/ > /iː, ɨː, uː/ ................................... 157 2.2.7 Breaking of éa /eː/ > /iә̯/ .................................................................... 158 Chapter 3 Manx reflexes of Gaelic ao(i) and ua(i) .......................................... 163 3.1 Introduction ............................................................................................... 163 3.2 Historical development of G. ao(i) and ua(i) ............................................ 163 3.2.1 ao(i) /әː/ .............................................................................................. 163 3.2.2 ua(i) /uә̯/ ............................................................................................. 165 3.3 Summary of Manx developments.............................................................. 165 3.4 Lexical items with ao(i), ua(i), agh etc. .................................................... 166 3.4.1 ao > /әː/ .............................................................................................. 167 3.4.2 aoi > /әː/ (>/eː/) .................................................................................. 169 3.4.3 ua > /ɨә̯/, /ɨː/ ........................................................................................ 170 3.4.4 uai > /әː/ (>/eː/), /ɨә̯/ (>/ɨː/) ................................................................. 172 3.4.5 ua(i) > /uә̯/, /uː/, /oː/ ........................................................................... 173 3.4.6 ua(i): items with variable reflexes ..................................................... 175 3.4.7 agh, adh > /әː/ .................................................................................... 176 3.5 3.4.7.1 eab ‘attempt’ ............................................................................... 177 3.4.7.2 eairk ‘horn’, G. adharc ............................................................... 178 3.4.7.3 oyr ‘reason, cause’, G. adhbhar.................................................. 178 3.4.7.4 abane ‘ankle’, EIr. odbrann, ScG. adhbrann ............................. 178 3.4.7.5 ymmyd ‘use’, G. adhmad ............................................................ 179 3.4.7.6 G. Raghnall (N. Rǫgnvaldr) in Roonysvie and Crennell ............ 179 Previous accounts of Manx ao(i), ua(i) ..................................................... 180 3.5.1 Rhŷs ................................................................................................... 180 9 Aspects of the historical phonology of Manx 3.5.1.1 ao > ȳ [әː] .................................................................................... 180 3.5.1.2 Allophonic realizations of ao ...................................................... 181 3.5.1.3 ua > ůy [ɯә̯], üy [ɨә̯], iy [iә̯] ....................................................... 182 3.5.1.4 aoi and uai > œ̄ [ә̞ː] ..................................................................... 185 3.5.2 Marstrander (1932) ............................................................................. 186 3.5.2.1 ē ̣ [eː] < G. aoi, uai ....................................................................... 187 3.5.2.2 ē [ɛː] < G. ao ............................................................................... 187 3.5.2.3 ī [iː] < G. ua ................................................................................ 188 3.5.2.4 î [ɪː].............................................................................................. 188 3.5.2.5 îә [ɪә̯] < G. ua(i), ao .................................................................... 189 3.5.2.6 ei [ɛi̯ ] < G. ao .............................................................................. 189 3.5.2.7 ū [uː] < G. ua............................................................................... 190 3.5.2.8 ‘ø’ ................................................................................................ 190 3.5.2.9 Analysis....................................................................................... 193 3.5.3 Jackson (1955).................................................................................... 194 3.5.4 Wagner (1969).................................................................................... 198 3.5.5 Broderick (1986) (HLSM III) .............................................................. 201 3.5.5.1 3.5.6 3.6 A quantitative analysis of data from HLSM (II) .......................... 205 Front rounded realizations? ................................................................ 208 Written evidence ........................................................................................ 209 3.6.1 Representation of ao(i), ua(i) in Phillips ............................................ 210 3.6.1.1 Overview ..................................................................................... 210 3.6.1.2 ua(i) in Phillips: preceding consonant conditioning and lexical diffusion 215 3.6.2 10 Representation of ao(i) and ua(i) in CM orthography ....................... 218 3.7 Instrumental data ....................................................................................... 225 3.8 Fronting and unrounding of ua(i) in Manx and Ulster Irish ...................... 230 Aspects of the historical phonology of Manx 3.8.1 /ɨә̯/ ~ /uә̯/ in feayr ‘cold’ (G. fuar): register, dialect, idiolect............. 232 3.8.2 hooar, dooar ‘got, found’ (G. f(h)uair, d’fhuair) .............................. 233 3.9 Other developments ................................................................................... 234 3.9.1 New diphthongs: ao(i), ua(i) + vocalized fricatives .......................... 234 3.9.1.1 ao(i) + /i̯ / ..................................................................................... 235 3.9.1.2 ua(i) + /i̯ / ..................................................................................... 237 3.9.1.3 Interpreting the developments of ao(i), ua(i) + /i̯ / ...................... 238 3.9.1.4 /ai̯ / ~ /әi̯ / contrast ........................................................................ 241 3.9.1.5 lheiy ‘calf’, G. laogh ................................................................... 243 3.9.1.6 leoie etc., G. ua(i) + /i̯ / ................................................................ 244 3.9.1.7 /әi̯ /, /әːi̯ /, (/ɯːi̯ /) > /ei̯ / ................................................................. 246 3.9.1.8 /әi̯ /, /әːi̯ /, (/ɯːi̯ /) > /iː/ .................................................................. 246 3.9.1.9 ao(i) + /u̯ / .................................................................................... 247 3.9.2 /eː/ > /әː/ in Scottish Gaelic and Manx ............................................... 248 3.9.3 geay ‘wind’, G. gaoth ........................................................................ 251 3.9.4 riyr ‘last night’, G. aréir, araoir ........................................................ 252 3.9.5 cleaysh ‘ear’, G. clua(i)s etc. ............................................................. 252 3.9.6 keayrt ‘time, occasion’, G. cuairt....................................................... 253 3.9.7 seaghyn ‘sorrow, affliction, trouble’ .................................................. 254 3.9.8 deayrtey ‘pour’, G. dort, doirt, dórt, duart ........................................ 255 3.9.9 smooinaghtyn ‘think’, G. smuain, smaoin ......................................... 255 3.9.10 Manx derivatives of G. draoi, druadh ............................................... 256 3.9.11 Unstressed and shortened ao(i), ua(i) ................................................ 257 3.9.11.1 Pre-tonic ao, ua > /w/ ................................................................. 259 3.9.11.2 unnane, annan ‘one’, G. aonán .................................................. 260 Chapter 4 Sonorant consonants and associated developments ...................... 261 11 Aspects of the historical phonology of Manx 4.1 Introduction ............................................................................................... 261 4.2 Rhotics ....................................................................................................... 262 4.2.1 4.2.1.1 /rʲ/ in Classical Manx .................................................................. 263 4.2.1.2 /rʲ/ in clusters ............................................................................... 264 4.2.1.3 /r ~ rʲ/ in pre-terminal Late Manx................................................ 265 4.2.1.4 /r ~ rʲ/ in terminal Late Manx ...................................................... 268 4.2.1.5 Initial /rʲ/ ...................................................................................... 269 4.2.2 Early loss of /r/, /rʲ/ ............................................................................. 271 4.2.2.1 /rs, rʃ/ > /s/ (?/ʂ/).......................................................................... 272 4.2.2.2 Other cases of early rhotic deletion ............................................ 272 4.2.3 Weakening and loss of /r/ Late Manx ................................................ 273 4.2.4 Fortis /R/ ............................................................................................. 276 4.2.5 Other realizations of rhotics ............................................................... 276 4.3 Laterals ...................................................................................................... 277 4.3.1 Introduction ........................................................................................ 277 4.3.2 Lateral contrasts in Late Manx ........................................................... 277 4.3.2.1 Merger of /L/ and /l/ .................................................................... 281 4.3.2.2 Alveolar /ḻ/ .................................................................................. 281 4.3.2.3 Initial lenition of laterals ............................................................. 284 4.3.3 4.4 12 /r ~ rʲ/ .................................................................................................. 262 Lateral contrasts in Early Manx ......................................................... 284 Coronal nasals............................................................................................ 285 4.4.1 Introduction ........................................................................................ 285 4.4.2 Contrasts in coronal nasals in Late Manx .......................................... 285 4.4.3 Evidence for /nʲ/ > /ṉ/ ......................................................................... 287 4.4.4 /N/ and /n/ > [n̪ ], [n̠ ] ........................................................................... 290 4.4.5 /ṉ/ representing fossilized initial lenition ........................................... 291 Aspects of the historical phonology of Manx 4.4.6 4.4.6.1 /Nʲ/ in final unstressed syllables.................................................. 295 4.4.6.2 Other developments of /Nʲ/ ......................................................... 297 4.4.7 Unstressed final /an/, /әn/, /әnʲ/ .......................................................... 297 4.4.7.1 verbal nouns in -yn, -in ............................................................... 299 4.4.7.2 -in ................................................................................................ 301 4.4.7.3 -an ............................................................................................... 302 4.4.7.4 Stressed reflexes of G. -éan, -án ................................................. 306 4.4.8 4.5 Reflexes of G. /Nʲ/ .............................................................................. 292 Summary of developments in coronal and velar nasals ..................... 308 Preocclusion .............................................................................................. 308 4.5.1 Introduction ........................................................................................ 308 4.5.1.1 Cross-linguistic typology ............................................................ 311 4.5.2 Descriptions ....................................................................................... 312 4.5.3 Written evidence ................................................................................ 318 4.5.4 The origins of Manx preocclusion: previous hypotheses................... 322 4.5.4.1 Rhŷs (1894) ................................................................................ 323 4.5.4.2 Chaudhri (2007) .......................................................................... 323 4.5.4.3 Ó Maolalaigh (2014b)................................................................. 329 4.5.4.4 Broderick (2018e) ....................................................................... 332 4.5.5 4.6 The origin and spread of preocclusion ............................................... 332 4.5.5.1 Typological and phonetic considerations ................................... 332 4.5.5.2 Generalization and reanalysis of preocclusion ........................... 334 4.5.5.3 Preocclusion as a synchronic prosodic process .......................... 335 Vowel lengthening and diphthongization before coda fortis sonorants .... 336 4.6.1 In monosyllables ................................................................................ 336 4.6.1.1 -ionn /iN/ > /in/, /uːn/ .................................................................. 337 13 Aspects of the historical phonology of Manx 14 4.6.1.2 -im /im/ > /iːm/ ............................................................................ 337 4.6.1.3 -inn /iNʲ/ > /iŋʲ/ (/in/) ................................................................... 337 4.6.1.4 -ill /iLʲ/ > /iːlʲ/, /ilʲ/ ....................................................................... 338 4.6.1.5 -eam /em/ > /am/ ......................................................................... 338 4.6.1.6 -eim /em⁽ʲ⁾/ > /em/, /im/, /eːm/ (?)................................................ 338 4.6.1.7 -eann /eN/ > /oːn/, /au̯ n/, (/uːn/, /on/) .......................................... 339 4.6.1.8 -einn /eNʲ/ > ?/eːnʲ/ ...................................................................... 341 4.6.1.9 -eall /eL/ > /oːl/, (/ol/) ................................................................. 341 4.6.1.10 -earr /eR/ > /ɛːr/ .......................................................................... 342 4.6.1.11 -eang /eŋɡ/ > /eŋ/, /aŋ/ ................................................................ 342 4.6.1.12 -am /am/ > /am/ ........................................................................... 342 4.6.1.13 -ann /aN/ > /oːn/, /au̯ n/, (/uːn/) .................................................... 343 4.6.1.14 -ainn /aNʲ/ > /eːnʲ/, /ai̯ nʲ/ .............................................................. 343 4.6.1.15 -all /aL/ > /oːl/, (?/au̯ l/) ............................................................... 344 4.6.1.16 -aill /aLʲ/ > /alʲ/, /elʲ/ .................................................................... 345 4.6.1.17 -arr /aR/ > /ɛːr/ ............................................................................ 345 4.6.1.18 -a(i)ng /aŋɡ/ > /aŋ/ ...................................................................... 346 4.6.1.19 -om /om/ > /oːm/, ?/au̯ m/ ............................................................ 346 4.6.1.20 -oim /om⁽ʲ⁾/ > /em/ ....................................................................... 347 4.6.1.21 -onn /oN/ > /oːn/, /au̯ n/, (/on/)..................................................... 347 4.6.1.22 -oinn /oNʲ/ > /eːnʲ/ ....................................................................... 348 4.6.1.23 -oll /oL/ > /oːl/, /au̯ l/.................................................................... 348 4.6.1.24 -oill /oLʲ/ > /olʲ/ ........................................................................... 348 4.6.1.25 -orr /oR/ > /oːr/ ........................................................................... 349 4.6.1.26 -ong /oŋɡ/ > /oŋ/.......................................................................... 349 4.6.1.27 -um /um/ > /um/ .......................................................................... 349 Aspects of the historical phonology of Manx 4.6.1.28 -uim /umʲ/ > /iːm/ ........................................................................ 350 4.6.1.29 -unn /uN/ > /uːn/ ......................................................................... 350 4.6.1.30 -uinn /uNʲ/ > /unʲ/, /inʲ/, /iŋʲ/ ........................................................ 350 4.6.1.31 -uill /uLʲ/ > /uːlʲ/ .......................................................................... 351 4.6.1.32 -uing /uŋʲɡʲ/ > /wiŋʲ/ .................................................................... 351 4.6.1.33 Summary ..................................................................................... 351 4.6.1.34 Dialectal variation ....................................................................... 352 4.6.2 Voiced homorganic rhotic clusters..................................................... 353 4.6.3 Other medial clusters.......................................................................... 357 4.6.4 Paradigmatic uniformity..................................................................... 359 Chapter 5 5.1 Suprasegmental phonology ............................................................. 361 Stress ......................................................................................................... 361 5.1.1 Previous accounts of stress shift and long vowel shortening ............. 363 5.1.1.1 O’Rahilly (1932) ......................................................................... 363 5.1.1.2 Jackson (1955) ............................................................................ 364 5.1.1.3 Ó Baoill (1980) ........................................................................... 364 5.1.1.4 Broderick (1986) ......................................................................... 365 5.1.1.5 Ó Sé (1991) ................................................................................. 367 5.1.1.6 Green (1997) ............................................................................... 371 5.1.2 An explanation for stress shift targeting heavy-heavy items ............. 372 5.1.3 Initial syllable shortening in items with forward stress ..................... 373 5.1.4 Long vowels arising from vocalization of fricatives.......................... 375 5.1.4.1 5.1.5 Exceptions................................................................................... 380 First syllable conditioning: further details ......................................... 381 5.1.5.1 Heavy sonorant clusters .............................................................. 381 5.1.5.2 Exceptions................................................................................... 382 15 Aspects of the historical phonology of Manx 5.1.5.3 Stress-shift in items with fricative vocalization in initial syllable: 384 5.1.5.4 Verbal nouns in -ail, -eil, -al (G. -áil)......................................... 384 5.1.5.5 Items with original heavy third syllables .................................... 385 5.1.6 Quality of post-tonic shortened vowels .............................................. 387 5.1.7 Irregular stress in reeriaght ‘kingdom’ .............................................. 387 5.2 Apocope ..................................................................................................... 387 5.3 Syncope ..................................................................................................... 391 5.4 Epenthesis .................................................................................................. 392 5.5 Vowel shortening and lengthening ............................................................ 397 5.5.1 Vowel shortening ............................................................................... 399 5.5.2 Vowel lengthening ............................................................................. 406 5.6 Vowel nasalization .................................................................................... 408 Chapter 6 Conclusion ......................................................................................... 411 6.1 Topics covered ........................................................................................... 411 6.2 Progress made ............................................................................................ 412 6.3 By-products ............................................................................................... 417 6.4 Prospects for future research ..................................................................... 418 Bibliography ........................................................................................................... 421 Primary sources .................................................................................................... 421 Secondary sources ................................................................................................ 422 16 Aspects of the historical phonology of Manx List of tables Table 1 Page 64 Table 18 Table 19 Table 20 Table 21 Table 22 Table 23 Table 24 Table 25 Table 26 Table 27 Table 28 Table 29 Table 30 Table 31 Table 32 Examples of variant spellings in Phillips’ MS (after Thomson 1953) a /a/ > /e/ a /a/ > /o(ː)/ (/u(ː)/) a /a/ > /a/, /o/ ai /a/ > /a/ ai /a/ > /e/ ai /a/ > /o/ ea /e/ > /e/ ea /e/ > /a/ ea /e/ > /o/ Incidence /e/ > /e/ and /e/ > /a/, /o/ in the lexicon by following consonant Incidence of ea /e/ > /e/ and /e/ > /a/, /o/ in the lexicon by preceding consonant Combined conditioning effect of preceding and following consonant on G. ea /e/ in Manx ea /e/ > /i/ o /o/ > /o/ o /o/ > /a/ Incidence of o /o/ > /o/, /a/ in the lexicon by preceding consonant (cluster) o /o/ > /u/ oi /o/ > /o/ oi /o/ > /a/ oi /o/ > /e/ io /i/ > /u/ ui /u/ > /u/, /(w)i/ ui /u/ > /wa/ Morphophonological alternations /a/ > /i/, /u/ Morphophonological alternations /o/ > /u/, /i/ OIr. air-, auró /oː/ > /oː/ etc. ói /oː/ > /oː/ ó /oː/ > /ɛː/ ói /oː/ > /ɛː/ a(i)th /ah/ > /aː/ 134 135 136 136 138 140 142 142 144 146 149 150 151 152 155 Table 33 éa /eː/ > /eː/ 159 Table 2 Table 3 Table 4 Table 5 Table 6 Table 7 Table 8 Table 9 Table 10 Table 11 Table 12 Table 13 Table 14 Table 15 Table 16 Table 17 100 101 103 105 107 110 111 112 116 117 119 121 122 125 129 131 17 Aspects of the historical phonology of Manx Table 34 Table 35 Table 36 Table 37 Table 38 Table 39 Table 40 Table 41 Table 42 Table 43 Table 44 Table 45 Table 46 Table 47 Table 48 Table 49 Table 50 Table 51 Table 52 Table 53 Table 54 Table 55 Table 56 Table 57 Table 58 Table 59 Table 60 Table 61 Table 62 Table 63 Table 64 Table 65 Table 66 Table 67 Table 68 Table 69 18 éa /eː/ > /iә̯/ ao /әː/ > /әː/ aoi /әː/ > /әː/ (>/eː/) ua /uә̯/ > /ɨә̯/ (>/ɨː/) uai /uә̯/ > /әː/ (>/eː/), /ɨә̯/ (>/ɨː/) ua(i) /uә̯/ > /uә̯/ (>/uː/), /oː/ ua(i) /uә̯/ > /ɨә̯/ (>/ɨː/), /әː/ or /uә̯/ (>/uː/), /oː/ agh, adh /aɣ/ > /әɣ/ > /әː/ and other developments Realizations of G. ao in HLSM II Realizations of G. aoi in HLSM II Realizations of G. ua in HLSM II Realizations of G. uai in HLSM II Spelling of G. ao(i), ua(i) in Phillips, by initial letter of orthographic sequence Breakdown of orthographic categories <e> and <y>, taking into account following vowel characters Spelling of G. ao(i) and ua(i) in Phillips Orthographic representations of /iː/, /iә̯/, /eː/, /uː/ in Phillips (A–C, Thomson [1953] glossary) Spelling of G. ua(i) in Phillips by place of articulation of preceding consonant ao(i) + /i̯ / Gaoidhealg, Gaoidhilg ua(i) + /i̯ / (excluding items with synchronic /uәi̯ /) Interpretation of Rhŷs’s descriptions of /ai̯ /, /әi̯ /, /әːi̯ / etc. ao(i) + /u̯ / seaghyn Pre-tonic ao(i), ua(i) Pre-tonic ao, ua > /w/ Incidence of palatalized and non-palatalized realizations of Gaelic laterals and coronal nasals in data from HLSM II inn /ˈiNʲ/ > /iŋʲ/, /inʲ/ Other instances of /Nʲ/ > /ŋʲ/ Summary of developments of post-tonic unstressed /Vn⁽ʲ⁾/ Unstressed reflexes of G. -ín Some examples of Manx -an < G. -án Unhistorical -an Unhistorical -yn < G. -án -an < G. /Vːnʲ(ә)/ Manx -ean /eːn/ < G. -éan (Mod.Ir. -eán) Representation of preocclusion in MNHL MD 900 MS 08307 (ed. Broderick 2015) 160 167 169 170 172 174 175 176 206 206 206 207 211 212 212 215 216 235 237 237 241 247 254 258 260 287 294 294 297 301 303 303 305 305 307 319 Aspects of the historical phonology of Manx Table 70 Table 71 Table 72 Table 73 Table 74 Table 75 Table 76 Table 77 Table 78 Table 79 Table 80 Table 81 Table 82 Table 83 Table 84 Table 85 Table 86 Table 87 Table 88 Table 89 Table 90 Table 91 Table 92 Table 93 Table 94 Table 95 Table 96 Table 97 Table 98 Table 99 Table 100 Table 101 Table 102 Table 103 Table 104 Table 105 Table 106 Table 107 Table 108 -ionn -im -ill -eam -eim -éim -eann -einn -eall -earr -eang -am -ann -ainn -all -aill -arr -a(i)ng -om -oim -onn -oinn -oll -oill -orr -ong -um -uim -unn -uinn -uill -uing Predominant CM/LM reflexes of stressed short vowels before final fortis sonorants Dialectal splits in stressed short vowel + final fortis sonorant sequences Representation of dialect variants in Bible orthography Vowel developments before rd, rn, rl Vowel developments before other medial clusters Long vowel shortening Vowel lengthening shown in orthography 337 337 338 338 338 339 339 341 341 342 342 342 343 343 344 345 345 346 346 347 347 348 348 348 349 349 349 350 350 350 351 351 351 352 352 353 357 400 407 19 Aspects of the historical phonology of Manx List of charts Chart 1 Chart 2 Chart 3 Chart 4 Chart 5 Chart 6 Chart 7 Chart 8 Chart 9 20 Realizations of G. ua(i), ao(i) in HLSM II Spelling of G. ao(i), ua(i) in Phillips, by initial letter of orthographic sequence Spelling of G. ao(i) and ua(i) in Phillips Spelling of G. ao(i), ua(i) in Phillips, with <e(V)y(V)> and <ye(V)> combined Spelling of G. ua(i) in Phillips by place of articulation of preceding consonant F1 tracks for front and central long vowels (corresponding to vowel height), all speakers, Irish Folklore Commission recordings F2 tracks for front and central long vowels (corresponding to backness), all speakers, Irish Folklore Commission recordings Means for each category with 95% confidence intervals, all speakers, Irish Folklore Commission recordings, showing F1 (height) and F2 (backness) Incidence of palatalized and non-palatalized realizations of Gaelic laterals and coronal nasals in data from HLSM II Page 207 211 213 214 216 226 226 227 287 Aspects of the historical phonology of Manx List of figures §0.8 Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 Figure 14 Map of the Isle of Man Examples from Phillips’ MS (cf. Table 1) Stressed vowels in Classical Manx Unstressed pretonic vowels in Classical Manx Unstressed post-tonic vowels in closed syllables Unstressed word-final post-tonic vowels ә-diphthongs i-diphthongs u-diphthongs Consonants in Classical Manx Main developments of G. ua(i), ao(i), agh in Manx Pairwise comparisons using t-tests with pooled standard deviation for Irish Folklore Commission data, F1 (height) Pairwise comparisons using t-tests with pooled standard deviation for Irish Folklore Commission data, F2 (height) Summary of developments of coronal and velar nasals Spectrograms showing preocclusion Page 31 64 91 92 92 93 93 94 94 95 166 229 229 308 309– 311 21 Aspects of the historical phonology of Manx 22 Aspects of the historical phonology of Manx Prefatory notes 0.1 Transcription practices 0.1.1 Phonetic and phonological transcriptions In my own transcriptions I follow the usual convention of giving material intended as a phonological or phonemic representation in slanted brackets / / and narrower phonetic transcriptions in square brackets [ ]. Conventions of the International Phonetic Alphabet are generally used, for example in using the yod diacritic ʲ for palatalization rather than the prime symbol ′ which has been commonly used in Celtic Studies. Certain fully palatal or post-alveolar “slender” consonants are transcribed with the conventional unitary symbols ʃ ç j, even in the phonological transcriptions, in preference to sʲ xʲ ɣʲ. However, in historical discussions the symbols L Lʲ N Nʲ R Rʲ are retained from conventional Celticist practice for the fortis or tense sonorants (Chapter 4), both for clarity, and because their exact realization is not always clear. Diphthongs and triphthongs are shown as follows: ai̯ , iu̯ , iә̯ uә̯, ɛːi̯ , iәu̯ etc. In transcriptions of Scottish Gaelic, the unaspirated stops are transcribed /b, d, ɡ/ and the aspirated stops /p, t, k/, rather than /p, t, k/ and /pʰ, tʰ, kʰ/ respectively, in keeping with practice for Irish and Manx, where the distinction is conventionally regarded as one of voicing, although in all Gaelic varieties the primary distinction is likely to be one of aspiration rather than voicing. Stress is generally only marked when it does not fall on the first syllable. Other conventional symbols utilized include: C = (broad / non-palatalized) consonant Cʲ = slender / palatalized consonant V = vowel Vː = long vowel 23 Aspects of the historical phonology of Manx # = word boundary . = syllable boundary > = becomes (diachronic sound changes) * = hypothetical form 0.1.2 Transcriptions cited from other authors Phonetic and semi-phonetic transcriptions cited from other authors, primarily Rhŷs, Marstrander, Jackson, Wagner and Broderick, are given in bold type (silently added within quotations) and without brackets, e.g. ghœ̄ ñey, ɡøːnʹə. The original transcriptions are reproduced as closely as typographically possible. Where this is considered helpful, these may be converted into my own transcription immediately after, e.g. [ɣәːnʲә], including within quoted passages. My interpretations of other scholars’ transcriptions are for guidance only, and should not be considered definitive. 0.1.3 Examples cited in original orthography Examples from Manx and other languages in the original orthography are given in italics, except for orthographic units which are given in angled brackets, e.g. <ýa>, <eay>. Manx lexical items are cited as far as possible in the standardized form as they appear in the Manx Bible and/or Cregeen’s and Kelly’s dictionaries (§1.6.8), but spelling variants are given where these provide additional information, or where there is no clear standard form. Where diacritics in the Phillips manuscript (§1.6.3) occur between or over two adjacent vowels (according to the interpretation of Moore and Rhŷs [1895], Thomson [1953]), they are transcribed here on both vowels, so <íí> rather than <iˊi>. 24 Aspects of the historical phonology of Manx 0.2 Citation of Gaelic forms Gaelic phones and forms, where the focus is on their diachronic development and not on their phonetic or phonological value at any given time, are given in Gaelic (Early Modern or Classical Irish) orthography in italics, e.g. ao, ua, aoi, é. Where parallel forms flanked by broad or slender consonants are referred to, this is often shown by adding the diacritic vowel symbols in brackets, e.g. ao(i) (= ao /әːC/ or aoi /әːCʲ/), -(e)aghadh (= -aghadh /Cәɣәɣ/ [>Ir. /Cuː/, ScG. /Cәxәɣ/, Manx /Caxә/] or -eaghadh /Cʲәɣәɣ/ [>Ir. /Cʲuː/, ScG. /Cʲәxәɣ/, Manx /Cʲaxә/). Cognates cited for comparative purposes are given in their Early Modern Irish or Classical Gaelic form (usually following the spellings in Dinneen’s Dictionary),1 except where otherwise stated. These forms are what is primarily meant by ‘Gaelic’ or ‘G.’. Here I follow the practice of Jackson (1955: 7), although I do not subscribe to his reductive notion that these forms represent a ‘Common Gaelic’ (‘that stage of the Goedelic branch of the Celtic languages immediately preceding its break-up into Irish, Scottish and Manx Gaelic, while they were still one undifferentiated speech’), which has been rightly problematized by a number of scholars, including Gillies (1994), Ó Buachalla (2002), Ó Maolalaigh (1996; 2008a) and Ó Muircheartaigh (2015) (cf. §1.4). Nevertheless, in most cases, the Manx forms can be understood as deriving from something close to the forms represented by the Early Modern Irish orthography. Where relevant, Scottish Gaelic forms, contemporary Irish forms, or Early (Old or Middle) Irish are also given, the latter usually following the headwords in eDIL. Occasionally Manx developments, attested or hypothetical, are transliterated into Gaelic orthography for illustrative purposes; these are marked *. Also ‘Gaelic’ forms are sometimes given for illustration which are not actually attested outside Manx, e.g. 1 Except that for clarity I mark vowel length on eó and iúi, and use the spelling -(e)aghadh of the verbal noun ending in preference to -(i)ughadh (Manx -aghey, ScG. -(e)achadh, Caighdeán Ir. -(i)ú). 25 Aspects of the historical phonology of Manx *iascóir,2 Manx eeasteyr for usual Ir. iascaire, ScG. iasgair (which could represent either form) (cf. Ó Sé 1991). 0.3 Names of Gaelic varieties Although the terms ‘Old Gaelic’ and ‘Middle Gaelic’ have been used by some scholars (e.g. Clancy 2010: 351; Ó Maolalaigh 2013: 42), and are more accurate in the sense that they refer to varieties used throughout Ireland, Scotland and presumably Man rather than Irish alone, the more conventional and widely-used terms ‘Old Irish’ and ‘Middle Irish’ (collectively ‘Early Irish’) have been retained here (cf. Ó Muircheartaigh 2015: 8–9). In this thesis the term ‘Goidelic’ is not used except in quotations from other authors, and ‘Gaelic’ refers either to the Gaelic languages as a whole, or specifically to the standardized written varieties of Early Modern Gaelic/Irish which were used in Ireland and Scotland from the thirteenth to the seventeenth century (§0.2), and which formed the basis of the orthographic conventions used in Ireland until the adoption of the Caighdeán Oifigiúil in the mid-twentieth century (Ahlqvist 1994). Although orthographic forms belonging to, or close to, the Caighdeán Oifigiúil, are occasionally given as illustrating Irish developments, generally such forms are avoided as they obscure historical developments and represent specifically Irish developments.4 Hence, of the standard modern Irish–English reference dictionaries, Dinneen is more often cited than Ó Dónaill. 2 The National Terminology Database for Irish <tearma.ie> gives iascóir for ‘fishkeeper’, but this is presumably a neologism. But cf. the variant iascadóir for iascaire given by Dinneen. 4 And sometimes exclude even some Irish dialects; see Ahlqvist (1994: 52). 26 Aspects of the historical phonology of Manx 0.4 Linguistic terminology Conventional linguistic terminology is generally used without special explanation, although less commonly used or specialized terms are defined and contextualized as appropriate. Certain terms from traditional Celticist analyses of Gaelic phonology are retained in order to refer to abstract categories of sounds persisting throughout periods and varieties, when their synchronic phonetic and phonological analysis may vary, and in order to avoid taking positions on theoretical questions when this is not immediately relevant to the discussion. Most notably, the terms ‘slender’ and ‘broad’ in reference to consonant quality are retained, in preference to ‘palatalized’ and ‘non-palatalized’, both because the former terms are less cumbersome and are well-known within Gaelic studies (and popular discourse), and because it has been argued that in some dialects, namely those of Munster, velarization is more important in distinguishing between velarized (broad) and non-velarized or palatalized (slender) consonants (Ó Maolalaigh 1997: 36); and ‘palatalization’ is not necessarily a very accurate term for the articulation of the Irish slender labials (Oftedal 1963: 73–4; McKenna 2001). In addition, the slender category contains both palatalized and fully palatal consonants. On similar grounds, the traditional terms ‘fortis’ and ‘lenis’ are retained for the distinction between sonorants maintained in some Gaelic dialects and earlier varieties (also called ‘tense’ and ‘lax’ [e.g. Archangeli et al. 2011]). This should not be confused with the more widespread cross-linguistic use of ‘fortis / lenis’ in contemporary phonetics and phonology to refer to e.g. aspiration contrasts in stops. 0.5 Citation of reference works Citation of academic works follows the conventional author–date system, except for certain frequently-cited key sources, which are given by author only or by abbreviation of the title, as shown below. Hence ‘Jackson’ always refers to Jackson (1955), unless 27 Aspects of the historical phonology of Manx otherwise indicated. Full references for these abbreviated citations are given in the bibliography. Works cited by author only: Cregeen, Cr. Dinneen Dwelly Jackson Kelly, K. MacBain Marstrander Ó Dónaill O’Rahilly Rhŷs A Dictionary of the Manks Language (1835), Wheeler ed. (2018) Irish–English Dictionary (1927) Gaelic–English Dictionary (1911) Contributions to the Study of Manx Phonology (1955) Fockleyr Manninagh as Baarlagh [Manx dictionary] (1866) An Etymological Dictionary of the Gaelic Language (1911) ‘Det Norske Landnåm på Man’ (1932) Foclóir Gaeilge–Béarla (1977) Irish Dialects Past and Present (1932) Outlines of Manx Phonology (1894) Works cited by initials of title: EDD eDIL GOI HLSM LASID LEIA OED PNIM SGDS English Dialect Dictionary, Wright (1898–1905) Electronic Dictionary of the Irish Language Grammar of Old Irish, Thurneysen (1946) Handbook of Late Spoken Manx, Broderick (1984–86) Linguistic Atlas and Survey of Irish Dialects, Wagner (1958–69) Lexique étymologique de l’irlandais ancien, Vendryes et al. (1959–96) Oxford English Dictionary Placenames of the Isle of Man, Broderick (1994–2005) Survey of the Gaelic Dialects of Scotland, Ó Dochartaigh (1994–97) All translations of works in languages other than English are my own. 0.6 General abbreviations and symbols # – word boundary * – unattested or hypothetical form _ – segment(s) in question ? – doubtful or uncertain conclusion, reconstruction etc. /abc/ – phonological / phonemic transcription [abc] – phonetic transcription abc – phonetic transcription in non-IPA system, quoted from external sources abstr. – abstract noun 28 Aspects of the historical phonology of Manx adj. – adjective AV – Authorized Version (King James Bible) C – consonant; broad consonant Cʲ – slender / palatalized consonant CM – Classical Manx (18th century) comp. – comparative (and superlative) cond. – conditional Cr. – Cregeen’s dictionary (Cregeen 1835) dat. – dative dial. – dialect, dialectal ed. – edition, edited by EIr. – Early Irish (Early Gaelic) (=Old and Middle Irish) EM – Early Manx (17th century, Phillips’ prayer book) Eng. – English f. – feminine f. – folio (leaf in manuscript) fn. – footnote Fr. – French fut. – future G. – Gaelic (Goidelic), Early Modern Irish / Gaelic (see §0.4), ‘Common Gaelic’ gen. – genitive Goi. – Goidelic (in quotations from Rhŷs), = G., Gaelic H – heavy syllable ˈH – stressed heavy syllable impv. – imperative invar. – invariable IPA – International Phonetic Alphabet Ir. – Irish J: – data in HLSM from Jackson (1955) (with speaker initials) K. – Kelly’s dictionary (Kelly 1866) L – light syllable ˈL – stressed light syllable L1 – first language L2 – second language l., ll. – line(s) len. – lenition, lenited Lh. – Manx vocabulary collected for Edward Lhuyd (Ifans and Thomson 1980) LM – Late Manx (19th – 20th century) LSM – Late Spoken Manx (Broderick’s term for the speech of the 20th-century informants) m. – masculine MIr. – Middle Irish (Middle Gaelic) 29 Aspects of the historical phonology of Manx Mod. – Modern MS, MSS – manuscript(s) N – north, northern Manx n. – neuter n. – noun OIr. – Old Irish (Old Gaelic) p., pp. – page(s) part. – participle Ph. – John Phillips’ prayer book manuscript and orthography pl. – plural (1pl. = first person plural, etc.) pret. – preterite r – recto (front side of leaf in manuscript) S – south, southern Manx s.v. – see under entry (sub verbo) ScG. – Scottish Gaelic sg. – singular (1sg. = first person singular, etc.) usu. – usual(ly) V – phonetic or phonological vowel; orthographic vowel symbol v – verso (reverse of leaf in manuscript) vn. – verbal noun voc. – vocative vol. – volume W:N, W:S – data in HLSM from Wagner (1958–69), northern or southern informant(s 0.7 Abbreviations of names of speakers and fieldworkers Examples from the last native speakers (§1.6.9) are marked with the initials of the speaker as given by Broderick (HLSM I: xxvii–xxviii), e.g. NM for Ned Maddrell, JTK for John Tom Kaighin (see also map §0.8). All such data are from Broderick’s dictionary (HLSM II) unless otherwise stated. These abbreviations are also used in presenting data from Jackson and Wagner (whether directly cited or via Broderick) rather than their own abbreviations of the informants’ names. ‘J’ means that the example is taken from Jackson (e.g. J:EK = example from Eleanor Karran noted by Jackson), and ‘W’ means Wagner (his data are marked only S ‘south’ or N ‘north’). Wagner also includes data from Marstrander which is labelled M. 30 Aspects of the historical phonology of Manx 0.8 Map of the Isle of Man Shown are parish5 boundaries, major settlements, and certain other places mentioned in the thesis. The informants whose speech is transcribed in Broderick’s Handbook of Late Spoken Manx (HLSM) are located on the map by place of upbringing, so far as this can be determined (following Broderick 2018a).6 Their initials are given as listed 5 The parishes date back at least to the late Middle Ages, and are still (together with more recently established town and village authorities) the basis of local government at the present day. <https://www.gov.im/media/1351687/map-local-authorities.pdf> [accessed 21.09.2019] 6 * born in Lezayre but brought up in Lonan (Broderick 2018a: 164); ** born(?) and brought up initially in Liverpool (Broderick 2018a: 146); *** born Jurby (Broderick 2018a: 142); **** resident at Ballaskeig Beg, Maughold, when visited by Marstrander; no further information (Broderick 2018a: 181). 31 Aspects of the historical phonology of Manx in HLSM (I: xxvii–xxviii) (§0.7). The traditional (pre-1796) north-south administrative division (running along the central ridge of mountains) is shown (Broderick 1999: ix) which has a bearing on dialect, although the status of Maughold in particular is ambiguous,7 and Broderick (HLSM I: xxvi) treats Thomas Christian (TC) as a northern speaker. 7 <http://www.isle-of-man.com/manxnotebook/parishes/parishes.htm> [accessed 21.09.2019]. 32 Aspects of the historical phonology of Manx Chapter 1 Introduction 1.1 Purpose and scope of the thesis The purpose of this thesis is to elucidate some of the hitherto poorly understood aspects of the diachronic development of Manx phonology, both from earlier varieties of Gaelic and within the attested period of written and recorded Manx, and to situate these developments in the wider context of Gaelic dialectology and historical linguistics. The lack of systematic analyses of the linguistic features of Manx has been recognized as a serious hindrance to Gaelic studies. For example, Ó Maolalaigh (1997: 5, 11), in his doctoral thesis on the historical phonology of Gaelic short vowels, states the difficulties of working with the existing research on Manx and the ‘raw’ phonetic data which it presents: Manx dialects have not been referred to in the core chapters […] for practical reasons, the main ones being (i) the absence of a monograph on a single dialect or dialect area of Manx and (ii) the difficulty of comparing the mass of raw phonetic Manx data to the phonological data of Irish and Sc[ottish] G[aelic] dialects […] [a]lthough the evidence of Manx is crucial for a full understanding of the development of Gaelic. (Ó Maolalaigh 1997: 5) All accounts of Manx exhibit a phenomenal degree of phonetic diversity which is difficult at the present state of research to analyse structurally. The inclusion of such ‘raw’ data in a minute phonological study like the present would be futile. (ibid.: 11) The sometimes overlooked importance of Manx within Gaelic linguistics has been pointed out by Thomson (1960: 116; 1969: 178) (see also §1.6.2): Despite the late date at which Manx first appears in a written form it has […] the special advantage of never appearing in the usual Gaelic orthography. Had it done so it would no doubt be as coyly uninformative about the beginnings of svarabhakti as the other more conventional dialects are. (Thomson 1960: 116) 33 Aspects of the historical phonology of Manx Apart from the intrinsic interest of the development of an isolated branch of East Gaelic, the non-traditional orthography allows us to observe soundchanges which are masked by conventional Gaelic spelling […], and with regard to grammar and meaning early Manx can shed light on that early period of Scottish Gaelic when writers of the languages still felt bound to the standards of literary Irish. (Thomson 1969: 178) In the spirit of the titles of Rhŷs’s Outlines of the Phonology of Manx Gaelic (1894) and Jackson’s Contributions to the Study of Manx Phonology (1955), it is recognized that the present thesis is only a partial and incomplete treatment of the topic. It has not been possible to deal with every aspect of Manx historical phonology here, but it is hoped that the thesis goes some way to filling the gap noted by Ó Maolalaigh. There is a particular focus on the development of the vowel system (Chapters 2, 3, also §4.6), along with developments of the sonorant consonants (Chapter 4), which have a particular bearing on the vowels, and prosodic phenomena (particularly stress) (Chapter 5). With regard to sources and methodologies I have throughout the thesis particularly focused on two previously neglected sources: (a) the extensive early descriptions of native Manx speech by Rhŷs (1894) (§1.5.2), and (b) the rich but difficult to interpret evidence of the Manx orthographies (§§1.6.1–1.6.7). I have also applied quantitative methods to existing bodies of data, which is largely new in the study of Manx,8 together with instrumental phonetic analysis of some of the recordings of the terminal speakers. Given the time-consuming nature of these methodologies, it has not been possible to apply them to every area where they might prove useful, but it is hoped that these analyses indicate what can be achieved and provide a basis for future research (§§6.2, 6.4). 8 Although see Ó Sé (1991) (§5.1.1.5), also Thomson (1969) and Broderick (2011), where some quantitative data are provided. See also Max Wheeler’s recent papers which apply corpus methodologies to various aspects of Manx morphology and syntax. <https://sussex.academia.edu/MaxWheeler> [accessed 29.08.2019]. 34 Aspects of the historical phonology of Manx 1.2 Descriptive and theoretical concerns It is hoped that the topic of this thesis will be of interest to a wide range of scholars, including those who specialize in descriptive and theoretical phonology, as well the broad fields Gaelic or Celtic Studies. I have therefore sought to keep the discussion as accessible as possible, and to avoid detailed discussion of issues of phonological theory except where essential to the argument being made. This thesis takes as its starting point the above observation that insufficient fundamental descriptive data on Manx phonology has hitherto been available, and that the first task of the researcher, in such circumstances and with limited space, is to present as much data as possible in an intelligible fashion, with analysis limited to initial and cautious interpretation of the data using a minimum of basic theoretical concepts defined as broadly and noncommittally possible. For example, when I refer to ‘phonemes’, the reader should understand in broad terms what is conventionally understood by such a concept in phonology, but I do not endorse any of the particular rival analyses of what exactly a ‘phoneme’ is, or take a position on whether the existence of such units can in a final analysis actually be justified. Similarly, in my discussion of the history of stress patterns in Gaelic (§5.1.1.6), I cite Green’s (1997) analysis of this topic, referring in general terms to typological generalizations such as the observation that unstressed heavy syllables are dispreferred, but without detailing Green’s particular Optimality Theory analysis of the phenomena in question, which can and have been analysed in a number of other frameworks. My approach thus has regard to the tension between descriptive and theoretical approaches mentioned by Hayes (1995: 5): I often found that it was precisely by moving beyond theory-centred writings to the original sources on which they were based that the data could be found to support a sharply different analysis. It is only natural that theorists, pressed for space, will focus on the data most relevant to their own analyses. (Hayes 1995: 5) 35 Aspects of the historical phonology of Manx There is, of course, also a risk that spurious or contradictory suppositions about the data will be made if insufficient attention is given to theoretical concerns. It is therefore my hope that the data presented here will be built on by future scholarship, which will subject them to more detailed and rigorous theoretical analysis where appropriate. 1.3 Historical background of Manx Scholars have conjectured that Gaelic entered the Isle of Man around the same time as the expansion into Scotland (by around AD 500) (Jackson 1953: 173; Williams 1994b: 739; Broderick 2009: 305), although later dates have also been suggested such as the eighth or ninth century (Watson 1926: 172–4). For an assessment of the options, see Thomson (2015). Ogham inscriptions dating from the fifth to the seventh century are the first attestations of Gaelic writing in the island (Broderick 1999: 13). A Brythonic language was apparently spoken before or alongside Gaelic, as evidenced by the bilingual and biscriptal Latin / Brythonic and Gaelic Knock-y-Dhoonee stone (c. 600) (Jackson 1953: 173; Thomson 2015: 241–3), and possibly the placename Hentre if this represents Welsh hen dref ‘old settlement’ (PNIM I: xxiii). Thomson (1992: 100; 2015: 252) also suggests that Manx, along with Scottish Gaelic, shows signs of a Brythonic substrate, e.g. in the verbal system. Whether Gaelic survived the Norse period or was reintroduced has been a matter of debate, especially on the basis of place-name evidence (Marstrander 1932; Gelling 1971, 1991; Megaw 1976; Fellows-Jensen 1983, 2015; Thomson 1983, 2015). I have argued (Lewin 2017a: 164–6, 171–3) that survival is more likely — and, at any rate, that the specific sociolinguistic circumstances did not exist which would result in a significant Norse substrate in Manx grammar, contrary to the hypothesis of Williams (1994b: 737–41). There is some evidence of the participation of Man in a wider culture of Gaelic learning at an early period (Macquarrie 2015), but this seems to have been disrupted first by Norse invasion (c. 900) and especially later when the island came under rule by English magnates in the fourteenth century. An early bardic praise poem to the King of Man, Rǫgnvaldr (Raghnall, Reginald) Guðrøðarson (Ó Cuív 1953; Clancy 1998: 36 Aspects of the historical phonology of Manx 236–41; Macquarrie 2015: 297–300; Etchingham et al. 2019: 123–96) appears to attest to a mixed Norse-Gaelic culture in the island, but it is not clear to what extent the Irish literary tradition was established in the island. Given Raghnall’s involvement in the politics of the wider Irish Sea world (McDonald 2007), the poem could well have been composed and performed at the court of an ally, relative or subordinate in Ireland or Scotland. If the Gaelic literate tradition was present in the medieval period, no trace of it has survived, and it does not seem to have been known in the period when it became necessary to write Manx for religious purposes after the Reformation.9 The first continuous Manx prose text, Bishop Phillips’ manuscript translation of the Anglican prayer book, is dated to around 1610 (Thomson 1953), and has an orthography diverging in several respects from the later eighteenth-century system (§1.6.3). Both systems are based to a large degree on contemporary English orthographic conventions. The first printed text is a bilingual catechism from 1707, and the orthography used in this volume was gradually developed through the eighteenth century, culminating in the completion of a Bible translation in 1773. Later texts include hymn books (Lewin and Wheeler 2019), religious tracts, and newspaper articles (Lewin 2014a). Literacy in Manx apparently became fairly widespread, as attested by the large quantity of manuscripts of carvals or religious ballads surviving from the mideighteenth century onwards, often composed and copied by ordinary people (Moore 1891). The carval and sermon manuscripts remain largely unstudied and are often in less standardized versions of the Manx orthography (see Lewin 2015b), providing valuable evidence of pronunciation. The eighteenth and nineteenth centuries also saw the collection of a significant number of Manx traditional songs (Moore 1896; Strachan 1897; Gilchrist et al.: 1924–26; Thomson 1960–62; Broderick 1981a; 1981c; 1982b; 1984a; 1984b; 1984c; 1990; 2015b; 2018e; 2018f; Ó Muircheartaigh 2016). 9 See Lewin (2015b: 83) for discussion of Rhŷs’s (33–4, 170) and Williams’ (1994b: 704–6) claims that Phillips’ orthography is based on a pre-existing Manx writing tradition, which I judge to be unlikely. 37 Aspects of the historical phonology of Manx The unmarked use of written Manx for vernacular purposes petered out in the middle of the nineteenth century as monoglots became scarce and language shift to English gathered pace (Thomson and Pilgrim 1988: 16–17; Broderick 1999: 27–30, 254; Lewin 2014a: i), and the composition and publishing of Manx texts from then on has been essentially an antiquarian or revivalist activity. 1.3.1 Periodization of Manx 1.3.1.1 Early, Classical and Late Manx For Manx the conventional periodization outlined e.g by Broderick (1999: 77) and Thomson (2015: 247) is used, i.e. Early Manx (seventeenth century, essentially the language of Bishop Phillips’ translation of the Anglican prayer book, c. 1610), Classical Manx (eighteenth century, the language of the Manx Bible completed in 1773) and Late Manx (nineteenth century). See also Lewin (2016a: 183). These are of course only vague labels of convenience and do not imply clear boundaries between the periods. Notably, the ‘Traditionary Ballad’, although preserved only in eighteenth-century manuscripts, shows some linguistic features more archaic than those of Phillips, in accordance with its presumed date of composition prior to 1520 (Thomson 1960–62), and the second-earliest known Manx prose text, a sermon from 1696 (Lewin 2015b), can be regarded as transitional between Early and Classical Manx. The Fenian Ballad Finn as Ossian (Broderick 1990; 2018f; Ó Muircheartaigh 2016), and the lament Baase Illiam Dhone on the death of William Christian in 1663 (Broderick 1981a), as well as other folk-songs, likewise show older linguistic features, although again transmitted in eighteenth and nineteenth-century manuscripts. As for Pargys Caillit, the Manx adaptation of Milton’s Paradise Lost, printed c. 1796 and previously believed to have been composed in his youth in the early 1770s by Thomas Christian (1754–1828), Vicar of Malew (1780–99) (Thomson 1995), strong 38 Aspects of the historical phonology of Manx internal and circumstantial evidence has recently been adduced10 to suggest that it was in fact composed in middle age (1730s or 40s) by his grandfather, also Thomas Christian, Vicar of Rushen (1713 to 1727) and of Marown (1734 to 1752). The poem shows a number of linguistic archaisms reminiscent of Phillips, Woods’ sermon, and Bishop Thomas Wilson’s bilingual catechism Coyrle Sodjeh, the earliest printed text (1707). Early eighteenth-century texts, including Pargys Caillit, Coyrle Sodjeh and the first edition of Matthew’s gospel (1748, although apparently translated c. 1722), are thus more linguistically archaic than texts from the second half of the century. To an extent which is difficult to assess, the choice of linguistically conservative or innovating forms is also likely to reflect dialect, idiolect, or register variation. The latter factor must always be borne in mind especially when considering texts from the late eighteenth and nineteenth centuries when literacy had become more established, and the archaizing influence of the Bible is sometimes to be suspected (cf. Broderick 1982a: 178–9). 1.3.1.2 The language of the terminal speakers In addition to the above periodization, I consider it important to distinguish, although not necessarily categorically, between the language of those born up until the early decades of the nineteenth century, who (in more isolated areas and marginalized socioeconomic strata at least) (Lewin 2019a: 79–82) were evidently Manx-dominant speakers with a full native command of the language, and those born late enough to be recorded in the mid twentieth century, all of whom are to be regarded, though to varying extents, as English-dominant ‘semi-speakers’ (cf. Dorian 1977) showing clear signs of ‘incomplete acquisition’ (Montrul 2008) (§1.6.9.1). 10 Max Wheeler, personal communication. 39 Aspects of the historical phonology of Manx Broderick (1999) does not always distinguish clearly between these two categories of speakers, classing both as ‘Late Manx’ and claiming to find some of the attrition or language shift features of the last speakers in the language of earlier writers, which Lewin (2017a: 189–91) disputes. The language of these last speakers (here ‘terminal speakers’) is here distinguished as ‘Terminal Manx’ (=Broderick’s ‘Late Spoken Manx’). ‘Late Manx’ here, unless otherwise qualified, refers to the speech of those born from the late eighteenth century to the mid-nineteenth century, which may show an intensification of language contact, but not necessarily the features of language obsolescence and incomplete acquisition found in the terminal speakers. 1.3.1.3 Revived Manx ‘Revived Manx’ is a variety spoken by a few hundred people today as a second language (or, in a few cases, as a first language acquired from second-language speakers). The development of this variety can be traced back to the late nineteenth century and especially the mid-twentieth century when a small number of enthusiasts sought out the last remaining traditional speakers (Broderick 1999: 172–87; Stowell 2005; George and Broderick 2009; Lewin 2015a; 2016c). This thesis is concerned only with developments in ‘Traditional Manx’, i.e. the Early, Classical and Late Manx periods referred to above, representing speech varieties passed down by uninterrupted intergenerational transmission from earlier periods of Gaelic (cf. Jackson: vi). For discussion of the linguistic differences between Traditional and Revived Manx see Lewin (2015a; 2016c). 1.4 The place of Manx within the Gaelic dialect continuum Manx shares features both with Irish and Scottish dialects, although it has generally been held that Manx is closer overall to Scottish Gaelic (O’Rahilly 128–40). Jackson (1951: 91–2) groups Manx and Scottish Gaelic together as ‘Eastern Gaelic’, branching off from an earlier ‘Common Gaelic’. While critiquing Jackson’s concept of ‘Common 40 Aspects of the historical phonology of Manx Gaelic’, Ó Buachalla (2002) continues to regard Manx and Scottish Gaelic as varieties of ‘Eastern Gaelic’, although he considers that the difference between northern dialects (Ulster Irish, Manx, Scottish Gaelic) and southern dialects (Connacht and Munster Irish) is more significant. Recently, this ‘tree model’ approach has been problematized by Ó Muircheartaigh (2015: 45): The classification of Manx and Scottish Gaelic together is not unproblematic, especially in light of evidence for an early fundamental structural feature (the development of eclipsis) shared by Irish and Manx to the exclusion of Scottish Gaelic […] In short, the tree model seems particularly ill-suited to a description of the position of Manx. (Ó Muircheartaigh 2015: 45) Manx has also been linked to Munster Irish, especially on the basis of shared phonological developments such as non-initial stress, as well as some shared lexis (Williams 1994b: 740–1). The ‘distinctiveness’ of Manx lexis has also been highlighted by Ó Muircheartaigh (2015: 75–76), on the basis of data from Elsie (1986) showing Manx basic lexis to be closer to Old Irish than to any other modern dialect. According to Ó Muircheartaigh, The situation of Manx, although covering a small geographic area, is particularly interesting for a variety of reasons. In historical linguistic terms, it is perhaps the Gaelic variety least suited to the strait-jacket of the Darwinian tree-model in which it has traditionally been analysed (O’Rahilly 1932; Jackson 1955; Broderick 2009). Given the mix of extremely archaic and innovative features one finds in Manx, along with its historical and geographic position as a centre, for a period at least, of traffic between Ireland, Britain and the Scandinavian north, it provides an interesting test case for an array of sociolinguistic theories discussed in this thesis. Most especially, it appears that Manx could be usefully analysed in terms of medieval new-dialect formation. (Ó Muircheartaigh 2015: 311) If it is accepted, however, that the ‘tree model’ is only really valid when one population is almost entirely isolated from another, it is no surprise that important isoglosses are found in various locations throughout the Gaelic-speaking area, and that no definitively unproblematic categorizations of Gaelic dialects can be attained. As Ó Muircheartaigh (2015: xx) has observed, commenting on Ó Buachalla (2002), ‘it 41 Aspects of the historical phonology of Manx would not be difficult to find twenty-three linguistic features linking any two parts of the Gaelic-speaking world’. Given that the Isle of Man is geographically both central in the Gaelic area (on the north-south axis and in terms of sea routes) and peripheral (in that it is located on the eastern edge, and has been socially and politically isolated from other Gaelic areas since the Late Middle Ages), it is also not surprising that Manx shows a number of highly distinctive and divergent features, as well as sharing features with diverse points of the dialect continuum. It is not clear that ‘new-dialect formation’ (implying a founder population of geographically diverse origins) is necessary to explain the features observed, although it is not implausible that this occurred. The Isle of Man would certainly have been within fairly easy reach of settlers from along the east coast of Ireland, Galloway, and the Hebrides and western Highlands. Further historical research, including historical genetics, as well as work on Manx personal and placenames, and those of neighbouring areas, may potentially elucidate this issue. Moreover, it is not clear that some shared features are necessarily the “same” feature at all. For example, diphthongization of historically short /e, a, o/ + /N/ is found in both Munster Irish and northern Scottish Gaelic, as well as the northern dialect of Manx (HLSM I: 161), in e.g. G. ceann ‘head’, Manx kione, S /kʲoːn/, N /kʲau̯ n/. Some form of lengthening, rounding or diphthongization in this position is widespread in Gaelic dialects (LASID I: 120; SGDS II: 165–6) as developments of what the Irish bardic grammarians called síneadh meadhónach ‘middle quantity’ (Greene 1952: 212; Mac Cárthaigh 2014: 168–71) before original fortis /N/.11 Phillips’ spellings (such as kian) suggest retention of a pronunciation [kʲaNː] in c. 1610, so the development of the forms represented by the eighteenth-century spelling kione must be fairly late, and it is not clear that it should be linked to the Munster or northern Scottish forms. It is not surprising that shared inherited structures give rise to a limited range of distinct outcomes in different dialects, which may nonetheless be parallel, independent developments. 11 Even in those dialects of Ulster and southern Scotland where the forms can be phonologically represented as /kʲɛN/ or /kʲaN/, there may be some phonetic lengthening of the vowel as well as the consonant, i.e. [kʲɛˑn̪ ˠˑ], [kʲaˑn̪ ˠˑ] (cf. Jones 2010: 61). See §4.5.4.2. 42 Aspects of the historical phonology of Manx Similarly, stress shift in Manx has been compared to the southern Irish development (Williams 1994b: 740), but is distinct in significant ways, such as its being conditioned by the length of the vowel in the preceding syllable (§5.1), the fact that unstressed syllables containing /ax/ attract stress in Munster dialects but not in Manx, and the fact that pretonic long vowels (at least in items which are synchronically monomorphemic) are shortened in Manx but not in Munster Irish (§5.1.3). It is interesting that developments in stress in Manx involve a combination of “Munster” stress shift and “Scottish / Ulster” vowel shortening, but whether this reflects dialect affinity or contact, or independent developments, is not immediately obvious (§5.1). Again, all Gaelic dialects inherited the violation of the weight-to-stress principle (whereby long vowels are disfavoured in unstressed syllables), and the options for eliminating this tension are limited (cf. Green 1997: 69–97). Another example is the ‘breaking’ of /eː/ to /ia̯ /, which outside Manx is found mostly in northern Scotland and in Munster. From orthographic evidence this would seem to have developed in Manx quite late, during the seventeenth century (§2.2.7). As in the other cases, it is unclear that this represents any particular relationship with other Gaelic dialects, not to mention that it is a recurring development from Old Irish onwards (McCone 1994: 89). Although further detailed consideration of the question of how to classify Manx within the Gaelic languages is beyond the scope of this thesis (but see §6.2), it is hoped that the data and analysis presented here will at least provide a clearer picture of some of the features present in Manx for scholars interested in comparative questions. 1.5 Review of previous scholarship Given that the evidence of Manx is both copious and valuable for Gaelic historical linguistics and dialectology, it is regrettable that the language has received so little scholarly attention, notwithstanding the best efforts of a small number of researchers. No comprehensive historical grammar, phonology or dictionary of Manx exists; the only works available are dated publications by amateur scholars such as Kneen’s grammar (1931) and dictionary (1938), or prescriptive revivalist works such as 43 Aspects of the historical phonology of Manx Fargher’s (1979) dictionary and Kewley Draskau’s (2008) grammar, as well as works from the late eighteenth and early nineteenth century by native scholars (Kelly’s grammar and dictionary, Cregeen’s dictionary). Otherwise, the only general academic descriptions are short chapters by Thomson (1984; 1992; 2000), Williams (1994b) and Broderick (2005; 2010). Much work of value is to be found in fairly obscure or local publications, even from the perspective of Celtic Studies. Notably, some of the most extensive linguistic notes on Manx are in editions or commentaries on texts by Thomson (1981, 1995, 1997, 1998) published in the Isle of Man by the language organization Yn Cheshaght Ghailckagh. Linguistic commentary is also to be found in Broderick’s articles, mostly editions of various texts (1981a; 1981b; 1981c; 1982a; 1982b; 1983; 1984a; 1984b; 1984c; 1990; 2011; 2017; 2018a; 2018b; 2018c; 2018d; 2018e), Thomson’s articles and editions in various journals and volumes (e.g. 1950; 1954–59; 1960; 1960–62; 1963; 1969; 1976; 1988; 1990; 1991; 1999), and my own work (Lewin 2011; 2014; 2015; 2016a; 2016b; 2016c; 2017a), as well as papers from a few other authors who have taken an occasional or inchoate interest in Manx, such as Ó Sé (1991), Breatnach (1993) and Ó Muircheartaigh (2016). Work on wider Gaelic linguistics has tended to ignore Manx, or to mention it only in passing, although as Ó Maolalaigh observes in the quotation given above, this does not necessarily reflect neglect or apathy, but rather the lack of readily available and interpretable descriptions to set beside Irish and Scottish data. A perception that Manx, the ‘Cinderella of Gaelic tongues’ (O’Rahilly 1932: ix; Thomson 1969: 177) has little of value to offer the scholar, and is merely an anglicized dialect which ‘hardly deserved to live’, to quote O’Rahilly’s (1932: 121) notorious evaluation, may also have played a role; see Lewin (2017a) for discussion. 1.5.1 Material collected for Edward Lhuyd The pioneering polymath and scholar of the Celtic languages Edward Lhuyd (1660– 1709) includes in a multilingual glossary in his Archæologia Britannia (Lhuyd 1707: 44 Aspects of the historical phonology of Manx 290–8) a selection of Manx lexical items in an idiosyncratic orthography (Thomson 1968), but these were apparently collected by William Jones, one of Lhuyd’s assistants (Ifans and Thomson 1980: 129), who did not understand the language; as a result the attempted phonetic representation is very approximate. Moreover, the choice of informant appears not to have been ideal (Thomson 1999: 390). Upon the discovery of Jones’ manuscript (National Library of Wales MS 13234A, pp. 73–128), the word list was edited by Ifans and Thomson (1980), and the linguistic information which can be gleaned from the data is analysed by Thomson (1999). Where this material is cited in the present thesis, the abbreviation ‘Lh.’ is used; this refers to Ifans and Thomson’s edition. 1.5.2 Rhŷs (1894) Despite its neglect by later generations of Celticists, Manx did not escape the notice and interest of the first Professor of Celtic at Oxford, the Welshman Sir John Rhŷs. His analysis of Manx historical phonology is published in his treatise Outlines of the Phonology of Manx Gaelic (1894), and is based on extensive fieldwork across all parishes of the island carried out during six visits between 1886 and 1893. The initial purpose of the visits was to study the Ogham stones of the island, but Rhŷs soon developed an interest in the vernacular language and folklore of the Manx people. In addition to his Outlines, significant amounts of data survive in notebooks now preserved in the National Library of Wales (Broderick 2016a; 2016b; 2018c; 2019; Lewin 2019a). Rhŷs’s work has a number of important advantages over that of later scholars, including the number, geographical range and birth-dates of his informants. Rhŷs records 107 potential informants in his notebooks, of whom 88 were interviewed (Broderick 2018c: 45). These were largely native speakers of an earlier generation than the terminal semi-speakers encountered by Marstrander, Jackson and Wagner and transcribed by Broderick (HLSM). Many were born in the 1810s to 1840s, in contrast with the terminal speakers who were born from the 1840s to 1870s (Broderick 1999: 54–66). 45 Aspects of the historical phonology of Manx Rhŷs’s informants would be more likely to be Manx-dominant or balanced bilinguals. They are reported as having features such as consistent h-prefixation after certain clitics (Rhŷs: 72), in contrast to the terminal speakers’ limited use of this feature (HLSM I: 23), extensive vowel nasalization (Rhŷs: 31–48; Lewin 2019a: 82–9), and grammatical gender concord (Lewin 2019a: 79–82). In most cases, those interviewed by Rhŷs would have been raised in a largely Manx-speaking community, socialized among Manx-speaking peers, and would have continued to use the language regularly for a substantial portion of their lives (albeit perhaps less regularly by the time of Rhŷs’s visit owing to the changing sociolinguistic situation). In contrast, the terminal speakers were raised a few decades later in communities already undergoing rapid language shift, with the language widely stigmatized as a marker of backwardness (Broderick 1999: 35–7). While they may have used Manx (actively or passively) in their youth with the older generation, peer group socialization was largely in English (Lewin 2014b, 2017a: 191–3; Miller 2007, and see §1.6.9 for further discussion), and their speech shows clear signs of uneven and incomplete acquisition (§1.6.9.1). Rhŷs’s data are thus particularly valuable in providing details obscured or unavailable in the speech of the later terminal speakers. In addition to his principal informants, Rhŷs relates that he engaged in briefer exchanges with many more Manx speakers as he travelled around the island. This seems to have given him a “feel” for what were the most common forms, perhaps leaving him less susceptible to the idiosyncrasies of individual informants in giving a general overview of the language. Rhŷs is, however, also careful to note any unexpected features and the circumstances in which his data were collected, which are frequently of use in analysing the material. When he is uncertain of the articulation of a particular sound, he notes his uncertainty, and explains his thought process for coming to his conclusions. Rhŷs was well-aware of his lack of training as a phonetician and readily admits it: In attempting to deal with the Manx vowels, I have had to classify them as best I could according to their effect on my ear; for I rarely could ascertain with any precision how they are formed. I should have been glad to have described them in the exact terminology with which Dr. [Henry] Sweet’s works on English philology have made us familiar; but convinced as I am that my ear has not been 46 Aspects of the historical phonology of Manx trained—under no circumstances probably could it have been trained—to appreciate the nice distinctions which English phonologists think it requisite to draw between closely related vowels, I have abstained from the attempt to follow their example. Even if I escaped blundering hopelessly in such an effort, it would only tend to make the reader fancy that I am blest with a power of discrimination which I cannot claim in the matter of phonetics. (Rhŷs: 1) Jackson (4) criticizes the accuracy and intelligibility of Rhŷs’s work: Sir John Rhys was the pioneer, with his “Outlines of the Phonology of Manx Gaelic” [...] This gives a good deal of information, but it is not always very intelligible, and in some cases one suspects his accuracy. (Jackson: 4) Rhys and Kneen both give the equivalents of üː [yː], üːə, or üə, but it may be doubted whether they really knew what this means […] and what they heard was probably uː, uːə, uə, or ʎː [ɯː], ʎːə, ʎə. (Jackson: 48) Jackson (48) seems to confuse Rhŷs’s description of Manx reflexes of G. ua (probably [ɨә̯, ɨː] or similar), with those of ao and uai / aoi ([әː] or [eː]), which Rhŷs carefully distinguishes (§3.5.1).12 That Rhŷs may admittedly have confused rounded and unrounded vowels here may be a more widespread problem of early phonetic descriptions of Gaelic (§3.5.6), and his descriptions are at least useful in determining which vowel sounds were alike and which were distinct, even if other evidence is needed to confirm their exact quality. Nevertheless, the weaknesses of Rhŷs’s work must be borne in mind in making use of his descriptions. Many of these can be overcome by judicious and patient analysis of the text in context, but nevertheless may have contributed to later scholars’ difficulty in making use of it. These weaknesses include the following: 12 For another problem with Jackson’s dismissal of Rhŷs, cf. Broderick’s (1999: 74) comment ‘[i]n view of the abundance of native Manx speech material now available, Jackson’s [4] view that claims for North/South dialectal differences do not stand up to examination cannot be endorsed.’ 47 Aspects of the historical phonology of Manx (a) As noted above, phonetics was not Rhŷs’s primary area of expertise and it was not one in which he had any formal training. In addition, the science in general was in its infancy, as was its application to the Gaelic languages. As noted by Sommerfelt (1959: 51), Rhŷs’s work was only the second publication on the phonetics and phonology of a Celtic language ‘which was up to the standards of contemporary science’, and the first major phonetic description of any Gaelic dialect. (b) His transcriptions are inconsistent in various ways. Frequently, and without indication, items are left partially in their orthographic form with only certain phones, often only the one under discussion, being transcribed. Rhŷs seems to have tried as far as possibly to avoid the use of symbols not yet introduced in the text of the book, rather than using a consistent notation all the way through his treatise and providing a key or index (as Jackson does). (c) Despite his disavowal of overly narrow transcription (Rhŷs: 1, quoted above), he sometimes seems to attempt to draw just such overly narrow distinctions, only to disregard or forget about them later on. (d) Vowel length is indicated by a macron over long vowels, and sporadically by a breve over short ones. However, frequently, no indication of length is given at all, especially if the vowel character already has a diacritic (e.g. ů, ü). The length intended can, however, often be deduced from the context (e.g. whether the item is cited in a section discussing a long or a short vowel). (e) Sounds are often described by comparison with other languages, principally English, Welsh, French and German, as well as Irish and Scottish Gaelic. It is, however, often unclear exactly what sound is meant by these comparisons, or what precise varieties are intended. (f) Sometimes different indications of a phone’s value seem to contradict one another, and it is difficult to work out what Rhŷs probably heard, at least from his descriptions alone. His ȳ (§3.5.1) is a case in point. 48 Aspects of the historical phonology of Manx (g) Rhŷs’s work is written in rather flowery and anecdotal continuous prose, with many digressions, and there is no index, which makes it less accessible than would be desirable, and may have contributed to Jackson and other scholars’ neglect of it. 1.5.3 Strachan (1897) In 1883 Strachan phonetically transcribed the Manx folksong Ec ny Fiddleryn13 from Thomas Kermode (1825–1901), a fisherman of Bradda near Port Erin. (Bradda was one of the strongholds of Manx noted by Rhŷs.) By the time of publication in the first issue of Zeitschrift für celtische Philologie (1897), he states that he has modified the notation to agree with Rhŷs’s system: I wrote down the song phonetically as well as I could, but, as I have unfortunately had no special training in phonetics, my attempt is at the best only an approximation. However I read the song as I had written it down to two other Manxmen and it was intelligible to them. As to symbols, ə, æ, ů, y, and ȳ have been used as in Prof. Rhŷs’s Outlines of Manx Phonology, though I am not quite sure that ȳ is exactly the long sound of y. (Strachan 1897: 54) 1.5.4 Marstrander (1932) In his long article on the Norse place-names in the Isle of Man, Marstrander includes a brief overview of Manx phonology, based on data from six informants (three from the north, including Peel, and three from the south), born between 1846 and 1854. The different phones listed are categorized according to the Gaelic (or Norse) sounds of which they can be a reflex. The transcription is broader and somewhat simpler to interpret than that of Wagner and Broderick, and perhaps more accurate than that of Jackson in certain respects (e.g. Marstrander shows long diphthongs, which are not noted by Jackson). In addition to his published materials, Marstrander left a considerable amount of recorded and transcribed material (Manx National Heritage Library MSS 5354–57B) (Broderick 2018d). The original spoken material is 13 See Broderick (1984c) for an edition of manuscript versions of this song with discussion and notes. 49 Aspects of the historical phonology of Manx incorporated by Broderick (HLSM) and some also appears in Wagner (LASID). Little use has so far been made of the large amount of transcriptions of Thomas Christian reading aloud, mostly from the Bible, apart from a short article discussing Christian’s vowel phonology (Thomson 1976). 1.5.5 Kneen (1931, 1938) J. J. Kneen was an amateur Manx scholar and revivalist active from the end of the nineteenth century to the 1930s. Although unable to pursue a formal academic career, he was recognized for his contributions to Manx and place-name studies with an honorary degree from the University of Liverpool and an award from the government of Norway. His two principal works on the Manx language, his grammar (completed in 1910, but not published until 1931), and his dictionary (published 1938), both contain phonetic information. The grammar contains a brief guide to Manx pronunciation in a phonetic transcription close to IPA (with some idiosyncrasies e.g. æ = [ɛ], ü = [y]), while the dictionary contains a transcription based on English orthography. Jackson (4) claims that ‘Kneen made use of phonetic symbols of the usually accepted types, but it is doubtful whether he fully understood them’. Despite the above comment, Jackson regards the dictionary as the more reliable of Kneen’s works for pronunciation, ‘if properly interpreted’, and Jackson includes a number of references to Kneen throughout his work. More fundamental questions may be raised over the degree to which Kneen’s works can be regarded as independent sources and the amount of exposure he himself had to vernacular Manx. It seems he had some degree of contact with native speech (Kneen 1927), but his ideas on the language were certainly coloured by his knowledge of written Manx and his archaizing tendencies, and his knowledge of Irish, as seen in the grammar (cf. Jackson 1955: 5), which is in many respects, according to Thomson (1969: 189), ‘merely a transcript of the Christian Brothers’ Irish grammar’. None of 50 Aspects of the historical phonology of Manx Kneen’s transcriptions are presented as directly representing the speech of a particular informant or informants. In the dictionary it seems that more frequently-occurring items are more likely to reflect genuine spoken usage, whereas less common words sometimes reveal Kneen’s uncertainty about pronunciation. For example, boght ‘poor’ (G. bocht) is transcribed as bawkh [bɔːx], with characteristic Late Manx lengthening of the vowel (see §5.5.2) and loss of the stop in the final cluster /xt/ (HLSM II: 38), while most other words ending in <ght> are transcribed with retention of the final /t/.14 A further example is Kneen’s erroneous transcription of maynrey ‘happy’ (G. méanar, EIr. mo-génar) as mahnra [maːnrә] rather than expected /meːnrә/, attested mendrɑ̜ (HLSM II: 293). Here he is apparently led astray by the ambiguous orthographic sequence <ay> (§§1.6.3, 1.6.4.4), and pronunciations with [aː] are subsequently found in the later revived language (Lewin 2016c: 45). Some of the information given by Kneen is clearly based on other sources, especially Rhŷs. Kneen’s (1931: 29–30) section on ‘Dialect’ in the grammar is a near word-forword, unacknowledged reproduction of parts of Rhŷs’s (160–1) corresponding section. Kneen also uses some of the same terminology as Rhŷs, such as ‘mouillé’ for ‘palatalized’ or ‘slender’ (Kneen 1931: 38), and quotes Rhŷs on the subject of secondary lenition (Kneen 1931: 39). As Thomson (1969: 189) warns, ‘Kneen’s description of the language should not be relied upon except where it is independent of its source or other evidence confirms it’. In view of the foregoing, Kneen’s data are not used in this thesis. 14 E.g. toshiaght ‘beginning’, transcribed as tozhakht, where the final orthographic t is probably a hypercorrection on the model of other nouns in -aght, cf. G. toiseach, tosach (Phillips mostly has spellings without t, e.g. tossiagh, once tossiaght). None of the terminal speakers have final [xt] (HLSM II: 454). Kneen also gives the gender incorrectly as feminine, on the pattern of most nouns in -aght, whereas it is masculine in Cregeen, and in Irish and Scottish Gaelic, earlier neuter. 51 Aspects of the historical phonology of Manx 1.5.6 Carmody (1947, 1954) Carmody’s first work on Manx, Manx Gaelic Sentence Structure in the 1819 Bible and the 1625 Prayer Book (1947), contains some highly dubious phonetic material obtained from a woman in California who had attended some Manx classes (cf. Jackson: 6). His subsequent article (1954) contains phrases and sentences in a broad phonetic transcription noted from five of Jackson’s informants as well as one learner or semi-speaker. The material is occasionally garbled and mistranslated, presumably owing to Carmody’s limited knowledge of Manx. There is no discussion of the diachronic development of Manx sounds; nor is any phonological outline or phonemic inventory given besides a few introductory remarks. The work is thus of limited usefulness, and is not used in this thesis. See Broderick (2018b) for further discussion of Carmody’s material. 1.5.7 Jackson (1955) Jackson’s data are derived from seven informants (four in the south, three in the north, born 1852–1877) (Jackson: 2–3) whom he interviewed on a ‘hurried trip’ (SGDS I: 36) over Christmas 1950–51. Unfortunately, Ned Maddrell, ‘the youngest and much the most fluent and alert of the surviving speakers’, was in hospital during most of Jackson’s stay and he was able to visit him only on the very last day (Jackson: v–vi). The trip was a pilot for the Linguistic Survey of Scotland (SGDS I: 36, 53), and as a consequence most of the data collection involved a questionnaire eliciting single words. Wagner, in a review of the book, criticizes this approach: Professor Jackson confines himself to phonology. But, in a case like this, one ought to present as much linguistic evidence, i.e. sentences and phrases, as possible. This was definitely the last chance to hear native Manx spoken. He says (p. v): ‘I took with me a questionnaire already prepared to cover the phonology of Manx from a historical point of view.’ In the light of contemporary method this approach must be considered antiquated or ‘neogrammarian’, but it proves quite successful in the case of Manx. (Wagner 1956: 107) 52 Aspects of the historical phonology of Manx Jackson defends his approach partly on practical grounds: Only two speakers had any real fund of continuous narrative material, in the form of little anecdotes or verses; and the inaccessibility of their homes, the number of distracting casual visitors present, and the fact that of the two one is blind and the other very old, made in their case an insuperable barrier to the accurate recording of phonetic texts other than single words and brief phrases. (Jackson: v) The work begins with a short description of the vowel and consonant sounds encountered in the data and an explanation of the phonetic symbols used. The body of the work is diachronic, arranged according to the phonology of ‘Common Gaelic’ (essentially Classical Early Modern Irish) (§0.4) and comparing the Gaelic sounds with their Manx reflexes as attested in Jackson’s data. There is, however, very little discussion of earlier stages of Manx between the presumed common Gaelic ancestor and the speech of the terminal speakers, apart from sporadic references to Rhŷs, Kneen and Marstrander. Sometimes the Manx reflexes are compared with those in other contemporary Gaelic dialects. There is little consideration of the evidence provided by the orthographies, even when these lead quite clearly to conclusions opposite to those reached by Jackson. For example the orthography distinguishes <iu> e.g. iu ‘drink’ (ibh), <eeu, ieu> e.g. screeu ‘write’ (scríobh) and <ieau, eeau> e.g. cleeau ‘chest’ (cliabh), from G. /i/, /iː/ and /iә̯/ respectively + /u̯ /. In Late Manx these are mostly found as monophthongal [uː]. Jackson (72–3) claims to have heard only short diphthongs, and projects the twentieth-century century realization of all three of these as /uː/ back to an earlier period and suggests an early shortening of long /iː/ in íobh, íomh, causing the short */iv/ and long */iːv/, */iә̯v/ to fall together before vocalization of the fricative. The orthography on the other hand, as with other vowel + fricative sequences, clearly suggests otherwise, and there is no motivation for the shortening posited. For other combinations, e.g. ábh, ámh > /ɛːu̯ /, /ɛː̃ ũ̯/, Rhŷs and Marstrander give clear synchonic evidence of long diphthongs, although it is possible such length contrasts were in the process of breaking down (cf. §3.9.1.7). 53 Aspects of the historical phonology of Manx 1.5.8 Wagner (1969) Wagner visited the island six months before Jackson and interviewed all but one of Jackson’s informants. The data are given in an appendix in vol. 4 of the Linguistic Atlas and Survey of Irish Dialects (LASID) and consist of five short texts plus a considerable body of individual words, phrases and sentences arranged in a glossary under English head-words. The individual informants are not noted, but each word or sentence is labelled N ‘north’ or S ‘south’. Also some data from Marstrander are included, labelled ‘M’. Wagner’s transcription is extremely narrow, to the extent that one doubts the reality (or significance) of some of the minute distinctions made in, say, central vowels or degrees of palatalization and velarization. As also noted by Broderick (2017: 51), there are occasional signs that Wagner was influenced by cognate Irish forms. For example, in niᵇm fɑːkən hu meːrəx nee’m fakin oo mairagh ‘I will see you tomorrow’ (LASID IV: 184), hu seems to represent Irish thú /huː/ rather than Manx oo /u/, in which the preponderance of evidence points to /h/ having been entirely lost at an early date.15 Similarly, there are several examples of initial unstressed schwa representing Irish particles and prefixes generally considered to be lost in Manx, e.g. tɑ ən gridn ə ˈgiri ta’n ghrian girree ‘the sun is rising’ (Ir. …ag éirghe) (LASID IV: 186), tɑ ən kidn tʹʃət əˈʃtʹaː ta’n keayn cheet stiagh ‘the sea is coming in’ (Ir. isteach) (LASID IV: 187). Perhaps these apparent schwas are phantoms arising in the mind of the transcriber from the transition between different consonants, or from speech discontinuities, in conjunction with Irish-based expectations. In this thesis forms from Wagner are generally presented as given by Broderick who incorporates them into his dictionary (HLSM II), although Wagner’s transcriptions of the ao(i) and ua(i) vowels are discussed separately (§3.5.4). 15 Although see HLSM I: 23. 54 Aspects of the historical phonology of Manx 1.5.9 Broderick (1984–86) Broderick’s Handbook of Late Spoken Manx contains the largest body of transcribed spontaneous speech from the terminal speakers, most of it transcribed by Broderick himself from the various sound recordings (HLSM I: xv–xxvi), but also incorporating material from Marstrander, Jackson and Wagner, converted by Broderick into his own transcription system, which is largely that of Wagner in LASID (HLSM I: 1). Broderick’s data are presented in the texts in vol. 1, mostly narrative monologues and snippets of verse, as well as throughout the grammar in vol. 1 and the dictionary in vol. 2. These contain material from the texts in vol. 1, as well as from dialogues which are not transcribed separately.16 Volume 3 is an analysis of Manx phonology, comprising a synchronic section, and a diachronic section modelled on Jackson (1955), but drawing on Broderick’s much larger corpus of data. Broderick usually follows Jackson’s interpretation, as in the case of the development of íobh, íomh, iabh, iamh (§1.5.7). Broderick’s dictionary (HLSM II) is the main source of data from the terminal speakers in the present thesis. Broderick’s conception of ‘phonemes’ and ‘allophones’ is not always clear, and much use is made of the concepts of ‘free variation’ and ‘wild allophony’, when particular divergent forms may be occasional rather than usual realizations, speech errors, or restricted to particular lexical items or reflect diachronic changes rather than synchronic variation (cf. Lewin 2017a: 187–8). According to Broderick (HLSM III: xxxv), it may not even be possible to arrive at a clear picture of a phonological system from the Terminal Manx data: In circumstances such as these, where variation is more often the rule than the exception, a classical phonemic analysis as seen in Ternes (1973) is not really applicable to L[ate] S[poken] M[anx], and it has either to be adapted or considerably modified, or abandoned altogether and something else put in its place, to make some sense of the messy picture of LSM. The spread of phonetic realizations arising from different fieldworkers and the breakdown of communicative competence means that a satisfactory assignment of particular 16 Some of the recorded dialogues between terminal speakers or between terminal speakers and revivalists are transcribed orthographically, albeit with some inaccuracies, in Manx National Heritage’s Skeealyn Vannin (2003), which consists of the material recorded by the Irish Folklore Commission in 1948. 55 Aspects of the historical phonology of Manx sounds to appropriate phonological units is difficult and has given rise to a number of instances of overlapping realizations […] which may be more apparent than real. (HLSM III: xxxv) Sometimes Broderick’s claims of loss of Classical Manx phonological contrasts may be overstated, as in the case of /eː/ and /ɛː/, where quantitative instrumental analysis of data from the terminal speakers allows us to demonstrate that merger had likely not occurred (§§2.2.3, 3.7). We may compare Ó Cléirigh’s assessment of the situation of Manx phonology with reference to Jackson’s work: It may perhaps be impossible ever to offer a complete structural interpretation of Manx owing to the peculiar position it occupies as a spoken language. Indeed, it may justly be argued that the scant linguistic material available precluded anything but the phonetic approach, which Professor Jackson has employed. (Ó Cléirigh 1961: 142) Although we should certainly be aware of the limitations of the data and the informants, we should nonetheless perhaps be rather sceptical of treatments which go too far in seeing Manx as sui generis and not susceptible to the kind of systematic analyses brought to bear on other languages. The development of theoretical models to describe processes of language shift and death in the decades since the pioneering works of Dorian (1981, 1989), Dressler (1981), Sasse (1992a; 1992b) and others means that these difficulties may not be so insurmountable as Ó Cléirigh and Broderick may have supposed. 1.5.10 Toponymic and onomastic evidence Place and personal names constitute a valuable source for the study of Manx phonology, including the period prior to the continuous texts of the seventeenth century onwards (PNIM I: xvii–xviii). The rent rolls or setting books of the Lordship of Man (1506–1911) are a particularly rich source of place-name data (ibid.: xviii), in 56 Aspects of the historical phonology of Manx addition to deeds of sale, estate plans, Ordnance Survey name books, maps, etc. (ibid: xix–xxii). For place-names we have an exhaustive survey in seven volumes (PNIM: Broderick 1994–2005), including a brief analysis of linguistic implications (PNIM I: xxii–xxxvii), and chronological discussions of place-names from the pre-Scandinavian period onwards (PNIM VII: 337–83). There are also earlier works by Kneen (1925–28) and Marstrander (1932). Personal names are less well-served by up-to-date treatments. Moore (1903), Kneen (1937) and Quilliam (1989) all contain etymological speculation which ranges from the sound to the implausible in terms of historical phonological analysis. These topics are clearly in need of reassessment. These sources are outside the primary focus of the present thesis, but are referred to at certain points where appropriate. 1.5.11 Summary In summary, the general descriptions of Manx phonology over the past century have focused on data gleaned from the terminal speakers (or semi-speakers). The diachronic element has mostly consisted of comparison with earlier stages of the Gaelic languages (mainly Old and Early Modern Irish). Since Rhŷs’s work on tracing developments from the Manx of Phillips’ prayer-book to the speech of his own informants, there has been little consideration of the internal historical phonology of Manx, nor its place in wider Gaelic dialectology and historical linguistics. In particular, the large amount of evidence available in the form of the corpus of Manx writing from the seventeenth century onwards has been under-utilized. Thomson (1953: preface), at the time of writing of his dissertation on the morphology and syntax of Phillips’ prayer-book, seems to have considered Rhŷs’s work sufficient until such time as more data should be available, and he never produced an in-depth analysis of Manx historical phonology, beyond two short articles (1960; 1976), his brief descriptions of the language in edited volumes (1984; 1992; 2000), and sporadic notes in other works. 57 Aspects of the historical phonology of Manx 1.5.12 Recent linguistic literature Apart from descriptive works within the broad ambit of Celtic Studies, very few works have considered Manx phonology in the international linguistic literature. Pickeral (1988–90) gives a brief analysis of Terminal Manx phonology from a generative perspective. Ó Sé’s (1991) paper on stress shift and vowel shortening in Manx is an important example of how quantitative methods can be illuminating, and uses concepts such as dialect contact and lexical diffusion (§5.1.1.5). Green (2006) examines Manx initial consonant lenition and medial lenition from the perspective of Optimality Theory, alongside consideration of mutations in other Celtic languages. The same author’s doctoral thesis (1997) examines ‘The Prosodic Structure of Irish, Scots Gaelic, and Manx’, and provides an analysis of the motivation from stress shift and vowel shortening in Manx and other Gaelic dialects (§§5.1.1.6, 5.5.1). Chaudhri’s doctoral thesis (2007: 39–43) on the development of the Cornish consonantal system discusses Manx preocclusion in comparison with that of Cornish (§4.5.4.2). 1.6 Primary sources 1.6.1 The written corpus and the orthographies Two main orthographies have been used for Manx (Thomson 1960: 116–8): that of Bishop Phillips’ manuscript translation of the Book of Common Prayer (c. 1610) (Moore and Rhŷs 1895; Thomson 1953; Wheeler 2019), and the system used in the eighteenth and nineteenth centuries. The latter begins with in Bishop Wilson’s catechism Coyrle Sodjeh (1707),17 and standardized to a significant extent in the printed prayer book translation of 1765 and the Bible completed in 1773, as well as in Kelly’s and Cregeen’s dictionaries. Although both are based predominantly on English models with little or no influence from the Gaelic orthography used in Ireland and Scotland, these two systems appear 17 Although elements of the Classical Manx system are found already in Woods’ sermon manuscript from 1696 (Lewin 2015b). 58 Aspects of the historical phonology of Manx to be independent of one another to a large extent (although see below §1.6.3). One transitional text has recently come to light (Lewin 2015b) which appears to deliberately incorporate elements of Phillips’ system into an early version of the later orthography. In addition, through the eighteenth and nineteenth centuries, nonstandard orthographies diverging to differing extents from the standard of the Bible are used in manuscripts of carvals, sermons, folk-songs and the writings of Edward Faragher (1831–1908) (Broderick 1981b, 1982a). 1.6.2 Scholarly views on the Manx orthographies Celtic scholars have tended to take a censorious view of the Manx orthographies, seeing them only as cutting Manx off from its sister languages and hindering scholarship, as well as being ambiguous and inconsistent. The following views are typical: Phillips and his successors, indeed, removed the reproach that it [Manx] was an unwritten language; but in so doing they encumbered it with an orthography which was hardly more fitted to represent its sounds than the orthography of Early Modern Irish would have been. (O’Rahilly: 120–1) Manx orthography is an English monstrosity which obscures both pronunciation and etymology. (Jackson: 108) Despite fundamental deficiencies and diverse inconsistencies, the result may have served the purposes for which it was devised. From a philological viewpoint, however, it had the regrettable effect of imposing on Manx a wholly inappropriate spelling which obscured its historical relationship with its congeners and discouraged scholarly interest in its investigation. (Breatnach 1993: 2) Such views have also been predominant among activists and amateur scholars in the Manx revival movement (Ó hIfearnáin 2007; Lewin 2017a: 177–8; 2017b): for example, the lexicographer Douglas Fargher (1979: vi) describes the orthography as ‘a historical abomination’. 59 Aspects of the historical phonology of Manx Thomson (1984: 307) (see also §1.1) gives a more balanced assessment, weighing up both the representational deficiencies of the Manx orthographies compared with the Irish-Scottish system, as well as some advantages (cf. Russell 1995: 229; Broderick 2010: 306–7): The English conventions mean that the radical and lenited or nasalized consonants lack the visible connection shown in Gaelic spelling, but the spelling has the advantage for the linguistic historian of showing the vocalization of fricatives and such new developments as svarabhakti vowels, and lengthening or diphthonging in monosyllables before unlenited liquids and nasals when these are not shown in the traditional orthography. The system is rather weak on the indication of palatalization, though better in this respect than the similar nonstandard orthography of Scottish Gaelic, based on Middle Scots usage. The conventions of English and Manx orthography, have, however, grown apart, and it by no means follows that Manx pronunciation is immediately apparent to the English reader. The spelling, moreover, has developed an iconic element, in that words of similar or identical pronunciation are as far as possible deliberately spelt differently. (Thomson 1984: 307) For illustration of these strengths and weaknesses of the orthography, see §1.6.4. For sociohistorical and ideological aspects of the question of Manx orthography and related issues, see also discussion in Sebba (1998) and Ó hIfearnáin (2007) as well as Lewin (2017a; 2017b). 1.6.3 Phillips’ Prayer Book translation and orthography The earliest continuous prose text to survive in Manx is a translation of the Anglican Book of Common Prayer by John Phillips, a Welshman who served as Bishop of Sodor and Man from 1605 until his death in 1633 (Thomson 1969: 178). The translation has been dated to between 1604 and 1610 (Moore and Rhŷs 1895: xi; Thomson 1953: 3), and survives in a mostly complete manuscript from 1625–1630 (Thomson 1953: 4), now preserved in the Manx National Heritage Library bound in two volumes as MSS 3 and 4. The text never reached print for liturgical use, but was edited by Moore and Rhŷs (1895), and collated in a glossary by Thomson (1953; 1954–59), who also provides some corrections and additions to the Moore-Rhŷs edition. Wheeler (2019; 60 Aspects of the historical phonology of Manx 2020) has recently prepared new diplomatic and critical editions of Phillips’ psalter. The text is the main representative of ‘Early Manx’ (§1.3.1.1), although John Woods’ sermon of 1696 also belongs to the seventeenth century and shows linguistic and orthographic affinities with Phillips’ text (see below and Lewin 2015b), and some of the ballads and folksongs also date to this century or earlier, though preserved in eighteenth and nineteenth-century manuscripts (Thomson 1960–62; Broderick 1981a). Thomson (1953: 3, 6) provides evidence that Phillips’ liturgy is a translation of the 1604 version of the Prayer Book. The manuscript of the liturgy contains a large number of emendations dating to after 1662, including updates to the names of royal persons mentioned in prayers. I have shown (Lewin 2015b: 50–1) that many of these emendations are in the hand of John Woods (c. 1665–1739), who was from 1695 to 1700 chaplain of Castletown and master of the town’s grammar school, where he was himself earlier educated. It is likely that it is during this time that he came into contact with the Phillips manuscript and that he had enough esteem for the orthography in it that he attempted to make use of it, albeit inconsistently, in his own sermon writing. Woods’ interest in Phillips’ orthography seems to have been the exception among the seventeenth-century clergy. Contemporary comment from 1610/11 suggests that it was poorly received by the clergy, since ‘it is spelled with vowells wherewith none of them are acquainted’ (Moore and Rhŷs 1895 I: xii). In 1663, Bishop Barrow commented that ‘there is nothing either written or printed in their language, which is peculiar to themselves; neither can they who speak it best write one to another in it, having no character or letter of it among them’ (Butler 1799: 305), while William Sacheverell in his 1702 Account of the Isle of Man claims that Phillips’ text ‘is scarce intelligible to the clergy themselves’ (Cumming 1859: 15). It has consequently been claimed that Phillips’ text had little impact on later writing in the language, and was largely independent of the later orthography (and any early form of that orthography in use in Phillips’ time, at least for writing personal and place names) (Thomson 1960: 116–8). Certainly there is no clear evidence that Phillips’ text was ever widely read or appreciated. Woods’ interpolations prove only his own personal interest in the text — perhaps he hoped to update it and prepare it for 61 Aspects of the historical phonology of Manx publication — and not that the manuscript prayer book was in active or widespread use after 1662. The fact that the surviving manuscript is a slightly later copy, rather than an original draft, shows that copies were made, but as this is still in Phillips’ lifetime, it may prove only his own enthusiasm for the project. According to Thomson (1953: 6), ‘[t]here seems no reason to doubt Phillips’ explicit statement that by 1610 he had in person translated the P[rayer] B[ook] into Manx; we know that he was competent to preach in the language, and was noted for doing so’, and certain orthographic and morphosyntactic features may point to a degree of Welsh influence (Thomson 1953: 7, 10; 1960: 118). On the other hand, Thomson points to some minor differences in the language of the psalter which may suggest the hand of a native Manx speaker in this part of the work. Some anomalies in the manuscript may be copying errors, which according to Thomson (1953: 12) may point to an English or possibly Welsh scribe who did not understand the text, such as <j> for initial <g> (on the basis that <g> before front vowels would be pronounced /dʒ/ in English), and occasional confusion of and <y> — a possible instance of influence from Welsh, where u (earlier /y/) and y (/ɨ/) have fallen together as /ɨ/ or /i/ (Morris-Jones 1913: 13; Jones 1982). On the other hand, some variation can be attributed either to contemporary phonological variation and change, or to confusion of similar symbols (Thomson 1953: 12; 1969: 181–2), e.g. the occurrence of both initial <kn> and <kr>, and this should be borne in mind when making arguments based on Phillips’ orthography. Thomson describes Phillips’ orthography as follows: The principal problem connected with Bishop Phillips’ Prayer Book is that of its orthography. It is plainly very unlike standard literary Manx […] While its consonantism is very similar to later Manx, the vowels are very different indeed, and appear to rest substantially on the “Continental” values, giving this older Manx in part a greater similarity to its related languages than the modern orthography does to the eye, at least. The two systems seem to be quite distinct […] instances of distinctively modern spellings in the P[rayer] B[ook] are extremely rare. Neither does the old orthography survive as a competitor to the new. In short, one might suppose that we have two different attempts to write down Manx, which largely coincide in consonantism, having taken the same model [i.e. English], but diverge in vocalism, having chosen different standards. (Thomson 1953: 8) 62 Aspects of the historical phonology of Manx It is this use of the ‘continental’ vowel qualities (i.e. roughly the values of the vowels in Latin, Spanish, Italian, Welsh, or Middle English before the Great Vowel Shift) which was probably most off-putting to the rest of the clergy, who may have been familiar only with English and Latin pronounced in the unreformed English fashion. Phillips, on the other hand, had his native Welsh and, in addition, probably a broader education: He had three or four different systems known to him, English, Welsh, Latin, probably Greek, perhaps some modern languages, and his obvious course was to frame a system for himself out of the material he had. This, it seems, is what he did. (Thomson 1953: 9) In addition to the vowels, there are some minor differences from the later system in consonant representation. For example, Phillips generally uses <k> in all positions, apart from <ck> word-finally, whereas the later orthography generally has the English distribution of <c>, <k> and <ck>; <g> represents both /ɡ/, /ɡʲ/, later <g>, and lenited /ɣ/, later <gh>, while <gh> is used for initial /x/, /ç/, for later <ch>. However, the most notable feature of Phillips’ orthography is its wide range of variation. An individual word may be spelled in a dozen different ways (including variation in diacritic placement), although most of the variation is minor and a clear pattern is often observable. As an example, the following fairly frequent items are taken from Thomson’s glossary (1953, 1954–59) of the text (Table 1), with discussion of the orthographic variants below. Note that the number of occurrences is given in brackets, and that Thomson presents lenited forms in their radical form (so several instances of klyesh etc. appear as ghlyesh in the manuscript). Some manuscript examples are given below. 63 Aspects of the historical phonology of Manx Table 1. Examples of variant spellings in Phillips’ MS (after Thomson 1953) bannaghey ‘bless’, G. beannaghadh bannaghey banaghy (2) banaghe (2) (impv.) banigi bannigi banni (fut. 1sg.) banníím (pret.) vani (part.) baníít (2) baniit banníít (8) cliwe ‘sword’, G. claidheamh (OIr. claideb) klêiu (2) kleiu (3) kleìu klêîu kléiu klieu (5) klyeiu (pl.) kleiuyn kleinyn (3) klyêinyn cleaysh ‘ear’, G. clua(i)s klyesh kluass kluash klúash (3) klúas (2) klúæsh (2) (pl.) klyasyn (4) klýasyn (2) klýassyn klúashyn Figure 1. Examples from Phillips’ MS (cf. Table 1)18 y ghlyesh ‘his ear’, Matthew 26:51, f. 84v y ghluass ‘his ear’, Mark 14:47, f. 96r rish y ghléiu ‘with the sword’, Acts 12:2, f. 223r. y ghlêiu ‘his sword’, Matthew 26:51, f. 84v y ghluash, y ghlúash ‘his ear’, Luke 22:50-51, f. 106r. (a) Note the ‘continental’ value of vowels, e.g. ban(n)igi for later bannee-jee (Ir. beannaígí) /banidʒi/. (b) Final /ә/ may be <ey>, <e> or <y>, usually only <ey> in the later orthography. (c) Note the variety of diacritics and their placement over the sequence <eiu>. (d) Note the occurrence of both <eiu> and <ieu> – such variation in the order of elements is frequent. (e) Note initial <kl> for later <cl>, and lenited <gh> for /x/, later <ch> (in manuscript, not shown in Thomson’s glossary entries). (f) /s/ and /ʃ/ are not consistently distinguished by Phillips in non-initial position (where <sh> = /s/ but <s>, <ss> = /s/ or /ʃ/) (Wheeler 2019: 4–5), perhaps reflecting the non-phonemic nature of the [s], [ʃ] contrast in Early Modern Welsh, as suggested 18 Reproduced with permission of Manx National Heritage Library (MNHR MS 3). 64 Aspects of the historical phonology of Manx by Thomson (1953: 10). In later Manx G. cluas ‘ear’ is found with a generalized dative form cluais, CM cleaysh, replacing the nominative (§3.9.5), but the ambiguity of Phillips’ orthography means that it is not clear which is intended in the Prayer Book. (g) As in the later orthography, and in Welsh, initial consonant mutation is shown by symbol substitution rather than a diacritic mark or letter as in the Irish-Scottish orthography, e.g. vani ‘blessed’, later vannee (G. bheannaigh or ḃeannaiġ). (h) In medial and final position there is much variation between single and double consonant symbols, e.g. banni- and bani-. In the later orthography double consonants are usual to mark preceding short vowels (bannee), as in English (e.g. ‘banner’). The use of single and double consonant symbols appears to have no relation to the historical fortis-lenis contrast, although this may well have been retained in the Early Manx period in this position (see Chapter 4), e.g. Ph. benneylt, beneylt etc., later bennalt, bentyn ‘touch, belong’ (ScG. beanailt), as opposed to Ph. banni, bani etc. (G. beannaigh). (i) The occurrence of both <ya> and <ua> (and minor variants thereof) to represent the G. diphthong ua(i) may reflect the fronting of /uә̯/ to /ɨә̯/ which was apparently in progress in this period (§3.6.1.2).19 This is a case where Phillips’ orthography can shed additional light on the earlier stage of a phonological development. It is also an example of where Phillips is more accurate than the later orthography. The representations <ua> and <ya>, and their variants, clearly show the diphthongal nature of Manx /uә̯/, /ɨә̯/ in contrast to monophthongal /әː/, G. ao(i), represented in Phillips as <y>, <yy> and similar. This is in contrast with the later orthography, where G. ua and ao are represented by the same sequences of symbols, although there is reason to 19 As noted by Thomson (1953: 12) there is some apparent confusion of and <y> which may represent copying errors; however, this is unlikely to be the case with representations of G. ua, given that both symbols are widespread in this use, both are expected, and the patterning of their frequency seems to agree with a plausible account of conditioning factors for the fronting (§3.6.1). Moreover, this variation between and <y> is hardly found in representations of G. ao, where sequences involving <y> (and , <e>) predominate. It is thus clear that the variation between <ua> and <ya> is for the most part reflective of the original text. 65 Aspects of the historical phonology of Manx believe they remained distinct, e.g. CM meayl ‘bald’ (G. maol) and eayn ‘lamb’ (G. uan) (§3.6.2). (j) A variety of diacritic marks are used, most frequently the acute and circumflex accents, and in orthographic sequences consisting of multiple vowel symbols, these may be placed over any of the symbols, or may sit above two symbols (or at least, this is how the manuscript has been interpreted by Moore and Rhŷs, and Thomson). In this thesis, these are represented for typographical reasons by placing an identical diacritic over each vowel, e.g. <êê> should be read as <ee> with one circumflex diacritic over both vowel symbols. Diacritics appear to be intended to mark length and/or stress, and perhaps the most prominent element of a diphthong; however, their use is not consistent and they have not been found to be crucial to any arguments in this thesis. Thomson reproduces them (by hand) in the typescript version of his glossary found in his 1953 thesis, but considers them unimportant enough to omit from the published version of the glossary (1954–59). They are reproduced here, following Thomson (1953), for completeness. In one important respect Phillips’ text did have a lasting influence on later Manx writing. It appears that the psalter in his Prayer Book was adapted for the eighteenth century Prayer Book and Bible translations: The Psalms were taken from the English Prayer Book version, which had not been superseded in 1662 by the A.V. text, and were incorporated unchanged into the O[ld] T[estament] when it appeared, so that the Manx Bible here has the same text and verse numbering as the Prayer Book. […] The translator of the Psalms must have had access to Bishop Phillips’ version; despite the modernisation of the spelling and the continual tinkering with the vocabulary the similarity of the two versions is too great to be coincidental and Phillips’ translation may be said substantially to have lived on in the Manx Prayer Book as long as the latter continued in use. (Thomson 1979: [ii]) Given that Phillips’ text survived, was known and was put to use in the later translation project, despite never meeting with general approbation among its intended audience, it might reasonably be expected that it would have some influence on the later Classical Manx orthography, even if the latter was substantially an independent creation. In fact, 66 Aspects of the historical phonology of Manx this appears to be the case. To begin with, the use of <y> for schwa is distinctively non-English, probably of Welsh origin (Thomson 1953: 11), and occurs in both systems. The use of <ey> to represent /ә/ in final and sometimes medial position in the CM orthography is also not a usual feature of English orthography, where it would be expected to represent final /i/ (e.g. ‘valley’), but is found in Phillips. Although not the most common representation for final /ә/ in the Prayer Book, <ey> was perhaps generalized in the later orthography to avoid confusion with the more frequent English final <y> = /i/, which is sometimes retained with this value in certain loanwords, such as torrity ‘authority’, and personal names. Another case of likely retention of an element of Phillips’ system is the spelling aynin forms of the preposition ‘in’ (G. ann). In the Classical Manx orthography <ay> usually represents long vowels /eː/ or /ɛː/, but ayn was pronounced /oːn/ in the south and /uːn/ in the north (HLSM II: 16). In Phillips <y> often occurs as the final element of sequences of vowel symbols, and could apparently serve as a mark of length, a feature of Northern Middle English and Older Scots orthography (Vikar 1922; Kniezsa 1997). The spelling ayn makes more sense in the Phillips orthography if we assume than ann was still pronounced /aN/ at this period (cf. Ph. kian ‘head’, CM kione, G. ceann), with some degree of conditioned lengthening of the vowel (§4.6.1). The spelling ayn- may have been retained in the later orthography to provide a common representation for the variety of positional and dialectal realizations of the morpheme [oːn, uːn, on, un, әn, o, u, ә]. The use of <ay> in ayrn ‘part’ (G. earrann) and tayrn ‘pull’ (G. tarraing), and in forms of ec ‘at’ (G. ag), aym (1sg., G. agam), ayd (2sg. G. agad) (LM /em/, /ed/, perhaps earlier */aːm/, */aːd/?) may also be instances of retention of this orthographical feature from Phillips. The use of <y> and as length markers more generally in both Manx orthographies can be traced to northern English spelling conventions, although the use of in particular to mark following palatality is a Manx innovation; the resemblance to Irish-Scottish spelling conventions may be coincidental. A final example is the CM spelling dty for the 2sg. possessive G. do, d’, t’. This appears in Phillips as tdhy (6 instances), thdy, tdy (2), ta, t’ (before vowels, 5 instances) as well 67 Aspects of the historical phonology of Manx as dy (5) and thy (4). As discussed in Lewin (2015b: 72) in relation to Woods’ consistent use of tdy in his sermon, the idiosyncratic spelling of the initial consonant of this possessive is probably prompted by the existence of two allomorphs still found in other Gaelic dialects, namely /dә/ before a following consonant and /t/ prefixed to a following vowel. Despite the reversal of the order of the consonant symbols, this orthographic device seems too peculiar for there not to be a link between the similar usage in the two systems. The spelling dty was likely brought into the CM system with the above motivation, and also to distinguish this item from the many other functors dy /dә/.20 These survivals of usage from Phillips’ orthography can all be understood as serving particular representational needs, where no English convention could easily be adapted. As with Woods’ much wider adoption of elements of the earlier orthography (Lewin 2015b: 51, 53), there is no sense, however, of any interest in adopting broader principles of Phillips’ system, such as the ‘continental’ vowel values, or the representation of G. ua(i) and ao(i). Evidently those who developed the Classical Manx system were on the whole content to use English-based conventions familiar to them, even when these are more cumbersome or ambiguous than Phillips’ usage. 1.6.4 The Classical Manx orthography and variants 1.6.4.1 Characteristics of the Classical Manx orthography Some of the characteristics of the Classical Manx orthography will now be illustrated, in order to help the reader appreciate the challenges encountered in interpreting the orthographic evidence in the rest of the thesis. 20 Whose number had in fact increased since the Early Manx period owing to phonological developments (Jackson: 92; HLSM III: 91–2; Lewin 2016a: 174). 68 Aspects of the historical phonology of Manx 1.6.4.2 Homophones differentiated by spelling Homophones are often deliberately differentiated by spelling (Thomson 1984: 307; Thomson and Pilgrim 1988: 4): there can be no guarantee that words of different meanings but sounding the same will be written similarly; in fact, there seems to have been a policy to ensure that they were not! (Thomson and Pilgrim 1988: 4) This is understandable when it is considered that Manx literacy was for the purposes of fluent (usually native) speakers, people who already knew how the language was pronounced; distinguishing clearly between similar or homophonous lexical items would thus be more important than accurate phonological representation. Examples of such homophones, following Thomson and Pilgrim, are as follows: olley ‘wool’ (gen.) (G. olla) ollay ‘swan’ (G. eala) both /olә/ leigh ‘law’ (ScG. lagh) leih ‘forgive’ (G. loghadh) both /lәi̯ / Even etymologically identical items may have different spellings to signify different senses: lieh ‘half’ (G. leath, leith) er-lheh ‘apart, special(ly)’ (G. ar leith) /(erʲ) ˈlʲeː/ feanish ‘witness, evidence’ (G. fiadhnaise) fenish, fênish ‘presence’ /feːnәʃ/ marish ‘with’(ScG. maille ri) mârish ‘with him’ /mɛːrʲәʃ/ 1.6.4.3 One sound, several spellings There is thus a considerable amount of variation in the way a particular phone or sequence may be represented, especially in the vowels and diphthongs. The following 69 Aspects of the historical phonology of Manx are a selection of examples of orthographic representations for certain phones, some of which also further illustrate homophony with distinct spelling: /eː/ <ea> <ay> <ey> <eh> <ai> <aiy> <eai> <e> <e_e> /kreː/ /kreː/ /ʃeː/ /ʃeː/ /neːm/ /feːr/ /feːlʲә/ /feːnәʃ/ /feːm/ crea ‘creed’ (G. créadha, cré) cray ‘clay’ (G. cré) shey ‘six’ (G. sé) sheh ‘hide’ (G. seiche) naim ‘uncle’ (Eng. dialect ‘eme’) faiyr ‘grass’ (G. féar) feailley ‘festival’ (G. féile) fenish ‘presence’ (G. fiadhnaise) feme ‘need’ (G. feidhm) /oː/ <oa> <oy> <oi> <oe> <o>, <ô> <o_e> /noː/ /oːr/ /toːnʲ/ /oː/ /oːnә/ /kʲoːn/ noa ‘new’ (G. nuadh, nódh) oyr ‘reason’ (G. adhbhar) thoin ‘bottom’ (G. tóin) oe ‘grandson’ (G. ó) oney, ôney ‘innocent’ (G. ónna) kione ‘head’ (G. ceann) /ɛːi̯ / <aie> <aaie> <aih> <aigh> /trɛːi̯ / /fɛːi̯ / /ɡrɛːi̯ / /ɛːi̯ / traie ‘shore’ (G. tráigh) faaie ‘home field’ (G. faithche) graih ‘love’ (G. grádh) aigh ‘luck’ (G. ádh) 1.6.4.4 One spelling, several sounds The same, or similar, orthographic sequences may also represent distinct sounds. <oi(e)> /o/ /oː/ /әi/, /iː/ /oːi/ /orʲ/ /noːdʲ/ /rәi/, /riː/ /boːi̯ rʲә/ oirr ‘edge’ (G. oir) noid ‘enemy’ (G. námhaid) roie ‘run’ (G. rith) boirey, boïrey (G. buaidhreadh) <ay> /e/ /eː/ /ɛː/ /aː/ /oː/, /uː/ /em/ /kʲeː/ /slɛːnʲtʲ/ /aːrn/ /oːn/, /uːn/ aym ‘at me’ (G. agam) (§1.6.3) kay ‘mist’ (G. ceó, dative ciaigh) slaynt ‘health’ (G. sláinte) ayrn ‘part’ (G. earrann) ayn ‘in him, in’ (G. ann) (§1.6.3) 70 Aspects of the historical phonology of Manx <eay> /әː/ /ɨә̯/, /ɨː/ /iә̯/, /iː/ /eː/ /blәːst/ bleayst ‘husk, egg-shell’ (G. blaosc) /kɨә̯n/, LM /kɨːn/ keayn ‘sea’ (G. cuan) /friә̯l/, LM /friːl/ freayll ‘keep’ (G. friotháladh) /lʲeːrʲ/ leayr ‘clear’ (G. léir) <ea> /eː/ /ɛː/ /ɨә̯/, /ɨː/ /o/, /e/ /e/ /reː/ /ɡɛːrʲә/ /lɨә̯x/, /lɨːx/ /folax/ /edax/ rea ‘ram’ (G. reithe) gearey ‘smile, laugh’ (G. gáire) leagh ‘reward’ (G. luach) feallagh ‘ones’ (G. ?eallach) eaddagh ‘clothes’ (G. éadach) Note that diacritics (diaeresis and circumflex) are used in some texts, notably the Bible, to make some distinctions, albeit not entirely consistently, for example roie ‘run’ /rәi/, /riː/ v. roïe ‘before’ /rõːĩ/. However, they are absent in other texts, and do not play a large role in the orthography. 1.6.4.5 English v. ‘continental’ vowel values Long vowel representations usually have their English value, i.e. <ee> = /iː/, <oo> = /uː/, <i_e>, <ie>, = /ai̯ /, but may also have a ‘continental’ value, as in feme /feːm/ ‘need’ (G. feidhm) above (presumably deriving from a more conservative pronunciation of English). This gives rise, for example, to the potentially confusing pairs such as the following: mian /miә̯n/ ‘desire’ (G. miann, Ph. mian, mían, miæn) Mian /mai.an/ ‘Matthew’ (G. Maitheán, Ph. Mein) kere /kʲeːrʲ/ ‘wax’ (G. céir, Ph. kéeir, kǽyr, kǽir) kere /kʲiːr/ ‘comb’ (G. cíor, Ph. kiyr)21 The ambiguity is perhaps deliberately exploited in the spelling hene ‘self’ (G. féin), which the evidence of rhyme shows can be realized either /heːn/ or /hiːn/ (cf. the Scottish variants fhéin and fhìn) (Thomson 1995: 116; Lewin 2015b: 74), with the /iː/ pronunciation apparently becoming predominant in Late Manx (HLSM II: 220). 21 These items may have been semantically associated through kere-volley ‘honeycomb’ (G. cíor mheala). 71 Aspects of the historical phonology of Manx 1.6.4.6 Representation of palatalization Slender consonants for which a similar phone is found in English generally have a specific representation in the Manx orthography corresponding closely to the English convention, i.e. <sh> /ʃ/, <ch>22 /tʃ/ (G. /tʲ/), <j> /dʒ/ (G. /dʲ/), <y> /j/ (G. /ɣʲ/) (also <ghi>). Otherwise palatalization is most commonly indicated by the placing of before or after the consonant symbol. #Cʲ /nʲart/ /lʲabi/ /kʲuːnʲ/ /mә çoːn/ /ɡʲat/ ‘strength’ (G. neart) ‘bed’ (G. leaba, dat. leabaidh) ‘calm’ (G. ciúin) ‘my head’ (G. mo cheann) ‘gate’ (G. geata) VCʲV s’taittyn, -in troiddey bainney theinniu ooilley quallian, quaillan cuirrey erriu muickey s’buiggey /s tatʲәnʲ/ /trodʲә/ /banʲә/ /tenʲu/ /ulʲә/, /ulʲu/ /kwalʲan/ ‘pleases’ (G. taitin) (s’taittyn lhiam ‘I like’) ‘chide’ (G. troid) ‘milk’ (G. bainne) ‘thaw’ (ScG. taineamh) ‘all’ (G. uile) ‘whelp’ (G. coileán) /kurʲә/ /erʲu/ /mukʲә/ /s buɡʲә/ ‘invite’ (G. cuireadh) ‘on you’ (G. oirbh) ‘pig’ (gen.) (G. muice) ‘softer, softest’ (G. is buige) Cʲ# /patʲ/ /kredʲ/ /taˈluːnʲ/ /suːlʲ/ /uːrʲ/ /uːɡʲ/ ‘plague’ (ScG. pait) ‘believe’ (G. creid) ‘earth, land’ (gen.) (G. talmhain) ‘eye’ (G. súil) ‘earth’ (G. úir) ‘cave’ (ScG. ùig) 22 niart lhiabbee kiune23 my chione giat paitt creid thallooin sooill ooir ooig Which is ambiguous with the use of <ch> to represent /x/ in initial position, except that the latter can usually be recognized as a mutation of initial /k/, /kʲ/. This ambiguity is resolved by the use of <çh> for /tʃ/ in the 1866 edition of Kelly’s dictionary and in some subsequent revivalist publications. 23 Note here the lack of clear marking of the final slender /nʲ/, attested as k′uːn′ (EK) (HLSM II: 255). 72 Aspects of the historical phonology of Manx may also redundantly24 occur after orthographic units which already show palatality: but but but shiaght shassoo chiamble chaghter /ʃaxt/ /ʃasu/ /tʃambәl/ /tʃaxtәrʲ/ ‘seven’ (G. seacht) ‘stand’ (G. seasamh) ‘temple’ (G. teampall) ‘messenger’ (G. teachtaire) jiooldey yiooldey jalloo yalloo /dʒuːldә/ /juːldә/ /dʒalu/ /jalu/ ‘refuse’ (G. diúltadh) ‘refuse’ ‘picture’ (G. dealbh) ‘picture’ In clusters, adjacency to an orthographically marked slender consonant can be taken to indicate that the other consonant is also palatalized:25 ushtey ashlish aigney /uʃtʲә/ /aʃlʲәʃ/ /aɡʲnʲә/ ‘water’ (G. uisce) ‘dream’ (G. aisling) ‘mind, will’ (G. aigne) can sometimes indicate a slender consonant followed by /u(ː)/, apparently based on its use in English items such as ‘lute’, ‘acute’, ‘mute’ etc. lhune kuse /lʲuːn/ /kʲuːs/ ‘beer’ (G. lionn) ‘a few’ (G. ciumhas) The following are minimal pairs contrasting by palatalization, distinguished orthographically by : att aitt /at/ /atʲ/ ‘swell’ (G. at) ‘funny’ (G. ait) 24 However the presence of here may be motivated by the fact that the lenited form of these items, spelt with initial <hi>, may have initial /hj/ or /ç/. The may perhaps represent a glide after the radical initial consonant of the kind encountered in Scottish dialects (e.g. Ternes 2006: 28–38), although such glides are not reported in Late Manx. Rhŷs (73–5) notes lexical and dialectal variation between /ç/ and /h/ as lenition of initial /ʃ/ and /tʃ/, as well as hypercorrection from /has/ hass ‘stood’ (G. sheas) (never */ças/, according to Rhŷs) to unlenited ‘sassoo’ /sasu/ (G. seasamh). 25 Note that some clusters can show variation, e.g. [slʲ] and [ʃlʲ] as in slieau ‘mountain’ (G. sliabh) (Rhŷs: 157–8, HLSM III: 118). 73 Aspects of the historical phonology of Manx meeley meeilley /miːlә/ /miːlʲә/ ‘soft’ (G. míonla, míolla) ‘mile’ (G. míle) dooney dooin /duːnә/ /duːnʲ/ ‘close’ (vn.) (G. dúnadh) ‘close’ (stem) (G. dúin) shooyl shooill /ʃuːl/ /ʃuːlʲ/ ‘walk’ (vn.) (G. siubhal) ‘walk’ (stem) (G. siubhail) cabbyl cabbil /kabәl/ /kabәlʲ/ ‘horse’ (G. capall) ‘horses’ (G. capaill) Certain representations are ambiguous, since elements of an orthographic sequence can indicate either palatalization, or vowel length / quality, or both, as in: fainey faitagh bainney daaney baney /fɛːnʲә/ /fatʲax/ /banʲә/ /dɛːnә/ /bɛːnә/ ‘ring’ (G. fáinne) ‘shy’ (G. faiteach) ‘milk’ (G. bainne) ‘bold’ (G. dána) ‘white’ (pl.) (G. bána) In fainey the long vowel length is shown by the single <n> following <ai>, and the can be taken as indicating slender /nʲ/ also. In bainney shows palatalization and the double <nn> indicates a preceding short vowel. Fainey contrasts with baney, where the absence of indicates a broad /n/, and the single <n> indicates a preceding long vowel. The spelling daaney is clearer, with two signals of a long vowel, the digraph <aa> and the single <n>. Faitagh, however, is not immediately clear; a knowledge of G. faiteach, or reference to transcriptions of native speech (HLSM II: 156),26 is necessary in order to be confident of the vowel length. Cregeen’s alternative spelling fashagh, showing medial voicing and fricativization of /tʲ/ > [dʲ] > [ʒ],27 would also help here, although it would be misleading taken on its own (since <sh> usually 26 Even this evidence is not unambiguous, since we have to reckon with the tendency in Late Manx to lengthen certain short vowels (§5.5.2). There is a short vowel in fɑt′ʃɑ ̣x from three speakers (TC, JW and HK), as well as in the abstract noun faitys ‘shyness’ fɑt′ʃəs (TC), but a long vowel from one speaker (TT): fɛːt′ɑx (HLSM II: 156). 27 With <sh> being the nearest available orthographic representation for [ʒ]. 74 Aspects of the historical phonology of Manx indicates underlying /ʃ/). In many cases there is no clear indication as to whether a consonant is broad or slender, as the following cases illustrate: enney genney glenney /enʲә/, /enә/ /ɡenʲә/ /ɡlenә/ s’glenney meinney gien gennal /s ɡlenʲә/ /menʲә/ /ɡʲen/ /ɡʲenal/ ‘recognition’ (G. aithne) ‘scarcity’ (G. gainne) ‘clean’ (vn.), ‘clean’ (adj. pl.) (G. glanadh, glana) ‘cleaner, cleanest’ (G. is glaine) ‘meal’ (gen.) (G. mine) ‘cheer’ (G. gean) ‘cheerful’ (G. geanamhail) There is no indication that enney,28 s’glenney and genney have /nʲ/, while glenney and gennal have /n/, whereas in meinney palatalization is marked by . Slender /ɡʲ/ is clearly shown by in gien, but not in its derivative gennal (g′e̜ nɑ ̣lː NM, HLSM II: 192), which has no and thus is not clearly distinguished from the broad /ɡ/ in genney (from goan, goaun ‘scarce’, G. gann). The slender consonant can be marked more clearly in the lenited form, since <yi> as well as <ghi> can represent lenition of /ɡʲ/, thus in the Bible we have yien as well as ghien, and yennal as well as ghennal (cf. Thomson 1995: 133). In unstressed syllables may indicate /CʲәCʲ/, /CʲәC/ or /CәCʲ/, and even /CәC/ (§4.4.7.2). CʲәCʲ fakin cuishlin /fakʲәnʲ/ /kuʃlʲәnʲ/ ‘see’ (G. faicsin, ScG. faicinn) ‘vein’ (G. cuisle, dat. cuislinn) CәCʲ shiaghtin Mannin /ʃaxtәnʲ/ /manәnʲ/ ‘week’ (G. seachtmhain) ‘Isle of Man’ (G. Manainn) CʲәC cadjin claig(g)in mwyllin /kadʲәn/ /klaɡʲәn/ /mulʲәn/ ‘common’ (G. coitcheann) ‘scalp’ (G. cloigeann) ‘mill’ (G. muileann) 28 Depalatalization may have occurred in this word (§4.4.3). 75 Aspects of the historical phonology of Manx CәC sheeabin Manninagh eddin ynrican, ynrycan /ʃiә̯bәn/ /manәnax/ /edәn/ /inrәkan/ ‘soap’ (ScG. siabann) ‘Manx’ (G. Manannach) ‘face’ (G. éadan) ‘only’ (G. aonracán) 1.6.4.7 Redundant symbols The Classical Manx orthography is replete with letters which are redundant or largely so. As Lewin and Wheeler (2017) observe, this tendency seems to increase with the standardization of the orthography during the eighteenth century: Where they differ, Coyrle Sodjey [the first printed book in Manx, 1707] is usually simpler, with fewer superfluous letters [than later texts]: thus initial lhis rarely found, so we have leid (lheid), liasagh (lhiasagh), liat (lhiat), lie (lhie), liettal (lhiettal), liggey (lhiggey); -eea- implies two syllables, so we have Creesteeaght, gimmeeaght, etc., but jeaghyn (jeeaghyn), gearree ~ gearee (geearree); other ‘simpler’ spellings are: baas (baase), callit (caillit), cheel (cheeill), coal (coayl), deartey (deayrtey), deney (deiney), feasley (feaysley), foar (foayr), foas (foays), freall (freayll), geashtagh (geaishtagh), gol (goll), janoo (jannoo), loart (loayrt), meigh (meiygh), raadjin (raaidyn), reyn (rheynn), seihl (seihll), talloo ~ tallow (thalloo), treshteil (treishteil); tregeil ~ treggeil is much more frequent than the modern treigeil, though this is also found. (Lewin and Wheeler 2017: preface) In this list we can see that some of the later additions add clarity, so keeill ‘church’ (G. cill) indicates the final slender /lʲ/ (as opposed to ambiguous keel, and so with callit and caillit, G. caillte). The later spelling <eea> in geearree (G. ag iarraidh) and jeeaghyn (G. déachain)29 shows /iә̯/ (§2.2.6) more clearly than <ea>, which more usually represents /eː/ (as well as /ɛː/, /әː/, /ɨә̯/). Similarly, <eay> in feaysley (‘release’, G. fuascladh) and deayrtey (‘pour’, G. dórtadh, duartan) more clearly indicates /ɨә̯/ than <ea> does (§3.6.2). However, it is unclear how rheynn30 (‘share, divide’, G. roinn) is better than reyn (neither indicate the final slender /Nʲ/, /nʲ/ or /ŋʲ/, §4.4.6), or why it is useful to add a silent <e> to baase /bɛːs/ ‘death’ (G. bás). The English final 29 Although jeaghyn could also indicate retention of monophthongal /eː/ here, as also indicated by spellings in Phillips, and in Woods’ sermon of 1696 (Lewin 2015b: 75) (§2.2.7). 30 For <rh> see §4.2.1.5. 76 Aspects of the historical phonology of Manx <e> is used as a marker of vowel length in Manx, e.g. bane /bɛːn/ ‘white’ (G. bán), but it is also very widespread where length of the preceding vowel is shown by other means, as in baase. Note that there is no danger of this being taken to mean final /ә/, which is always <ey> or very occasionally <ay>, <ah>; see especially coyrle and Baarle in the list below, which have loss of final schwa (§5.2). Other examples of superfluous final <e>: baare sheese coyrle Baarle jymmoose /bɛːr/ /ʃiːs/, /siːs/ /kõːrlʲ/ /beːrl/ /dʒiˈmuːs/ ‘top’ (G. barr) ‘down’ (G. síos) ‘advice’ (G. comhairle) ‘English language’ (G. Béarla) ‘wrath’ (G. díomdha + as) 1.6.5 Late spellings (especially Cregeen) While the biblical orthography became a standard which was followed, to a greater or lesser extent, by most subsequent writers, later spellings of items not found in the eighteenth-century texts (often for secular or modern concepts, and everyday life) may diverge from the conventions of the Biblical orthography, sometimes showing later phonological developments (Thomson 1999: 402). This is notable in Cregeen’s dictionary (1835), for example, where there are forms showing the medial lenition of /s/ to [ð] and /dʲ/ to [ʒ], which is never indicated in the Biblical orthography:31 gadyree, -ey shuddyr Breeshey fashagh /ɡasәrә/ /ʃisәr/ /briːdʲә/ /fatʲax/ ‘heat’ (of bitches) (ScG. gasradh) ‘scissors’ (G. siosúr, ScG. siosar) ‘Bridget’ (ScG. Brìghde) ‘shy’ (G. faiteach) In the Biblical orthography the diphthong /iә̯/ (G. ia, é) is consistently represented as <eea>, <ia>, and kept distinct from the monophthong /iː/ <ee>, <eey>, <eei>. Similarly /uә̯/ (G. ua), where it retains its back quality, is consistently written <ooa>, 31 Apart perhaps from luddan-mea (Job 41:32) ‘phosphorescence on the surface of the sea’, if this is the same as lossan ‘luminous particles seen in the sea by night, and on fish that are not dry, in the dark; the aurora borealis or northern lights’ (Cregeen) (G. losán). Cregeen and Kelly both have separate entries for lossan and luddan(-mea). 77 Aspects of the historical phonology of Manx <ua>, contrasting with monophthongal /uː/ <oo>, <ooy>, <ooi>, <ooh>, <u_e>. In Cregeen, however, there are spellings which apparently represent the changes /iә̯/ > [iː] and /uә̯/ > [uː] (§2.2.6): keeir lheegh hooir /kʲiә̯r/ /lʲiә̯x/ /huә̯r/ ‘dark’ (G. ciar) (K. keear) ‘ladle’ (G. liach) ‘forebode, threatened’ (G. tuar) The examples keeir and hooir also illustrates another feature of Cregeen’s usage which may indicate a sound change; the superfluous before a historically broad /r/ might also be taken as evidence of loss of the contrast /r ~ rʲ/ (§4.2.1). However, Cregeen seems not to have fully understood the use of in the Biblical orthography, and thus inserts it in many items adjacent to broad consonants: bwoid boiddagh brooightooil jooigh mooin /bod/, /bud/ /bodax/ /bru(ː)xˈtuːlʲ/ /dʒuːx/ /muːn/ ‘penis’ (G. bod) ‘stingy person, churl’ (G. bodach)32 ‘belch’ (G. brúchtghail) ‘greedy’ (G. díbheach) ‘urinate’ (G. mún) Sometimes Cregeen introduces such spellings instead of, or alongside, the Biblical form: ‘broigh, or broghe’ /broːx/ ‘dirty’ (G. broghach) (Bible broghe) ‘hioll, or hoyll’, thiolley /tolә/, /hol/ ‘bore, pierce’ (G. tolladh) (Bible hoylley)33 There are however, some such forms in the Bible itself such as druight ‘dew’ (G. drúcht), seyir ‘carpenter’.34 Metathesis of digraphs involving is quite common, 32 This spelling is perhaps intended to distinguish this sense from the homophonous (and probably etymologically identical) boadagh ‘cod’ (cf. Dwelly s.v. bodach). 33 Thomson (1995: 120–1) notes in his edition of PC that ‘hiolley […] is Cregeen’s spelling, but […] the source of the palatalisation is obscure’. It is much more likely that there is no palatalization (which in any case would be expected to be represented as <ch>), and that this is simply an instance of Cregeen’s tendency to introduce redundant , perhaps in this case following PC 1796 which has hioalley. In the Bible the spelling is hoylley (it so happens that only lenited forms are found). It is possible that the spelling is influenced by the lenited forms of shiaulley ‘sail’, shiolteyr ‘sailor’, and perhaps also by the variation in some items between /h/ and /ç/ as the lenition of radical /ʃ/ and /tʃ/ (Rhŷs: 74–5). 34 Both singular and plural in the Bible, where one might expect singular seyr and plural seyir. The motivation may be to distinguish this sense from seyr ‘free’, and dooinney-seyr ‘gentleman’ (G. saor). 78 Aspects of the historical phonology of Manx especially in manuscripts and sometimes in print, for example <ia> and <ai>, <io> and <oi> (Broderick 1982a: 180; 1984: 166). This usually seems to involve modification of sequences unknown or rare in English to those found more commonly in English. 1.6.6 Interpreting the Manx orthographies: summary of difficulties As observed above, native Manx speakers would of course have known how to pronounce the language, so the inconsistencies and ambiguities of the orthography would not have caused major problems (although see §1.6.9.1 for spelling pronunciations among the terminal speakers). For the scholar of Manx with a knowledge of other Gaelic varieties, they do not usually cause much trouble either, since the etymology of many items is readily apparent, as in the items presented as illustration above. In any case, the standard Manx orthography is in many respects considerably more systematic than its critics have acknowledged, as discussion throughout this thesis will show. Despite incorporating redundant features of English orthography such as silent final <e>, those who devised the Manx orthography succeeded in making considerable innovations to represent phones not found in English, including the complex system of short and long diphthongs (§§1.5.7, 3.9.1), vowels such as /ɛː/, /әː/ and /ɨә̯/, /ɨː/, and (not always so successfully) palatalized consonants. Problems arise, however, when dealing with ambiguous spellings of items (a) where the etymology is less clear, or entirely obscure, (b) where evidence in the form of recordings, phonetic transcriptions or descriptions and variant spellings is unavailable or inconclusive, (c) where variant forms in the other Gaelic languages mean that it is not possible to determine with certainty which variant the Manx form represents, or (d) where it is suspected that the Manx form may be an irregular development. In such cases, it may not be possible to reach a firm conclusion. 79 Aspects of the historical phonology of Manx 1.6.7 Non-standard orthographies In addition to the two main orthographies discussed above, various degrees of departure from the biblical standard are found in eighteenth and nineteenth century texts, both printed and especially manuscripts. Some of these diverge very substantially from the standard. The manuscript corpus is extensive, consisting of carvals (religious ballads), folksong manuscripts (e.g. Broderick 2015, see §4.5.3), Bible translation drafts, sermons, and Edward Faragher’s writings. Faragher at least appears to have been an avid reader of the Manx Bible, but nevertheless seemingly had little interest in conforming to its orthography in his own usage (Broderick 1982a: 178–9). Much of this corpus has yet to be subjected to in-depth scholarly analysis, although as noted by Thomson (1960: 116–7) this would no doubt be rewarding: For linguistic purposes these carvals, especially in their manuscript form, are probably more important than anything else, for the books were written and copied by native speakers with no very accurate recollection of the standard spelling […], and many points of phonological interest are illustrated in their free spellings and in the rhymes. (Thomson 1960: 116–7) In this thesis details of non-standard manuscript spellings will be discussed where relevant details have come to my attention, but a full consideration of the information contained in these texts awaits a future treatment. 1.6.8 Dictionaries Manx scholarship is fortunate in having two principal Manx–English dictionaries compiled by native speakers during the period when Manx remained widely spoken, those of John Kelly (1750–1809), compiled towards the end of the eighteenth century, and Archibald Cregeen (1774–1841), a generation later in the 1830s. Although both works have deficiencies and omissions, they nevertheless provide us with a more complete knowledge of the language’s lexis than would be the case from the corpus of 80 Aspects of the historical phonology of Manx texts alone (cf. Thomson 1990: 444), and are therefore crucial sources in establishing the distribution of phonological developments across the lexicon. 1.6.8.1 Kelly (1866) John Kelly, who as a young man assisted in copying, editing and proof-reading the Manx Bible (Thomson 1969: 185–6), began his grammar and two dictionaries in 1766 ‘for the instruction of […] the Rev. Dr. Hildesley, Lord Bishop of Sodor and Mann; and […] to assist and direct my fellow labourers and myself in that arduous and important work, the translation of the Manks Bible’ (Kelly 1804: iv). Early drafts of these works survive in Manx National Heritage Library MS 1477, with a later draft of the Manx–English dictionary in MS 1045–7A. The grammar was printed in 1804, but the Manx–English dictionary did not appear until it was published posthumously in 1869, edited by William Gill. Kelly also authored a ‘triglott’ dictionary of English to Manx, Scottish Gaelic and Irish (Thomson 1969: 205–6). Printing of this began in 1809 but came to nothing in a fire in the printing shop. The manuscripts survive (Manx National Heritage Library MSS 2045, 51), and the English and Manx columns of this were utilised by the Manx Society to form the English–Manx half of the 1866 publication (Wheeler 2020). Thomson (1990: 447) estimates that ‘in round figures Cregeen presents the reader with a little more than 6000 words; Kelly has most though not quite all of these and adds rather more than another 4000, making a total of about 10,500 in all’. The reasons for the greater number of entries in Kelly, according to Thomson, are that he gives a large number of derivatives, mostly ones which are logically possible but which may not have been in use. Furthermore, many items are taken from previous Gaelic dictionaries, often showing an incomplete understanding of Gaelic orthography; and ‘[t]here are a few cases of etymological guesswork giving rise to supposedly independent words’ (Thomson 1990: 450). 81 Aspects of the historical phonology of Manx Of the latter two categories, many are easy to spot. For example Kelly has items rane ‘stanza’ and raneyder ‘poet’, allegedly from G. rann35 but also linked by him with arrane ‘song’ (G. amhrán); G. ruaig is claimed to have given rueg, but the word is not attested elsewhere, the spelling <ue> is unusual, and G. ua more commonly gives a fronted vowel in Manx (which, as Thomson suggests, might be spelled *reayg or *reaig);36 taishbyn ‘reveal’ is shown to be a borrowing by the retention of from Shaw’s taisbeun and the lack of -ey for G. -adh; G. ughdar ‘author’ is given by Kelly as ughtar, which in Manx orthography would be interpreted as */uxtәr/, whereas this word, if it existed in Manx, would be expected to give *oodyr /uːdәr/. Other probable borrowings have more plausible orthographic forms, such as doghys ‘hope’ and dooill ‘desire’, corresponding to G. dóchas and dúil respectively (Thomson 1990: 452), and their status can only be surmised by their absence from the corpus, and/or by similarities with definitions and cognates in Shaw. In some cases it may not be possible to be entirely sure whether an item in question was in use in Manx or not. For this reason Kelly’s dictionaries have been approached with caution, and Cregeen has been taken as the basic source for lexical information. Data from Kelly is given where it backs up or complements Cregeen, or otherwise seems likely to be genuine. Despite the problems illustrated above, further research into Kelly’s work remains a desideratum, as it certainly contains much valuable material not found elsewhere. 1.6.8.2 Cregeen (1835) The title page of Cregeen’s dictionary states that it was printed in 1835, although it may not in fact have appeared until 1837 (Wheeler 2018: x). The dictionary was reprinted by Yn Cheshaght Ghailckagh three times in the twentieth century, and has more recently been edited by Wheeler (2018) in a digital format with the headwords reordered to group cognate, derivative and inflected forms, together with an G. rann would give */roːn/ or */rau̯ n/ in Manx (§4.6.1), not */rɛːn/; cf. attested ronney /ronә/ ‘portion’ (G. rannadh) and ronneeaght /roni.axt/ ‘foolish song, raving’ (G. rannaigheacht). 36 As Thomson also notes, however, part of the definition differs somewhat from Shaw, so we cannot exclude entirely the possibility that this is a genuine Manx word, even if the spelling and sense given by Kelly have been influenced by the external source. 35 82 Aspects of the historical phonology of Manx introduction and notes. Wheeler’s edition forms the basis for the use made of the dictionary in this thesis. Cregeen is overall much more reliable than Kelly as a source of linguistic data, his main weakness being his own spellings of items not found in the religious literature (§1.6.5), and the evident frequent inaccuracy of the stress placement which is marked on almost every polysyllabic headword (§5.1.1), as well as the sometimes obscure definitions (Wheeler 2018: iii–iv). Thomson (1990: 447) summarizes the virtues of Cregeen’s dictionary as follows: Cregeen has a reputation for being a reliable witness in linguistic matters. He collected his material from written sources and from the spoken language, and within the limits of his time and resources he can be described as a scientific lexicographer. If he offers a few popular etymologies he does not usually let them dictate his spelling, which is fully traditional even when, as he occasionally observes, he disagrees with it and thinks it could be improved. […] So on the whole Cregeen gives an impression of sobriety and reliability. (Thomson 1990: 447) 1.6.8.3 Revival era English–Manx dictionaries Finally, in the twentieth century two English–Manx dictionaries were published by prominent figures in the revival movement, Kneen (1938) and Fargher (1979). The weaknesses of Kneen’s pronunciation guidance have been discussed above (§1.5.5). Fargher’s is a self-consciously prescriptive work incorporating large numbers of unacknowledged neologisms, many of them borrowings from Irish and Scottish Gaelic, frequently adapted in a way which lacks philological rigour and consistency (see Lewin 2017b). It is also a fairly unselective compendium of the material in Cregeen and Kelly (including uncritical inclusion of the latter’s spurious borrowings and inventions of the type illustrated above §1.6.8.1). Fargher’s dictionary is thus of little use for scholarship of the traditional language, although it does incorporate some lexical items noted by the author from the last native speakers, rendered into Manx orthography. 83 Aspects of the historical phonology of Manx 1.6.8.4 Pitfalls in the use of Manx dictionaries Reliance on the dictionaries, in the absence of a detailed study of the primary sources, has sometimes led scholars astray. For example, Ó Baoill (1978: 281) reports a Manx form dooill ‘desire’ corresponding to G. dúil on the strength of Kelly, but this item is likely a borrowing as noted above. Similarly, Grannd (2000: 16) claims, referencing Kneen (1970: 66), that G. spéir ‘sky’ (ScG. speur) ‘seems to be the word used in Manx’ (speyr). However, this appears to be unknown outside dictionaries,37 the normal word being aer (also ‘air’) (Bible; HLSM II: 4), agreeing in fact with the most common term in Scotland according to Grannd (adhar). Ó Maolalaigh (2013: 65) quotes Fargher (1979: 287) that dy-chooilley is an ‘archaic spelling’ for dagh ooilley ‘every’ (G. gach uile, ’chuile etc.) when in fact the former is the standard Classical Manx spelling, and the latter an example of revivalist antiquarian spelling. Ó Maolalaigh discusses two examples from Fargher dagh ooilley cor hiaghtin and [er] dagh vod, the former of which appears to be Fargher’s own invention38 and the latter is an obscure form from Kelly.39 Such data (certainly the former) cannot be taken to have a bearing on linguistic features of the traditional language. These examples show that all of the dictionaries are to varying extents unreliable as reference works. Kelly and Cregeen, however, are highly important as primary sources. 37 The word speyr is in Cregeen, glossed ‘the sky’, so is likely to be genuine, but nevertheless does not seem to have been the most usual word for ‘sky’. It is absent in Kelly, who gives only aer and niau (‘heaven’, G. néamh) for ‘sky’ in his English–Manx dictionary. 38 Prefixed G. corr- ‘(the) odd, occasional’ does not seem to occur productively in Manx, although as an element meaning ‘rounded, pointed; remote’ it is found in place-names (PNIM VII: 427–8) and there is a derived adjective corragh ‘tottering, weak; touchy, capricious’ (G. corrach). Kelly gives derreylaa as ‘every other day’, cf. indara la ‘every other day’ (eDIL s.v. 1 dara), but this is not given by Fargher, who does however have gagh derrey Doonaght ‘every other Sunday’. *Gach uile c(h)orr- is not found in other Gaelic varieties for ‘every other…’. 39 In Kelly’s manuscript (MS 1045–57), but not the printed work (s.v. er-dagh-vod), the note ‘Ir. Ar gach mhead’ is given. 84 Aspects of the historical phonology of Manx 1.6.9 Native speech: recordings and transcriptions Details of the recordings and transcriptions made in the twentieth century of native speech by fieldworkers, both professional and amateur, are provided by Broderick (HLSM I: xv–xxiii, III: xi–xxxiii; 1999: 54–75; 2017; 2018b; 2018d). All of these data were collected from elderly informants, most of whom lived largely in isolation from other Manx speakers and had not used the language for many years, and who had grown up in communities already experiencing rapid language shift to English. For biographical information on the speakers see Broderick (2017; 2018a). In this thesis two main sources of native speech data are used: the transcriptions given in Broderick’s Handbook of Late Spoken Manx (1984–86), especially the dictionary (vol. 2), and the audio recordings made by the Irish Folklore Commission, published on CD in 2003 (Manx National Heritage). The latter were investigated instrumentally using Praat (Boersma and Weenink 2015), the first use of such a methodology with Manx audio data (§§2.2.3, 3.7, 4.5). Further recordings exist, notably those made by the Linguistic Survey of Scotland, and by the local language organization Yn Cheshaght Ghailckagh and other amateur language enthusiasts, with copies held in the Manx National Museum and the School of Scottish Studies. Some of these are transcribed by Broderick in HLSM, but it is desirable that the original audio material should be investigated instrumentally in future research. The transcriptions made by Marstrander of texts read aloud by terminal speakers also deserve further attention (see Thomson 1976; Broderick 2018d). There are also some very early recordings from the first decade of the twentieth century (Broderick 1999: 54; Trebitsch and Remmer 2003; Manx National Heritage 2017), which are occasionally referred to in the present work, although the material is limited, mostly read or recited, and the sound quality poor. 1.6.9.1 The Manx of the ‘last native speakers’ Although conventionally referred to without qualification as ‘the last native speakers’ of Manx (e.g. Broderick 2017), the individuals recorded in the twentieth century all 85 Aspects of the historical phonology of Manx show signs of varying degrees and types of incomplete proficiency in the language, at least at the point in time when they were recorded, including phonetic variability, gaps in lexis and idiom, and lack of control of complex morphosyntactic phenomena such as initial consonant mutation and grammatical gender. Broderick (1999: 77–158) catalogues these features extensively, but does not come to an altogether clear conclusion as to how the speakers should be classified (ibid.: 5–6). He ascribes the observed features primarily to rustiness and lack of practice in the language, using Menn’s (1989) term ‘rusty’ speaker, as well as to discomfort caused by ‘the presence of the microphone and recording machine’ rather than ‘any short-comings on their part’ (Broderick 1999: 6). In Broderick (2017: 54) this position is restated more forcefully and the same factors are given as explaining apparent linguistic weaknesses: I am of the opinion that all fifteen of our speakers are to be regarded as ‘full’ (i.e. ‘formerly fluent’) speakers of Manx. That is to say, they had gone through the gamut of the language during their formative years (their pronunciation is consistent with what is to be expected), but that there is clearly some loss to be seen is due, in my view, not to imperfect learning when young, but to lack of use in later life. (Broderick 2017: 54, original emphasis) However, I have argued (Lewin 2014b; 2017a: 191–3; 2019a: 81–2) that rustiness or language attrition during the lifetime of the speaker are insufficient to explain some of the features observed. The terminology is not entirely settled in this field (‘semispeaker’, ‘terminal speaker’, ‘younger fluent speaker’, ‘weaker speaker’, ‘reduced speaker’ or ‘post-traditional speaker’ etc.),40 but it is clear that incomplete acquisition of certain linguistic features by upcoming generations is a pervasive feature of language shift and minoritization situations (Dressler 1985: 12; Sasse 1992b: 62–63; Montrul 2008)41 — as also in other situations with suboptimal levels of linguistic exposure and peer-group socialization, such as the case of second-generation bilingual immigrants and heritage speakers (Carroll 1989; Polinsky 2006; Unsworth et al. 2014; 40 For discussion in the Manx context see Broderick (1999: 4–11; 2017) and Lewin (2017: 143), and for the development of the terminology in wider scholarship see e.g. Dorian (1977, 1981, 2010), Dressler (1981), Schmidt (1985), Campbell and Muntzel (1989), Sasse (1992a; 1992b), Grinevald and Bert (2011), Lenoach (2012: 21–25), Ó Curnáin (2012). 41 See for example Ó Curnáin (2007; 2012) for detailed data from a contemporary Irish-speaking community currently experiencing language shift. 86 Aspects of the historical phonology of Manx Montrul 2008; 2015). A strong argument can be made that incomplete acquisition is required to explain the absence or lack of control of features such as initial mutations and grammatical gender concord with inanimate nouns among the last speakers of Manx. The latter in particular is well maintained in written Manx of the seventeenth to nineteenth centuries, including in the stories and reminiscences of Edward Faragher (1831–1908), and is also found in some but not all of Rhŷs’s informants,42 but is absent in the speech recorded from terminal speakers such as Ned Maddrell.43 Although the difficulties of comparing speech and writing should always be borne in mind, I have argued (Lewin 2017a: 191–3) that it is implausible either that gender could have been maintained by the earlier generations in writing only — presumably by conscious learning — or that the complete loss of these features should have occurred owing to rustiness in otherwise ‘full’44 native speakers. It is much more plausible, and agrees better with our cross-linguistic understanding of the processes of language shift, that grammatical gender, initial consonant mutations etc. were normal parts of the linguistic competence of earlier generations of ‘full’ native speakers, albeit complex, opaque, late-acquired features particularly dependent on rich input and socialization in the language for complete acquisition (cf. Nic Fhlannchadha and Hickey 2017), and that an insufficiency of such input and peergroup socialization is responsible for the gaps in the competences of terminal speakers such as Maddrell, despite high levels of conversational fluency and confidence, and complete or near-complete acquisition of other components of the language. Indeed, it 42 Rhŷs notes maintenance of historical gender concord with inanimate feminine nouns in speakers from poorer socioeconomic backgrounds and more remote communities which remained strongly Manxspeaking longer than others, but there may be failure to observe gender concord in individuals from less isolated communities or higher social strata, that is, members of social networks within which language shift had occurred earlier and more extensively by the time of Rhŷs’s fieldwork (Lewin 2019a: 79–82). 43 It also appears that Maddrell, usually regarded as ‘the last native speaker’ of Manx, was a sequential bilingual who acquired English first and only later acquired Manx from an elderly relative, starting between the ages of two and a half and five (HLSM I: 463, 467–8; Broderick 1999: 75; 2017: 44–5; Lewin 2014b: 17–8). See e.g. Meisel (2009) and Unsworth et al. (2014) for major differences between simultaneous and sequential bilinguals, even when the age of onset of acquisition of the L2 is relatively young. 44 In Menn’s (1989) sense. 87 Aspects of the historical phonology of Manx would be much more remarkable if speakers from this stage in the language shift process did not display symptoms of incomplete acquisition. It is curious that Broderick (2017) does not examine the incomplete acquisition scenario further, nor reference the voluminous empirical literature on bilingualism and language acquisition under conditions of minoritization and language shift, and that he is so keen to foreclose the possibility that incomplete acquisition is part of the explanation for the linguistic features of ‘Late Spoken Manx’. Indeed, comments such as the following, which assume that a speaker will automatically have a competence in a language equivalent to that of the source of input, show a lack of consideration of cognitive and social aspects of acquisition in a situation of language shift, and indeed for intergenerational language change in general: Although English was, according to himself, Maddrell’s home language for the first five years of his life, he was then (c. 1882) allocated to live with a greataunt who apparently had little or no English […] because he was brought up with a great-aunt born in the first decade (1809) of the 19th century, his Manx, unlike that of his peers, would be of that vintage. (Broderick 2017: 45) This is not to say that the ‘vintage’ of the great-aunt’s Manx would make no difference to Maddrell’s acquisition. She was clearly Manx-dominant or monolingual, and herself acquired Manx when there was little English in the community (cf. Miller 2007), and so Maddrell may well have acquired features from her more successfully than he would have done if his own parents had spoken Manx in his presence, as they would likely have been weaker ‘post-traditional’45 speakers to some degree. As Broderick (2017: 45) observes, Maddrell ‘is the only one of the last fifteen speakers who makes use of the inflected synthetic tenses of the verb […] and distinguishes between the imperfect and conditional forms of the verb ‘be’.’46 It is to be expected in cases of reduced acquisition that some features may be acquired much more fully than others, 45 Cf. Ó Curnáin (2007: 59–60) for the decline in proficiency of Iorras Aithneach Irish speakers since 1960, with moderate effects initially in the first generation of ‘post-traditional speakers’ giving way to more severe impacts of ‘reduced’ acquisition in younger speakers. 46 For the latter feature, see Broderick 2011. 88 Aspects of the historical phonology of Manx and it is often unpredictable which features will be better acquired and by which speakers (cf. Dorian 2010: 107, 269). Broderick’s (2017) discussion of the biographical details and census data of the terminal speakers (while otherwise highly useful), also reveals the limits of his sociolinguistic analysis, notably when he rather mechanically dates the switch from Manx to English in a particular relationship or household from the introduction of a non-Manx-speaking member, or a presumed decision by parents not to speak Manx to a new child. While such junctures are indeed likely to represent termini post quem by which point use of Manx in a given situation had ceased, or at least significantly reduced, we cannot assume that Manx was used with any regularity between any given two bilingual Manx speakers at an earlier stage. Rather, the available information on the Manx situation, with rapid language shift accompanied by widespread stigmatization (Broderick 1999), and parallel examples from other language communities, would suggest that many younger bilinguals used English as their normal language of peer-group social interaction, even if Manx had been their home language during childhood and adolescence, and continued to be used (actively or passively) with older members of the community. For example, Maddrell’s parents apparently habitually spoke English together despite being able to speak Manx (HLSM I: 463; Broderick 2017: 43–45). Indeed, the already-established status of English as the normal language of interaction between a young couple may well be a significant part of the reason they failed to transmit the language to the next generation. 1.6.9.2 Features of Terminal Manx phonology In the phonology, features such as apparent confusion of palatalized and nonpalatalized consonants (Broderick 1999: 85–90), loss of coda /r/ (HLSM III: 113) and intrusive /r/ (HLSM II: 267; §4.2.3), lack of vowel nasalization (Jackson: 63–4; HLSM III: 147; §5.6) and perhaps increased variability in vowel realization and tendencies towards merger (Broderick 1999: 81–83; but see §2.2.3, 3.5.5.1, 3.7) are likely to reflect incomplete acquisition under conditions of language shift. In contrast, Rhŷs 89 Aspects of the historical phonology of Manx (31–48) provides evidence that vowel nasalization, for example, was widely maintained among his informants. There are also some spelling pronunciations, e.g. feoh ‘hate’ (G. fuath), expected /fɨә̯/, where <eo> is one of the orthographic representations of the fronted reflex of the G. diphthong ua (§3.6.2); this is attested as expected fiːə from TT, but with an evident spelling pronunciation fẹːo (TC, HK) and fjoː (JW) (HLSM II: 165), and aigh ‘luck’ (G. ádh), expected /ɛːi̯ /, realized as ɛːx (TC) with misinterpretation of orthographic <gh> as a representation of /x/. These attest to the persistence of a degree of literacy (in a religious context) during the period of language shift, as well as lexical contraction as less frequent or higherregister items were partially forgotten. Nevertheless, it is clear that the speech of the terminal speakers preserves a great deal of the patterns of the Manx of earlier generations, including features such as preocclusion which are scarcely attested in written sources (§4.5). If approached judiciously, with quantitative methods, and with an eye to the comparative data provided by earlier written material, the recordings of the terminal speakers remain a rich and important source of linguistic data. 1.7 Outline of synchronic phonology of Classical Manx Throughout the thesis, and especially in the tables of lexical data, phonological reconstructions are provided which aim to represent the likely pronunciation of the language in the middle of the eighteenth century (i.e. the spoken language which forms the basis of the standardized Bible orthography). The purpose of these reconstructions is primarily to assist readers in orientating themselves with regard to the complex and possibly unfamiliar orthographies and other sources of data presented, including the ‘mass of raw phonetic […] data’ (Ó Maolalaigh 1997: 5) of the fieldwork sources. They should not be taken as primary data for further analysis without careful consideration of the original sources and scholarly assumptions on which they are based. 90 Aspects of the historical phonology of Manx Individual transcriptions represent varying degrees of confidence on the part of the author (the most doubtful forms are flagged with a question mark). They also may rest on assumptions and research outwith the topic areas covered explicitly in this thesis (especially with regard to the consonants). It should be noted that the broad phonemic transcriptions are in some instances quite far removed from possible Late Manx phonetic realizations, especially with regard to medial consonants (cf. Thomson 1984: 314–5), e.g. cassyn ‘feet’ (ScG. casan) is represented phonemically as /kasәn/ but could be realized [kasәn], [kazәn] or [kaðәn] (HLSM II: 60–1), also with optional lengthening of the stressed vowel [kaːzәn] etc. (§5.5.2).47 1.7.1 Stressed vowels Figure 2. Stressed vowels in Classical Manx High Mid-high Mid-low Low Front Central Back i iː e eː ɛː (?ɨ) (ɨː) (?ә) әː u uː o oː a aː Notes (a) /ɨː/ apparently occurs in Late Manx, arising from the monophthongization of the diphthong /ɨә̯/ (§2.2.6). (b) /aː/ has a restricted distribution (§2.2.4). (c) All vowels can apparently be phonemically nasalized (§§1.7.8, 5.6). (d) It is unclear whether short /ә, ɨ/ exist as shortened reflexes of G. ao(i), ua(i) (§3.9.11). 47 Only the earliest stages of this development are shown in the CM orthography and represented accordingly in the phonological transcription. Thus it is assumed that cassyn remained underlyingly /kasәn/, but that CM cabbyl ‘horse’ (G. capall) had become underlyingly /kabәl/ (> [ka(ː)bәl], [ka(ː)βәl], [ka(ː)vәl]), for older EM /kapәL/ (>[kapәL], [kabәL]), Ph. kapyl (2), kabyl (2). 91 Aspects of the historical phonology of Manx 1.7.2 Unstressed vowels in pretonic position Figure 3. Unstressed pretonic vowels in Classical Manx Front High Mid Mid-low Low Central Back i (iː) u (uː) e (eː) (ә әː) o (oː) (ɛː) a Notes (a) The original quality of pretonic vowels seems generally to be preserved judging by the orthography, although it is not always represented consistently (§5.1.3). (b) Original pretonic schwa in words such as G. amuigh ‘out’ (G. /әˈmuɣʲ/ > Manx /mui̯ /) is generally lost (but see əˈniːs neese, G. aníos [HLSM II: 321]), except in proclitics, as in my hie ‘my house’ /mәˈhai̯ / (ScG. mo thaigh). (c) In proclitics only /ә/ and /a/ occur, the latter in some cases representing historical */aː/. Long vowels may be preserved in Early Manx based on Phillips spellings ma, mý, má alongside my for G. má ‘if’, and dâ, dá, da, occasionally dy for contraction of G. do and ag with the third person possessives, G. dá, agá etc. (Lewin 2016a: 174; Ó Maolalaigh 2019). Gaelic má is written my in the Classical Manx orthography, but seems to have been pronounced /ma/ (HLSM II: 311–2; Cregeen s.v. mannagh), or confused with myr ‘as’, as shown the pronunciations given in HLSM and Edward Faragher’s spelling mor (Broderick 1982a: 180). Examples: with /ә/ dy, y ‘of’, ‘to’, verbal noun particle (G. prepositions do, de > a); my ‘my’ (G. mo); dty ‘your’ (G. do); e ‘his, her, its’ (G. a); nyn ‘our, your, their’ (G. ár, bhur, a); dy ‘that’ (subordinator) (G. go), dy, gy ‘to’ (G. go). With /a/ cha ‘not’ (Ulster and ScG. cha(n) < G. nocha(n) < OIr. nícon); nagh ‘not’ (G. nach); my ‘if’ (see above) (G. má); mannagh ‘if not, or not’ (G. má + nach). 1.7.3 Unstressed vowels in post-tonic preconsonantal position Figure 4. Unstressed post-tonic vowels in closed syllables Front Mid Low (e) Central ә a Minimal pairs: /slatәn/ slattyn ‘rods’, ScG. slatan /slatan/ slattan ‘small rod’, G. slatán 92 Aspects of the historical phonology of Manx /elʲәn/ ellyn ‘manners, behaviour’, G. aileamhain (§5.1.4.1) /elʲan/ ellan ‘island’, G. oileán, ScG. eilean, earlier ailén Note: (a) Post-tonic /a/ represents original long vowels, such as /aː/ in e.g. arran ‘bread’ /aran/ < /araːn/ (G. arán), and original short a before /x/ in the endings -agh /ax/ and aght /axt/. (b) There may be unstressed /e/ in -er (G. < -óir) (§5.1.6). 1.7.4 Unstressed vowels in post-tonic final position Figure 5. Unstressed word-final post-tonic vowels High Mid Front i Central Back u ә Minimal triplets: /mari/ marree ‘will kill’, G. marbhaidh /maru/ marroo ‘dead; to kill’, G. marbh, marbhadh /marә/ marrey ‘sea’ (genitive), G. mara /erʲi/ erree ‘fate’, G. airidh (Thomson 1981: 148) /erʲu/ erroo ‘ploughman’, G. aireamh /erʲә/ errey ‘burden’, G. eire, OIr. aire 1.7.5 Diphthongs and triphthongs In principle all of these can be nasalized, e.g. laue /lɛː̃ ũ̯/ ‘hand’ (G. lámh), feer vie /ˌfiːr ˈvãĩ̯ / ‘very good’ (G. fíor mhaith) (§§1.7.8, 5.6). Figure 6. ә-diphthongs High Front Central Back iә̯ ɨә̯ uә̯ Notes: (a) The split of ua into /ɨә̯/ and /uә̯/ is discussed in §§3.3, 3.4.3–6, 3.8. 93 Aspects of the historical phonology of Manx (b) In Late Manx these usually become long monophthongs /iː ɨː uː/ except in final position and before /x/ (§2.2.6). Figure 7. i-diphthongs Front (iәi̯ ) High Mid-high Mid-low Central ei̯ eːi̯ ɛːi̯ (ɨәi̯ ) әi̯ әːi̯ Back ui̯ uːi̯ uәi̯ oːi̯ ai̯ Low Notes: (b) These primarily arise from vocalization of fricatives to /i̯ /, e.g. G. maith ‘good’ /maθʲ/ > /mahʲ/ > /mai̯ / Manx mie; ScG. lagh ‘law’ /Lәɣ/ > /lәi̯ / Manx leigh; G. cloiche ‘stone’ (gen.) /kloçә/ > /klojә/ > /kloːi̯ / Manx cloaie;48 G. tráigh ‘shore’ /traːɣʲ/ > /trɛːi̯ / Manx traie. (b) Earlier */iәi̯ / usually gives /eːi̯ / (as in jeigh /dʒeːi̯ / ‘close’, G. iadh) or /eː/ (e.g. blein /blʲeːnʲ/ ‘year’, G. bliadhain) but is possibly retained in feeaih ‘deer’ (G. fiadh). (c) /ɨәi̯ / (> [ɯːi̯ ]?) may have been distinctive for some speakers in items such as creoi ‘hard’, G. cruaidh; leoie ‘ashes’ (G. luaith), but otherwise merges with /әːi̯ / (§3.9.1). (d) Earlier */oi̯ / has apparently merged with /ai̯ /, as in criy /krai̯ / ‘gallows’ (G. croich), lhiy ‘colt’ /lai̯ / (ScG. loth > *loith). Figure 8. u-diphthongs Front High Mid-high Mid-low Low iu̯ iːu̯ iәu̯ eu̯ eːu̯ ɛːu̯ Central ?әːu̯ Back ?oːu̯ au̯ (a) These primarily arise from vocalization of fricatives to /u̯ /, e.g. G. gabh ‘take’ /ɡav/ > /ɡau̯ / Manx gow; G. scríobh ‘write’ /skrʲiːv/ > CM /skriːu̯ / > LM /skruː/ Manx screeu; G. snámh ‘swim’ /sNaːṽ/ > /snɛː̃ ũ̯/ Manx snaue; G. rabhadh ‘warning’ /Ravәɣ/ > /rawә/ > /raːw(ә)/ > /rɛːu̯ / Manx raaue.49 (b) One would expect a contrast between /ɛːu̯ / and /eːu̯ /, but there is little evidence of this in the CM orthography, e.g. laue, raaue, snaue etc. = /ɛːu̯ /, /ɛː̃ ũ̯/, but also fraue ‘root’ (G. fréamh), A(a)ue ‘Eve’ (G. Éabha). Compensatory lengthening, cf. Donegal Irish: ‘In several instances α:i arises by the contraction of two syllables caused by the quiescence of intervocalic th, bh, gh, dh, e.g. brα:i, ‘hostage, prisoner’, M.Ir. brage (this word is also used to mean unfilled ears of corn)’ (Quiggin 1906: 58). 49 Compensatory lengthening, cf. Donegal Irish rabhadh roːwə, roːuw ‘warning’ (Quiggin 1906: 18). 48 94 Aspects of the historical phonology of Manx (c) /әːu̯ / would be expected from ao(i)bh/mh, but the evidence is unclear as to whether this was kept distinct from /au̯ /, e.g. crouw ‘shrub’ (G. craobh), but note monosyllable in noo /nuː/ ‘saint; holy’ (G. naomh). Similarly ?/oːu̯ / might be expected in loau ‘rot, rotten’ (G. lobhadh, lobhtha) (via compensatory lengthening, as in raaue above), and the spelling may indicated a contrast with low ‘allow’. (d) Earlier */ou̯ / has merged with /au̯ /, as in bouyr /bau̯ r/ ‘deaf’ (G. bodhar), towse /tãũ̯s/ ‘measure’ (G. tomhas). (e) /uv/ may give monophthongal /uː/ as in the southern pronunciation of doo /duː/ ‘black’ (G. dubh), diphthongal in northern Manx /dau̯ / (§4.6.1.34). 1.7.6 Consonants Figure 9. Consonants in Classical Manx Labial Stop / affricate Nasal Fricative Lateral Rhotic Semivowel p b m f v w Dental / alveolar t̪ d̪ ṯ d̠ tʲ dʲ (n̪ ) n̠ nʲ s (l̪ ) l̠ lʲ r rʲ Retroflex (ʂ) Postalveolar tʃ dʒ ʃ Palatal Velar Glottal k ɡ kʲ ɡʲ ŋ xɣ ç ŋʲ h j Notes (a) Manx agrees with Scottish Gaelic in lacking synchronic palatal labials (Oftedal 1963; Ó Maolalaigh 1997: 27, 44, 65–72; Ternes 2006: 27–43), which is particularly evident in items such as bannaght /banaxt/ ‘blessing’ (G. beannacht, Ir. /bʲan̪ әxt/), where Manx lacks the glide found in Scottish Gaelic varieties with /e/ > /a/ in beannachd /ˈbjan̪ әxk/. Glides (assumed here to be segmental /j/) are found in a few items with original G. eó, eabh, iú (e.g. bio ‘alive’ /bjoː/ G. beó; mioyr ‘mental faculties’ /mjoːrʲ/ G. meabhair; feeu ‘worth’ /fjuː/ G. fiú). But cf. foays ‘benefit’ /fau̯ әs, foːs/ (HLSM II: 171), Phillips fiâuys /fjau̯ әs/. There is sometimes /j/ > [lʲ] in bio in Late Manx > bl′oː (HLSM II: 31). That there was a palatalization contrast in the past, at least in medial and final position, is suggested by examples such as kemmyrk /ˈkemәʲrkʲ/ ‘refuge’ (G. coimirce), where the change /o/ > /e/ in the first syllable is difficult to account for without assuming earlier */mʲ/ (§2.1.6.1). Similarly, a residue of a contrast /v ~ vʲ, ṽ ~ ṽʲ/ is retained in Early Manx pairs such as Phillips dou, deyf /dãu̯ , dẽv/ ‘ox, oxen’, Classical Manx dow, dew /dãu̯ , dẽu̯ / (G. damh, daimh). (b) Gaelic /tj/ and /dj/ seem to have been realized as affricates [tʃ dʒ] initially and medially and as palatalized stops [tj] and [dj] finally. In final position these seem to be contrastive with /tʃ/ and /dʒ/, the latter occurring in loanwords, e.g. native aitt ‘funny’ (G. ait) and probably borrowed (?Eng. ‘botch’) spotch ‘joke’ (cf. de Bhaldraithe 1945: 36). Following the native intuition that apparently underlies the orthographic 95 Aspects of the historical phonology of Manx distinction, initial [tʃ, dʒ] from Gaelic /tʲ/ and /dʲ/ are synchronically grouped with final [tʃ, dʒ] (< English /tʃ, dʒ/) as phonemes /tʃ, dʒ/, contrastive with /tʲ/ and /dʲ/ from Gaelic /tʲ/ and /dʲ/ in final position. (c) The distinction between /d̪ / and /d̠ / is given on the strength of Cregeen (vi, §7–8), and the existence of similar contrasts between dentals in native words and alveolars in borrowings in other Gaelic dialects. (d) In most of the reconstructions given throughout the thesis only a two-way contrast in the coronal nasals and laterals is assumed, but this is probably an oversimplification: there is evidence of a three-way lateral contrast in some Late Manx speakers, and possibly the same for the coronal nasals. However, the distribution of these is not entirely clear. See Chapter 4 for discussion. (e) The retroflex sibilant /ʂ/ is posited on the strength of Jackson’s (1955: 125–6) claim to have heard it as a reflex of historic /rs, rʃ/ clusters where the /r/ is deleted and not written in the orthography (except in ersooyl ‘away’, G. air siubhal). If this [ʂ] did exist, it was apparently not analysed as /rs/ because it is consistently written <s(s)>, rather than <rs>, in items such as claasagh ‘harp’ (G. cláirseach), essyn ‘doorjamb’ (G. ursann), as ‘says, said’ (ScG. arsa), fesst ‘spindle (G. fearsaid), Phillips kuys, kus ‘course’ (later doublet coorse) (§4.2.2). (f) [v] and [w] may be allophones. There is apparently free (?) variation between [v] and [w] in forms of the preposition veih / voish (weih / woish) /vei wei vuʃ wuʃ/ (G. ó > 3sg.m. uaidh, voish < cf. rish, lesh etc.). However, there are a few lexical items with apparently fixed [v] or [w] (see under ‘v’ and ‘w’ in Cregeen). There is evidence of apparent substitution of [v] for [w] in nineteenth-century Manx English in a satirical article (Mona’s Herald, 20.06.1834), e.g. vell, vife, velfare, vondering. (g) Preoccluded or prestopped nasals and laterals are regarded as free variants of their non-preoccluded versions (§4.5). (h) The process of secondary lenition whereby e.g. peccagh ‘sinner’ /pekax/ can be realized [peɡax, peɣax, pejax, pei̯ ax] is analysed here as allophonic (see above). 1.7.7 Stress Stress usually occurs on the first syllable (§5.1). In the transcriptions it is ordinarily only marked if it occurs on a non-initial syllable. A synchronic long vowel in a noninitial syllable is always stressed. 96 Aspects of the historical phonology of Manx 1.7.8 Vowel nasalization Where it is considered that phonemic vowel nasality was likely present in Classical and Late Manx, or when there is positive evidence that it was so (e.g. from Rhŷs’s descriptions), stressed50 vowels are marked with a tilde in reconstructed phonological forms throughout this thesis. Both elements of a diphthong are marked as nasal. However, given the incompleteness of our knowledge of nasalization in Manx (Lewin 2019a: 82–9), no indication of the presence of absence of nasalization is to be taken as a claim that this was the case, unless explicitly stated. In most cases, vowel nasality will have no bearing on the question at hand. The marking of nasalization is primarily for the purpose of alerting the reader that vowel nasalization was a more substantial part of the Manx phonological system than has previously been assumed (Jackson 63– 4; HLSM III: 147; Ó Maolalaigh 2003a: 129), at least in the language of pre-terminal speakers. Nasalization is assumed to be present on stressed vowels adjacent to synchronic nasal vowels (cf. Rhŷs: 31), but is not indicated in the transcriptions. This is probably strictly inaccurate, since detailed studies of other Gaelic dialects show the existence of exceptions in which phonemic vowel nasalization is absent despite an adjacent neighbour consonant (cf. Ó Curnáin 2007: 324–33), e.g. Applecross Gaelic muir /mur/ ‘sea’, but muc /mũxk/ ‘pig’ (Ternes 2006: 104). This entails that vowel nasalization in such environments cannot simply be a phonetic consequence of the neighbouring nasal consonant. There is no clear data on this point for Manx.51 50 Phonemic nasalization is rarely attested on unstressed vowels according to Ó Curnáin (2007: 292–3); see also Ternes (2006: 111). According to Rhŷs (33), ‘wherever Manx has an unaccented u for a Goidelic ămh, I can find no trace of nasality attaching to the Manx vowel of the present day’. 51 ‘M in accented syllables beginning with it (or with v as its continuator) induces nasality. […] What exceptions there may be to it I could not say in a comprehensive or decided fashion’ (Rhŷs: 31). This tantalizing comment suggests that Rhŷs may have suspected that there were indeed such exceptions. Rhŷs (31–4) is much more definite about the existence of nasalization in items such as laue /lɛː̃ ũ̯/ ‘hand’ (G. lámh), troo /trũː/ ‘envy’ (G. tnúth), lenited vooar /vũә̯̃r/ ‘big’ (G. mhór). 97 Aspects of the historical phonology of Manx 1.7.9 Presentation of data In tables of lexical data throughout the thesis items are presented as follows: (a) The lemma in standard eighteenth-century orthography (Bible and/or Kelly’s, Cregeen’s dictionaries), together with variant spellings (with sources) where these add additional information. (b) A comprehensive list of spellings from the Phillips manuscript, as given in Thomson’s (1953) glossary. (c) Reconstructed Classical Manx (eighteenth-century) phonological transcription (§1.7). (d) The closest Gaelic cognate(s) (§0.3), i.e. Early Modern Irish forms, with other Irish, Scottish Gaelic, and Early Irish forms where relevant. (e) English translation or explanation. (f) A comprehensive list of occurrences in Broderick’s dictionary of terminal speech (HLSM II), with speaker initials (§0.7). For reasons of space, (f) HLSM data are omitted where the other evidence is sufficiently clear, and where the transcription data are not felt to add any additional or unexpected information. Only the relevant orthographic or transcription segment(s) are given in (b) and (f), except where it is felt desirable to give the whole word. In-text citations of data follow similar presentational practices. 98 Aspects of the historical phonology of Manx Chapter 2 The Manx vowel system According to Ó Maolalaigh (1997: 2–3), long vowels have been more stable in Gaelic dialects than their short counterparts, which have a greater ‘susceptibility to change’ ‘[b]ecause of their relatively short duration […] and their tendency to assimilate in quality to the consonantal environment’. This may be the case; but, as we shall see, there have also been significant changes in the Manx long vowels (§2.2), often with similar links to consonant environment to those seen in the short vowels (§2.1). Nevertheless, the bulk of the present chapter deals with the short vowels. The most complex developments concerning the long vowels and diphthongs involve reflexes of G. ao(i) and ua(i), to which a separate chapter (3) is devoted. New vowels and diphthongs arising from the vocalization of historical fricatives are also omitted for reasons of space, but are mentioned at §§1.7.5, 3.9.1. 2.1 Short vowels 2.1.1 a /a/ > /a/, /e/, /o/ G. a is mostly retained as /a/, e.g. annym /anәm/ ‘soul’ (G. anam), gastey /ɡastә/ ‘nimble’ (G. gasta), marroo /maru/ ‘dead’ (G. marbh), saggyrt /saɡәrt/ ‘priest’ (G. sagart). In some items there is raising to /e/ or backing and rounding to /o/. 2.1.1.1 a /a/ > /e/ Several items have variant spellings indicating a realization with /e/ alongside /a/; categorical raising is rarer. 99 Aspects of the historical phonology of Manx Table 2. a /a/ > /e/ Phillips CM pronunciation /bradax/, /bredax/ etymology English bradach thievish braddagh (Bible, Cr.), bred(d)agh (Cr.) derrey a (12) /derә/ dara the one derrey e (13), é /derә/ nó go dtara until ennal a (6), á /enal/ aná(i)l breath glen a (4) /ɡlen/ glan clean /ket/, /kat/ cat cat /kek/, /kak/ cac excrement /kesmad/ coiscéim footstep /skel/, /skelal/ scal beam, ray; disappear /skadan/, /skedan/ /teʂәn/ scadán herring tarsainn across kayt (Cr.), cat, caht (K.) 52 keck (Cr., K.), cac, cackey (K.) kesmad scell, skell; skellal skeddan, scaddan (Cr.) tessen a a (3) The motivation for the raising in these items is not clear, although it may be observed that all except one have a following coronal consonant. In the case of glen and ennal the following nasal consonant may be responsible; raising in items such as these is also found in some south-western Scottish dialects (Holmer 1938: 40; 1957: 48; 1962a: 5– 6, 74; Jones 2010: 85–90; Scouller 2017: 50–1). In several items only <a> is found in Phillips, which suggests that the change is not very old. There are also a number of loans from English with /e/ for English /æ/, such as blest ‘blast’, clesp ‘clasp’, edd ‘hat’, gless ‘glass’. Apparently this vowel was perceived by HLSM (II: 242): kɑtʹ TC, kʹe̜ t EKh, ke̜ tʹ JW, kʹet, kʹɛt NM, kʹɛt, ke̜ t HK. The forms of the singular and plural seem to be confused in the Manx of the terminal speakers, with apparently free variation of palatalized and non-palatalized consonants both initially and finally. Forms in /e/ seem to predominate in the singular, but TC has /a/, in accordance with Kelly’s spelling and Lhuyd’s form Chat (Ifans and Thomson 1980: 135). Cregeen’s spelling kayt could be interpreted as reflecting variant forms /kat/ and /ket/, since <ay> may represent /e/ in aym ‘at me’ (G. agam), ayd ‘at you’ (G. agat, agad) but /a(ː)/ in ayrn ‘part’ (G. earrann) etc. (§1.6.4.4). The reflexes of G. cat are irregular in a number of Ir. and ScG. dialects. 52 100 Aspects of the historical phonology of Manx Manx speakers as being closer to Manx /e/ than /a/ in the English varieties with which they were in contact (as also in Welsh; cf. Parry-Williams 1923: 24–7). 2.1.1.2 a /a/ > /o(ː)/, /u(ː)/ In the following items the change /a/ > /o(ː)/ (or occasionally /u(ː)/) seems to be complete in Classical Manx, as shown by standard spellings with <o>, <oa> etc. in the Bible. Unrounded /a/ is often indicated by Phillips in these items,53 so the change must have been in progress in the seventeenth and eighteenth centuries. Table 3. a /a/ > /o(ː)/ (/u(ː)/) Phillips aunlyn, oanlyn, oalyn (Cr.) ayns, ayn, aynym, aynyd, aynjee, aynin, ayndoo boal, boalley, voalley boandey CM pronunciation /oːnlәn/, /õːlәn/, /au̯ nlәn/ etymology English annlann gravy, soup, relish /oːn/, /uːn/, /o(n)s/, /onәm/ etc. ann in, in him / it, in me, in you, in her, in us, in you (pl.), in them /boːl/, /boːlә/, /voːlә/ /boːndә/ balla wall band (of iron etc.) boandyrey, boandyrys boayl oy (‘bondservant’) baintri /boːndәrә/ Ir. banda, ScG. bannd banaltra a, o /boːl/ ball place bollag o, a (2) bollan cloan coayl croan; cronney; kranghyr (Ph.) au (3), áu (4), aú a (6), á, à (2), ia (croan) a (2), ó; (kranghyr) ai, a (7), ay, á nurse /bolaɡ/ ballóg, ScG. ballag skull /bolan/ ballán rockfish /kloːn/, /klau̯ n/ clann children /koːl/ call lose /kroːn/, /krau̯ n/; /kronә/; /kronxәr/ crann; crannadh; crannchor mast, tree; lot, fate, portion 53 However, <all> in e.g. dall, CM doal ‘blind’ (G. dall) or fallsy, CM foalsey ‘false’ (G. fallsa) could possibly represent /o(ː)l/ (cf. English all, ball). In Ph. fallaghy etc. ‘hide’, CM follaghey, the /o/ is etymological, and it is unclear whether Phillips’ <all> represents /ol/, or a form equivalent to ScG. falach. On the basis of English orthography, however, <all> = /ol/ is perhaps less likely in polysyllables. 101 Aspects of the historical phonology of Manx doal, pl. doaley; dalley, dallagh, dolley, doalley, dollan a (3), (dall(agh)ey) a (2), á /doːl/, /doːlә/, /dalә/, /dalax/, /dolә/, /doːlә/ dall, dalla; dalladh blind o /dolan/ dallán winnowing-fan doltey, doltanys doo-oallee a (3) /doltә/, /doltanәs/ /duː ˈoli/ dalta damh allaidh ward, adopted child; adoption spider drunt au /drunt/ drannt, dranndal gums foaddey a (6) /fodә/ kindle, light foalsey a (8) /foːlsә/ (f)adódh, (f)adughadh, ScG. fadadh fallsa a /foldәr ʲ /; /jaːrn foldәrax/ ScG. fàladair mower; scythe /folәdʲ/, /folәkʲ/ ScG. fallaid S /ɡoːn/, N /ɡau̯ n/ /ɡolˈtuәi̯ / gann ‘dry meal put on a cake to bake or clap it out’ (Cr.) scarce gal + tuaith rainbow gorley /ɡoːrlә/ galar disease loaghtyn (Bible, Cr.), loghtan (K.) lossey /loxtәn/ ScG. lachdann, Ir. lachna, lacharnach tawny brown grey o (6) /losә/ lasadh blaze, flame o (3), ó /losәrʲ/ lasair flame /moːl/ mall slow, poorly, bad /moːndax/ manntach blunt, stammering /modә/ madadh dog /molkә/ ScG. malcadh macerate, rot /molaxt/ mallacht; mallaghadh curse /moli/ eyebrow folder, foldyr; yiarn foldyragh follyd, follick goan, goaun goll-twoaie lossyr (Ph., PC) moal a moandagh moddey o (3), ŏa, a (8) molkey mollaght, mollaghey o (6), a (1), (mollaghtoil) o (2) mollee () false /noːl/, /hoːl/ mala, malaidh, malaigh anall, thall oaldey /oːldә/ allta wild oalsum (Cr.), ousym (K.) oanluckey oghsan a (12), á /oxsan/ Norse halsband > *allsam annlacadh, adhlacadh ach(mh)asán etc. tie on cattle a (6) /oːlsәm/, /oːsәm/, /au̯ sәm/ /oːnlәkә/ ollish a (2) /olәʃ/ allas sweat noal, hoal 102 á, a over (adv.) bury rebuke Aspects of the historical phonology of Manx ollym (Cr.) /olәm/ alum olt, oltey ay (2), a (5) /olt/, /oltә/ alt member, organ oltaghey a (2) /oltaxә/ altaghadh salute, greet ommidan, ommijagh, ommijys ronneeaght a (11) /omәdan/ fool /roni.axt/ amadán, amaideach, amaideas rannaigheacht /ronә/ rannadh ronney foolish song ronsaghey a3 /ronsaxә/ rannsaghadh portion, share, division search sollagh, sollaghey, sollaghys sollan a (9) /solax/, /solaxә/ salach, salaghadh dirty /solan/ salann salt sondagh /sondax/ sanntach greedy Sostyn (Cr.), Sausin, Sacsin (K.), Socsyn (FRC, Trad. Ballad etc.) sporran /sostәn/, /soksәn/, /sosәn/ Sacsa etc. England /sporan/ sparán, ScG. sporan purse stholley /stolә/ stalla stall, station (Cr.) stoandey /stoːndә/ ScG. stannd standish stronnagh, stronnal toghtey, toaghtey (K.) /stronax/, /stronal/ /toxtә/ srann snort tachtadh choke a (2) In the following items spellings are variable in the eighteenth- and nineteenth-century texts and dictionaries, suggesting a continuing change in progress. In all of these items the vowel precedes original lenis /l/ or /n/ (possibly /N/ in Onnee, cf. ScG. Anna /aNә/). Table 4. a /a/ > /a/, /o/ Albin, Nalbin, Nolbin (Cr.), Alpin, alban, olban, alpan, olpan (K.);54 Albinagh (Cr.), Alpinagh (K.) CM pronunciation /(n)albәnʲ/ /(n)olbәnʲ/, /albәnax/, /olbәnax/ etymology English HLSM Alba, Albain; Albanach Scotland; Scottish, Scot a, ɔ̜ NM, ə, ɔ̜ JK, nø TC nɔ, no̤ ⁱ HB, nɑ JW, na NM, TK, nɑ̜ TK, (adj.) ɑ JW, u̜ JK 54 Kelly’s spellings with derive from etymological fancy, but the variation between /a/ and /o/ forms is probably genuine. 103 Aspects of the historical phonology of Manx boannoo, bainniu (Cr), bannoo (K.) Colloo (Cr.), Calloo (K.) maynagh, monnagh, monnaghan (all K.) Olister, Alister (Cr.), Allastar (K.); Callister (surname) Onnee (Cr.) sannish, sonnish (Cr., K.) tallagh (Cr., Bible), tollagh (K.) /banu/, /bonu/, /banʲu/ banbh, bainbh /kalu/, /kolu/ half-grown pig Calf of Man /manax/, /monax/, /monaxan/ manach /(k)olәstәrʲ/ (Mac) Alastair /oni/ ɑ̜, a TK, ɑː, ɔ̜, ɑ̜ NM monk; (monnaghan) ‘fat greasy fellow’ (K.) ɔ̜ NM, TK Ann /sanәʃ/, /sonәʃ/ sanas, sanais whisper /talax/, /tolax/ ScG. talach murmur, grumble, complain Further items may belong here, such as thalloo ‘land’, G. talamh, found with back realizations of the vowel from most of terminal speakers (ɑ, ɑ ̣ TC, a HB, ɑ, a JK, ɑ̜, ɔ NM, a J:EK, J:CW, ɑ W:NS, HLSM II: 446). In such cases we may have an as yet incomplete near-merger /a/ > /o/. Backing and rounding of /a/ occurs mainly before the velarized sonorants /L, l, N/, the velar fricative /x/ and after labials /b, m, f, p/. In a couple of cases there is possibly influence from semantically and phonetically similar items, i.e. Socsyn, Sostyn, Sausin ‘England’ (G. Sacsa), cf. Nalbin, Nolbin, Albin, Olbin ‘Scotland’ (G. Alba), Loghlyn ‘Norway’, (G. Lochlainn); lossey ‘blaze, flame’ (G. lasadh), lossyr ‘flame’ (G. lasair), cf. lostey ‘burn’ (G. loscadh). The change is almost categorical before historical intervocalic /L/ and /N/ (see also §§4.6.1.13, 4.6.1.14) but only incipient before /l/ and /n/ which may reflect the fact that the fortis and lenis sonorants were kept separately until at least the Early Manx period. Before coda /L, N/ the development to /oː/ (or /uː/) may be via diphthong /au̯ / rather than via /a/ > /o/. 104 Aspects of the historical phonology of Manx 2.1.2 ai /a/ > /a/, /e/ EIr. ai (often > G. orthographic oi [Ó Maolalaigh 1997: 156–63; Ó Baoill 2012]) gives Manx /a/ and a raised reflex /e/ in roughly equal proportions. There are also some nonstandard spellings which suggest the existence of a further raised reflex /i/ in some lects, such as mirriu ‘dead’ (pl.) for usual merriu.55 2.1.2.1 ai /a/ > /a/ Table 5. ai /a/ > /a/ Phillips CM etymology English aggindagh a (8) /aɡʲәndax/ aigeantach eager, willing aigney a (25) /aɡʲnʲә/ aigne mind, will aile angil (3), ángil, angill (2), angyl, aínill ay (4), áy, a (2) /ãĩl/ aingeal fire /anʲdʒәs/ aitheantas, aithint acquaintance /ãĩl/ aingeal angel /atʲ/ ait funny aithne commandment ScG. aird area ainjys ainle a (11), ái (2), ai, á aitt anney (§4.4.3) a (7) /anә/ ard (pl. ardjyn) a, á /aːrd/ argid a (9) /arʲɡʲәd ʲ / airgead silver, money arkys a /arʲkʲәs/ airc misfortune /aːrnʲ/ airne sloe /arʲә/ aire watch /arʲәʃ/ aithris jeering arn arrey a (15), ăy̆ , ay, á (2), ay̆ arrish () arrys a (7) /arʲәs/ aithreas repentance ashlish a (2) /aʃlʲәʃ/ aisling dream, vision atchim a (7) /atʃәm/ fear bainney a /banʲә/ eitim, OIr. e(i)tim, aitim56 bainne /banʲu/ bainbh young pig bainniu milk 55 Noted in MS of 1 Thessalonians. Thomson (1953: 153) and Broderick (HLSM II: 14) derive this tentatively from an unattested compound of time ‘tepidity, warmth, softness, weakness, cowardice, fear’ (Dinneen), ScG. tioma, Manx chymmey ‘pity’. However, as suggested by Williams (1994: 734), it seems more likely that it is a semantic development of OIr. etim, etaim, aitim ‘spring, leap (?); thrust… chance, opportunity; breach (?)… in Laws applied to a species of pledge’ (eDIL) which in later periods may have senses ‘danger, hazard; a hazardous effort; chance, opportunity; a sudden spring’ (Dinneen s.v. eitim), ‘danger, hazard’ (Dwelly s.v. eiteam). 56 105 Aspects of the historical phonology of Manx balley a (8) bashlagh bashtey blashtyn caillagh (K.), keyllagh, keyhlagh (Cr.)58 cashtal a (13) 57 a a (2) clash /balʲә/ baile town /baʃlʲax/ baisleach douse, splash /baʃtʲә/ baisteadh baptize /blaʃtʲәn/ blaiseadh taste /kalʲax/, ?/kelʲax/ cailleach hag, old woman, nun, dryad (Cr.) /kaʃtʲal/ ScG. caisteal castle /klaʃ/ clas, clais furrow dash a /daʃ/ dais heap er ash a (5) /erʲ ˈaʃ/ ar ais fakin, (fut.) vaik, naik, (cond.) vaikagh, naikagh farkiagh a (26), á, aî, ăi, ai (4) /fakʲәnʲ/, /vakʲ/, /nakʲ/ faicsin, ScG. faicinn coming to light, blossoming etc. (cf. Lewin 2016c: 96–7) see a (8), á (2) /farʲkʲax/ faircsin wait gailley /ɡalʲә/ gaile stomach glashtin /ɡlaʃtʲәnʲ/ ScG. glaisteag goblin */madʒәn/ maidean morning /madʒә/ maide stick a (2) /naʃtʲ/ naiscthe betrothed paitt a (5) /patʲ/ pait plague palchey a (7), ai /palʲtʃә/ pailte plenty prash prass (4) /praʃ/ ScG. prais brass saynt ai (12), ái (3), âi (2), (sayntoil) ai (3) a (4) /sanʲtʲ/ sain(n)t lust, covetousness /tarʲәʃ(ax)/ tairise tender (tashey ‘compassion’) a (5), (tashlys ‘moistness’) a a (6), á /taʃ/ tais; taise; taisleach damp /taʃtʲә/ taisceadh treasure madjyn (Ph.) a maidjey nasht tarrish, tarrishagh (Ph.) tash (Cr.), taaish (K.) tashtey 57 Only one occurrence of blastchyn (=CM blashtyn), other forms appear to show blas- rather than blais-, i.e. blassyght, blasghy, blasaghtyn, although /s/ amd /ʃ/ are not always distinguished in the Phillips orthography. 58 Cregeen appears to assume a derivation from keyll ‘forest’ (G. caill, coill). 106 Aspects of the historical phonology of Manx 2.1.2.2 ai /a/ > /e/ Table 6. ai /a/ > /e/ Phillips brellish ec; echey; eck (ec) a (27), (echey) a (5), (eck) e (3), ĕ (2), æ, a eggey eirin(n)agh ié e (2) elley e (6) etymology English brailis, braichlis wort /ekʲ/; /eɡʲә/; /ekʲ/ aig; aige; aici at; at him; at her /eɡʲә/ oige, EIr. aicde, aice aireamh web /elʲaɡ/ faileog hiccup /elʲan/, ?/alʲan/ oileán, EIr. ailén, , ScG. eilean eile, EIr. aile island oileamhain, EIr. ailemain aimsir behaviour /erʲәnʲax/ ellag ellan CM pronunciation /brelʲәʃ/ /elʲә/ 59 farmer other ellyn ellyn, elúyn /elʲәn/ emshyr, emshir enney, enn; ennaghtyn a (5) /emʃәrʲ/ (enn) e (7), (ennaghtyn) e (3), ea, a (3), æ, iêa e (5), æ (3), æ, é, ey, (enmys) e (9), æ e (9), (3sg.m.) e (3) /en⁽ʲ⁾(ә)/; /en⁽ʲ⁾axtәn/ aithne recognition; perceive, feel /enәm/ ainm name /erʲ/ air on /erʲi/ airidhe60 fate ennym er erree errey weather /erʲә/ eire, EIr. aire burden erroo (pl.) érynyn /erʲu/ aireamh ploughman gedjey e, ei (2) /ɡedʒә/ oide, EIr. aite foster father geid ey (4), e (2), ey̆ /ɡedʲ/ goid, EIr. gait steal geinnagh ǽ, e /ɡenʲax/ gainmheach sand genney, gennid (er-)gerrey; gerrid e (4), eà, ea, æ /ɡenʲә/, /ɡenʲәdʲ/ gainne scarcity (er-gerrey) e (9), (gerrid) a (2), e (7), æ æ, e (11), é, ey, æy /ɡerʲә/, /ɡerʲәdʲ/ gaire near; short, soon /ɡerʲәm/ gairm crowing gerrym 59 60 For the latter form with final stress /eˈlʲuːn/, see §5.1.4. See O’Rahilly (1927: 13–4). 107 Aspects of the historical phonology of Manx keiyt, kiyt (Cr. pl.), chett (K. gen.)61 kelk kellagh e (6) kennip kerraghey /ketʲ/, /kitʲ/ cait cats, cat (gen.) /kelʲkʲ/ cailc chalk /kelʲax/ coileach, EIr. caileach cainb cock punish /kerʲu/ coireaghadh, EIr. caircairbh carp (pl.) /kenʲәp/ e (13), ey, o (?e), æ (4), é (2) kerriu /kerʲaxә/ hemp keyll ǽi, é, e (2) /kelʲ/ coill, EIr. caill forest, wood merriu êî, e (3), ei (3), ĕ, ĕi, æ /merʲu/ mairbh dead (noun pl.) /reʃ/ rais seed (gen.) /renʲax/ raithneach fern saick, seick /sekʲ/ saic sacks skerin /skerʲinʲ/ scair splice, scarf (Cr.) s’melley, meillid smerg /s melʲә/, /melʲәdʲ/ is maille /s merʲɡʲ/ is mairg slower, worse; meanness woe /s tenʲә/ ScG. as taine thinner /terʲu/ tairbh bulls /tenʲu/ ScG. taineamh thaw /trenʲә/ tairnge; tairngeadh, ScG. tàirng, tarra(n)g nail, to nail resh rhennagh (kellagh rhennee) e (3) a (6), á s’theinney terriu, teirroo (both Bible) theinniu e (2) treiney, also Cr. treinney, treinnit ei Manx agrees with Scottish Gaelic and Ulster Irish in keeping original ai /aCʲ/ (often spelled oi in Ir. and ScG.) and original EIr. oi /oCʲ/ distinct (Ó Maolalaigh 1997: 153– 63, 202–9), with the former being generally found as /e/ in Manx in those items where raised forms occur in the other dialects. Ó Maolalaigh (1997: 202–4) shows that raising of /a/ to /ɛ/ in Scottish Gaelic occurs mostly ‘before palatalised apical consonants’ and ‘most commonly in the vicinity of nasals’ or ‘when the vowel is nasalised’. Raising to /e/ ‘occurs mostly in absolute initial position and is particularly common before the palatalised apical //lʹ//. The development //a// > /e/ is also attested following the velars /g k/ in the prepalatal keːtʹ, kɑ ̣ːitʹ TC, kʹɛt, kʹe̜ t, EKh, kït, kɪt HK (HLSM II: 242). These forms and the spelling variants may imply two variants /ketʲ/ and /kitʲ/. Speakers’ uncertainty about these forms may be reflected in the development of a regular plural *kaytyn ke̜ tən JW, kʹetːən NM; see also the singular (fn. 52). 61 108 Aspects of the historical phonology of Manx position, particularly in Arran and Kintyre dialects’ (Ó Maolalaigh 1997: 204), while ‘the raising of //a// to /ɤ/ (/o/?) occurs mostly in words of the shape C __ Cʹ where C = /k g/ and C = /dʹ lʹ rʹ/’ (Ó Maolalaigh 1997: 212). In Irish the development /a/ > /e/ C_Cʲ occurs in the following environments (Ó Maolalaigh 1997: 152): (1) # __ Cʹ Cʹ = [+coronal] for most examples, e.g. air, aige, (aileán), aile, aileamhain, aideachas (2) C __ Cʹ C = [–velarised] (i.e. /t d s r/ […]), Cʹ = [+ coronal] mostly e.g. sair, saidhbhir, traigh (Connacht dialects especially) (3) C __ Cʹ mostly C = [+velar] , Cʹ = [+coronal] but also following certain velarised consonants, e.g. caileach, cair, gairm, gaid, gaile, gairid; traigh, laigh It has not previously been noted that Cʹ in almost all words which illustrate the development //a// > /e/ share the features [+coronal] [+voice] and include /lʹ rʹ dʹ/ but apparently not /Lʹ Nʹ/. It is also significant that the development is common throughout Irish dialects in words containing absolute initial //a//. (Ó Maolalaigh 1997: 152) The environments in which the change /a/ > /e/ occur appear to be similar in Manx to those where raising is found in Scottish Gaelic and Irish: absolute initial position, preceding slender coronals, especially sonorants /Lʲ, lʲ, Nʲ, nʲ, rʲ/, following /ɡ, k/, and in nasal environments. Spellings with <a> in Phillips in echey (G. aige), emshyr (aimsir), s’merg, (is mairg) and gerrid (gairid) suggest that this development was still in progress in the seventeenth century.62 The consistent spelling of G. a(i)g ‘at’ as ag in Phillips may represent unraised /aɡʲ/, or possibly maintenance of the historical simple preposition ag (OIr. oc), which has otherwise been replaced by aig (from the 3sg.m.) throughout Gaelic dialects (Williams 1994a: 462), and in Classical Manx (where ec shows the same devoicing of final /ɡʲ/ found in aspick ‘bishop’, G. easpaig etc.). 62 Although Thomson (1953: 7) suggests that Phillips’ representation of emshyr with initial <a> reflects Welsh amser. 109 Aspects of the historical phonology of Manx NM’s form alʹan (HLSM II: 145)63 for ellan ‘island’, Lh. alyn, and also some [a] spellings and pronunciations in place-names (PNIM),64 may represent survival of original /a/ in EIr. ailén, otherwise universally raised (Ir. oileán, ScG. eilean), and otherwise with e̜ , ɛ, ə, e in Manx terminal speech. 2.1.2.3 ai /a/ > /o/ Three cases of rounding of ai /a/ to /o/ have been identified (cf. the much more widespread development /a/ > /o/ before broad consonants, §2.1.1.2), all of which may be ascribed to the preceding labial consonant (cf. Ó Maolalaigh 1997: 205–6, 223). In the case of OIr. Maire ‘Mary’, rounding is universal in Gaelic dialects, and boireann ‘female’ in well-established in Scottish Gaelic. Table 7. ai /a/ > /o/ Phillips bodjal o (9) CM pronunciation /bodʒal/ bwoirryn /borʲәn/ Moirrey /morʲә/ etymology English ScG. baideal, Eng. battlement baineann, ScG. boireann ScG. Moire, Ir. Muire, EIr. Maire cloud female Mary 2.1.3 ea /e/ > /e/, /a/, /o/ G. /e/ before broad consonants may retain65 its original mid height or be lowered to /a/, sometimes with subsequent backing and rounding to /o/. In a few cases raising to /i/ is found. Some items show variation between reflexes, especially between /a/ and /e/, sometimes reflected in variant spellings. In the following tables, and the calculations based on them, items are categorized according to the most common variant or the variant reflected by the standard spelling. 63 Also aⁱlʹan (JTK), unless this is influenced by the English word. PNIM I: 175, III: 160, 162, 239, 269–70, 381, 384 (north); VI: 59, 329–30 (south). 65 I follow Ó Maolalaigh (1997: 289) in assuming, as the most parsimonious account, retention of original /e/, rather than lowering to /a/ with subsequent raising. 64 110 Aspects of the historical phonology of Manx Table 8. ea /e/ > /e/ Phillips beg; beggan e (14) CM pronunciation /beɡ/, /beɡan/ ben e (8) /ben/ bean woman benn, bentyn bennalt bher e (12), ey /bentәn/, /benәlt/ /ber/ bean touch. belong; flap bior, bear spit breb; breban a breab, breabán kick; (breban) dried snot speckled; trout, mackerel breck, brack (Cr.), breac (K.)67 creg e (5), è dress etymology English beag small /breb/; /breban/66 /brek/, /brak/ breac /kreɡ/ creag rock /dres/ dreas bramble, briar edd æ /ed/ nead nest eddyr, edyr e (5) /edәr/ eadar between, whether edyr e (4) /edәr/ eadar at all /fed(an)/ fead, feadán pant, whistle /fer/ fear man, one /feʂt/ fearsaid spindle /freɡәrt/ freagairt answer fed, feddan fer e (59), é, ie (2), y (3) fess, fesst freggyrt a (19), e 68 gennish e (2), ea /ɡʲenәʃ/ gien, gennal a (14), ia /ɡʲen/ gean cheer gleck e /ɡlʲek/ gleac wrestle, struggle greddey /ɡredә/ gread grill, roast, toast guess /ɡʲes/ geas spell, charm /kʲeru/ ceathramha quarter /kʲebә/ ScG. ceaba spade kerroo æy, æ kiebbey geanas barren, infertile kied a /kʲed/ cead permission lieckan ie /lʲekan/ cheek lhieggey; lhieggal meddyr, mheddyr mennee ie, ia /lʲeɡә/; /lʲeɡal/ EIr. leccan, leccond leag /medәr/ meadar pail, wooden vessel /meni/ meana(i)dh awl mess ea (8), éa /mes/ meas fruit smessey a (5) /smesә/ is measa worse, worst mestey, mastey a (2), á /mestә/, /mastә/ meascadh mix, churn fall; fell, overthrow e̜ HK, a NM (breban). Bible mostly breck, one instance of brack. 68 See fn. 85. 66 67 111 Aspects of the historical phonology of Manx mettey, meddey (K.) sniessey /metә/ meata a (2) /snʲesә/ is neasa tender, delicate, soft, cowardly nearer, nearest pecca e (25), é /pekә/ peacadh sin preban (K.) /preban/ preabán patch of land, cloth prest /prest/ preas cupboard screb /skreb/ screab scab scred /skred/ scread gasp /ʃelɡ/ sealg hunt /ʃelɡ/, /ʃolɡ/ sealg milt; stomach, guts e (7) /ʃelu/ sealbh herd e (4), a (13), é, ey, ia (5), iǽ, æ S /ʃen/, N /ʃan/ (HLSM II: 398) /snʲeŋ/ sean old sneadh nit sniengan (fn. 82) streng /snʲeŋan/ seangán ant /streŋ/ sreang string strepey, strebin (Cr.) /strepә/ ScG. streap, Ir. dreap struggle, wrestle CM pronunciation /aɡәl/ etymology English eagal fear /aɡlәʃ/ eaglais church agh /ax/ each horse arragh /arax/ earrach spring /aspәrt/ easpart(a) askaid /askәdʲ/ neascóid vespers, evening prayer boil asney /asnә/ /aspәkʲ/ easna, also E.Ir. asna easpag bishop /asaɡ/ easóg weasel /asi/ easbhaidh harm astan /astan/ eascann eel astyrt /astәrt/ eascairt uproot /aːrn/ earrann part shelg, sheilg69 shelg (K.), chiolg (Cr.) shelloo shen e (2), ǽ snieng (fn. 82) Table 9. ea /e/ > /a/ Phillips aggle a (22) agglish a (9), á (2), æa, ea, e (4) asbyrt aspick a (2) a, yn ia (2) assag assee ayrn 69 a ay (3), áy (2), a (4), á, æa The spelling sheilg (Bible and Cregeen) may indicate G. seilg. 112 rib Aspects of the historical phonology of Manx baght /baxt/ beacht bangan /baŋan/ beangán /banaxt/; /banaxә/, /bani/ /brakә/ beannacht, beannaghadh, beannaighthe breacadh blessing, bless, blessed /braxtan/ breachtán bread with butter etc. /tʃax/ teach hiding place teachtaire; teachtaireacht teaglaim messenger; message gather bannaght; bannaghey, bannee brackey (Cr.) a (26) braghtan chagh ia (2), a, iá observation, discernment branch sharpen chaghter; chagheraght chaglym a (12), ia (7) /tʃaxtәrʲ/; /tʃaxtәrʲaxt/ /tʃaɡlәm/ chiamble ia (6), a /tʃambәl/ teampall temple /tʃarә/ tearadh, turadh dry spell a (5) /tʃas/; /tʃasaxә/ teas heat; fever a (6), á (2), ia /klʲaxtә/ cleachtadh custom chiarrey chiass; chiassaghey cliaghtey cliass (Cr.) /klʲas/ cleas ‘the same fate’ (Cr.) /krax/ creach spoils, prey drappal /drapal/ dreap climb fam; famlagh, famyragh feam; feamnach stem of seaweed; seaweed fannag /fam/; /famlax/, /famәrax/ /fanaɡ/ feannóg crow fanney /fanә/ feannadh flay farbaghey /farbaxә/ fearb inflame farbyl /farbәl/ earball tail /farɡ/, /ferɡ/ fearg anger /farnә/ fearn black alder /ˌfarәsˈtai/ fearas tighe housekeeping a (5), (len.) ia, îâ a /fastәr/ feascar evening, afternoon /ɡʲal/ geal bright (giare) ia (3), (yiare) iar, (giarraghey) ia, (giarey) (v.) iá (2), ia (2), a, á, ay, iâ, (pret yiare) gáre, jarr, iár (n.) ia, a (6), (participle) a (3), á /ɡʲɛːr/, /ɡʲarә/; /ɡʲɛːrә/, /ɡʲarә/ gearr; gearradh short; cut /ɡʲaran/ gearrán worthless horse /ɡʲastәlax/ geastal charitable; charity cragh farg (Bible, K., Cr.), ferg (Cr.) farney éa, e, a (3), á, æ ay (2), a, (fargoil) e, a farrys-thie fastyr gial giare, (pl.) giarrey; giar(r)ey giarran giastyllagh a (7), (len.) ia 113 Aspects of the historical phonology of Manx giat ia (4), a /ɡʲat/ geata gate hannah a (4) /hanә/ cheana already hiar, har, shiar, niar jagh ia (6), a, ià, iæ̆ ear east ea, a (5) /h(j)ar/, /ʃar/, /nʲar/ /dʒax/ deach- went jaghee ia /dʒaxi/ deachmha tithe ?/dʒaxәn/ deacon /dʒalu/ EIr. dechon (later deoch-) dealbh /dʒarɡ/ dearg be able /dʒaru/ dearbh very, indeed, even jasdil, jasdyl /dʒastәlʲ/ deasgabháil Ascension jastan /dʒastan/ deascán jastee /dʒasti/ deasca(idh) ‘a course or row of ling or heather laid on the ground from the hand of the puller’ (Cr.) yeast jaghin (Cr.), joghan (K) jalloo a (11)71, iă jarg jarroo a (16), ia (4), ía, iâ, but jeru ‘affirm, certify, prove’72 picture jialg, jolg a /dʒalɡ/, /dʒolɡ/ dealg thorn, knitting needle jiarg a (5), á, ia /dʒarɡ/ dearg red jiass, yiass, ass a (3) deas south kialg a (4), á, (kialgoil) a /dʒas/, /jas/, /as/ /kʲalɡ/ cealg deceit /kʲaltәr/ cealtair, -ar ceangladh kiap /kʲaŋlә/, /kʲõːlә/ /kʲap/ unmilled woollen cloth tie, connect ceap block kiare /kʲɛːr/ cearr left /kʲark/ cearc hen /kʲarkәl/ ScG. cearcall circle kialter kiangley kiark ia (3), a (8), á ia kiarkyl kiarroo ia, a (3) /kʲaru/ ceathramhadh fourth kiart ay (2), iâ, ia (6) /kʲart/ ceart right, correct lhiabbee ia (3), a, iă /lʲabi/ leaba(idh) bed lhiaght /lʲaxt/ leacht tomb, couch lhiannan /lʲanan/ leannán lover 71 Instances of jallunyn, jallúnyn probably = jalloonyn /dʒaˈluːnәn/ ‘idols’ (G. dealbhán). Further ta jeru ayms ‘I am certified’, ry-ieru ‘instantly’, rŷ ieru ‘earnestly’, ry ieru ‘seriously’, ră ieru ‘earnest’. If jeru does not simply represent dearbh(adh) with /e/ rather than /a/, with the phonological distinction perhaps serving to distinguish the functions (cf. mastey ‘among’ and mestey ‘mix’), we may have a by-form *deirbh(eadh) or abstract noun deirbhe. Ry-ieru etc. could perhaps be linked with G. dáiríre etc. (cf. eDIL s.v. darírib). 72 114 Aspects of the historical phonology of Manx lhiannoo ia (10), iá (3) /lʲanu/ leanbh child lhiantyn ia (6), iá, ie (2), iæ /lʲantәn/ lean cleave, adhere, stick /lʲargax/ leargach slope lhiargagh s’lhiass; lhiassaghey lhiastyn ia (2), îâ /lʲas/, /lʲasaxә/ leas(aghadh) need; atone, improve ia (3), a (2) /lʲastәn/ dleastanas owe, debt lhiattee ia (2), iea /lʲati/ leataobh side /manan/ meannán mannan marran kid, young goat 73 /maran/ ScG. mearan mistake, mistaken mastey a (26) /mastә/ i measc among niart iá (2), ía (3), ia (5), (gniartoil etc.) (47)74 penaȳs, peynans75 /nʲart/ neart might, strength /panәs/, /penәs/ /raxt/ peanas penance reacht stubbornness /rastax/ ScG. reasgach blustery; hoarse pannys (Cr.), pennys (K.) raght rastagh shaghey a (15), ia /ʃaxә/ seachad past shaghney a (6) /ʃaxnә/ seachnadh avoid, spare (er-)shaghryn a (7), iâ /ʃax(ә)rәn/ seachrán astray shallid a /ʃalәd/ sealad moment /ʃaŋ/ seang lank, lean shang shanstyr ia (4) /ʃanstәr/ cf. sinnsear elder, elders share /ʃɛːr/ is fearr better, best sharragh áy (4), ay (3), niarr, âŷ a /ʃarax/ searrach foal sharroo ia (3) /ʃaru/ searbh bitter shassoo a (21) /ʃasu/ seasamh stand /ʃaslax/ seasclach bentgrass (shaliygh) a /ʃal/ seal, sealaigheacht turn; (shaliygh) reason (Ph.) ay, ia (5), a (3) /ʃaxt/ seacht seven shaslagh shayll (Bible, Cr, K.), shall (K.) shiaght shiaghtin ia, a (4) /ʃaxtәnʲ/ seachtmhain week shiast a /ʃast/ seasc dry, barren /smarә/ smearadh grease /snʲaxtә/ sneachta snow smarrey sniaghtey a 73 Cf. Thomson 1998: 132; Ó Maolalaigh 2014a. Forms of ooilley-niartal ‘almighty’ not given in full by Thomson (1953). 75 This may represent a late borrowing of English penance, or remodelling under its influence, rather than retention of G. peanas. 74 115 Aspects of the historical phonology of Manx stiagh ia (10), ía, ya /stʲax/ isteach in trass, tress (Cr.) vaght (K., PC) é, e (9) /tras/, /tres/ treas third /vaxt/ i bhfeacht ever yiarragh ia /jarax/ dear- would say Table 10. ea /e/ > /o/ Phillips etymology English Boaldyn CM pronunciation /boːldәn/ Beal(l)taine May Day chiollagh /tʃolax/ teallach hearth /tʃon/, /tʃonә/ teann(adh) tight; press /foːl/ feall deceit chionn; chionney foall (K., PC) ia (6), ià, ĭă, a (4), ay gioal; giall, gialdyn /ɡʲoːl/, /ɡʲoːldәn/ geall pledge; promise glione ia (7), a (12), á; (yiall) iall (2), iáll, jall (pl.) glantínyn /ɡlʲoːn/ gleann valley joan a (2), à /dʒoːn/ deann dust /dʒolәs/ dealas greed jollys kione ia (12) /kʲoːn/ ceann head kionnaghey ia (10), a (6) /kʲonaxә/ ceannacht buy lhionney /lʲonә/ leanna beer (gen.) molg, mylg /molɡ/, /milɡ/ mealg milt /molaɡ/ buoy; satchel (K.) mollag molley a /molә/ meallóg, ScG. mealag meala molley a (5), à, o, áy, á /molә/ mealladh deceive mongey /moŋә/ meangadh smile; shear lhuss-nymoal-moirrey (B.), lus ny moyl Moirrey (Cr.), luss-nymoal-moirree (K.)76 ollagh /lus nә moːl morʲә/ ScG. lus nam meall móra mallows /olax/ eallach cattle ollaghan /olәxan/ treadle of spade ollay /olә/ (e)alchaing, ScG. ealachainn eala polley; pollan /polә/ pealladh; peallán mat, stick together; saddle-cloth 76 a a (5), nan ialagh honey (gen.) swan Apparently with reanalysis / folk etymology involving Moirrey ‘Mary’ (G. Moire). 116 Aspects of the historical phonology of Manx shoggyl /ʃoɡәl/77 spollag /spolag/ seagal, ScG. var. seogal, Ir. var. siogal spealóg rye chip 2.1.3.1 Analysis Ó Maolalaigh (1997: 277–89) investigates the development ea > /a/ in terms of following consonantal environment in Scottish dialects. Although his study looks at the prevalence of the development across dialects, rather than within the lexicon of a single variety, similar results obtain for Manx. Excluding /m/, for which there was only one token, the development to /a, o/ (and subsequent developments)78 occurred in 100% of items where the vowel is followed by the sonorants /L/, /N/, /R/ and the fricative /x/ — all (historically) velarized or velar consonants. The percentage of items exhibiting this development was also above 80% preceding the alveolar sonorants /r/ and /l/ (Table 11). There was also a very high percentage (77.4%) for /e/ > /a/ preceding the sibilant /s/, but, as argued below, this may reflect the fact that a large proportion of these items are vowel-initial. For the other consonants the picture is more mixed and there is no obvious pattern, and for /b/, /k/, /t/, /p/ and /m/ there are 5 or fewer items. According to Ó Maolalaigh (1997: 283), the reflex /a/ is most prevalent in Scottish dialects in the environments __ /N, l, x, L, R, rt/, ‘thus implying that the most conducive environment for the lowering of original //e// to /a/ in ScG has been before velarised consonants and the velar fricative /x/’ (Ó Maolalaigh 1997: 283). In view of the results in Table 11, this would appear to be also the case in Manx. Table 11. Incidence of ea /e/ > /e/ and /e/ > /a/, /o/ in the lexicon by following consonant e a, o total %e % a, o L 0 8 8 0.0% 100.0% m 0 2 2 0.0% 100.0% N 0 11 11 0.0% 100.0% R 0 8 8 0.0% 100.0% x 0 20 20 0.0% 100.0% r 4 19 23 17.4% 82.6% 77 Also /u/ in some place-names with spellings shuggle, shugil etc. (e.g. PNIM II: 180, III: 38, 235). Including further developments to /o/, /oː/, /au̯ / before /L, N/ (§§4.6.1.9, 4.6.1.13) and to /ɛː/ before /R/ (§4.6.1.10). 78 117 Aspects of the historical phonology of Manx l 3 14 17 17.6% 82.4% s 7 24 31 22.6% 77.4% p 1 2 3 33.3% 66.7% t 1 2 3 33.3% 66.7% ŋ 3 4 7 42.9% 57.1% ɡ 4 4 8 50.0% 50.0% n 7 5 12 58.3% 41.7% k 3 1 4 75.0% 25.0% b 4 1 5 80.0% 20.0% d 8 0 8 100.0% 0.0% In certain Ulster dialects ea may be realized as /e/ (Ó Baoill 1978: 303–5; Ó Dochartaigh 1987: 75–82; Ó Maolalaigh 1997: 244–5) before /ɡ, d, s, h/. Ó Dochartaigh suggests this gives a clue as to the conditioning factors of the lowering /e/ > /a/ which is the rule in other Irish dialects: One may presume that the historical change of /e/ to /a/ has come about through the increasing prominence of what must have been an a-like on-glide to the following neutral consonant. We might reasonably expect this glide to be most prominent in those circumstances where a sonorant consonant follows, that is consonants such as /l n r/ where the secondary articulation is of considerable auditory prominence and hence more capable of influencing the preceding vocalic element. This means that in the case of /d/ and /s/, these segments, with their fairly neutral secondary articulation, have preserved the low-mid front articulation of the vowel where it has been modified in the more sonorant environments. (Ó Dochartaigh 1987: 77) This is a reasonable hypothesis and would agree with the Manx data, where the sonorants /L l N R r/ have been noted as especially favouring lowering of preceding /e/ to /a/, /o/, and /d/ and /ɡ/ especially are among the following consonants favouring /e/. Although not a sonorant, /x/ likely also had a prominent on-glide, or rather formant transition, as seen also in diphthongization of /iː/ in keeagh /kʲiә̯x/ ‘breast’ (G. cíoch) and preservation of original diphthongs ia, ua before /x/ (§2.2.6). Ó Maolalaigh and Ó Dochartaigh do not examine preceding consonant (cf. Ó Dochartaigh 1987: 80), but for Manx at least this seems also to be relevant (Table 12). 118 Aspects of the historical phonology of Manx Table 12. Incidence of ea /e/ > /e/ and > /a/, /o/ in the lexicon by preceding consonant e a, o total %e % a, o dʒ 0 13 13 0.0% 100.0% h 0 2 2 0.0% 100.0% R 0 2 2 0.0% 100.0% tʃ 0 9 9 0.0% 100.0% v 0 1 1 0.0% 100.0% f 2 11 13 15.4% 84.6% vowel 3 16 19 15.8% 84.2% Lʲ 2 10 12 16.7% 83.3% ʃ 4 15 19 21.1% 78.9% kʲ 3 11 14 21.4% 78.6% ɡʲ 2 6 8 25.0% 75.0% lʲ 1 3 4 25.0% 75.0% p 1 3 4 25.0% 75.0% m 6 10 16 37.5% 62.5% b 5 4 9 55.6% 44.4% Nʲ 3 2 5 60.0% 40.0% r 12 5 17 70.6% 29.4% There is less of an obvious pattern when it comes to conditioning by preceding consonant. If categories with 5 or fewer items are excluded the following hierarchical ordering is in evidence for incidence of ea > /a, o/ by preceding consonant: 100% >80% >70% >60% >40% >20% dʒ, tʃ f, vowel, Lʲ kʲ, ɡʲ, ʃ m b r Most of the consonants in the top three percentage bands are palatalized or palatal. The consonants /m/, /b/ and /r/ (as well as /f/) are assumed to be non-palatalized in Manx80 It may be that the development /e/ to /a/ after /dʒ, tʃ, Lʲ, kʲ, ɡʲ, ʃ/ represents dissimilation 80 I.e. there is no palatalization contrast in labials (Jackson: 66) and /r/ was not palatalized in initial clusters (§4.2.1.2). 119 Aspects of the historical phonology of Manx between the palatal quality of the consonant and the height and backness of the following vowel, which can be explained as hypercorrection (e.g. Ohala 1981; 1993). The apparently anomalous position of initial vowels and /f/ can be explained by the presence of /Nʲ/ in the proclitic definite article (e.g. ayns yn arragh ‘in the spring’,81 ayns yn astyr ‘in the evening’, cf. ScG. anns an earrach, anns an fheasgar). The majority of items with preceding /Nʲ/ as part of the same morpheme (snieng, sniengan, sniessey) have /e/ (although sniessey has /a/ variants in Phillips and in Late Manx speech, HLSM II: 332), as against two (niart, sniaghtey) with /a/.82 This is too small a sample to come to any conclusions about /Nʲ/. It is likely, however, that all things being equal /Nʲ/ would favour lowering to /a/ as with other palatal(ized) consonants. It is perhaps significant that of the three items which have initial /e/ rather than /a/, two of them are the preposition eddyr ‘between’ (Ir. idir, ScG. eadar) and the adverb edyr ‘at all’ (ScG. idir, OIr. etir),83 which of course cannot be preceded by the article. The third item is the noun edd ‘nest’ (G. nead), which has misdivision of the initial /Nʲ/. The /e/ quality here may be due to the final /d/, which is the only following consonant which categorically conditions /e/ (in eight items). In addition, as far as preceding /f/ is concerned, 9 out of 11 of the items have a following consonant which is one of those which strongly favour /a/ (/L N R r/). Table 13 shows the combined effect of preceding consonant (slender or broad, with vowel and /f/ included under slender in accordance with the conclusions of the preceding paragraph) and following consonant (belonging to the set /L l N R r x/ or not). As can be seen, the combination of slender preceding consonant and following /L l N R r x/ strongly favours /a, o/, with almost 90% of items in this category showing this development. The only category with a majority of /e/ reflexes (74.2%) consists of those items in which both the preceding and following consonant consonant favour 81 The palatalized /nʲ/ is sometimes shown orthographically by prefixed ni- in e.g. yn niarragh (Exodus 34ː21). 82 The development of snieng (G. sniodh, sneadh) and sniengan (G. seangán) is irregular and the two items seem to have influenced one another. The highly nasal environment (nasal consonants on either side of the vowel) may furthermore have served to maintain (or restore) the mid height – compare raising of ea /e/ to /i/ below. Niart and sniaghtey both have following consonants /r/ and /x/ which categorically favour lowering to /a/. 83 Originally the preposition ‘between’ with 3sg. neuter pronoun (eDIL s.v. etir). 120 Aspects of the historical phonology of Manx retention of /e/. For the two other combinations /a, o/ is favoured in 85.7% and 76.6% of items respectively, although in the category /b, h, m, p, R, r, v/_/L l N R r x/ there are only seven items. We may tentatively conclude that the conditioning factors for the development /e/ > /a, o/ are stronger than those conditioning retention of /e/. Moreover, several of the conditioning factors for /a, o/ appear to be categorical (/L, l, N, R, r, x/_, _/tʃ, dʒ/, and #_ with the exception of eddyr, edyr, edd), whereas none of the conditioning environments for /e/ have /e/ in 100% of cases, except _/d/ (eight items, one of which, kied, G. cead, has <a> in Phillips). Table 13. Combined conditioning effect of preceding and following consonant on G. ea /e/ in Manx following consonant preceding consonant dʒ, f, gʲ, j, kʲ, Lʲ, lʲ, Nʲ, tʃ, ʃ, vowel b, h, m, p, R, r, v b, d, g, k, m, n, ŋ, p, s, t LlNRrx a, o 56 88.9% a, o 36 76.6% e 7 11.1% e 11 23.4% total 63 a, o 6 85.7% a, o 8 25.8% e 1 14.3% e 23 74.2% total 7 47 31 That the development ea /e/ > /a/ is of considerable antiquity, at least in some environments, is shown by the development -earr /eRː/ > /aRː/ > /aːr/ > /ɛːr/, where the development to /a/ must precede the lengthening before /R/ which is found in all modern Gaelic dialects (O’Rahilly: 50), as well as the Manx development /aː/ > /ɛː/ (§4.6.1). Orthographic evidence in the form of the appearance of the spelling <ea> in Gaelic, as well as evidence from Anglo-Norman spellings of Irish names, suggest that this development goes back to the thirteenth century or earlier (McManus 1994: 346– 7; McCone 1996: 141). 2.1.3.2 ea /e/ > /o/ All cases of rounding ea /e/ > /a/ > /o/ (or /oː/, /au̯ /) occur before /L, N, l/, apart from shoggyl (G. seagal), for which the velar /ɡ/ may offer a tentative explanation. Before /L/ rounding is categorical even when /L/ is medial, e.g. chiollagh ‘hearth’ (G. teallach), Boaldyn ‘May’ (G. Beal(l)taine), whereas with /N/ it is categorical only in 121 Aspects of the historical phonology of Manx monosyllables, e.g. chionn ‘tight, fast’ (G. teann), kione ‘head, end’ (G. kione), but bannaght ‘blessing’ (G. beannacht). In Phillips all these items have spellings indicating /a/, apart from one instance of molley ‘deceive’ (G. mealladh) with <o>. The development of rounding (and lengthening or diphthongization §4.6.1) is therefore a relatively recent development compared with the development /e/ > /a/, which was already the predominant reflex of G. ea in Phillips. It is assumed that forms in /o/ developed via /a/, and where there are by-forms, variation is between /a/ and /o/,84 except in a handful of cases (see shelg, chiolg; mylg, molg; mingey, mongey above and §2.1.3.3). 2.1.3.3 ea /e/ > /i/ A small number of items have /i/ from ea, mostly adjacent to a nasal consonant. This nasal conditioning is also found in Irish and Scottish dialects (Ó Maolalaigh 1997: 249, 290). The vocalism of shyrgaghey ‘wither’ (G. seargaghadh) may reflect the influence of inflected forms of the noun and adjective searg (unattested in Manx).85 Table 14. ea /e/ > /i/ Phillips CM pronunciation /iŋax/ etymology English eangach train of nets i (2) /miŋә/ meangadh pinch; purloin y (13), ý (2) /mixәnʲ/ mercy milt wither ingagh mingey; myngyraght (Cr.), mingyraght (K.) myghin mylg, molg /milɡ/, /molɡ/ ScG. meachainn, Ir. miochaire mealg shyrgaghey, shirkaghey (Cr.), shyrg (adj.) (K.) /ʃirɡaxә/ seargaghadh Cf. Ó Maolalaigh’s (1997: 291–2) discussion of the similar developments /e/ > /a/ > /o, ɔ/ in certain eastern Scottish dialects. 85 I.e. from a fossilized dative i siorg (hi seurc, i siurc, eDIL s.v. 1 serg) (or i seirg, i sirg if feminine as in Dinneen) or from genitive sirg as in Cf. fear sirg ‘consumptive’ (Ó Dónaill s.v. searg), ben sirg, fer siric etc. (eDIL). The eDIL entry includes an apparent example of reanalysis of sirg as an attributive adjective or noun, dia rob dall na bodar . . . no sirg (Trinity College Dublin MS 1336, 658c). A similar reanalysis of a genitive probably explains the Manx adjective gennish ‘barren, childless’ < bean gheanais ‘woman of chastity’. The simple form shyrg is only attested in Kelly (as an adjective). 84 122 Aspects of the historical phonology of Manx 2.1.3.4 Lexical diffusion As noted above, a number of items show variation between /e/ and /a, o/, especially in Early Manx. In some cases Phillips’ spellings are at variance with later evidence, or there is spelling variation in the dictionaries and Classical Manx texts, or variation is securely attested in spoken data (including at least one clear dialect isogloss in shenn ‘old’). Most of the items showing variation in the later language (breb ‘kick’, G. breab; breck ‘speckled; trout’, G. breac; mestey ‘mix’, G. meascadh; trass ‘third’, G. treas) belong to the set of items with two conditioning factors for /e/, and except in the case of trass, <e> is the standard or most frequent spelling. The fact that these items have variants with /a/ is further evidence that the development to /a/ is the dominant reflex of G. ea, and that there has been ongoing lexical diffusion in this direction, even in the environments most resistant to the development. The occurrence of forms in Phillips apparently showing /e/ after /Lʲ/, and in the case of the initial vowel in agglish ‘church’ (G. eaglais), may show that the change was not as well established in these environments at this period, while freggyrt ‘answer’ (G. freagairt), gien ‘cheer’ (G. gean) and kied ‘permission’ (G. cead) with /a/ show more progressive forms not found in the later standard language (cf. ooashley, §3.4.6). 2.1.3.5 Semantic splits between /e/ and /a/ variants In a few cases the different reflexes of G. ea in the same or related etymological items have developed differing meanings,86 e.g. mastey, maskey ‘among’ (G. i measc), but mestey ‘mix’ (G. meascadh). Mastey is also found in Cregeen and the Bible (Proverbs 27:22) for the verb, apparently with the specialized meaning of ‘churn’ (perhaps influenced by G. maistreadh, although this is unattested in Manx). Cregeen apparently attests to a split between breck ‘speckled’ and brack ‘trout’ (G. breac). Kerroo ‘quarter’ has /e/ but kiarroo ‘fourth’ has /a/ (both G. ceathramhadh). It is likely that both mingey ‘pinch’ and mongey ‘smirk’ are reflexes of G. meangadh. If Ph. jeru 86 For this phenomenon, cf. Dillon (1953). 123 Aspects of the historical phonology of Manx represents dearbhadh ‘confirm’ (§2.1.3, fn. 72), this would be a split with jarroo ‘very, indeed’, G. dearbh. Such semantic splits are consistent with lexical diffusion. 2.1.4 ei /e/ > /e/, (/i/) This is mainly retained as /e/, including notably in certain items where raising to /i/ is widespread in Irish and Scottish dialects (Ó Maolalaigh 1997: 249–52, 292–3), such as: mec ‘sons’, OIr. maicc, G. meic, Ir.,ScG. mic mennick ‘often’, G. meinic Ir. minic, ScG. minig meshtey ‘drunkenness’, G. meisce, Ir. meisce, misce, ScG. misg (raising to /i/ apart from in Munster [Ó Maolalaigh 1997: 249]) gennee ‘germ’, gientyn ‘conceive’, G. gein, Ir, ScG. gin For Manx meinn ‘meal’, Early Irish already shows variation between men and min (apparently under the influence of mi(o)n ‘small’, eDIL s.v. men, min), with the latter becoming the general Irish and Scottish Gaelic form. Cregeen’s form merre in merre-cheilley ‘s. f. deadness of wit or sense’ may represent an abstract form *meire rather than usual G. mire ‘madness’ (abstract noun from adj. mear), although this could also be G. meirbhe ‘dullness, weakness, folly’ or mairbhe ‘deadness’.87 Manx has /i/ for usual G. ei /e/ in jir ‘will say’, Ph. jirr (Ir. deir); forms with -dir- are sometimes found in Early Modern Irish, confirmed by rhyme (eDIL s.v. as-beir; Bergin 1946: 175–6). Bink ‘bench’ may represent Eng. dialectal bink (EDD), rather than being a raising of the ScG. form being. The spelling chingey-jee ‘ringworm’ (Cregeen; chenney-jee Kelly) (Ir. t(e)ine dhia(dh), ScG. teine-dé) may represent raising as in Ir. tine (from teine ‘fire’), or is perhaps a result of the destressed position. The vowel /e/ is maintained in other derivatives of teine, such as chen(n)ey taarnee ‘lightning flash’ (PC l. 456; Kelly s.v. chenney, tienney) (G. teine toirnighe), chentyn 87 It is unclear whether the final <e> is an exceptional representation of /ә/, or whether we have a form /merʲ/ with apocope. 124 Aspects of the historical phonology of Manx ‘flash(es)’ (Ezekiel 1:14; Cregeen s.v. chent), cheinjean ‘bonfire’ (Cregeen) (G. teinteán). 2.1.5 o /o/ > /o/, /a/, (/u/) G. o most commonly gives Manx /o/, but there is also lowering to /a/ in many items, as also found especially in Scottish and northern Irish dialects (Ó Maolalaigh 1997: 329). Table 15. o /o/ > /o/ Phillips boayrd boyn coayr coayr coggyl coghal (Cr.), coghyl (K.) colbagh colbey couyl (Bible), coll, cohll (Cr.), coull (K.) collagh commee connagh-nygiark conney cor (er chor erbee) corkey corp corrag corragh corran corree correy; correyder cosney cost, costal CM pronunciation /boːrd/ etymology English bord table, board /boːn/ /koːr/ /koːr/ /koɡәl/ /koxәl/ bonn corr corr cogal cochall heel odd heron, bittern, crane tares caul /kolbax/ /kolbә/ /koːl/, /kau̯ l/ colpach colpa, calpa coll heifer calf, body, trunk hazel /kolax/ /komi/ /ˌkonax nә ˈɡʲark/ collach comaidh conach stallion, boar common participation henbane /konә/ /erʲ ˌxor erʲˈbiː/ connadh cor gorse as fuel at all /korkә/ /korp/ /koraɡ/ corca, coirce corp ScG. corrag o (9) kurryder /korax/ /koran/ /kori/ /korә/ corrach corrán corraighe cortha oats body crook of hand, forefinger tottering; capricious sickle anger, angry sowing (gen.); sower o (7), u o /kosnә/ /kost/ cosnadh oy (3), óy, oỳ, oa, o (2) ŏy o win, earn cost 125 Aspects of the historical phonology of Manx crockan croghey cron cronnaghey cronk, crongan cront crossan crossey crottag croym; croymmey doyn (Bible), dhoan, dhone (both Cr.) doarn dobberan doccar dolley (Bible, Cr.), d(h)ulley (Cr., K.) donney dorraghey dorrys doss drogh droghad drolloo drommey droun, drone, droyn (Cr.); dronnagh fockle foddey; foddeeaght follaghey; folliaght follan 88 o (3), a, á o (4) /krokan/ /kroxә/ /kron/ /kronaxә/ /kroŋk/ crocán crochadh cron cro(th)naghadh cnoc crock, pitcher hang scar, stain discern, perceive hill /kront/ /krosan/ /krosә/ /krotaɡ/ /kroːm/, /kromә/ Eng. crosán crosadh ScG. crotag crom; cromadh knot coral crucify curlew stooped; stoop /doːn/ donn brown /doːrn/ /dobәran/ /dokәrʲ/ /dolә/ dorn dobrón docair, dochar88 dolaidh fist grief, mourning stress of labour lack /donә/ /dorәxә/ /dorәs/ /dos/ /drox/ /droxәd/ poor, mean, foolish dark door bunch, cluster bad bridge /drolu/ /dromә/ /droːn/; /dronax/ dona dorcha doras dos droch drocha(i)d, droichead drolamh droma dronn; dronnach o (11), ô o (8), a (2), ay, (foddeeaght ) a (2), o (4) a (23) /fokәl/ /fodә/ focal fada, EIr. fota word far, long; longing /folaxә/; /foli.axt/ folach, folaghadh hide; secret a /folan/ folláin, E.Ir. follán, fallán, ScG. fallain wholesome a, o o (4) o (14) (croymmey) o (3), ou, ú, óy, u, oy o (4), o o (4) o (9) o (11) o (4) o (12) pot hooks back (gen.) hump; humped These two items (eDIL s.v. doccair, dochor) seem to be confused in Phillips, where both medial <k, kk, ck> and <gh> are found without differentiation of sense (Thomson 1953: 195). As far as is known, forms with medial /x/ are later unattested. Note that medial lenition of both /k/ and /x/ could give [ɣ]. 126 Aspects of the historical phonology of Manx follym a (2) /folәm/ folt frogh gob goggan goll, doll (Ph.) gonnagh óo gorrym gortey; gortaghey hoght kurn (Cr.), curn (K.) lhome lhon lhong lhongey lhott logh loght (ny) lomarcan lomman lommyrt; loamrey lorg lostey moggyl moghey moghrey o (8), ó, óy, oy (2) (gonnit) o (3) o (4), ó o (2), oy, ó, oi (2) oy o (8) o (6), io, oy o (2), oy (5) y (10), o (3)90 o y (2), oy, a, u (2), ú o (5) o (3) o (2) mohlt mol mollagh monney moylley ó, o (3), ô, é, è o (26), ò, oa empty /folt/ /frox/ /ɡob/ /ɡoɡan/ /ɡol/ folamh, EIr. folomm89 folt Eng. frough gob gogán dol hair dry rotten beak, gob noggin, piggin go /ɡonax/ gonach sore /ɡorәm/ /ɡortә, ɡortaxә/ gorm gorta; gortaghadh blue famine; hurt /hoxt/ ocht eight /koːrn/, ?/kuːrn/ corn can /loːm/ /lon/ /loŋ/ /loŋә/ /lot/ /lox/ /loxt/ /lomәrkan/ lom lon long longadh lot loch locht (ina) lomracán bare blackbird ship meal wound lake crime alone /loman/ /lomәrt/ lomán lomairt; lomradh scorching wind shear; fleece /lorɡ/ lorg staff /lostә/ /moɡәl/ /moxә/ burn mesh early /monә/ loscadh mogall moch, EIr. i mocha91 mochthráth, mochéirghe92 molt mol moth(al)lach, ScG. molach monadh /molә/ moladh /mox(ә)rә/, /moxәrʲi/ /molt/ /mol/ /molax/ morning mutton nave, hub of wheel hairy, rough much, any, nothing; kind, sort, manner praise 89 O’Rahilly 1942b: 191–2. Phillips’ spellings with <y> perhaps represent a variant form with loim-, luim-, or simply o > /u/? 91 eDIL s.v. mocha, mucha. 92 These two forms are possibly blended in Manx (Lewin and Wheeler 2019: 92). 90 127 Aspects of the historical phonology of Manx noght Nollick (Cr.), Nodlick (SW), Ollick (Cr.); Ullick, Ullig (K.) oard oardaghey oarlagh obbal obbee obbyr oghlish ogh(y)rish olk ollan orrym, ort, orrin, orroo osney pobble pott poyll rockey roddan rolley scoarnagh scobbey scolbey scollag scoltey scrobbaghyn soccar, socker sock soddag sollys (er) son sonney soylley sponk spot sproght sthock surn, sorn 93 o (2) o (4) /noxt/ /no(d)lәkʲ/ anocht Nodlaig, Nollaig tonight Christmas o o (14), ó /oːrd/ /oːrdaxә/ /oːrlax/ /obal/ /obi/ /obәrʲ/ /oxlәʃ/ ord ordaghadh ordlach ob ubaidh obair ScG. achlais, EIr. ochsal fochras olc olann orm etc. hammer order inch refuse, deny charm, enchantment work armpit o (11) o o (10) o (6) o (7), ó o (2), ó o 16, ôŷ (2), oy o (3), ó o (10), ó (2) o o (4) a o (3) o (43), ō, oy, oȳ ou o o See O’Rahilly 1927: 22–3. 128 /ox(ә)rәʃ/ /olk/ /olan/ /orәm/, /ort/, /orәnʲ/, /oru/ /osnә/ /pobәl/ /pot/ /poːl/ /rokә/ /rodan/ /rolә/ /skoːrnax/ /skobә/ /skolbә/ /skolaɡ/ /skoltә/ /skrobaxәn/ bosom bad, evil wool on me etc. sigh people pot pool bulge, pucker, cockle rat roll, roller throat snack, repast chip, break shell boy, stripling split, burst crop of bird /sokәrʲ/ /sok/ /sodaɡ/ /solәs/ /(erʲ) ˈson/ osna pobal Eng. poll rocadh ScG. rodan rolla scornach ScG. sgobadh scolb scológ scoltadh scrobán, Sc sgrob, sgròban, Eng. crop socair soc sodóg solas ar son /sonә/ /solә/ /spoŋk/ /spot/ /sproxt/ /stok/ /soːrn/ sona soladh sponc spot sprocht93 stoc sorn happy enjoyment tinder spot vexation, sulks stock fire-place in kiln ease, leisure ploughshare cake, bannock bright for Aspects of the historical phonology of Manx tholtan (Cr., Bible), tultan (K.) thorran tobbyr (Bible, Cr.), tubbyr, tubbir (Cr.), tobbar, tubbar, tubbyr (K.) toghyr tonn, toayn, touyn (Hymns) torragh tost towl; thoylley troggal trome trostey (torr) torr̀ , (torradh) torry, ghorey o (3) /toltan/, /tultan/ *tolltán ruin /toran/ torr; torrán heap /tobәr/, /tubәr/ tobar laver, baptismal font /toxәr/ dowry wave óy, (pl.) onnyn (3) /ton/, /toːn/, /tau̯ n/ ScG. tocha(i)r, Ir. tachar, tochar tonn o (2) o (2), óy, oy o, óy, óu o (22), ó oy (6), ôŷ (2), oy̆ , ó, ou, u, ú, o, o o (9) /torax/ /tost/ /toːl/; /tolә/ /troɡal/ /troːm/ torrach tost toll; tolladh ScG. togail, trogail trom pregnant silence hole; pierce lift, raise heavy /trostә/ troscadh fast Table 16. o /o/ > /a/ Phillips etymology English bass CM pronunciation /bas/ bos, bas palm brasnag /brasnaɡ/ brosna faggot, firebrand /brasnaxә/ provoke brasnaghey o (6) caggey a (13) /kaɡә/ ScG. brosnachadh, G. brostaghadh copóg, capóg, ScG. capag cogadh callin a (17) /kalәnʲ/ colainn body cappan a (10) /kapan/ copán cup Cargys karús, karus (2), karrýus (3), karryûs /karɡәs/ Carghas, EIr. Corgus94 Lent cabbag /kabaɡ/ dock, bloodwort war, fight 94 This item is pronounced with unlenited /ɡ/ in northern Irish dialects (Dinneen s.v. carghas; Quiggin 1906: 138–9; Ó Súilleabháin 1953: 104). The CM form is presumably equivalent to this Ulster form Cargas, whereas Phillips’ form represents Carghas, with vocalized */әɣә/ > /uː/ and stress shift as in Munster Irish (represented by Dinneen as carraigheas, i.e. with /ˈiː/ rather than /ˈuː/). See similar by- 129 Aspects of the historical phonology of Manx casley, cochaslys95 cass a (9), á /kaslә/ cosmhail like a (3), o /kas/ cos, ScG. cas foot cassid a (7) /kasәdʲ/ castey a /kastә/ quell, defeat cayrn a (2) /kaːrn/ EIr. cosaít, G. casaoid cosc(adh), ScG. casg corn clag klaggyryght /klaɡ/ clog, ScG. clag bell, clock clagh a (12) N /klax/, S /klox/ /koːr kratax/ cloch, ScG. clach crotach stone /kraŋk/, /kroŋk/96 /daraɡ/ cnag knock *dorghóg fishing line OIr. dor, doraid + adóir fogas, ScG. fagas fochaid doorkeeper, porter coar-crattagh cronk, crank darrag darreyder (Cr.) faggys /darәderʲ/ a (5) /faɡәs/ faghid a (3) /faxәdʲ/ farennym97 /farenәm/ farrar(ey) /farәrʲә/ fasney /fasnә/ fasscadagh (Cr.) ?/faskәdax/98 fast, fastagh /fastax/ for-, ScG. far- + ainm foraire, ScG. faraire foscnamh, ScG. fasgnadh foscadh, ScG. fasgadh cf. G. foscadán, ScG. fasgadan, < OIr. foscatae fosc sap /sap/ sop fastee a (2) /fasti/ horn, trumpet snipe near disdain, mockery, contempt nickname wake, vigil winnow shelter umbrella quiet, pensive, modest wisp forms with pargys ‘paradise’, G. parrdhas (§5.1.4). Another example of non-initial /ɣ/ > /ɡ/ in Manx is cleigh ‘hedge’ (G. cladh), pronounced with a final stop by some speakers, e.g. kle̜ g HK, TC, kleG JK, JTK, but kløi JW, kle̜ i EKh, HK, klaːi NM (HLSM II: 84). 95 But cf. cossyllagh ‘middling’ (G. cosamhlach), and other cases of G. comh- (except chammah ‘as well’, G. chomh maith, and unstressed cha ‘so’, G. chomh, also cho, Cregeen. 96 ‘written in the Manks Scriptures cronk; but as crank is the sound used, and as cronk rather confounds it with cronk (hill), this is inserted’ (Cregeen). The only instance in HLSM (II: 100, s.v. crankal) has e from TT, presumably representing the form with /a/, /ã/. The Manx forms may represent conflation of G. cnag (itself likely borrowed from Eng. knack, knock or related Scandinavian forms, cf. LEIA s.v. cnag) with a later doublet borrowing of Eng. knock. 97 And other instances of this prefix (Cregeen s.v. far-chail). 98 Cregeen indicates stress on the second syllable (i.e. ?/faˈskɛːdax/), possibly erroneously (§5.1), since the cognate forms would lead us to expect initial stress. However, the word may have been altered under the influence of scaa ‘shade, shadow’ (G. scáth), cf. scaa-liaghee ‘umbrella’ (Cregeen). 130 Aspects of the historical phonology of Manx skabbag (Cr.) /skabaɡ/ scobóg, giobóg /tamaɡ/ tomóg ‘a lock or handful of green flax’ (Cr.) bush, thicket tramman /traman/ tromán elder tree wass /was/ abhus, a-bhos below, down thammag u 2.1.5.1 Conditioning environments for /o/ > /a/ According to Ó Maolalaigh (1997: 330–1), the change /o/ > /a/ in Irish is particularly prevalent following initial /f/, following velar consonants, and in initial position. In Scottish Gaelic the most frequent preceding consonantal environments are /f/, /k/ and /kr, kL/, in that order (ibid.: 350). The situation is similar in Manx. Of the items which show the development, 11 out of 32 (34.4%) have a preceding velar stop /k/, with a further 4 items showing /k/-initial clusters. 8 out of 32 items (25.0%) have initial /f/,99 making this the second-most important environment. The same pattern is apparent if we count all lexical items with either /o/ > /o/ or /o/ > /a/ by preceding consonant (or cluster), as shown in Table 17. Only sets with at least five items are shown. Again preceding /f/ and /k/ are shown to be the most conducive environments for the development /o/ > /a/. All other sets have zero items, or only one or two, with /a/. Table 17. Incidence of o /o/ > /o/, /a/ in the lexicon by preceding consonant (cluster) o a total %o %a f 6 8 14 42.9% 57.1% k 21 10 31 67.7% 32.3% d 9 2 11 81.8% 18.2% sk 5 1 6 83.3% 16.7% (s)kr 11 2 13 84.6% 15.4% s 8 1 9 88.9% 11.1% t 8 1 9 88.9% 11.1% dr 5 0 5 100.0% 0.0% ɡ 6 0 6 100.0% 0.0% l 12 0 12 100.0% 0.0% m 8 0 8 100.0% 0.0% # 12 0 12 100.0% 0.0% 99 Although three of these items, fastee, fast(agh), fasscadagh are etymologically related and may or may not have been synchronically semantically associated. 131 Aspects of the historical phonology of Manx The conditioning environments likely represent (a) dissimilation between labial /f/ and round /o/ and (b) dissimilation between velar /k/ and back /o/ (the latter assuming a non-back realization of /a/ as attested in Late Manx and in Scottish Gaelic and Ulster Irish) (cf. Ó Maolalaigh 1997: 331). With regard to following consonantal environment, Ó Maolalaigh (1997: 331, 350) finds that the most favourable environments are ‘r >> l, s >> L’ in Irish and ‘l >> r >> s, g’ in Scottish Gaelic. The most favourable environment is _/s/ in Manx (12 out of 32 items, 27.5%), followed by _/r/ (5 out of 32, 15.6%). 2.1.5.2 Diachronic development of /o/ > /a/ and Gaelic dialectological context The evidence of Phillips shows that for most items which are spelt with <a> in Classical Manx, the /a/ realization was already stable in Early Manx. Orthographic evidence from Middle and Early Modern Irish shows that the change /o/ > /a/ was established or underway by the end of the twelfth century (Ó Maolalaigh 1997: 376– 7), with orthographic variation being most frequent after /f/ and /k/: The similarity of environments for this change in Irish and ScG is striking and suggests that the change //o// > /a/ may be an old one, perhaps dating back to the so-called period of Common Gaelic. […] The geographical distribution of the change //o// > /a/ establishes an important isogloss which separates Munster from other Irish dialects and also from those of ScG: […] the lowering of //o// > /a/ is all but unknown in Munster dialects. This suggests clearly that the development had a northern locus. This provides us with yet another early phonological development separating northern from southern Gaelic dialects. (Ó Maolalaigh 1997: 364–5) The evidence presented above shows that Manx clearly fits in this northern dialect group with regard to this development. Brasnaghey ‘provoke’ (ScG. brosnachadh) consistently has <o> in Phillips, so the development to /a/ is presumably recent, and also lacks the obvious conditioning environments. Brasnag ‘firebrand’ (G. brosna) seems to show the same 132 Aspects of the historical phonology of Manx development.100 The development to /a/ may be the result of analogy, perhaps with G. bras ‘boastful, great, forceful’, although this item is not attested in Manx so far as is known. Cass ‘foot’ (G. cos) and thammag ‘bush’ (G. tomóg) have one instance each in Phillips101 with <o> and respectively. These may represent alternative developments, although palaeographical uncertainties concerning the copying of the vowels preclude firm conclusions from individual examples (cf. Thomson 1953: 11– 2). The appearance of <a> in follaghey, folliaght, follan and follym may simply be cases of the use of <al(l)> to represent /ol/, noted elsewhere in Phillips. Cayrn ‘horn, trumpet’, and curn ‘can’ may show a semantic split between realization of G. corn. The realization of clagh ‘stone’ (G. cloch, ScG. clach) with /o/ in southern Manx and /a/ in the north is recognized as a dialect shibboleth by Rhŷs (161) (cf. HLSM II: 80). 2.1.5.3 foddey ‘far, long’, G. fada, OIr. fota G. foda, fada ‘far, long’, OIr. fota, is the only item where the development /o/ > /a/ has occurred in all modern Irish and Scottish dialects (Ó Maolalaigh 1997: 365). However, Classical and Late Manx unexpectedly has /o/ in this item. Phillips’ spellings suggest variation between /a/ and /o/, however. It is possible that the development to fada occurred in Manx as in other Gaelic dialects, and that the occurrence of /o/ is in fact an instance of the secondary raising of /a/ > /o/ found in certain items, including after labials in foaddey ‘kindle’ (G. fadadh) and moddey ‘dog’ (G. madadh) (§2.1.1.2). Orthographic variation in these items in Phillips shows that this was a later change still 100 There may be parallel development of these items with substitution, especially in Scottish Gaelic, of /n/ for /t/ in brostaghadh, perhaps through metaphorical association of incitement and kindling of firewood, and/or by association or confusion between brosna, a poetic metre (eDIL s.v. brosna, brosnach), and brostaghadh ‘a (poetic) incitement’? The two may be etymologically related (MacBain s.v. brosdaich), although LEIA (s.v. brosnae, brostaid) suggests it is more likely that the latter is a borrowing. 101 In fact both instances of cass in Matthew 18:8, f232r in the Phillips MS, appear to have o, with the second instance emended to a in a superscript interpolation. Owing to inconsistency in marking the /s ~ ʃ/ contrast (Wheeler 2019: 4–5), Ph. dáa chos could potentially represent the historical dual dá chois, rather than dá chos / dá chas; however, the parallel dá láu shows no sign of slenderized láimh (evident in dative er lǽf stei ‘within’, G. ar láimh istigh). 133 Aspects of the historical phonology of Manx in progress in the seventeenth century, whereas the evidence discussed here shows that /o/ to /a/ is in general a much earlier development. The comparative sodjey (G. is foide) and the abstract noun fodjeeaght (G. *foididheacht) (with parallel foddeeaght, G. fadaidheacht) may have also had an effect in preserving or restoring /o/ in foddey. 2.1.5.4 /o/ > /u/ Spellings of certain items show raising of /o/ to /u/, but this development is variable; most such items have spellings with both and <o>. Table 18. o /o/ > /u/ Phillips cumir (Cr.) CM pronunciation /kumәrʲ/ etymology comair, cuimir monadh English ScG. close, concise, tidy, compact mountain (Ph.) muyne uy (2), u */munә/ (N)ollick (Cr.), Ullick (K.) thammag (Bible, Cr., NBHR), thummag (NBHR) thummid (Cr.), thummyd (K.) tobbyr (Bible, Cr.), tubbyr, tubbir (Cr.), tobbar, tubbar, tubbyr (K.)103 tooran, thurran (Cr.)104 o (4) /(n)o(d)lәkʲ/, /ulәkʲ/ /tamaɡ/; /tomaɡ/, /tumaɡ/102 /tumәd/ ScG. tomadach /tobәr/, /tubәr/ tobar laver, baptismal font /turan/ torrán, cf. ScG. turr ‘tomb, large heap’ (Dwelly) tonnóg, ScG. tunnag *tolltán round corn stack thunnag (Cr.), tonnag (K.) tholtan (Cr., Bible), tultan (K) u o (3) /tunaɡ/, /tonaɡ/ /toltan/, /tultan/ Nodlaig Christmas tomóg bush, thicket; shoal of fish (NBHR) bulk, size duck ruin Both reflexes attested in HLSM: /a/ tɑ̃ ̣məg, tɑ̃ ̣mɑg TC, tamagʹ (HB); /o/ to̜ mɑ ̣g (HK), (pl.) to̜ mɑgən. There is apparently a dialect split here between N /a/ (TC, HB), and southern /o/, /u/ (HK, TT, NBHR). 103 Influenced by Eng. ‘tub’, Manx tubbag? 104 A variant of thorran ‘heap’ (G. torrán), influenced by toor ‘tower’ (G. túr), as implied by Cregeen? 102 134 Aspects of the historical phonology of Manx The favourable environments for the development to /u/ appear to be adjacency to nasal consonants, labials, and after /t/. Note also spellings in Phillips in cosney ‘win, gain’ (G. cosnadh), croymmey ‘stoop’ (G. cromadh), lorg ‘staff’ (G. lorg) and trome ‘heavy’ (G. trom) (§2.1.5). Spellings with for usual <o> also appear in certain other texts, such as the manuscript of 1 and 2 Thessalonians, where we find buggey for boggey ‘joy’ (G. bogadh), uddagh etc. for oddagh (G. féad), lummyrkyn for lomarcan ‘alone’ (G. lomracán). 2.1.6 oi /o/ > /o/, /a/ OIr. oi is retained as /o/ in the majority of cases. Table 19. oi /o/ > /o/ Phillips etymology English coirrey CM pronunciation /korʲә/ coire coillar /kolʲar/ coiléar caldron, kettle, furnace collar conning o /konʲәnʲ/ coinín rabbit coshey o (2) /koʃә/ coise foot (gen.) coyin /konʲ/ coin greyhounds /kroʃ/ crois cross /dolʲi/ doiligh difficult /dorʲәnʲ/ doineann storm dronney /dronʲә/ droinne hump (gen.) sodjey /sodʒә/ is foide further, furthest fodjeeaght /fodʒi.axt/ *foididheacht /loʃtʲ/ loisc, loiscthe ‘the distance of the furthest arrow shot in archery, farness’ (Cr.) burn, burnt /ponʲar/ boy /skolʲ/ ScG. ponach, poineach scoil /ɡәn ˈtort/ toirt heed /tolʲu/, /tolʲtʃәnʲ/ toill, tuill deserve /totʲ/ toit the whole crosh o (8) doillee dorrin o (7), y losht o (5), oi (2), oy pohnnar (Cr.), ponniar (K.) schoill, schoillar (gyn-)tort o (3) toilliu, toilchin o (4), oi (6) 105 toit (K.) school 105 ‘the whole, as yn slane toit jeu’ (Kelly), cf. ScG. ‘Toit. Whole entire’ (Shaw 1780), and figurative use of ceo in Irish. 135 Aspects of the historical phonology of Manx toshiaght o (13), ó /toʃax(t)/ toiseach beginning toshtal, hoshtal o (6), oi, oy /toʃtʲal/ left troiddey ó, o (2) /trodʲ/ ScG. toisgeal, G. lámh shoiscéala (O’Rahilly 1927: 23) troid /troʃtʲ/ troisc fast trosht scold, chide, quarrel There are some cases of lowering and unrounding of oi to /a/, all following initial /k/, /kl/, /kr/ (Table 20). The conditioning environments are thus comparable to those for o discussed above. Table 20. o /o/ > /a/ Phillips cadjin a (5), ia CM pronunciation /kadʒәn/ cagliagh a (3) /kaɡʲlʲax/ coigríoch border caigney a /kaɡʲnʲә/ chew casherick a (9) /kaʃәrʲәkʲ/ cognamh, cognadh, ScG. caigneadh coisricthe casht ái /kaʃtʲ/ coisc quell, defeat clashtyn a (21), á (2) /klaʃtʲәn/ MIr., Ir. clois(t)in etc., ScG. claistinn cloigeann hear scalp croiceann skin etymology English toir < tabhair give, put, send (fut.) claiggin crackan /klaɡʲәnʲ/ kraghyn, kráckyn 106 ?/krakʲan/ etymology English coitcheann general holy There are also a few instances of oi > /e/: Table 21. oi /o/ > /e/ Phillips der107 CM pronunciation /derʲ/ erriu e (5), o (2) /erʲu/ oirbh on you kemmyrk kemmyrk (3), kymmyrk, kymmirk, kemmirk, kæmmirk (2) /kemәrʲkʲ/ coimirce, OIr. commairge refuge 106 107 Perhaps = variant crocann, cracann (eDIL s.v. croiccenn) with broad medial /k/. Probably influenced by the independent form ver (G. bheir). 136 Aspects of the historical phonology of Manx kómrick, komrick, kæmrick, kommirk, kémmirk perkyn (Cr.) /perʲkʲinʲ/ poircín, ScG. poircean scoilte porpoise sloinne, (vn.) sloinneadh, OIr. sloindemain stoirm surname skeilt, scelt éi, ái, ai sliennoo (see §4.4.6.2) slonniu, slonniú /slʲenu/ sterrym ey, y (3) /sterʲәm/ cloven storm 2.1.6.1 kemmyrk ‘refuge’, G. coimirce etc. This is found with com- in the earliest sources and later coim- (eDIL s.v. commairge), with palatalization of m apparently spreading from the cluster /rʲkʲ/ (also /rʲɣʲ/, /rʲç/). Phillips’ forms with both <o, ó> and <e, é, æ, y> may represent co-existence of both variants. The form with /e/ which survives in Classical Manx is more likely to represent oi than o, since there are no other cases of o > /e/. However, this raises the question of the existence of slender */mʲ/ (cf. Ó Maolalaigh 1997: 65–72) (see §1.7.6 for a brief discussion of slender labials more generally). The spellings with y may represent raising to /i/ or /u/, although there are other cases where Phillips apparently uses <y> for short /e/, such as sterrym here. We may raise the possibility that Manx had a short vowel /ә/, /ɤ/ analogous to that found in the larger vowel inventories of Scottish dialects, but the overall evidence for this does not seem strong, except perhaps for cases of shortening of original ao(i) (§3.9.11). 2.1.7 i, io /i/ The Gaelic vowel /i/, like /u/, is mostly stable in Manx. In the prepalatal environment it almost invariably, and before broad consonants (G. io) predominantly, remains as /i/. However, i and io are frequently differentiated in Manx spelling with mostly used for i /iCʲ/ and <y> often appearing for io /iC/. This may indicate (allophonic) backing of /i/ before broad consonants, or simply be a device to mark the quality of the following consonant. 137 Aspects of the historical phonology of Manx <y> brynneraght ‘flattery’ (Ir. brionnal), chymmey ‘compassion, pity’ (ScG. tioma, time), chymmylt ‘around; foreskin’ (timcheall), chymney ‘will, testament’ (tiomna), chymsaghey ‘gather’ (tiomsaghadh), chyndaa turn (tiontódh), chyrrys ‘tour, journey’ (turas), cryss ‘belt’ (crios), fynnican ‘egg-white’ (*fionnacán, cf. Ir. gealacán, ScG. fionnagan ‘crowberry’), fyn-ruy ‘brown-haired’ (fionnruadh), fynney ‘fur’ (fionnadh), fynneraght ‘cool breeze’ (fionnuaracht), fys ‘knowledge’ (fios), fysseree, fyssyree ‘foreknowledge, prescience’ (fiosraighe), jyst (K.), juist (Cr.) ‘dish’ (ScG. diosg), kimmagh (Bible), kymmagh (Cr.) ‘criminal’ (ciomach), kyndagh ‘guilty, because of’ (ciontach), kys ‘how’ (cionnas), kynn (K.) ‘love, affection’ (cion), myn ‘fine, small’ (mion), mynlagh ‘the fine of meal or flour’ (Cr.) (mionlach), mynnagh ‘guts’ (mionach), mynney ‘curse’ (mionnadh), mynthey ‘mint’ (mionta), myskid (Bible, Cr.), myskit (Cr.), miskid (K.) ‘malice’ (mioscais), shynney lhiam ‘I love’ (is ionmhainn liom), shynnagh ‘fox’ (sionnach), skyrraghtyn (Bible, Cr., K.), skirraghtyn (Cr.) ‘slip, slide’ (sciorradh), slyst ‘coast, border’ (slios), spyrryd ‘spirit’ (spiorad), yllagh (Bible, Cr.), ullagh (K.) ‘cry’ (iolach), ymmodee ‘many’ (iomad), ymree ‘behove’, ymmyrch ‘need’ (ScG. imir, iomair, cf. imirt, iomairt),108 ymmyrkey ‘bear, carry’ (iomchor, iompar), ymmyrt ‘row’ (iomramh), yndyr ‘graze’ (Ir. inbhear, ScG. ionaltair, ionaltradh), yngyn (Bible, K.), ingin (Cr.) ‘nail, hoof’ (ionga, iongain), yngyr (Cr., K.), ingyr (Cr.) ‘pus’ (iongar), ynnyd ‘place’ (ionad), ynrick ‘righteous, upright’ (ionraic), ynsaghey ‘learn, teach’ (ScG. ionnsachadh), yskid ‘shank, hough, ham’, (iosca(i)d, EIr. esca(i)t), scryss (Bible, Cr.), scriss (Cr.) ‘bark, peel, shaving’ (scrios) drig ‘drip’ (driog), gimmagh ‘lobster’ (giomach), grindeyr (Bible, Cr.), grinder (Cr., K.), grindeyr (Bible), grynder (Cr.) ‘mocker’ (?greann, Ir. gliondar, Eng. grin), imbagh ‘season’ (EIr. imbocht, imbach), pibbyr (Cr.), pebbyr (K.) ‘pepper’ (piobar); shimmey ‘many’ (is iomadh), skibbylt ‘nimble, light of foot’ (sciobalta), spinney (Cr.) ‘elasticity’ (spionnadh); yindys ‘wonder’ (iongantas) 2.1.7.1 io /i/ > /u/ In some items, spellings with seem to indicate that backing seems to have led to merger with /u/: Table 22. io /i/ > /u/ Phillips etymology English bluight CM pronunciation /blʲuxt/ bliocht milch burley /burlә/ biolar cress giucklagh /ɡʲuklax/ giolcach broom 108 See O’Rahilly 1931: 57. 138 Aspects of the historical phonology of Manx kiuttagh (Bible, Cr.), kyttagh (Cr.), kiutagh, kytagh (K.) jummal /kʲutax/ ciotach left-handed /dʒumal/ diomailt waste rught (K.) /ruxt/ riocht sprite, spirit shuilgey (Cr.) /ʃulɡә/ ScG. siolcadh nibble shutternee (Bible, Cr.), shussarnee,109 shutternee (K.) sluight /ʃutәrni/ seitreach, siotrach neigh, bray u (3), iu /slʲuxt/ sliocht smyrr /smur/ /ulәnʲ/, /ilәnʲ/ smior, OIr. smiur iothlainn progeny, descendants; amount marrow /urlә/ iolar smuir, smuirr110 uhllin (Cr.), yllin (K.)111 urley i y, ý stackyard, haggard eagle Spellings of jummal, smuir and urley in Phillips suggest these items retained /i/ in Early Manx. Late variation in spelling and attested pronunciations of certain items shows that the change was not settled. Conditioning factors appear to be adjacency to labial /m/ (rounding assimilation), velar or (formerly) velarized consonants /k, kʲ, ɡʲ, x, l or L, R, r/ (backness assimilation). 2.1.8 u, ui /u/ Before broad consonants G. u remains as /u/, e.g. muc ‘pig’, G. muc bun ‘bottom’, G. bun 109 For <ss> representing /t/ > [ð], a realization which can also represent underlying /s/, see §1.6.4, 1.6.5. smur (TC), smör (JW) (HLSM II: 420). It is unlikely that these /u/ forms represent continuation of OIr. iu, and the Phillips spellings suggests the regular development of G. io; possibly influenced by smua(i)s, smúsach (although unattested in Manx), smooirlagh ‘broken bits, fragments’ (Cregeen) (G. smúrlach). 111 Data from HLSM (II: 468) suggests /i/ and /u/ variants: ɔlinʹ (TC), o̤ lïn (JTK), u.lʹɪn (JN) (confusion with uillin ‘elbow’?), e̜ linʹ, ɪlɪnʹ (JK). 110 139 Aspects of the historical phonology of Manx lugh ‘mouse’, G. luch The vowels /u/ and /o/ are generally kept distinct, although variant spellings suggest there may have been a tendency towards (?near) merger in some lexical items or lects, especially in the direction /o/ > /u/ (§2.1.5.4). There are a couple of attested variants apparently showing /u/ > /o/, although the /o/ may in fact be original, as in loaghee ‘mice’ (G. luch, but historical oblique stem loch-112) (1 Samuel 6. 4, 5, 11, 18), alongside lugh(ee) elsewhere in the Bible, and Edward Faragher’s spelling coammal for usual cummal ‘dwell, hold’ (ScG. cumail, but G. congbháil) (Broderick 1981b: 141). Manx appears therefore to lack the more general tendency towards lowering of /u/ to /o/ found in Irish dialects (Ó Maolalaigh 1997: 379–83), and agrees with Scottish Gaelic in this respect (ibid.: 400). Before slender consonants G. ui usually remains as /u/, spelled <ui>, , <ooi>, agreeing again with Scottish Gaelic rather than Irish, e.g. cooid /kudʲ/ ‘part, goods’, G. cuid ooilley /ulʲu/ ‘all’, G. uile fuirraghtyn /furʲaxtәn/ ‘wait, stay’, G. fuireacht duillag /dulʲaɡ/ ‘leaf’, G. duilleóg tushtey /tuʃtʲә/‘understanding’, G. tuigse Certain items have spellings which may indicate ui > /(w)i/, although the interpretation is not always clear: Table 23. ui /u/ > /u/, /(w)i/ Phillips bwinnican cluinn, clynn, clinn 112 ŭy, ŭy̆ , uỳ, ui (10), uíí, úi, iu, uy (2) CM pronunciation /bwinʲәkan/ /klinʲ/, /klunʲ/ etymology English cf. G. buidheac(h)án, EIr. buidén, ScG. buidhean, buidheagan, Ir. buinne? cluin(n)- yolk hear Cf. goan /ɡoːn/ ‘words’ (G. goth-), singular goo /ɡuː/ ‘word’ (G. guth). 140 HLSM Aspects of the historical phonology of Manx döt EKh, de̜ tʹ HK, do̤ t JK, dïtʹᶴ NM driːm NM, driᵇm JK, drim TC, driːᵇm J:EK, J:JTK kʹipː, kʹip JK, kʹip JW dhyt ui (9), ûi (2) /ditʲ/, /dutʲ/ duit to you dreeym drym, dry̆ im, dryim, (pl.) drymmyn, drimmyn /driːm/ druim back kip i /kip/ whip mimmey114 i (3) /mimә/ ScG. cuip, Ir. fuip < Eng. whip?113 muime mane wiŋ TC /mulʲәn/ mong, muing, ScG. muing muileann mill /mulʲarʲ/ muilleóir miller mulʹinʹ TC, mulʹən JK, W:N, mïlʹjən W:N, mwɪlʹən J:EK, mulʹənən JTK, mo̤ lʹi, mulʹi EKh, mo̤ lʹən, mulʹən NM mülʹɛ, mulʹɛː HB /miʃ/, /muʃ/? *muis, ScG. mu about, about him, it /kwiɡʲal/ distaff mwing mwyllin /mwiŋʲ/ ui mwyllar mysh ymmish (5), immish (3), ymmysh quiggal godmother /kwiŋʲ/ sym quing (3), kuing i G. coigeal, coigeál, ScG. cuigeal cuing /sim/ suim sum toiggal ui (16), ŭi /tiɡʲal/ tuig- understand trimshey, trimshagh i (7), yi (2), y /trimʃә/, /trimʃax/ ? *truimse < trom sorrow, sad quing yoke ɪ JK, JTK, HB, NM, W:N, W:S, ᵘẹ, ø TC ï TT, ĩ JW In some cases it is not entirely clear whether the reflex should be analysed as /u/ or /wi/, and there may have been synchronic variation. Alongside bwinnican, mwing, 113 Possibly from Eng. ‘kip’ ‘[t]he hide of a young or small beast […], as used for leather’ (OED), rather than ‘whip’? 114 But cf. mummig ‘mother’, G. muime + -óg, Ph. mummug. 141 Aspects of the historical phonology of Manx mwyllin above, we have buinnagh ‘diarrhoea’ (G. buinneach) (?/bunʲax/, /bwinʲax/), muickey ‘pig’ (gen.) (G. muice) (?/mukʲә/, /mwikʲә), muinney ‘mesentery, entrails’ (G. muinne) (?/munʲә/, /mwinʲә/). 2.1.8.1 ui /u/ > /wa/ In certain items ui develops to /wa/, which is probably to be interpreted as dissimilation from roundness (/m/) and backness (/k/). In mwannal it seems to be associated with depalatalization of /nʲ/ (§4.4.3). Table 24. ui /u/ > /wa/ Phillips mwannal115 muínal, muynal CM pronunciation /mwanal/, /munal/ etymology English HLSM muinéal neck mɔnəl, mɔnɑ ̣l TC, mɔnɑl EKh, mwanəl J:EK, wanal EL, vɔne̜ l TC, mwanəl, mɑnəl W:S, wɑ̜̃nələn TC kwalʹən J:EK, kᵘölʹjən W:S, kwɑlʹənən TC kwelʹjɑɡ, kwelɔg, kwalʹjən W:N quallian, quaillan (both Bible) quaillag (Cr.), carchuillag (Cr.), carwhaillag (K.), char-chuilag (Bible) quellan /kwalʲan/ cuileán whelp, pup /kwalʲaɡ/, /karˈxulʲaɡ/ cuileóg, ScG. car-chuileag fly, gnat 2.1.9 Morphophonological alternations /a/, /o/ > /i/, /u/ Morphophonological alternations involving final palatalization (genitive, plural, comparative forms, abstract nouns) may give /u/ or /i/ from roots with /a/ (Table 25) or /o/ (Table 26) (cf. Ó Baoill 2012). Note that such alternations have spread to some 115 Bible MSS: mwonnal (Exodus 13:13), wonnal (1 Samuel 4:18), mionnallyn (Judges 5:30), mhonnallyn (Judges 8:21, 26). The variation of spellings and realizations in HLSM may point to two by-forms with /wa/ and /o/. 142 Aspects of the historical phonology of Manx loanwords (e.g. block, cront, pot, spot). The front /i/ is more frequent in forms deriving from stems with /a/ than /o/: Table 25. Morphophonological alternations /a/ > /i/, /u/ bart inflection / derivation comp. syrjey, abstr. yrjid pl. buirht boayl pl. buill caayr (K.), cayr (Cr.) car clag pl. khyr116 (Cr.) pl. khyr (Cr.) pl. cluig croan pl. cruin crann mast ẹː NM, HK edd pl. idd ad hat e̜ dʹ HK garroo comp. s’girroo pl. glick garbh rough glac pl. glish, gen. gle(i)sh (Cr.) pl. keiyt, kiyt (Cr.), kitt (K.), gen. chett (K.) pl. ligg glas hollow of hand lock cat cat lag hollow lag weak, feeble mart beef stalk stalk ard glack glass kayt lagg lhag mart stalk abstr. lhuiggid (SW) pl. muirt (K.), muihrt (Cr.) pl. sthilk (Cr.) Phillips (sg.) e (gen.) yi etymology English HLSM ard high (yrjid) öː TT, ö JW, HK, ï JK, jö TC Eng.? G. beart? ball burden place carr wagon, car car knot, twist clag bell, clock ui NM, o̜ i HK, o̤ JW, TC see §2.1.2.2 116 Apparently khyr = cairr /kir/, or /kirʲ/ if slender rhotic restored by analogy (cf. Ó Curnáin 2007: 222) (see also §4.2.4). 143 Aspects of the historical phonology of Manx Table 26. Morphophonological alternations /o/ > /u/, /i/ boayrd inflection / derivation pl. blhick (Cr.) pl. blhuid (Cr.) pl. buird bock bog block blod bolg bwoid (Cr.), boid (K.) cronk cront Colloo corp doarn gob gorrym lorg mohlt, molt olk pot, pott poyll 144 Phillips etymology English Eng.; Ir. bloc, ScG. ploc block blade bord table, board pl. buick boc comp. s’buiggey pl. builg bog gelding, hegoat soft bolg belly bod penis cnoc hill Eng. knot Norse kalfr Calf of Man corp body gen., pl. durn (Bible), duirn (Cr.) gen., pl. gib dorn fist gob abstr. girmid, gormid, gorrymid (Cr.) luirg gorm mouth, beak, point blue lorg staff, stave molt wether olc evil ui (2) pota pot ui poll pool pl. bwhid (Cr.), buid (K.) pl. croink, crink, cruink (all Bible, Cr.) pl. cruint gen. Chylloo, Keylliu (Cr.) pl. kirp ui ui (4) y, yi, i lyrgyn, lurgyn (2), lúrgyn pl. muihlt (Cr.), muilt (K.) pl. uilk pl. pooiyt (Bible, Cr.), gen. phuyt (Cr.) pl. puill HLSM ᵘi TC, wï JW ï HK, i NM, W:N, ü TC ɪ JW Aspects of the historical phonology of Manx purt pl. puirt (Cr.), purtyn (Bible) port harbour, port sock pl. sick (Cr.) soc ploughshare spot pl. spuitt spot stott stoyll pl. sthitt (Cr.) pl. stuill towl pl. tuill Eng. spot, cf. G. spota Eng. stot, Old Norse stútr Eng. stool, cf. G. stól toll hole ûː W:N trome comp. s’trimmey, abstr. trimmid trom heavy (trimmid) ö TC, ï JW, (s’trimmey) ï TT bullock, steer ste̜ tʹ HK stool, chair 2.1.10 OIr. air-, aurAs Ó Maolalaigh (2003c: 163) notes, ‘words formed from the preposition / preverb air ‘before, for’ are well known for the variation which they exhibit in the vocalic initial from the Old Irish period onwards […] Variation in such words includes vacillation, to varying degrees in individual cases, between ai-, au-, e-, i-, u-, o-’. See also Pedersen (1909: 339–40), GOI (497–9), Greene (1976: 41), Breatnach (1994: 231–2), McManus (1994: 346). In Modern Irish and Scottish Gaelic the reflex of this is usually /u/, or /o/ as a secondary development of /u/ (Ó Maolalaigh 2003c: 164), although there may be also /e/ or /a/, which in some cases may represent survival of earlier air- (ibid.: 165). Ó Maolalaigh (2003c: 167) cites ‘slim evidence’ from Manx: Based on the slim evidence which survives in Manx, a small set of words containing original air- yield a and e, e.g. aarloo ‘ready’ (<Old Irish airlam), arryltagh ‘willing’ (possibly a derivation of Old Irish erail), arrym ‘respect’ (<*airrim), essyn ‘door-post’ (< Old Irish airsa): see [HLSM II] s.v. aarloo, essyn, arryltagh). An intermediate development to /u/ in such instances is not generally supported elsewhere in Manx historical phonology. (Ó Maolalaigh 2003c: 167) In fact a few more items may be cited, which further confirms that a front realization, apparently usually ea- (confirmed by spellings of arryltagh and ayrn showing slender n in proclitics) was usual in Manx, with /u/ or /o/ found only in Phillips’ form 145 Aspects of the historical phonology of Manx úyrrymagh etc. (perhaps representing a semantic split with arrym(agh)), and orraghey (Phillips orghyr). Table 27. OIr. air-, aurPhillips CM pronunciation /eːrlu/, /eːrlaxә/ etymology English OIr. airlam, aurlam; G. urlamh, earlamh ready; prepare, cook arbyl /arbәl/ tail arral (Cr.) /aral/ OIr. erball, G. eireaball, earball, urball etc.118 OIr. eráil, Ir. foráil, uráil, ScG. earail ? OIr. eráil etc. aarloo; aarlaghey a (17), ern iarlaghy, yn iarlaghey117 arryltagh a /arәltax/ arrym a (9), á,119 úy, u (5), y (5)120 /arәm/ ayrn, y Niarn (CS) ay (3), áy (2), a (4), á, æa /aːrn/ earish ie (5), ié (8), e, iǽ, êi /irʲәʃ/121 essyn orraghey /eʂәn/ orghyr /orәxә/ OIr. *airraim, G. uirrim, oirrim, urraim, ScG. urram OIr. airrann, G. urrann, earrann, ScG. earrann iris < OIr. airisiu ‘history’, OIr. airis ‘foreknowledge, tryst’122 OIr. airsa, G. ursa(in) OIr. airchor, G. urchar offer willing respect, reverence part time, weather doorjamb (arrow) shot Another item which apparently contains *aur, *air is G. turas ‘journey’ (eDIL s.v. turus, var. teros, terus, tirus) (Pedersen 1913: 600; Ó Maolalaigh 1997: 656), which gives two variant forms in Manx: chyrrys /tʃirәs/ (Bible, Cr., K., PC), thurrys /turәs/ (Cr.), nan jurryssyn, my hyrrys (Phillips), tʹʃɪrəs, tʹʃerəs NM (HLSM II: 80). Again we see a Manx tendency to favour front vowels found in variants in the earlier languages, 117 Ph. <n ia> = /nʲa/ G. ea-? ‘Despite its containing original iar- / er-, it seems fairly clear from synchronic variation and earlier literary sources that eireaball joined the class of words with initial air- at an early stage’ (Ó Maolalaigh 2003c: 167). 119 <a> forms: ‘reverence, worship, honour, obey, obedient, humble, reverend’. 120 forms: ‘great, greater, greatest, chief’. 121 i. EK, ɪ NM, i: JK (HLSM II: 139). 122 See Greene 1962: 112. 118 146 Aspects of the historical phonology of Manx but usually giving way to /u/ in modern Irish and Scottish dialects. There is also chiarrey ‘dry spell’ (G. tearadh, turadh < OIr. turad, taurad) (§2.1.3). 2.2 Long vowels The following sections discuss phones which were long vowels or diphthongs from the Old Irish period onwards, excluding ao(i) and ua(i) which are discussed in Chapter 3. 2.2.1 Stressed final vowels in monosyllables Breatnach (2003) argues that Old Irish did not permit final stressed short vowels, and that items such as seo ‘this’, de ‘of him’ and te ‘hot’ which have, or may have, short vowels in Modern Irish (and Scottish Gaelic) had long vowels in the earlier language (see also GOI: 32; Green 1997: 71–3). In his discussion Breatnach cites Manx jeh ‘of him’ and j’ee ‘of her’ with long vowels (HLSM I: 62). It appears in fact that all such items have long vowels in Manx. Cheh ‘hot’ (G. te, ScG. teth), is spelled chée (2 instances), chæ in Phillips, and all instances of this item in HLSM (II: 74) have long [eː]. Shoh ‘this’ (G. seo) has both long and short realizations (as well as forms which seem to represent /ʃox/, cf. ScG. dialectal seothach etc.) (HLSM II: 405), but this was probably determined by the varying degree of stress on this item. Final stressed vowels arising from loss of final fricatives also seem to be compensatorily lengthened, as in the ath class discussed below (§2.2.4). It thus appears that Manx has preserved, or restored, the Old Irish constraint against stressed final short vowels. We might compare the constraint in certain Scottish Gaelic dialects which requires the insertion of an unhistorical [h] after word-final short stressed vowels in pausa, e.g. /du/ dubh ‘black’ > [duh] (cf. Ternes 2006: 66–71). These can be regarded as examples of the ‘bimoraic norm’ (cf. Iosad 2016b; §4.5.5.2). 147 Aspects of the historical phonology of Manx 2.2.2 Fronting of /aː, oː/ > /ɛː/ One of the most striking developments in Manx is that G. /oː/ from ó, ói (not eó) often merges with G. /aː/, both being realized as fronted /ɛː/123 and spelled in CM <aa>, <a>, <a_e>, <ay>, <ai>, <ae> etc. There are also items where /oː/ is retained, however, and the conditioning factors for this phonemic split are not immediately obvious. Similar developments are found elsewhere in the Gaelic world, e.g. Gairloch is dàcha ‘it is likely’ (is dòcha) and pàg ‘kiss’ (pòg) (Wentworth 2005: 840), and the homophonic semantic splits noted by Dillon (1953: 323) in Lewis: ba chòire ‘it would be more generous’, ba chàra dhuit ‘it would be more fitting for you, you should’, fàd ‘sod of turf’, fo’n fhòid ‘in the grave’. Many such items had variants with á in Early or Classical Irish (Jackson: 41; Dillon 1953: 324), and where these are recorded in eDIL they are noted in the tables below: In Middle Irish ō was apparently an open vowel tending towards ā, and the Grammatical Tracts permit varying forms for a number of words in modern bardic poetry (Dillon 1953: 324) In the lists below obviously relatively recent loanwords, likely to postdate the change, are excluded. Some of less certain antiquity (e.g. floag, sole, fload) are included which possibly should be excluded; on the other hand that there are some older loanwords such as braag (G. bróg) and sharvaant (G. searbhónta etc.) which have undergone the change. Included below (Table 28) are items having /oː/ in most other Gaelic dialects, but which have /uә̯/, /uː/, /ɨә̯/ in Manx (mooar, deayrtey, poosey), or which may have /oː/ in Manx where other dialects generally have /uә̯/ (see also §3.4.5). Items which 123 The symbolization /ɛː/ has been chosen as best reflecting the dominant pronunciation in Late Manx, which may even verge on [eː]. However, some of the terminal speakers show a lower realization of /ɛː/, more like [æː, aː], which may be dialectal (HLSM I: 160, III: 123–4), and it is unclear how high and fronted this vowel was in earlier periods (see discussion below, §§2.2.3, 2.2.4). 148 Aspects of the historical phonology of Manx generally have /uә̯/ or developments thereof in both Manx and other Gaelic varieties are not listed here (see §§3.4.3–6). Table 28. ó /oː/ > /oː/ etc. Phillips bugogue (Cr.), boggoge (K.) CM pronunciation /buˈɡoːɡ/ etymology English Eng. buck + -óg? cf. ScG. bòcaidh, mucag? ScG. burdag, cf. Ir. burdán ? buck-thorn berry, hip burdoge (K.) /burˈdoːɡ/ cologe (K.) /koˈloːɡ/ croe, croae /kroː/ cró, EIr. croa, crao etc. croag /kroːɡ/ ?crobhóg, ScG. cròg, crobhag (MacBain) Ir. dórtadh, doirteadh, ScG. dòrtadh, vn. < dofortai, cf. Ir. duartan Eng. float deayrtey iâ, îy, ie, iy (2), iê, ue, ýa (4), ýy, ya, yá, ýy, ye er-fload /dɨә̯rtә/ /erʲ floːd/ small fish, minnow ‘a party, a faction, a league’ (K.) sheep-pen; eye of needle, notch of arrow claw pour, spill afloat floag /floːɡ/ Eng. flock? flake fo /foː/ fo under; under him, it gamshoge (K.)124 graynoge (K.)125 mooar /ɡamˈʃoːɡ/ gaimse, gaimseóg buffoon, mimic /graˈnʲoːɡ/ gráin(n)eóg hedgehog /muә̯r/ mór big, great /noː/ nuadh, nódh new /oː/ ó, ua grandchild noa oe ua (5), úa (10), úy, úay; (mooarid ‘greatness’) óy, oa óo (6), oo (2), ó 124 This item is found only in Kelly’s dictionary, which could make it suspect (§1.6.8.1). The item is not found in the earlier manuscript of the dictionary (MNHL MS 1477), but is found in the later MNHL MS 1045–7, with a reference to ‘Ir.’ gamal ‘a fool’ (cf. Shaw, s.v. gamal ‘[a] fool, stupid person’) (this reference is lacking in the published version). Kelly thus does not seem to have been aware of Ir. gaimse ‘simpleton’ or diminutive gaimseog, which perhaps makes it more likely that the Manx form gamshoge is genuine. 125 Possibly from Shaw (1780). 149 Aspects of the historical phonology of Manx peajeog (Cr.), pitteog (K.) peeoge piteóg miser, churl a puny, petty, tiny thing /poːɡә/ Ir. pióg ‘pie’ (Ó Dónaill), ScG. pitheag ‘young girl’ (Faclair Beag) póca bag /poːnrә/ póna(i)r, pónra bean(s) /puːsә/ pósadh marry raun /roːn/ rón seal scoadey, scoidan sole /skoːdә/ scód, Norse skaut sloop, smack /soːl/ threshold stoamey, stooamey thoree /stoːmә/, /stuә̯mә/ /toːri/ ScG. sòla, Eng. sole stuamdha poagey /piˈtʲoːɡ/, /piˈdʲoːɡ/ /pi.ˈoːɡ/ o (2) poanrey poosey trocairys ú (6), u (3), úy (2), uy o (5), ó, ô tóraidhe, eDIL táir (s.v. tóir) comely /troːkәrʲәs/ trócaire highwayman (Cr.), ‘overbearing person’, ‘headstrong child’ (Moore et al. 1924, s.v. tory) pity CM pronunciation ?/boːi̯ .ax/, /buәi̯ .ax/126 etymology English ScG. bòidheach, buaidheach; bòidhchead beautiful; beauty /koːrʲ/ cóir, eDIL cáir agreeable, kind Table 29. ói /oː/ > /oː/ Phillips bwaagh; bwoyid coar croiaght áa, áy, ay, úoia, (bwoyid) óiy, ói (3), oíí (2), óii, oîî, oi, oi, óy ?/kroːi̯ .axt/, ?/krәi̯ .axt/, ?/krai̯ .axt/ 127 < cró, cf. EIr. croaigecht, ScG. incest HLSM (II: 51) bwɑ̜ːiɑ ̣x HK, bɑ ̣iax SK, bɔːiax, JK, bɔiax J:EK. There are various semantic developments from the basic meaning ‘blood’ (Greene 1983), but the sense of ‘incest’ appears to be attested only in Manx; cf. EIr. croöige ‘heir’ and the abstract noun croaigecht (eDIL), and ScG. croidheach ‘dowry’ (Dwelly). Owing to the fluctuation of vowel length in these forms in other Gaelic varieties (early hiatus, later ó and o) and the ambiguity of Manx <oi>, the vowel length and quality in the Manx form is uncertain. If short we might expect /ai̯ / (§§1.7.5, 3.9.1, 3.9.1.4), though the <oi> spelling would suggest ortherwise. As far as is known, the word is a hapax legomenon found only in Leviticus 18:17 (and thence in Cregeen, Kelly), where it is inserted in the manuscript by the reviser for the translator’s original olkys (AV ‘wickedness’). The reviser writes criaght, correcting himself with superscript o to the spelling in the printed text, which may lend support to the */ai̯ / interpretation. 126 127 150 Aspects of the historical phonology of Manx doaie; doaieagh foaid (Cr., K.), faaid (K.)128 froaish /doːi̯ /; /doːi̯ .ax/ /foːdʲ/, /fɛːdʲ/ /froːʃ/ croidheach ‘dowry’ dóigh; dóigheach; eDIL dáig fó(i)d state, condition; decent sod of turf, clod /ɡloːrʲ/ próis, Fr. / Eng. prouesse, prowess (eDIL, LEIA) glóir glory, glorify /moːnʲ/ móin peat, turf /oːnʃәr/ óinseach harlot oaisht /oːʃtʲ/ fóisc, E.Ir. óisc, ScG. óthaisg yearling sheep snoaid, snooid /snoːdʲ/, /snuːdʲ/? gloyr, gloyraghey moain óy (9), oy (4), oe, óa, ó (6), oainjyr áy stroin (pl.) stroanyn thoin egotism, swagger snood /stroːnʲ/ sró(i)n nose, nostril /toːnʲ/ tó(i)n bottom CM pronunciation /ɛːɡ/; /sɛː/ etymology English óg, E.Ir. comp. oa young; younger, -est diminutive suffix Table 30. ó /oː/ > /ɛː/ orthography Phillips aeg; saa áy (7), ay (2), áa, a (2) (faasaag) áy (1) -age, -aig, -eig e.g. caabaig (Cr.), cabbage (K.) faasaag imleig liehbage (Cr.), liabage (K.) sornaig airh, oar (Trad. Ballad)129 bayr ay (2), áy (3), áe, au (2)130 /ɛːɡ/ (?and /eːɡ/, §4.4.7.4) /kaˈbɛːɡ/ suffix -óg ScG. càbag a cheese, slab of tallow etc. /feˈsɛːɡ/ /imˈlʲɛːɡ/, ?/-eːɡ/ /lʲeˈbɛːɡ/ féasóg imleóg leadhbóg beard navel flounder, fluke /sorˈnɛːɡ/ /ɛːr/ (or /ɛːrʲ/?, §4.2.1.3) sornóg ór sewer, drain gold /bɛːr/ bóthar, eDIL báthar bróg road Ir. claochlódh, ScG. caochladh, change braag áy, ay (2) /brɛːɡ/ caghlaa á (7), a, ghyghláym /kaˈxlɛː/ shoe 128 According to Kelly, faaid is an ‘a[djective]’ (i.e. genitive) ‘belonging or pertaining to turf or sod; consisting of sods’. All instances in HLSM (II: 170) have /oː/ apart from fɛːd (TT). 129 Thomson 1960–62 II: 64. 130 Thomson (1953: 7) suggests that Phillips’ spelling aur could reflect Welsh aur. 151 Aspects of the historical phonology of Manx charbaa chyrbâŷt /tʃarˈbɛː/ chyndaa á (7), a (5), (fut.) chyndai, hyndáys, hyndáym, (pp.) tyndáit, tyndait â (5), á (2), a (4), /tʃinˈdɛː/ da daah (§2.2.4) E.Ir. coímchloud, claechládh etc. tearbódh, eDIL terba, -úd, -ód tiontódh, eDIL tin(n)tád wean, excommunicate turn /dɛː/ dó to him /dɛː/ singe; daah-cree heartburn glare áy (7), ay (5), aa /ɡlɛːr/ dóghadh, ScG. dòthadh, dàthadh, E.Ir. dóüd, dód glór, eDIL glár paag, paagey ay (3), a (4), á /pɛːɡ/ póg kiss Parlane ay /parˈlɛːn/ Bartholomew language sharmane áy, áa, a /ʃarˈmɛːn/ sharvaant ay (5), áy (4) /ʃarˈvɛːnt/ Partholón, eDIL Parthalán ScG. ròlaist, MacBain cf. rigmarole, or ?rail searmóin, seanmóir searbhónta áy (2), a, á /soˈnɛːs/ saobhnós arrogance, ambition CM pronunciation /ɛːɡʲәdʲ/ etymology English óige youth /kɛːrʲ/, /aˈɡɛːrʲ/ cóir, éagóir, EIr. co(a)ïr, eDIL cáir right; wrong faaid, foaid /fɛːdʲ/, /foːdʲ/ sod ?neaynin (K.) ?/nɛːnʲәnʲ/ fód, eDIL fát, fád, fáid nóinín, ScG. neòinean Tríonóid raalish ?/rɛːlәʃ/ 131 sonnaase loose, empty talk sermon servant Table 31. ói /oː/ > /ɛː/ orthography Phillips aegid á (2), a (5) cair; aggair áy (12), ay (9), ây, áe, a (2) Trinaid ái (3), ai (17) /triˈnɛːdʲ/ daisy Trinity Conditioning factors are not immediately obvious in this fairly small set of lexical items, although as noted by Jackson (41) /oː/ is maintained in a number of items where there is an adjacent nasal vowel, where there may also be raising to /uː/ or /uә̯/ (e.g. mooar ‘big’, G. mór; moo ‘more, most’, G. mó). It is likely that the development 131 Cf. connaase ‘disdain, contempt’ (Cr.)? 152 Aspects of the historical phonology of Manx originated in a split between mid-high */oː/ (in nasal environments and elsewhere) and mid-low */ɔː/, with the latter later merging with low /aː/, with both later fronted to /ɛː/. The development to /ɛː/ is also more common in items which have á variants in eDIL. Other factors which may be implicated in the maintenance of /oː/ rather than /ɛː/ include preservation of /oː/ in high-register items such as trocairys ‘mercy’, gloyr ‘glory’, oar ‘gold’ (but later airh) (cf. lack of diphthongization in high-register items with /eː/, §2.2.7); analogy with the English or Latin/Romance cognate in the case of gloyr ‘glory’; cases where /oː/ varies with /uә̯/, or may on the basis of other Gaelic dialects have shown such variation in the past, such as stoamey, stooamey /stoːmә/, /stuә̯mә/ ‘beautiful, comely’, noa ‘new’ (G. nuadh, nódh etc.), oe ‘grandchild’ (G. ó, ua). There is evidence of lexical diffusion in earlier oar ‘gold’ alongisde later airh and co-existence of variants in foaid, faaid ‘sod’, and the suffixes -age, -og(u)e , G. fó(i)d, and semantic split between cair ‘right’ and coar ‘kind, decent’ (both G. cóir). Among the items with the G. diminutive -óg it is unclear what, if anything, the small group of nouns with /oː/ may have in common (bugogue, burdoge, gamshoge, graynoge, peajoge, peeoge), apart from the presence of a labial consonant in the first syllable of most of them? 2.2.3 /ɛː/ and /eː/ It is clear that /eː/ (<G. é etc.), and /ɛː/ were generally kept distinct in Early and Classical Manx, since the latter is fairly consistently spelled <aa>, <a>, <a_e> etc. in CM (<a, á, ay, áy> in Phillips), while /eː/ is usually <ea, e, e_e> (Ph. <e, æ> etc.), although some orthographic units are ambiguous, such as <ay> (cf. mayl /mɛːl/ ‘rent’, G. mál, but maynrey /meːnrә/ ‘happy’, G. méanar). A merger or near-merger /ɛː/ > [eː] is reported before /rʲ/ (Rhŷs: 6; Marstrander: 64) (§4.2.1.3) as in nearey ‘shame’ (G. náire), where a grapheme usually representing /eː/ rather than expected /ɛː/ appears; but expected <aa> is found in the adjective naareydagh ‘shameful’ (also neareydagh). For another possible case of interaction between /eː/ and /ɛː/, see §4.4.7.4. 153 Aspects of the historical phonology of Manx According to Jackson (24–5), /eː/ and /ɛː/ are contrastive in his informants, and this is the interpretation of Marstrander (62–4) also with the exception of the environment _/rʲ/ noted above. Rhŷs’s (3–4, 6) descriptions also support the existence of this contrast. However, Broderick (HLSM III: 50) claims there has been a (recent) merger in Late Manx: some ex[amples] show only allophonic variation indicative of close /eː/, while others have allophonic variants indicative of open /eː/. That is to say, that the pattern suggests a merging of two phonemes, i.e. of /eː/ and /ɛː/ into one, which I write here as /eː/. There is now no phonemic opposition between /eː/ and /ɛː/. (HLSM III: 50) Aside from the continued lexical conditioning noted by Broderick himself, an instrumental analysis of the speakers recorded by the Irish Folklore Commission suggests that there is a clear height difference between reflexes of CM /eː/ and /ɛː/ (for further details see §3.7). Impressionistically, the three vowels /iː/, /eː/ and /ɛː/ seem to the present author’s ear to be bunched quite tightly together in the front high area of the vowel space, which may explain the perception of merger. Broderick’s claim that /ɛː/ (and /әː/, /ɨː/) were no longer distinct in Terminal Manx has been accepted uncritically by some subsequent scholars. For example, Green (1997: 45) sketches a system of five long vowels for ‘Manx at the time of its extinction’, based on Jackson, Broderick (HLSM III; 1993) and Williams (1994b), rather than the seven or eight contrastive vowels which should be posited. Williams (1994b: 709) posits a tendency towards merger of /eː/ and /ɛː/, but nevertheless concludes they remained distinct in Late Manx. 2.2.4 a(i)th /ah/ > /aː/ Although historical /aː/ becomes /ɛː/ in Manx, /aː/ is reintroduced by the vocalization of final /h/ (OIr. /θ/, /θʲ/) in monosyllables in -a(i)th (cf. §2.2.1). 154 Aspects of the historical phonology of Manx Table 32. a(i)th /ah/ > /aː/ Phillips CM pronunciation /aː/, /naː/ etymology English HLSM ath, an ath /blaː/ ɑ ̣ː TC, naː NM, nɛː HK aː J:NM /braː/ ScG. blàth, G. bláith, E.Ir. mláith brath re-, second, next warm betray, reveal ɛː JW /kaː/ cath battle /daː/ dath, dathadh colour, dye ɛː TC, JW /ɡaː/ gath sting aː EKh, ɛː JW /raː/ rath success, prosperity scraa (Cr.), scrah (K.) /skraː/ scraith scraw, layer of sods on roof skraː JTK, skrɛː JW, pl. skrɛːɣən JW skah (Bible, Cr.), scah (Cr.), scaaghey (K.) /skaː/ scathadh, scoitheadh earmark; strong wind132 skʹɛː W:N aa, nah ná (4), na (2), náa (3) blah brah brá (3), bra, bráh (3), da vrâsyn, da vrǽsyn cah daah gah raah dâ, (daahghey) daghy, daghit gah, gagh, (pl.) gághyn rágh, rá, răh, rah, ráh (2) Blah ‘warm’ (G. mláith, bláith, ScG. blàth) is included here since it appears to have assimilated to this set of lexical items (i.e. > *blath), perhaps to distinguish it from blaa ‘flower’ (G. bláth), as suggested by Broderick (HLSM III: 124). The example of blah and blaa, and the similar spellings in Phillips for the ath set and for reflexes of á (and ó), as well as certain items in the CM orthography, and the overlapping pronunciations in some of the terminal speakers, suggest that there may have been no great phonetic contrast between /ɛː/ and the new /aː/, at least for some speakers. The use of <a(a)> for this vowel in the orthographies presumably reflects the 132 ‘a mark in the ear of sheep; a strong wind that sheds or shakes corn or fruit’ (Cregeen). 155 Aspects of the historical phonology of Manx situation in Early Modern English around the year 1600, when conservative [æː] and progressive [ɛː] realizations of Middle English /aː/ (<a>, <a_e> etc.) seem to have coexisted (Lass 2000: 83–4). The vowel [ɛː] in daah ‘colour’ (G. dath) may represent interference from daah ‘singe’ (G. dóghadh) (§2.2.2). 2.2.5 /oː/ and */ɔː/ According to Broderick (1999: 83; HLSM III: 54), a contrast between high-mid and low-mid back rounded vowels may also have existed in Classical Manx: As with original /eː/ and /ɛː/ more close realisations are restricted to some and more open to other items, suggesting two former contrasting phonemes /oː/ and /ɔː/ that could have given the following near minimal pair: boayl */boːl/ ‘place, spot’ G. ball : Boaldyn */bɔːldәn/ ‘May’ G. Bealltuinn. now realised as: boayl /b[oː]l/, /b[ɔː]l/, /b[ɔː]l/ Boaldyn /b[ɔː]ldәn/, /b[ɑ ̣ː]ldәn/, though both orig. /a/ + /l/. (Broderick 1999: 83) However, it is not clear why G. (e)all should be expected to give differing realizations in these items. If anything a contrast might be expected between reflexes of */aN/, */aL/ on the one hand and reflexes of historical /oː/ (and compensatorily lengthened /o/) on the other, but there seems to be no strong orthographic or other evidence for such a contrast. A comment by Rhŷs (1–2) hints that /ɔː/ may have been (re)introduced via English borrowings, but no details or examples are given:133 133 Perhaps e.g. walkal ‘walk’? 156 Aspects of the historical phonology of Manx o […] occurs pretty often both long and open, resembling the pronunciation of a or aw in the English words ‘all’ and ‘drawl;’ but this sound in Manx is recruited largely from borrowed words. (Rhŷs: 1–2)134 For the purposes of this thesis only one long vowel /oː/ will be assumed. 2.2.6 Monophthongization of /iə̯, ɨə̯, uə̯/ > /iː, ɨː, uː/ As noted by Jackson (50–1), Thomson (1976: 260–1) and Broderick (HLSM III: 139), the G. diphthong ia [ia̯ , iә̯], is most commonly realized as monophthongal [iː] in Late Manx (as is secondary /iә̯/ from G. éa /eː/, §2.2.7), thus merging with historical /iː/. Similarly, where the back quality of G. ua [ua, uә̯] is retained, this is smoothed to monophthongal /uː/ (Jackson: 53; Thomson 1963: 67; HLSM III: 139). However, in the Phillips and Classical Manx orthographies these are generally clearly distinguished: G. ia(i) í(o) ua(i) ú(i) Phillips <ia, ie> <i, ii> <ua> <u, uy> CM <eea, ia> <ee, eey, eei> <ooa, ua> <oo, ooy, ooi> By the early nineteenth century, however, monophthongal realizations appear to be indicated in certain spellings in Cregeen’s dictionary, such as keeir ‘dark’ (G. ciar); lheegh ‘ladle’ (G. liach), and hooir ‘forebode, threatened’ (G. tuar) (§1.6.5). Early indication of this development may be evident in the transcriptions for Edward Lhuyd made at the turn of the eighteenth century, such as phegil for CM feeackle, Ph. feakylyn, fiekylyn etc, ‘tooth’ (G. fiacal) (Thomson 1999: 395). The development of the fronted reflex of G. ua /ɨә̯/ is less clear, since both it and monophthongal G. ao /әː/ are represented indiscriminately by spellings such as <eay, ea, eo> etc. in Classical Manx (§3.6.2). However, they are more clearly distinguished 134 In the section dedicated to ‘Open o, long’, Rhŷs (9–10) gives only shoh ‘this’ (G. seo), ro ‘too’ (G. ró), and items with secondary lengthening of native short /o/, such as boght ‘poor’ (G. bocht) and boggey ‘joy’ (G. bogadh). 157 Aspects of the historical phonology of Manx in Phillips (ua = <ya> etc., ao = <y, yy> etc.) (§3.6.1), and the evidence of Rhŷs and his successors shows that they remained contrastive in Late Manx, with /ɨә̯/ tending to be smoothed to [ɨː] in line with the development of /iә̯/ and /uә̯/ (§3.5.1.3). Diphthongal realizations may be maintained, however, (a) before /x/, as in creagh ‘stack’ (G. cruach) (krɪːᵃx, Jackson: 53), jeeagh ‘look’ (G. déach) (dʒiːᵃx, Jackson: 32)135 and (b) when the diphthong synchronically occurs before a word boundary as a result of loss of final /h/, as in lheeah ‘grey’ (G. liath); theay ‘common people’ (G. tuath). In jee ‘god’ (G. dia), the historical word-final diphthong has been smoothed to /iː/ even in Phillips’ time (jih etc.),136 as also in the saint / parish name Malew /mәˈluː/ (G. MoLua) (PNIM VI: 143); but cf. booa /buә̯/ ‘cow’ (G. bó). In the case of following /x/, the CM orthography appears to indicate the development of a glide in keeagh ‘breast’ (G. cíoch) although HLSM (II: 244) has only iː (TT, JW). 2.2.7 Breaking of éa /eː/ > /iə̯/ The ‘breaking’ of long /eː/ preceding historically broad consonants to a diphthong /iә̯/ is found in a number of items in Classical Manx (with subsequent monophthongization to /iː/ in Late Manx).137 A similar development is well-known in Munster Irish (Ua Súilleabháin 1994: 482–3) and northern or central dialects of Scottish Gaelic (Jackson 1968), as well as in certain items in Ulster Irish (O’Rahilly: 194). In Manx it is not usually found in final position, whether original (so shey ‘six’ /ʃeː/, cf. northern ScG. 135 But cf. lheegh ‘ladle’ above, which appears to show monophthongization of G. liach. Although the diphthongal realization is apparently preserved in the spelling Yieeah (PC l. 580) and Dy bishee jeeah shiu ‘God prosper you’ (Kelly s.v. bishaghey). 137 According to Grannd (2000: 55) ‘[i]n Manx, according to Jackson 1955, this diphthongization does not occur and even the vowel in the word ceud, which seems to be realized as a diphthong everywhere in Gaelic Scotland, is realized in Manx as a long é’. This is clearly a misreading of Jackson (30–1), who in fact notes the two realizations of G. éa in Manx, comparing [iː] to the ‘i(ː)a (or iə)’ of Irish and Scottish dialects. As for kied ‘first’, notwithstanding Jackson’s kEːd (NM), kiːd (EK, JK), this seems to be usually reduced to a short vowel (HLSM I: 50), as indicated by the CM spelling (§5.5.1). 136 158 Aspects of the historical phonology of Manx sia /ʃia/) or via the loss of a final fricative (e.g. fea ‘quietness’ /feː/, cf. northern ScG. fèath /fia/).138 Table 33. éa /eː/ > /eː/ Phillips CM pronunciation /beːrl/ etymology English Béarla English language beasagh (Bible, Cr., K.), beysagh (Cr.) blakey /beːsax/ béasach compliant /blʲeːkә/ stare blean /blʲeːn/ /breːɡ/ spléachadh, ScG. spleuchdadh ScG. bleun, blian, Ir. bléin bréag /breːɡә/ bréagadh coax Baarle breag, breg éy, æy, ée, ǽ, é (2), éa (2), æ (2), e, a, áy, á breagey flank, groin lie cheh ée (2), æ (2) /tʃeː/ te, té hot cray æ (2) /kreː/ cré clay /kreː/ créadh creed /eːdaxә/, /e(ː)ˈdoːlʲ/ /edax/ (§§3.9.2, 5.5.1) /eːst/ éad jealousy éadach, ScG. aodach éasca clothes /edәn/ (§§3.9.2, 5.5.1) /feːr/ éadan, ScG. aodann féar face /faˈlʲeːs/, /faˈlʲeːʃ/ EIr. folés, ScG. faileas, faileus gleam of light /feː/; /feːax/, /feːx/ féath; féathach /ɡʲeːr/ géar rest, quietness; quiet sharp /ɡreːs/ gréas crea eadaghey; eadoil eaddagh æ, (eadoil) a (2) y (12) eayst íæ̂ , ía, ia, ea, ǽa (2) æ (2), e (4), y (7), é, ey éy (2), ǽy, ée eddin faiyr falleays (Cr., K.), failleish (Bible) fea; feagh geyre, gyere grease; greasee ée (4), ǽé139, e, æ, ee ǽy (2), éy, (yeyre) iǽ (2) moon grass ‘industry in making clothing’ (Cr.); shoemaker 138 One exception appears to be skeay, skeeah ‘vomit’ (G. sceith) (skiː TC, adj. skiːəx TC, skiːx JW HLSM II: 409). A further case may be meeaylys ‘fatness’ (Cr., Bible), meeaynlys ‘scum, fatness, grease; obtained by boiling flesh’ (K.), probably from G. méith, méath (itself found as Manx mea /meː/), as in yn joan jeant mea lesh meeaylys ‘their dust made fat with fatness’ (Isaiah 34:7), i.e. *méith|eamhl|as > *méal|as > *mialas with loss of perception of the first morpheme boundary. 139 Acute accent on each vowel character (Thomson 1953: 212). 159 Aspects of the historical phonology of Manx Jamys a (3) /dʒeːmәs/ Séamus James JeheineyCheays (Cr.), Jy-heiney Cheast (AG) jeig áy /dʒiˈhәːnʲә çeːst/ Aoine an Chéasta Good Friday ée (3), ee (3), éy, (yeig) iée, ie (4) iêê /dʒe(ː)ɡ/ déag teen /dʒiˈnʲeːr/ dinnéar dinner /kʲeːs/ buttock, ham /kʲed/, /çed/ ScG. ceus, eDIL ces 4, césán céad /çedәn/ céadna same /lʲeːs/ léas < Eng. lease /ˌlɛː̃ u mә ˈheːxt/ ?lámh um a h-éacht méar season (in ‘nightseason’) hand-suit140 jinnair kease kied, chied keddin, cheddin Ph. liǽs ie (6), iê, ei (5), êî e (3), æ (2), ie (3), ie iǽ, iæ̂ , ǽy, ǽa, éa, æ laue-my-height mair ée, éy (3) meydlagh raipey sheayney, sheaynt (Cr.); sheeaney, sheeant (K.) smeyr (Bible, K.), smair (Cr.) trean ay (3), áy (2), a (4), á (3) ǽy, ǽa, é, ǽ (3), e /meːr/ first finger /meːdlax/ (or /mәːdlax/?) (§3.9.2) /reːbә/ méadlach, ScG. maodalach slow-moving, unwieldy réabadh tear, rend /ʃeːnә/, /ʃiә̯nә/ séanadh, séanta bless; blessed /smeːr/ sméar /treːn/ tréan (black)berry, grape strong, valiant CM pronunciation /biә̯l/ etymology English béal mouth /jiә̯n/ éan, éanlaith young bird; birds /dʒiә̯xәn/ déachain look, show Table 34. éa /eː/ > /iə̯/ Phillips beeal eean, eeanlee jeeaghyn 140 ea (4), æa, éa (8), éy (3) ǽ (2), æ (2), ǽy (5), áy, êŷ, éy, ey a (4), ea (20), éa (2), á, (yeeagh-) ia (3), iá (4), yia (1), ía, a A legal term; see Thomson (1988: 141–2). 160 Aspects of the historical phonology of Manx keead ey (4), éy (3), ie (4), iæ, íe, iéy, ei keeaght /kʲiә̯d/ céad hundred /kʲiә̯xt/ céacht(a) plough skeeal ia /skʲiә̯l/ scéal story lheead îê, iæ, íæ /lʲiә̯d/ width lheeannee /lʲiә̯ni/ leithead, ScG. leud léana neeal, neealloo /nʲiә̯l/, /nʲiә̯lu/ néal(aibh) screeagh (Bible, Cr., K.), scragh (Cr.) /skriә̯x/ scréach meadow countenance; swoon, faint screech, scream Phillips’ spelling generally indicates /eː/ in those items where the later orthography consistently has <eea> for /iә̯/ (later /iː/). In a few cases such as skial ‘story’ (G. scéal, CM skeeal) it is possible that an early instance of /iә̯/ is indicated; however <a> may also represent /eː/ and the may indicate the palatalized nature of /kʲ/. Similarly, the spelling íæ̂ sk, íask, iask ‘moon’ (G. éasca, CM eayst) could possibly represent a form */iә̯sk/ (which would be more progressive than the later attested unbroken /eːst/), but it is likely that /eː/ is intended (especially in view of <æ̂ >), with marking the palatalization of the preceding n of the article. The breaking of /eː/ in Manx in those items in which it occurs thus appears to be a development of the seventeenth century, or at least, the conservative forms represented by Phillips’ spellings had become obsolete by the eighteenth century. The forms recorded in the data collected for Edward Lhuyd at the beginning of the eighteenth century correspond to the Classical Manx distribution, except that lèena seems to suggest /eː/ for later lheeannee ‘meadow’ (G. léana) (Thomson 1999: 394).141 That the unbroken form survived well into the eighteenth century is probably shown by the spelling lheantyn (editorially emended to lheayntyn) in the MS of Judges 20:33, which probably means /lʲeːntәn/, which is replaced by lheeantyn in the printed text.142 Apart from lheeanee, the only apparent examples of variation between /iә̯/ and /eː/ in the same item in Classical and Late Manx are sheayney (Cr., Bible), sheeaney (K.) ‘bless’ 141 Non-breaking: breag, cheh, cray, eayst, faiyr, geyre, lheeannee, mair. Breaking: beeal, eean, jeeagh, keeaght, neeal (Thomson 1999: 394). 142 This item is very common in place-names (PNIM), but usually spelled leany, leaney etc. which could be interpreted either way. 161 Aspects of the historical phonology of Manx (G. séanadh), and possibly screeagh ‘screech’ if Cregeen’s variant scragh means /skreːx/. In the Late Manx data there are realizations such as fiːɑ ̣ʳ (WQ, HLSM II: 157) for faiyr ‘grass’ /feːr/ (also fẹːəʳ JW, fẹːə NM, feːə JTK, fe̜ ːɑ̆ DC), and similarly for geyre ‘sharp’ (G. géar), but the pre-rhotic glide here is likely to be an artefact of English influence on the phonology of the terminal speakers (cf. Jackson: 118; HLSM III: 113) (§4.2.3). Conditioning factors for the breaking are not obvious, although most of the diphthongized items have following /l, n, x/, velar or (formerly) velarized consonants implicated in other vowel changes such as backing of ea /e/ > /a, o/ (§2.1.3). Register and/or lexical frequency may also play a role. The diphthongized items are mostly everyday terms of probable high frequency, whereas the two items with unbroken /eː/ followed by /n/, for example, are the probably less frequent trean ‘valiant, strong’ and blean ‘groin’, cf. non-breaking in ‘a loose group of ‘high register’ words’ in Scottish Gaelic (Bauer 2011: 362; cf. Robertson 1902: 89; Ó Maolalaigh 2001: 31) such as beus ‘manners’, ceusadh ‘crucify’, treun ‘strong’ (see also Dillon 1953: 322–3). Shortening of the vowel is found in certain of the unbroken items (§5.5.1), eaddagh (G. aodach), eddin (G. éadan), kied (G. céad ‘first’), keddin (G. céadna), jeig (G. déag), and may have predated the breaking. Eaddagh and eddin may additionally have the change éa > ao (as in Scottish Gaelic), and suggested by Phillips’s spellings with <y> (§3.9.2). There are also one or two presumably relatively recent (post-Great Vowel Shift) loanwords from English with eea, such as keeak ‘cake’ (Bible, Cregeen, Kelly)143 which may represent either the Manx development, or the similar diphthongization found in northern English dialects (EDD s.v. ‘cake’). 143 Spellings apparently showing monophthongal /eː/ are found in the Bible translation manuscripts, e.g. cheakyn (Leviticus 2: 4), cakyn (Leviticus 7: 12, 1 Kings 14: 3), kheak (Numbers 6: 19, emendation), keakyn (Judges 6: 19, emendation). 162 Aspects of the historical phonology of Manx Chapter 3 Manx reflexes of Gaelic ao(i) and ua(i) 3.1 Introduction The development of the Gaelic diphthong ua(i) (/ua̯ / or /uә̯/) and the long monophthong ao(i) */әː/ (<OIr. diphthongs /ai̯ /, /oi̯ /, /ui̯ /) is a complex and difficult area in Manx historical phonology.144 It is clear that Gaelic ua(i) in most lexical items has a fronted realization in Manx, that it is retained as back /uә̯/ (> LM /uː/) in others (spelled mostly <ooa> in the Classical Manx system), and is variable in still others (Jackson: 52–3; HLSM III: 139–40). The Classical Manx orthography tends to represent both ao(i) and the fronted reflex of ua(i) by a number of di- and trigraphs, especially <eay>, <ea>, <ey>, <eo>. Some of these orthographic sequences can also represent front vowels /eː/ (G. é), /ɛː/ (G. á). This appears to suggest that (fronted) ua and ao were pronounced similarly both to each other and to the front vowels. Phillips’ orthography tends to distinguish them more clearly, with <yy, y> for monophthongal ao and <ya>145 for diphthongal ua, but the use of <y> in both may suggest a degree of similarity between them (§3.6.1). 3.2 Historical development of G. ao(i) and ua(i) 3.2.1 ao(i) /əː/ G. ao(i) goes back to the Old Irish diphthongs áe, aí /ai̯ /, óe, oí /oi̯ / and uí /ui̯ /.146 The diphthongs /ai̯ / and /oi̯ /147 were in the process of merging already in the Old Irish 144 For a brief discussion of this topic, drawing on the research presented more fully here, see Lewin (2019a: 89–94). 145 And variants of the same with a range of optional diacritics. 146 The standardization of the placement of the diacritics here is an artefact of modern scholarship, intended to distinguish the diphthongs clearly from monophthongal ái /aːCʲ/ and ói /oːCʲ/ (Greene 1976: 26). Usage in the manuscripts is varied. The phonetic symbolization of these diphthongs is conjectural; the exact pronunciation is unknown. 147 /oi̯ / itself being a merger of earlier /oi̯ / and /ui̯ / (Cowgill 1967: 134–137; Greene 1976: 39–40). 163 Aspects of the historical phonology of Manx period, as shown by orthographic confusion (GOI: 42; Ó Máille 1910: 36; McCone 1996: 139), and seem to have become a long monophthong by the Middle Irish period (Pedersen 1909: 8; O’Rahilly: 31; Greene 1976: 40; Ó Murchú 1989b; McCone 1994: 92; Breatnach 1994: 233; Ó Maolalaigh 1997: 192), generally assumed to be some kind of central long vowel, here represented */әː/. This merged long monophthong comes to be spelt ao(i) in Classical Irish from the 15th century onwards (McManus 1994: 349), and in Modern Irish and Scottish Gaelic. There are also important interactions between reflexes of ao(i) */әː/ and agh, adh */aɣ/,148 especially in northern Irish, Scottish (and apparently Manx) dialects (Shaw 1971; Ó Maolalaigh 1997: 340; 2006; Lewin 2018). In modern varieties there are three main developments of ao(i) and agh (inc. adh etc.) (Lewin 2018: 159): (a) In southern Irish (Connacht and Munster) varieties ao(i) has merged with /eː/ and/or /iː/,149 while agh is usually a diphthong /әi̯ /. (b) In southern Scottish dialects, and apparently Manx, ao(i) remains a mid central vowel /әː/, may be fairly fronted and, according to some descriptions, may have weak rounding. Mergers with /eː/ and/or /iː/ are reported from certain speakers in Arran and Kintyre (O’Rahilly: 33; 709 1957: 8–10; 1962: 6–10; SGDS), and there is a tendency towards similar mergers in Late Manx, at least in aoi > /eː/ (§§3.4.2, 3.4.4, 3.5.1). Agh is also realized as /әː/ and thus merges with ao(i) (Dilworth 1996: 44; Ó Maolalaigh 1997: 227–228). (c) In Ulster and northern Scottish varieties ao(i) is realized as a high back unrounded vowel /ɯː/, which is contrastive with mid back unrounded /ɤː/ representing agh. In Ulster, both of these have tended to front and merge with /iː/ and /eː/ respectively, 148 Also eagh, eadh, ogh, odh. In Munster generally ao > /eː/, aoi > /iː/ (Ó Murchú 1989b: 144), while in Connacht both ao and aoi > /iː/. 149 164 Aspects of the historical phonology of Manx apparently fairly recently,150 but in northern Scotland they remain robustly distinct in most varieties.151 3.2.2 ua(i) /uə̯/ G. ua(i) derives from breaking of /oː/ in the Early Old Irish period (GOI: 39–41; McCone 1994: 89). There is a certain amount of interchange between ó and ua at all periods, including in Manx (§3.4.5). This diphthong is spelt ua(i) in modern Irish and Scottish and generally remains a diphthong [ua̯ ] or [uә̯]. It is, however, sometimes smoothed to [uː], [oː] in certain environments and lexical items in some Irish dialects (O’Rahilly 1932: 193; Quiggin 1906: 21; de Bhaldraithe 1945: 87; de Búrca 1958: 116; Mhac an Fhailigh 1968: 136; Ó Curnáin 2007: 91–2). In Scottish dialects there may be a phonemic split into /ua̯ / and /uә̯/ (Oftedal 1956: 97; Ó Maolalaigh 1997: 141). In Ulster ua(i) may be unrounded and fronted in certain environments (Quiggin 1906: 28–9; O’Rahilly 1932: 37; Wagner 1959: 77; Ó Dochartaigh 1987: 110–4). The similarity between this development and the fronted and unrounded reflex of ua(i) in Manx has been noted by scholars (Quiggin 1906: 29; Jackson: 52; HLSM III: 60; Ó Dochartaigh 1987: 112), but the conditioning factors seem to be quite different (§3.8). 3.3 Summary of Manx developments The main developments of these phones in Manx may be summarized as follows (Figure 10), according to the analysis in the present chapter: 150 For discussion of Ulster data see Ó Dochartaigh (1987: 114–21, 289–90). See also Ó Maolalaigh (1997: 670, 672), Maps 6, 8a, 8b. 151 For further discussion of the historical developments across Gaelic dialects, see Lewin (2018). 165 Aspects of the historical phonology of Manx Figure 10. Main developments of G. ua(i), ao(i), agh in Manx Gaelic ua ua̯ uai aoi әː ao әː agh aɣ > әɣ > әː Early / Classical Manx (C17–18) uә̯ ɨә̯ әː әː әː Late Manx (C19–20) әː uә̯, uː ɨә̯, ɨː әː әː әː (eː) Thus it is claimed that a distinct phoneme /әː/ survived throughout the attested periods of Manx, representing ao(i), agh and some reflexes of uai, with the proviso that a degree of allophonic variation gave rise to merger of /әː/, representing aoi and uai, to /eː/ in some speakers, as described by Rhŷs (25) (§3.5.1.4). The situation with original G. /ua̯ / is more complex. Some items retain back /uә̯/ (with a tendency towards monophthongization to /uː/ in Late Manx), in certain cases with by-forms with /oː/ (§3.4.5), while others show fronting to /ɨә̯/ (> /ɨː/) (§3.4.3) or /әː/ (> /eː/) (§3.4.4). Further items show both back and front by-forms, although one may be more commonly found than the other (§3.4.6). The regular development of uai with following slender consonant is /әː/, but there are exceptions where /ɨә̯/ is or may be found through paradigmatic analogy (cleaysh, G. clua(i)s; geayltyn, G. guailne) (§3.9.5) or other phonological developments (depalatalization of */rʲtʲ/ in keayrt, G. cuairt) (§3.9.6). Despite orthographical overlap, there seems to be little evidence of merger or confusion between /әː/ and /ɨә̯/, /ɨː/ (other than the above-mentioned aoi and uai items showing /әː/), although the development of a form /ɡɨә̯/ for geay ‘wind’ (G. gaoth) in certain dialects or idiolects (also expected /ɡәː/) is a notable exception (§3.9.3). 3.4 Lexical items with ao(i), ua(i), agh etc. The following tables (35–41) show most of the items with historical G. ao(i), ua(i), agh (etc.), with the exception of the new diphthongs arising from vocalized fricatives 166 Aspects of the historical phonology of Manx and certain other problematic lexical items which are dealt with in §3.9. 3.4.1 ao > /əː/ Table 35. ao /əː/ > /əː/ Phillips etymology English HLSM bleayst CM pronunciation /blәːst/ blaosc iːə TC, iː TT ceaghley /kәːxlә/ cleayn (Bible, Cr.), cleoyn (Cr.); cleayn(agh)ey /klәːn/, /klәːnә/ claochlódh, ScG. caochladh claonadh husk, eggshell change ĩː JW, iːə, iə̯ TC deyll (K.), tarroo-deyill (Cr.) deyr /dәːl/ daol enticement, propensity, seduction; entice, seduce, tempt beetle ýa (3), ýy (4), ýæ uæ, ý (2) /dәːr/ daor dear /dәːrә/ daoradh condemn eː, e̜ ː NM, ɛː JTK e̜ i JW ý (2), y, éy */dәːʂәn/ dearth /әːl/ ScG. daorsainn aol /әːnax/ aonach desert, waste /әːnәnʲ/ cf. ScG. aonan precipice /fәːn/ fao(i)n freoagh /frәːx/ fraoch wide, expansive, void, waste heather gaer /ɡәːr/ ScG. gaorr /ɡәː/, /ɡɨә̯/ gaoth; gaothach deyrey Ph. dysyn, cf. deyrsnys (Cr.) eayl eaynagh, eaynee eanin (Cr.), eaynin (Cr., K.) feayn geay; geayagh (§3.9.3) ýy yei, ýæi, ya (2), ýa (3), ýæ (3), ye, ua, úa lime short dung, ordure wind; windy ø, øː TC, yː, iː, uː, λː NM, eːⁱ SK iː HK, iːə JW, øː TC ᵘiː, ᵘi, iː (TC), iːə, iː JK, λːə, eː, eːə NM, λː HK, iː TCr, o̤ i, o-i W:S, ɪːᵊ J:JTK, HB 167 Aspects of the historical phonology of Manx inney-veyl éy, ýy (2), yy (2), ée /inʲә ˈvәː̃ l/ handmaid, servant girl rowan ö TC, iː TT iː JW, HK, NM, yː JW, uː HK i, y., ũ NM, yː, λː HK, i JW iː HK keirn /kәːrn/ inghean mhaol caorthann keyl /kәːl/ caol slender keyllys /kәːlәs/ caolas sound keynnagh /kәːnax/ caonach moss keyrrey, gen. keyrragh y (4), e (2) /kәːrә/, /kәːrax/ caora, caorach sheep kirree (§§3.911, 5.5.1) kyagh (K.)152 kirri, kiri /kiri/, ?/kirʲi/ sheep (pl.) /kәːx/ caoirigh, caoraigh, ScG. caoraich caoch leaum (Cr.), lheaym (K.) /lәːm/ laom meayl /mәːl/ maol one-eyed, blind sudden heavy shower of rain bald meayllee /mәːli/ maolaidhe hornless cow meoir /mәːr/ maor yː HK, NM, ɪ HK, iː NM, HK, e JK, ɛ i, iː NM, ï HK, ɪ TC ø TC e̜ ː TC, eː J:NM seihll; seihlt(agh) ýy (11), yy (5) /sәːl/, /sәːltax/ saoghal; saoghalta moar, bailiff, government officer world; worldly seyr y, ýy (3), éy (3), ey /sәːr/ saor free /sәːr/ saor carpenter /sәːrә/ saoradh acquit, free e̜ i TC, ẹː HK, JW, ẹː, eː, ëː, yː NM e̜ i JW /slәːdә/, /slәːdax/ slaodadh drag ø TC, ö JW seyir seyrey sleaydey; sleaydagh 152 ýy (3), ya eːⁱ, e̜ ː, JTK, öː, Eːⁱ, Eː, e̜ ː JK, o̜ ː, øː TC, Eː NM, e̜ ː EK, ɛː JːNM, eː JW, öː WːS ïː JW The spelling of this item is unexpected in the Classical Manx orthography (perhaps representing diphthongization before /x/?), and it occurs only in Kelly, but is present in the first draft of the dictionary (MNHL MS 1477) so may be genuine. The derivative kyaghan ‘mole’ (G. caochán), found only in Kelly’s later manuscript (MNHL MS 1045–47) and the printed dictionary, is more suspect, as stress shift would be expected in a heavy-heavy item. 168 Aspects of the historical phonology of Manx t(h)eaymey (Bible, K.), teaumey (Cr.) teaym /tәːm/ taomadh bail out water, teem /tәːm/ taom teayst /tәːst/ taos whim, fancy, fit dough teaystney /tәːsnә/ taosnadh ScG. aon un (§3.9.11) y (13), ý, ú (2), u (2) /un/ ɛː, e̜ ː HK, ẹː TT knead one oː JTK, ɛː, Eː JK, uː TK, o̜ ː EKh, u̜ ː J:JTK HLSM 3.4.2 aoi > /əː/ (>/eː/) Table 36. aoi > /əː/ (>/eː/) Phillips CM pronunciation /ɛːi (ɣ)әːlʲ/ etymology English áith aoil limekiln é (3), e (4) /dәːnʲә/ daoine men /әːʃ/ aois age eoylley yǽ, ýæ, ǽy, ey (2), êŷy, úe, ué ýy /әːlʲә/ aoileach dung feoilt(agh) ei (2), êy /fәːlʲtʲ/ faoilte(ach) generous feysht ey, yǽ (2), úea, uá, eay, úey (2), ýey, æy, ǽy, úe, uéy, yé, éye, éâ, ua, yey /fәːʃtʲ/ faoiside examine, question geaysh(t) /ɡәːʃ(t)/ gaoisid animal hair iː TC, ẹː JW Jecrean (Cr.), Jy-curain (K.) /dʒeˈkrәːnʲ/ Dia Céadaoin Wednesday ẹː WQ, HK, JW, ẹːⁱ NM aaie eayil, aaie gheayil deiney eash eː NM, JK, J:EK, e̜ ː JK, öː JK iː TK, HB, ẹː NM, JK ẹː JW, HK, eː TK fix (?) NM, fẹˈoːlʹtʹ, fẹˈoːlʹt (sic) TC 169 Aspects of the historical phonology of Manx Jeheiney (Cr.), Jy-heney (K.) é (2) /dʒeˈhәːnʲә/ Dia hAoine Friday Jerdein (Cr.), J’ardain (K.) keayney ey /dʒarˈdәːnʲ/ Diardaoin Thursday e, é, æ (2), ee, ŷe éy /kәːnʲә/ caoineadh weep, cry /kәːnʲ/ caoin kind, delicate /mәːʃ/ maois /rairʲ/ araoir, aréir, EIr. irráir mease, measure of herring last night /skәːlʲә/ scaoileadh spread, scatter /ˌtaru ˈdәːlʲ/ *tarbh daoil bull-worm keiyn (Bible, Cr.), keayn (Cr.), keain (K.)153 meaish (Cr.) meays (K.) riyr (see §3.9.4) skeayl(l)ey; skeayltagh e (6), é, ýa, y (2), ý, ýy, éy (2), îê, æ̆ y, ai, ey tarroo-deyill øː TC, ẹː HB, JW, WQ, eː NM øː TC, ei WQ, eːⁱ NM, ẹː HK ẹː HK, e. SK, eː JK, ẽː̣ e̜ JK, eː TL, mẹːs JW rɑːr, rɑ ̣iə̯r NM, rɑ ̣ːʳ, rɑ ̣ːiə̯r JW, rajə WːN ẹː TC, JW, HK iː JTK, eː TK, e̜ ːⁱ NM 3.4.3 ua > /ɨə̯/, /ɨː/ Table 37. ua /uə̯/ > /ɨə̯/ (>/ɨː/) Phillips beayn; cobeayn úa (4), ŷâ, ýa (2), ya beaynee úa beoyn cheayll, geayll creagh 153 ýy (7), yy (2), ý CM pronunciation /bɨә̯n/ etymology English HLSM buan i SK /bɨә̯ni/ buanaidhe eternal, lasting; coeternal reaper /bɨә̯n/ buan ? /xɨә̯l/, /ɡɨә̯l/ chuala /krɨә̯x/ cruach tendency, drift, instinct (Cr.), ‘necessity, fate’ (K.) heard rick, stack, heap iː TC y NM, ɪ, ᵘï TC, iː HK iː TC, JW, HK, iə̯ JK, i Thomson (1953: 248 s.v. KEYN) wrongly indicates that this word is found only in Phillips; in fact it is a securely-attested, if not especially common, item in Classical Manx. 170 Aspects of the historical phonology of Manx DC, NM, yː HK, iːə JTK creoghys, creoighys; creoghan (K.) ye, yey, ýo, úa, iy /krɨә̯xәs/ deayrtey iâ, îy, ýa (4), ýy (2), ue, ie, ya (2), yá (2), ýy, iy (3), ye, iê iy (3) /dɨә̯rtә/ ey (2), éy (2), ǽy (2), æy, eâŷ, eay, yea, yeă, ŷêa, yéa, ýea, ýe, êya ýa (5), ya (8), iâ, ia (2), ŷâ, ea, ue, yá ûŷ, ua (3), ýa, uá, úa (5), úua eaghtyr eayn feaysley, feayshil feoh cruas, cruadhas, cruachás; cruadhachán dortadh, doirteadh, *duartadh hardness, hardship; gadfly, harsh creditor pour, spill /ɨә̯xtәr/ uachtar surface /(ɣ)ɨә̯n/ uan lamb /fɨә̯slә/ fuascladh untie, release, relieve fẹːʃlə TT, fẹːl′ə HK /fɨә̯/ fuath hate iːə TT, ẹːo TC, HK, joː JW, ẹoː, eˈoː TC154 i EKh, W:N, W:S, yː NM, EK, HK, uː HK, λː NM, iː TT, JK, HB, ẹː JTK, öː J:EK, ɪː J:TL. ɪ J:JK iə̯ TC, iː Ekh, ẹː SK, i JK, öː J:EK, ï W:S i JK, TC, W:N, TK, W:S, e̜ HB, ɪː J:JTK, J:JK, J:HB, yː, JW λ, λː HK, ü, y NM, ɪ., i. W:S, ö, ɛ J:EK geayl ýy /ɡɨә̯l/ gual coal geaylin ýa /ɡɨә̯lәnʲ/ gualainn shoulder keayn ie, êy, ýa /kɨә̯n/ cuan sea 154 iː TC, üːᵊ, λː NM, i JTK ẹː TC, iː TT, JW eː EKh, iː JTK, HK, ɛ JːEK, λː NM All except TT’s are clear spelling pronunciations. 171 Aspects of the historical phonology of Manx leagh úa (7), uá /lɨә̯x/ luach leah ýa, iæ, iǽ, úa (3), ua (3), ýæ /lɨә̯/ luath leaystey /lɨә̯stә/ luascadh leaghyr, laghyr, leoighyr (Cr.), leeaghar (K.) leoh (Cr.) /lɨә̯xәr/ luachar /lɨә̯/ ash (gen.) /lɨә̯n/ luaith, gen. luatha Luan /skɨә̯b/ scuab brush, broom /slɨә̯st/ sluasaid shovel, fan /tɨә̯/ tuath common people etymology English HLSM clua(i)s ear i TC, JK, öː J:JTK, W:S eː HK, ẹː JW, NM eːï Ekh, ɪ HB, ɪ, ɪː HB, J:JTK, eːi SK, ɛː J:TL ẹː EKh, NM, HK i NM, ẹ., eː, iː HK iə̯ TC, i NM, W:S, iː NM Lhein, Lheiun (Cr.) skeab y (2) sleayst theay, theo (Cr.) ua value, reward early, soon iːə TC, JW, iː TT iːə JTK, yə NM, iː HK swing, rock, reel, stagger rushes iː TC, JW, leːʃt′ᶴə W:S Monday iː TC, JW, JK, HK, WQ, i TC, HB, i. JK 3.4.4 uai > /əː/ (>/eː/), /ɨə̯/ (>/ɨː/) Table 38. uai /uə̯/ > /əː/ (>/eː/), /ɨə̯/, (>/ɨː/) Phillips CM pronunciation N /klɨә̯ʃ/, S /klәːʃ/ cleaysh ye, ua (2), úa (6), úæ, ya (4), ýa (6) eayin iĕ (2), yĕ, yæ̆ /(ɣ/j)ɨә̯nʲ/, /(ɣ/j)әːnʲ/ uain lambs geayltyn ýy /ɡɨә̯lʲtʃәn/ shoulders g(h)eayney ûŷ, iæ /(ɣ)әːnʲә/ *guailtean, G. guailne, ScG. guailnean uaine green øː TC, ï W:S gleashagh(t) y, ýa (3), ya, yy ie (2), e, æy /ɡlәːʃax(t)/ gluaiseacht move ẹː JW, TT, eː NM, iː TC */әːntʃәs/ uain opportunity, leisure Ph. ienchys etc. 172 Aspects of the historical phonology of Manx í, iy, úey /dʒeˈlәːnʲ/ Dia Luain Monday iy (5), iý, íy (2), ŷi, yi, iê, eiy, yy (2), ýy, iŷ /kɨә̯rt/ cuairt time, occasion /lәːlʲ/ luadhail156 motion, power of movement /mәˈɡɨә̯rt/ ma gcuairt around, about neayr’s /n⁽ʲ⁾ɨә̯rәs/ ? an uair agus since treickney, treicknane (Cr.), tratney, traitney, tratneayn (K.) /trәːkʲnʲә/, /trәːtʲnʲә/; /trɨkʲˈnʲeːn/, /trɨtʲˈnʲeːn/ tuairgneadh, tuairgnín beat; beetle, mallet Jelhein (Cr.); Jy-luain (K.) Jelune, Jelhuin (Cr.)155 keayrt lheihll (Bible, Cr., K.), lheill (Bible), lheil (Cr.) mygeayrt iy (8), îŷ (2), îy (2), yi (2), iŷ, i (2), y, éy i JK, JTK ø TC, iə̯ JTK, e̜ ː, eː HK, ei JW, WQ, eⁱ, ẹː NM i HB, JK, iː JW, y NM, öː HK, ï HB iː TC, JW, TT iə̯ HB, iː, iᵊ, i, λː JK, i.ə, ĭə W.N, öː, ɪː NM, J:EK, J:TL, ü, y, ĕ, i, ï, e NM iː JW, ɪ NM, JK 3.4.5 ua(i) > /uə̯/, /uː/, /oː/ There are a number of items which categorically retain back /uә̯/ (>/uː/). Included are certain items such as mooar, booa, sto(o)amey which may also have /oː/ within Manx or in other Gaelic varieties; this generally appears to block development of fronting, although see deayrtey (§3.9.8). There are also a number of items with both fronted and unfronted variants (§3.4.5). Included here are those items with G. uaidh which retain back /uә̯i̯ /; for other diphthongs arising from ua(i) + vocalized fricatives, see §3.9.1. 155 156 Spelling based on supposed etymology. Apparently treated as > luail, or via */lәːi̯ .әlʲ/, cf. seihll (§3.4.1). 173 Aspects of the historical phonology of Manx Table 39. ua(i) /uə̯/ > /uə̯/ (>/uː/), /oː/ Phillips booa druaiaght (Bible), druaight (Cr.), druiagh(t) (Bible, K.), cf. drualtys, driualtys (Cr.) (§3.9.10) dwoaie; dwoaiagh, dwoaioil uó dûôi, duoi; (dwoaioil) duoiyl, duoióyl, duoióel groamagh (Cr., K., Bible), grooamagh (K.) gruaie CM pronunciation /buә̯/ etymology English HLSM bó, cf. dat. pl. buaibh; buachaille etc. cow buːə NM, HB, buɑ TC, (pl.) buːɑ ̣xən, bøuxən, bøuhən TC, bọːrən JTK /druә̯i̯ axt/, ? /drәːi̯ axt/ draoidheacht, druadh- enchantment /duә̯i̯ / duadh (Ir. m.), duaidh, ScG. duaidh (f.) hate, dislike ui JW, TT /ɡroːmax/, /ɡruә̯max/ gruaimeach gloomy, sorrowful, sullen oː JW /ɡruә̯i̯ / gruaidh cheek; grimace big, great ui TC mooar ua (5), úa (10), úy, úay /muә̯r/ mór, Munster Ir. muar smooinaghtyn (§3.9.9) ú (11), u (5), úi (2), ui, uy /smu:nʲaxtәn/ smuain, smaoin think, thought /stoːmә/, /stuә̯mә/ /tuә̯rәstal/ stuamdha comely tuarascbháil appearance, form, resemblance stoamey, stooamey tuarystal (Cr, K., Bible), tooarystal (SW) 174 uː TC, TK, TL, W:S, NM uːə TC, JK, NM, oːə NM, u̜ ː J:TL, J:EK, J:JTK, J:HB, u̜ ːə J:EK, J:JTK, J:HB u JK, HB, NM, uː TC, ï W:N, W:S, smaːᵊxin Myl õː, ũː TC, oː JW Aspects of the historical phonology of Manx twoaie uoi (2), ûôi, ûêi /tuә̯i̯ / tuaidh north twoaie; twoaiagh uôi (2), ûôi (2), uói, oi, ôi /tuә̯i̯ /, /tuә̯i̯ ax/ tuaith (Dinneen) beware; wary ui, u-i JK, NM, u̜ i J:EK, J:JTK, uːi W:N, ûːi W:S 3.4.6 ua(i): items with variable reflexes Table 40. ua(i) /uə̯/ > /ɨə̯/ (>/ɨː/), /əː/ or /uə̯/ (>/uː/), /oː/ Phillips CM pronunciation ? /klu:isaɡ/, /klәˈʃtʲeːɡ/ etymology English ScG. cluasag pillow, bolster, cushion úa, iýa /fɨә̯r/, fuә̯r/ fuar cold feayght, feayraght ya fuacht cold, coldness heose, seose, neose ua (6), úa (14), uá (4), u, ya (2) /fɨә̯xt/, /fuә̯xt/, /fɨә̯raxt/, /fuә̯raxt/ /hɨә̯s/, /huә̯s/, /hoːs/ thuas, suas, anuas up hooar, dooar, fooar (PC, Ph.) éy (12), ǽy (2), ey (3), æy (4), óy (2), eý, êŷ, ýy, uóe, ûêy /huә̯rʲ/, /hәːrʲ/ fuair got, found ooasle; ooashley ýa (10), yá, ya (5), ýæ /uә̯sәl/, /wusәl/; /uә̯ʃlʲә/, /wuʃlʲә/ uasal; uaisle noble, worthy; worship, honour clooiesag (Cr.),157 cleayshteig (K.) feayr HLSM uᵊ NM, iːə TC, TL, DC, HB, iː TT, iːĕ̜ HB iːə, i TC, uᵊ NM, uː HK öː JW, EL, DC, J:NM, u HB, TC, ëːs, ëi, ëː, eiː NM, yː EK, HB, oː EL, uː TC, JTK, ɔː J:EK uːə JTK, HK, ᵘuːə, ᵘuː, uː NM, uː, uːə W:N, e̜ ː HK u̜ ᵉʃl̠ i J:EK, miː ˈwuːʃl′ə TC 157 Cregeen’s form is apparently influenced by clooie ‘small feathers, fur’ (G. clúmh, ScG. clùimh, clòimh), as further suggested by his definition of clooiesag ‘a bolster of feathers’. 175 Aspects of the historical phonology of Manx Fronted reflexes of ooasle appear to be restricted to Phillips, (see §3.6.1.2) although ny seasley (PC 1796 l. 3424) may represent the comparative (Thomson 1995: 138). 3.4.7 agh, adh > /əː/ The interaction of adh/gh (excluding those items which give /Vi̯ /-diphthongs) with the ao(i) set in Manx is not immediately clear, primarily because few examples of the adh/gh set are found in which the development of historic /a/ can be determined. Several items found in other Gaelic varieties, such as adhradh, ladhar are not attested in Manx, and in other items the realization of adh/gh is obscured by later developments, as with oyr ‘reason’ (G. adhbhar) (§3.4.7.3). However, if <eo> in leodaghey ‘decrease, diminish’ (G. laghdaghadh) means /әː/ as in feoilt ‘generous’ (G. faoilte), eoylley ‘dung’ (G. aoileadh), freoagh ‘heather’ (G. fraoch), meoir (Cr.) ‘official’ (G. maor), and <eay> in reayrt means /әː/ as in keayney ‘crying’ (G. caoineadh) etc. (§3.6.2), then it seems likely that adh/gh became /әː/ and merged with ao(i) as in southern Scotland (Ó Maolalaigh 2006; Lewin 2018). TC’s l′ẹˈoːdɑ ̣xə is clearly a spelling pronunciation (§1.6.9.2). Other items are discussed below. Table 41. agh, adh /aɣ/ > /əɣ/ > /əː/ and other developments Phillips CM pronunciation etymology English HLSM leodaghey y (3) /lәːdaxә/ laghdaghadh eairk éyrk, erick (3), ǽrick, (pl.) érikyn, ærikyn, æyrkyn, ǽrkyn, ærkyn /әːrk/ adharc decrease, diminish horn l′ẹˈoːdɑ ̣xə TC (sic)158 ẹːʳk JW, eːɑ̜k, eːak, iak NM, öːṛk HK, elʹk W:N, pl. ẹːʳkən JW, e̜ rkən TC, ɛːʳkən, e̜ ːṛkən JK, ẹːakən NM, elʹkən W:N /әːrkan/ /rәːrt/ adharcán radharc lapwing view earkan reayrt 158 Spelling pronunciation. 176 rɛːt NM Aspects of the historical phonology of Manx eab /әːb/ /oːr/ ScG. aghaib (O’Rahilly 1926: 36) adhbhar reason ọːəṛ NM, o̜ ːr TC /imәd/ adhmad use Roonysvie /ˌru(ː)nәsˈvai/ *Raghnasbhagh Ronaldsway ɪməð JW, ẹməd, eməð, eməs, ɪmɪd ˈruːnəsˈvɑ ̣ːi JW Crennell /krenәl/ Mac Raghnaill (surname) oyr ymmyd óer (3), oer, óeyr (2), oeyr (2), oéyr (2), óyr (5) y, ý attempt, effort Although adharc and radharc seem to show /aɣә/ > /әː/, the more common development of agh, adh, ogh, iodh in intervocalic or final position is /aɣә/ > /[ә]ɣә/ > /әjә/ > /әi̯ /, as in the following items: cleigh, cleiy /klәi̯ / ‘hedge; dig’ (G. cladh), eiy /әi̯ / ‘foot lock’ (G. iodh), eiystyr /әi̯ stәr/ ‘halter’ (G. adhastar), eiyrt /әi̯ rt/ ‘chase, follow’ (G. adhairt), feiyr /fәi̯ r/ ‘noise’ (G. foghar), fuygh /fәi̯ / ‘wood’ (G. fiodh), leigh /lәi̯ / ‘law’ (ScG. lagh), leih /lәi̯ / ‘forgive’ (G. loghadh), reih /rәi̯ / ‘choose’ (G. rogha), teiy /tәi̯ / ‘choose’ (G. togha), oaie, oi /әi̯ / ‘face, front’ (G. aghaidh). 3.4.7.1 eab ‘attempt’ So far as is known, this item and its derivatives are attested only from Cregeen, who defines eab as ‘an attempt, effort, or push; to say or do some thing’, and eabee as ‘a person, &c. partly taught, formed or planned to some particular work or use’. He also gives a verb eabbey ‘attempt, &c. […] form or plan’, eabit ‘planned, formed, cut out, marked, &c’. O’Rahilly (1926: 35–6) links this to ScG. aghaib ‘attempt, essay, trial’ and compound form comhaib ‘contention’, which he derives from < *ad-od-be (O’Rahilly 1926: 36), and compares with the more frequent ScG. form oidhirp ‘effort’, M.Ir, aidirbe < *adair-ess-be, and Ir. fobha, fogha ‘attack, attempt’ < OIr. fubae, fo-ben. If O’Rahilly’s etymology is correct, eab presumably represents /әːb/. 177 Aspects of the historical phonology of Manx 3.4.7.2 eairk ‘horn’, G. adharc There are no <y> or <yy> spellings for this item in Phillips, and forms such as ǽrick, erick may represent a metathesized form *adhrac. A form adhraic is reported in South Uist and Barra (Campbell 1972: 218) and Tiree (Ó Maolalaigh 2008b: 520); compare also the Sutherland form represented as earag by Dwelly. A form /eːrk/ could develop via /ejәrk/, /eiә̯rk/ if adh followed the usual development in stressed coda position rather than the preconsonantal development as found in leodaghey. 3.4.7.3 oyr ‘reason, cause’, G. adhbhar G. adhbhar is found as oyr in Manx (oer, oeyr and similar in Phillips), apparently representing /oːr/ (cf. HLSM II: 344). It is impossible to tell (cf. Ó Maolalaigh 2006ː 58–9) whether this represents merely loss of /ɣ/, followed by vocalization of /v/ (i.e. > abhar, cf. Manx doagh /doːx/ ‘vat’, G. dabhach), or the same with compensatory lengthening (i.e. > ábhar, cf. Manx foayr /foːr/, G. fábhar, Eng. favour), or with modification of the quality of /a/ as found in Scotland, since faobhar seems to develop in the same way as items with a and á, to Manx foyr /foːr/, though also apparently with /uː/ from one speaker (Jackson 1955: 49, HLSM II: 176). 3.4.7.4 abane ‘ankle’, EIr. odbrann, ScG. adhbrann EIr. odbrann ‘ankle’ has a peculiar development in Manx to abane (Cregeen, Kelly), abbane (Bible) /aˈbɛːn/. Ó Maolalaigh (forthcoming b) posits a derivation of the modern forms from an earlier *adhbarn ( < EIr. odb ‘knot, lump’, Ir. fadhb ‘knotty problem’ + *sperno/ā ‘heel’). The (presumably early) loss of /r/ in an unstressed cluster in the Manx form would have a parallel in maggle ‘testicle’ (G. magairle) (§4.2.2.2), with the resulting /әn/ being remodelled under the influence of the diminutive ending -án (cf. §4.4.7.3), as in certain Scottish Gaelic dialects (ibid.: 262). It is also possible that the -án ending is an early substitution for earlier -arn if the latter had been reanalysed as a singulative or diminutive termination, as suggested by the 178 Aspects of the historical phonology of Manx Scottish variant adhbairne (Ó Maolalaigh forthcoming b: 261; cf. O’Rahilly 1931: 66– 9) and Ir. fadhbairne ‘lumpy object’. The stress shift obscures the development of the first syllable, but since non-initial stress is strongly correlated with the historical presence of a long vowel or diphthong in the first syllable (§5.1), it is likely that this was earlier */ˈәːbaːn/ *adhbán. 3.4.7.5 ymmyd ‘use’, G. adhmad Manx ymmyd ‘use’ apparently represents Irish adhmad and is attested with initial [ɪ] and [e] (HLSM II: 480). For the short vowel in a polysyllable, see §5.5.1; presumably the earlier form was */әːmәd/. 3.4.7.6 G. Raghnall (N. Rǫgnvaldr) in Roonysvie and Crennell The name Raghnall, N. Rǫgnvaldr, is not attested independently in Manx as far as is known, but is present in the Norse place-name Roonysvie, Runnusvei, Runesvie ‘Ronaldsway’, ˈruːnəsˈvɑ ̣ːi (JW, HLSM II: 509) (Norse Rǫgnvalds vað > G. *Raghnasbhadh) (PNIM VI: 161–4) and the surname Crennell (< Mac Raghnaill). The first vowel may apparently be long or short, presumably because the stress is on the final syllable. According to Broderick (PNIM VI: 164) ‘[t]here is no obvious reason for /uː/. The entry for 1770 [Runnusvei] suggests a short vowel, which would agree with its unstressed position’. However, we might also compare Manx un < aon, and Manx red, G. rud < réad, raod, both of which may represent special developments of /әː/, /eː/ under light stress. The short vowel in Crennell may be explained by the widespread phenomenon of stressed long vowel shortening in polysyllables (§5.5.1). 179 Aspects of the historical phonology of Manx 3.5 Previous accounts of Manx ao(i), ua(i) 3.5.1 Rhŷs Rhŷs (11–28) devotes considerable space to the Manx reflexes of G. ao(i) and ua(i), and evidently appreciated the complexity of the topic (Lewin 2019a: 94). His fieldwork notebooks include lengthy tables of realizations of items containing ao(i) and ua(i) from different informants arranged in parallel columns (Broderick 2018c: 52–6): The Manx reflexes of G[aelic] AO(I), UA(I) […] seemingly became a fetish with Rhŷs. Practically every speaker he interviews is asked questions on this aspect of Manx phonology. (Broderick 2018c: 47) Although Rhŷs’s descriptions are not without their ambiguities, they are an extremely important source for this topic, providing evidence of the pronunciation of more conservative, Manx-dominant speakers born in the early nineteenth century. 3.5.1.1 ao > ȳ [əː] According to Rhŷs (17), the regular realization of G. ao before a broad consonant in Manx is a phone ȳ, with examples given including keyrrey ‘sheep’ (G. caora), meayll ‘bald’ (G. maol), seihll ‘world’ (G. saoghal). This sound is stated to be ‘nearly identical with the eu in such French words as ‘jeune’ and ‘peur’’, which suggests [œː]. However, it is also defined as ‘the long vowel corresponding to the short one last discussed’. This short y is described as follows: By short y in Manx I mean the sound of North Cardiganshire pronunciation of the words yn ‘in,’ and yr ‘the:’ to my hearing it is identical with that of u in the English words ‘but’ and ‘gun.’ (Rhŷs: 14) 180 Aspects of the historical phonology of Manx The Welsh comparison suggests [ә],159 but the English presumably refers to RP [ʌ].160 The vowel in French jeune ‘young’ and peur ‘fear’ would be expected to be open-mid [œ] rather than close-mid [ø] in jeûne ‘fast’ (cf. Fougeron and Smith 1999).161 However, [œ] appears to be the value intended by Rhŷs for his œ, which he describes as ‘a rounded ĕ like the ö of the German words ‘brödchen’ and ‘hölle’’ (Rhŷs: 21). Since Rhŷs clearly intends a distinction between ȳ and œ̄ , the latter being more clearly defined, perhaps ȳ should be understood as a somewhat higher vowel. It is worth noting the following comment by Strachan (1897: 54), casting doubt on the identity of quality of ȳ and y, which would suggest that ȳ does not have quite the same quality as short [ә]: As to symbols, ə, æ, ů, y, and ȳ have been used as in Prof. Rhŷs’s Outlines of Manx Phonology, though I am not quite sure that ȳ is exactly the long sound of y. (Strachan 1897: 54) Notwithstanding this uncertainty, it will be assumed that ȳ = [әː] for the purposes of the following discussion. More important than the exact value of the vowel Rhŷs heard, or thought he heard, is his testimony that ao had a largely consistent realization as this ȳ, and showed no signs of falling in with /iː/ or /eː/, unlike aoi and uai (see below §3.5.1.4). 3.5.1.2 Allophonic realizations of ao Rhŷs (17, 21) notes realizations of eayl ‘lime’ (G. aol), with œ̄ , œ [ә̞(ː)] or ü [ɨː] (?), which he surmises represents the genitive aoil as in clagh eayl ‘limestone’, G. cloch According to Ball and Williams (2001: 35–6), Welsh /ә/ is somewhat fronted [ә̟] in both northern and southern varieties. 160 It is possible, however, that Rhŷs’s status as an L2 speaker of English, his L1 being North Ceredigion Welsh, may have affected his perception of English, notwithstanding his prominent position in British elite society. In Rhŷs’s lifetime RP /ʌ/ appears to have been predominantly a back vowel, although later twentieth-century accounts describe fronting in the direction of [ә] or [æ] (Bauer 1985: 67). 161 jeune [ʒœn], jeûne [ʒøn], peur [pœʀ] <https://www.collinsdictionary.com/dictionary/frenchenglish/> [accessed 26.08.2019]. 159 181 Aspects of the historical phonology of Manx aoil. Similarly, a realization of meayll ‘bald’, G. maol as vœl is noted ‘as if representing a dative feminine maoil in Irish’ (though he does not comment on the quality of the lateral). The short versions of these vowels are noted also in Gaelg ‘Manx Gaelic’ (G. Gaoidhealg), which Rhŷs (21) states he has ‘heard fluctuate between Gœlc and Gülc’ (§3.9.1.1). Of short ü he writes that ‘[i]t is met with in Manx mostly before l in such of the following instances as are monosyllables, but elsewhere it is more commonly long, as will be pointed out under the respective instances’ (§4).162 These realizations are presumably to be understood as allophones of /әː/. 3.5.1.3 ua > ůy [ɯə̯], üy [ɨə̯], iy [iə̯] According to Rhŷs (19), ‘ü represents the Goidelic diphthong úa, and the pronunciation fluctuates between ua, ů, ü, and i’. Although ů, ü, and i are presented here as monophthongs, transcriptions in the subsequent discussion suggest that they were often diphthongal, as in the historical ua, with a schwa off-glide. When monophthongal, it was usually long, but short in certain instances: It [short ü] is met with in Manx mostly before l in such of the following instances as are monosyllables, but elsewhere it is more commonly long. (Rhŷs: 19) Rhŷs describes the phonetic value of ü and ů as follows: By short ü is meant that occurring in the German word ‘dünn’ and ‘üppig.’… by ů I mean a sound considerably more rounded than ü, and reminding one more clearly of the u which is here the starting point. (Rhŷs: 19) Cf. Terminal Manx klɑːx yːl clagh eayl (JW) (also iːl, u:l), λːl (NM), eːilʹ (SK), ø(ː)l (TC) (HLSM II: 140), and kyːl keyl, G. caol (JW), xuːl (HK), also kiːl (JW, HK, HK) (HLSM II: 249). 162 182 Aspects of the historical phonology of Manx On the face of it, the only difference between ü and ů would seem to be the degree of rounding, i.e. ü = [y] and ů = [ʉ]. However, ‘reminding one more clearly of the u which is here the starting point’ may in fact refer to degree of backness. This is further supported by James Murray’s interpretation of Rhŷs’s pronunciation of these sounds which Rhŷs quotes as follows: He [Murray] describes the u of fuyr as originally ‘a high-back round vowel,’ which became successively ‘high-back mixed, like the Swedish u’; then ‘a high-front round vowel, like the German ü,’ and finally, owing to ‘the loss of the rounding,’ an open ī. (Rhŷs: 20)163 There are reasons to be doubtful about the descriptions of rounding here (§3.5.6), and we shall interpret the descriptions as referring to degrees of back to central unrounded vowels, so ů = [ɯ], ü = [ɨ]. This symbolization should not be understood as precluding a degree of rounding, however. Rhŷs gives the following examples of such realizations of G. ua: feayr ‘cold’, G. fuar: fūyr [fuә̯r], fůyr [fɯә̯r], füyr [fɨә̯r], fiyr [fiә̯r] beayn ‘lasting, eternal’, G. buan (no transcription given) keayn ‘sea’, G. cuan: kǖyn [kɨә̯n] kīy̆ n [kiә̯n] eayn ‘lamb’, G. uan: ghůy̆ n [ɣɯә̯n], ghüy̆ n [ɣɨә̯n], ghiy̆ n [ɣiә̯n] leagh ‘reward’, G. luach: lüagh [lɨa̯ x], lüygh [lɨә̯x] leah ‘early, soon’, G. luath: lůy̆ [lɯә̯], lüə [lɨә̯] theay ‘common people’, G. tuath: tüə [tɨә̯], tiə̯ [tiә̯] According to Rhŷs, (20), historical ua was monophthongal, often short, and always ů and ü before [l]: In the case of ua followed by l, the only pronunciations which I have heard are ů and ü (never i), and the vowel is now mostly cut short (Rhŷs: 20) 163 Rhŷs adds the cryptic comment ‘[h]ad he heard the u sound from a Manxman I am not sure that he would not have pronounced it to be simply an open u’. 183 Aspects of the historical phonology of Manx The following examples are given: geayl ‘coal’, G. gual: gůl [ɡɯl], gül [ɡɨl] cheayll, geayll ‘heard’, gůl [ɡɯl], gül [ɡɨl] geaylin ‘shoulder’, pl. geayltyn, G. guala(i)nn (see §3.9.5) Rhŷs (20) states that the realization of leagh ‘reward’ (G. luach) was monophthongal in the south lůgh [lɯːx], and he heard a realization of theay ‘common people’, G. tuath, as tü [tɨː] from ‘a native of Dalby’ (on the west coast south of Peel).164 The instances in Phillips of geayl ‘coal’ and cheayll / geayll ‘heard’ all have spellings which suggest a monophthongal pronunciation: one instance with <ýy> for the former and for the latter, seven instances of <ýy>, two of <yy> and one of <ý>. No spellings suggesting a diphthongal realization, such as <ya> or <ua>, are found for these items (see also §3.6.1.2 for discussion of Phillips’ orthographic representations). For geaylin only one instance occurs in Phillips, with <ya>, suggesting the regular diphthongal reflex of ua. The one instance of the plural has <ýy>, which may suggest the regular development of uai to /әː/. All the realizations of geayl ‘coal’ in terminal speech have monophthongal realizations, sometimes short. Southern gyːl (NM, EK and HK), guːl (HK) and possibly gλːl [ɡɯːl] (NM), göːl165 (J:EK) probably represent the sound that Rhŷs heard, but the most frequent realizations are with iː (EKh, TT, south; JK, HB north), ɪː (J:TL south), or short i (W:N/S) or ɪ (J:JK, north) (HLSM II: 188). There is also ẹː (JTK, north). For cheayll / geayll ‘heard’ we find kyl only from NM, xiːl from HK (both south) and kɪl, kuɪ̈ l (TC, north), and also ha ˈgɪl cha geayll (TC, W:S). For geaylin we find 164 Although possibly to be considered part of the north, according to the traditional administrative division (§0.8). 165 Jackson explains his ö as ‘a slightly retracted central ö, very poorly rounded, further back and more open than ø in French peu, and somewhat closer than œ in French pleure; though occasionally it is open enough to write œ.’ Broderick (HLSM III: 44) explains his use of ö as follows: ‘In certain circumstances /ә/ can be realized as [ö] or [öː] (i.e. articulated with a degree of retraction and lip-rounding)’. 184 Aspects of the historical phonology of Manx diphthongal iə̯ (TC), but also iː (EKh), ẹː (SK), i (JK, HK), ɪ̈ (W:S). In the plural we have iə̯ (TC), i (NM, W:N) and iː (NM).166 Strachan (1897) also gives monophthongal ů̄ for G. ua: χů̄l cheayll ‘heard’, G. chuala ků̄ᵈn keayn ‘sea’, G. cuan 3.5.1.4 aoi and uai > œ̄ [ə̞ː] According to Rhŷs (22–6), œ̄ is the regular reflex of both aoi and uai. The value of œ̄ is defined by reference to the short vowel, described as followed: By this I mean a rounded ĕ like the ö of the German words ‘brödchen’ and ‘hölle’: it occurs in Manx in a few words in which it is a shortening of long œ. (Rhŷs: 21) The vowel ĕ is defined as ‘Open e, short…so far as I can judge, that of e in the English words ‘get’ and ‘men’’ (Rhŷs: 5), i.e. [ɛ]. These descriptions would suggest that œ̄ represents front rounded mid-low [œː], but again the degree of rounding is unclear (§3.5.6), and œ̄ is therefore interpreted as a lowered version of ȳ [әː]. This is noted as varying with ē [eː] in at least some items. The following items with G. aoi giving œ̄ are given: seihill ‘world’ (gen.?), G. saogha(i)l): sœ̄ l [sә̞ːl] seyir ‘carpenter’ (gen.?), G. saoir: sœ̄ r [sә̞ːr]; also in Mac y Tœr [makәˈtә̞ːr] surname ‘Teare’, G. Mac an tSaoir eash ‘age’, G. aois: œ̄ sh [ә̞ːʃ], ēsh [eːʃ] deiney ‘men’, G. daoine: dœ̄ ñi̯ ey [dә̞ːnʲә], dēñey [deːnʲә], (voc.) ghœ̄ ñey [ɣә̞ːnʲә], ghēñey [ɣeːnʲә] feysht ‘examine, question’, G. faoiside (no transcription) skeayley ‘scatter, spread’, G. scaoileadh (no transcription) Jeheiney ‘Friday’, G. Dia hAoine: hœ̄ ney [ˈhә̞ːnʲә], hēney [ˈheːnʲә] 166 It should be noted that some of the attested forms show apparent confusion between the singular and plural, with non-palatalized [l] appearing in TC’s plural giəltən and palatalized [lʲ] in some of the singular forms such as SK’s gʹẹːlʹən. Confusion between broad and slender consonants is apparently common in terminal speech (Broderick 1999: 90). 185 Aspects of the historical phonology of Manx Jerdein ‘Thursday’, G. Diardaoin: Jyrdœ̄ ñ [dʒәrˈdә̞ːnʲ], Jydœ̄ ñ [dʒәˈdә̞ːnʲ] Rhŷs (23) states that œ̄ can also ‘represent […] uai or the umlaut of the Goidelic diphthong ua’. The following items are given: Jelhein ‘Monday’, G. Dia Luain: Jylœ̄ ñ [dʒәˈlә̞ːnʲ] ‘rhymes with Jardain pronounced Jy(r)dœ̄ ́ ñ’ keayrt ‘time, occasion’, G. cuairt (no separate transcription) mygeayrt ‘around’, G. má gcuairt: myʒœ̄ ́ rt, myʒi̯ œ̄ ́ rt [mәˈɡʲә̞ːrt] eayin ‘lambs’, G. uain: ghœ̄ ñ [jә̞ːnʲ]167 geayney ‘green’, G. uaine: ghœ̄ ñey [ɣә̞ːnʲә], ghēñey [ɣeːnʲә] The above suggests a merger between uai and aoi,168 which Rhŷs (25) himself postulates: From the foregoing instances it will be seen that the sound common to all the more regular representatives of uai and aoi is œ̄ which, however, does not always remain thus, in the case of open syllables like gheayney ‘green,’ deiney ‘men,’ and Jyheiney ‘Friday,’ one hears either œ̄ or ē: of these I should regard œ̄ as the older sound and ē as a modification of it. So we seem to have the two series uai, œ̄ , ē and aoi, œ̄ , ē: in other words, the two series converge at œ. (Rhŷs: 25) Rhŷs’s observations appear to show a merger in progress, with [œː] falling in with [eː] (from G. éi). 3.5.2 Marstrander (1932) Marstrander lists different phones according to the Gaelic phonemes of which they are a reflex. The data relevant to ua and ao are summarized below. 167 168 ‘with a palatal gh passing off into i̯ or the y of the English words ‘yield’ and ‘yes’’ (Rhŷs: 23). For some or most items; certain uai items may instead go with ua (§§3.9.5, 3.9.6). 186 Aspects of the historical phonology of Manx 3.5.2.1 ē ̣ [eː] < G. aoi, uai This is primarily a reflex of G. é etc., but may represent G. aoi, uai (Marstrander: 64). aoi deiney ‘men’, G. daoine: dēˈñə ̣ [deːnʲә] meaish ‘measure of herring’, Norse meiss, cf. ScG. maois: mēṣ̌ [meːʃ] skeayley ‘spread’, G. scaoileadh: skēˈɫə ̣ [skeːlʲә] Jeheiney ‘Friday’, G. Dia hAoine: dʹžẹ hēˈñə ̣ [dʲʒeˈheːnʲә] keayney ‘weep’, G. caoineadh: kēˈñə ̣ [ˈkeːnʲә] meiygh ‘soft, tender’, G. maoth: mē ̣ [meː]169 uai naightyn ‘news’, G. nuaidheacht, nuadhacht, nódhacht, cf. ScG. naidheachd: nēˈaχtən [ˈneːaxtәn] ̣ cleaysh ‘ear’, G. cluais: klēṣ̌ [kleːʃ] kɫēṣ̌ [klʲeːʃ] gleashagh(t) ‘move’, G. gluaiseacht: glēˈžaχ [ɡleːʒax] ̣ Marstrander (64) explicitly notes that before broad consonants ao becomes either ē [ɛː] (HK) or î [ɪː] (WQ). The forms with [eː] would represent the innovating reflex of aoi / uai noted by Rhŷs (§3.5.1.4), [eː] resulting from unrounding of [әː]. [kleːʃ] for cluais would represent Rhŷs’s (24) southern form. 3.5.2.2 ē [ɛː] < G. ao Long open ē [ɛː], predominantly representing G. á or ó, is also noted as a reflex of G. ao in two items (Marstrander: 63): teayst ‘dough’, G. taos: tēs [tɛːs] seihll ‘world’, G. saoghal: sēl [sɛːl] This realization is noted as a Cregneash feature. The Manx spelling suggests a Manx reflex maoith, perhaps attested in TC møiɑ ̣x meiyghagh (HLSM II: 297) < *maoitheach (cf. §3.9.1.1). 169 187 Aspects of the historical phonology of Manx 3.5.2.3 ī [iː] < G. ua Long close ī, as well as representing G. í, ia and ‘broken’ é (§2.2.7), is noted as a reflex of ua (Marstrander: 65): skeab ‘brush’, G. scuab: skīb [skiːb] feayr ‘cold’, G. fuar: fīr [fiːr] feayraght ‘coldness’, G. fuaracht: fīˈraχt [fiːraxt] keayn ‘sea’, G. cuan: kīdn [kiːᵈn] eaghtyr ‘surface’, G. uachtar: īˈχtə(r) [iːxtә(r)]170 3.5.2.4 î [ɪː] Under ‘i-lydene’ (‘i-sounds’), Marstrander (65–6) describes ‘[d]en senkede î’ as a frequent reflex of ao. This is described as ‘noe senket i omtrent som i irsk daoine i Kerry’ (‘a somewhat lowered i approximately as in Irish daoine in Kerry’).171 Presumably by this ‘lowered’ (‘retracted’?) vowel something like [ɪː] is meant. The following items are noted: geay ‘wind’, G. gaoth: gî [ɡɪː], also gîə [ɡɪә], giə̯ [ɡiә̯] eayl ‘lime’, G. aol: îl [ɪːl] keyl ‘slender’, G. caol: kîl [kɪːl] keyllys ‘sound’, G. caolas: kîˈləs [kɪːlәs] Marstrander also notes î from G. aoi, ua and secondary /iː/ (<uidhe): skeayley ‘spread’, G. scaoileadh: skʹîˈɫə [skɪːlʲә], also skēˈɫə ̣ [ˈskeːlʲә] keayn ‘sea’, G. cuan: kîdn [kɪːdn], also kīdn [kiːᵈn] guee ‘pray’ G. guidhe: gwî [ɡwɪː], also gwī [ɡwiː] 170 Marstrander (65) also lists cleayney ‘seduce, entice, persuade’ (klīˈñə [kliːnʲә]) here, apparently deriving this from cluain ‘deceit, flattery’ (which does not generally have a derived verbal form), although it is more likely from claoin, claonadh ‘incline, slant; pervert, lead astray’. It is possible, however, that the Manx form represents conflation of both words, as might be suggested by the palatalized [nʲ] noted by Marstrander, unless this is simply by analogy with the finite stem claoin, and by the alternative realization klîˈənə [ˈklɪәnә] (Marstrander: 74). 171 I.e ‘guta fada tosaigh [iː], beagán faoi bhun Guta Cairdineálta 1, liopaí leata […] beagán lárnaithe in aice le consan leathan’ (‘a long front vowel [iː], a little below Cardinal Vowel 1, lips spread […] a little centralized adjacent to a broad consonant’) (Ó Sé 2000: 22). 188 Aspects of the historical phonology of Manx Marstrander (66) notes that in all the words with î, open ī [iː] may also be heard. In addition, he suggests that ē ̣ [eː] and ē [ɛː] vary with î before both broad and slender consonants. It should be noted that Marstrander does not use î in his transcriptions of Christian reading from the Bible in 1928 (Thomson 1976). This may be because this sound was not noted from Christian, or because Marstrander had not yet noticed the distinction at this point. Cf. Marstrander’s introduction of the symbol ï part-way through his transcription of Christian (Thomson 1976: 256). 3.5.2.5 îə [ɪə̯] < G. ua(i), ao Marstrander (74–5) notes îə [ɪә] as a realization of ua(i) and ao: ua leah ‘early, soon’, G. luath: ʎîə [l̪ ɪә̯] geayl ‘coal’, G. gual: gîəl [ɡɪә̯l] eaghtyr ‘surface’, G. uachtar: îˈəχtər [ˈɪә̯xtәr] leagh ‘reward’, G. luach: ɫîəχ [lʲɪә̯x] creagh ‘turfstack’, G. cruach: krîəχ [krɪә̯x] uai mygeayrt ‘around’, G. mágcuairt mə gîˈərt [mәˈɡɪә̯rt] ao bleayst ‘husk, egg-shell’, G. blaosc: blîəst [blɪә̯st] geay ‘wind’, G. gaoth: gîə [ɡɪә̯] ?ao/ua cleayney ‘seduce’, G. claonadh, cluain: klîˈənə [ˈklɪә̯nә] (see fn. 170) 3.5.2.6 ei [ɛi̯ ] < G. ao Marstrander notes a diphthongal realization ei [ɛi̯ ] of ao before r in two items: deyr ‘dear’, G. daor: deir [dɛi̯ r] seyr ‘free; carpenter’, G. saor: seir [sɛi̯ r] This may be explained as an onglide before the [ɹ], noted elsewhere in Late Manx (though more usually as [ә], §4.2.3). However, a similar realization is recorded in seose ‘up’, G. suas, sëis, seiːs (NM, HLSM II: 389). 189 Aspects of the historical phonology of Manx 3.5.2.7 ū [uː] < G. ua Long tense back ū [uː] is noted as a reflex of ua in certain items (Marstrander: 69) where back realizations are well-attested: hooar ‘got, found’, G. f(h)uair: hūr [huːr] w̥ ūr [ʍuːr] seose ‘up’, G. suas: sūs [suːs] 3.5.2.8 ‘ø’ Marstrander (69) describes a vowel sound resembling Norwegian ø and occurring in various environments and being a reflex of various Gaelic vowels, including ao and ua. Both short and long versions of this vowel are noted. Foran dentaler, likvider og sibilanter får vokalene ofte i betonet stilling en modifiseret uttale, som for et norsk øre bryter på ø. Denne uttalen skyldes en senkning av vokalen. Fortungevokaler senkes til en midttungevokal av lignende karakter som eng. ea i earth, ja ikke sjelden til en baktungevokal som kan ligge like dypt som eng. u i but. [Before dentals, liquids and sibilants in stressed position the vowels often receive a modified pronunciation, which to a Norwegian ear resembles ø. This pronunciation is due to a lowering of the vowel. Front vowels are retracted to a central vowel of similar character to English ea in earth, indeed not seldom to a back vowel which can be as low as English u in but.] (Marstrander: 69) The exact quality of this vowel is not identified, perhaps deliberately owing to uncertainty on the part of the author, and the examples are not transcribed phonetically, but rather simply listed in their orthographic form. The comparison with Norwegian [ø] would suggest a rounded vowel, but English [ɜː] in earth and [ʌ] in gun are unrounded (at least in Received Pronunciation). It is worth noting that English /ʌ/ is generally adopted as /ø/ in loans into Norwegian, as in lønsj ‘lunch’, pønk ‘punk’. See Lewin (2018: 172–4; also §3.5.6 below) for discussion of confusion between front rounded and unrounded vowels in fieldwork on the Gaelic languages, including that of the Norwegian Nils Holmer. 190 Aspects of the historical phonology of Manx In line with the view adopted throughout this chapter that Manx reflexes of ao(i) and ua(i) were most likely unrounded, or at most only weakly rounded (cf. §§3.5.1.1; 3.5.5; 3.5.6), a neutral central vowel transcription will be adopted here and Marstrander’s ‘ø’-vowels will be transcribed as [ә] and [әː] here. According to Marstrander, they are found in the following items: [ə] ea edyr ‘at all’, G. eadar, idir jiarg ‘red’, G. dearg (also long vowel) jialg ‘thorn’, G. dealg, also dʹžolg [dʲʒɔlɡ], dʹžōlg [dʲʒɔːlɡ], pl. dʹžïlg [dʲʒɪlɡ] io fys ‘knowledge’, G. fios, also fïs [fɪs] kys ‘how’, G. cionnas, also kïs [kɪs] chirrym ‘dry’, ScG. tioram, Ir. tirim chymsagh ‘gather’, G. tiomsaghadh myn ‘fine’, G. mion burley ‘cress’, G. biolar shutternee ‘neigh’, Ir. siotrach, seitreach, ScG. sitir, sitrich smuirr ‘marrow’, G. smior éa, ao eddrym ‘light’, G. éadrom, ScG. aotrom éa, u red ‘thing’, G. réad, rud á, ái snaid ‘thread’, G. snáthaid garaghtee ‘laugh’, G. gáireachtach á mayl ‘rent’, G. mál slane ‘whole’, G. slán lane ‘full’, G. lán bane ‘white’, G. bán ui, (i)u dhyt ‘to you’, G. duit, dut rhyt ‘with you’, G. riut cur ‘put, send, give’, G. cu(i)r u surranse ‘suffer’, English sufferance oi stermagh ‘stormy’, G. stoirmeach 191 Aspects of the historical phonology of Manx o brott ‘broth’, G. brot ao sleayd ‘tow, drag’, G. slaod eayl ‘lime’, G. aol keyl ‘slender’, G. caol meayl ‘bald’, G. maol keirn ‘rowan’, G. caorthann aoi Jardain ‘Thursday’, G. Déardaoin éa, ao feddyn / geddyn ‘get, find’, ScG. faotainn uai (Je)lune (sic) ‘Monday’ /i/ < /iː/ < ighea aigh cheet ‘come’, G. tigheacht mainshter ‘master’, G. maighstir [əː] ea fer ‘man, one’, G. fear aarloo ‘ready’ G. earlamh, OIr. aurlam etc. (§2.1.10) jiarg ‘red’ (also short vowel) ei keird ‘craft’, G. ceird ao teayst ‘dough’, G. taos seihll ‘world’, G. saoghal freoagh ‘heather’, G. fraoch172 aoi (Jy)heiney ‘Friday’, G. Dia hAoine keayney ‘cry’, G. caoineadh skeayley ‘spread’, G. scaoileadh eoylley ‘dung’, G. aoileach eash ‘age’, G. aois deiney ‘men’, G. daoine feysht ‘question’, G. faoiside 172 This item is mentioned separately at the end of the paragraph: ‘Stundom også foran andre konsonanter. Jeg har hørt freoagh „lyng‟ uttalt med samme senkede vokal’(‘Sometimes also before other consonants. I have heard freoagh ‘heather’ pronounced with the same retracted vowel’). It is unclear whether Marstrander means a short or long vowel in this word. Rhŷs (18) notes that he usually heard a short vowel in freoagh. 192 Aspects of the historical phonology of Manx ua seose ‘up’, G. suas /aː/, /ɛː/ <(e)a baare ‘top’, G. barr baarney ‘gap’, G. bearna ard ‘high’, G. ard giare ‘short’, G. gearr, also gʹēr [ɡʲɛːr] éa Baarle ‘English’, G. Béarla abha goayr ‘goat’, G. gabhar loayrt ‘speak’, G. labhairt adha eairk ‘horn’, G. adharc 3.5.2.9 Analysis It is unclear whether the rather wide range of following consonantal environments listed by Marstrander (69) (dentals, liquids, sibilants, as well as [x] in freoagh) is relevant in conditioning the appearance of the ‘ø’ vowel phone. Many of those items which are not reflexes of ao(i), especially when the vowel is long, are followed by [r], which is probably relevant (§4.2.3). The large number of items with G. io reflects the tendency across Gaelic dialects for backing of /i/ before broad consonants, reflected in Manx orthography by the use of <y> (§2.1.7). Finally, the class of items slane, lane, bane and perhaps mayl probably represent centralization in Northern Manx of á [ɛː] with shortening and prominent preocclusion noted by Rhŷs (160) and Broderick (HLSM I: 161) (§4.5.2). Most of the remaining items represent G. ao(i). The eight items with aoi and also Jelune (uai) would represent Rhŷs’s œ, œ̄ [ә̞(ː)]. Otherwise, Marstrander records aoi and fronted uai as giving ē ̣ [eː], with one example of î [ɪː] (skeayley) and îə [ɪә] (mygeayrt). This general picture of aoi / uai > [әː] > [eː] is in line with Rhŷs’s findings. For ao, we have five items giving a short ‘ø’ vowel and two the long version. Three of the short realizations are before [l], where Rhŷs (21) also notes a short vowel. Other realizations of ao are î [ɪː] (four items), ē [ɛː] (two items), ei [ɛi̯ ] (two items, both with 193 Aspects of the historical phonology of Manx following r) and îə [ɪә̯], iə [iә̯] (three items, two of which may show interference from ua). Although Marstrander’s data apparently show more variation in realizations of ao than Rhŷs’s, including more overlap with aoi, it should be noted that higher realizations represent only ao, but not aoi, and î is only found for ao, apart from one instance of aoi. It is likely that Marstrander’s î corresponds to Rhŷs’s ȳ, although the variation of realizations in Marstrander, and the variety of descriptions of î / ȳ in both authors suggest variation in the articulation, as well as uncertainty on the part of the fieldworkers. For ua, Marstrander notes mostly high front or centralized unrounded vowels, which may or may not be diphthongized: ī [iː] (five items), îə [ɪә] (five or six items), î [ɪː] (one item). These would correspond to Rhŷs’s ī [iː], ü [ɨː]. Marstrander’s data appear to agree broadly with Rhŷs’s descriptions: ao > [әː], [ɪː] ua > [ɪә], [ɪː], [iә̯], [iː] aoi / uai > [әː] > [eː] It is striking that Marstrander’s data, when collated and examined, agrees so well with Rhŷs’s, given that Marstrander gives no consideration in his description to grouping Manx sounds according by phonemes or diachronic developments, and does not discuss Rhŷs’s data. 3.5.3 Jackson (1955) Jackson finds the development of ao(i) in Manx to be ‘varied and confused’: In Manx the history of ao and aoi is varied and confused; previous writers have noted chiefly sounds which mean öː and eː or ɛː, rarely ɪː or uː […] The curious thing is that several different stages in the history of a sound should appear contemporaneously in the language of one small island, as they seem to do. (Jackson: 47–8) He records ‘the same bewildering variety’ with ua(i), and suggests that ao(i) and ua(i) may have fallen together. 194 Aspects of the historical phonology of Manx In Manx, there is the same bewildering variety [with ua(i)] as with ao and aoi; indeed the two might be said to have fallen together if it were not that there is also ɔː from ua, and that ö-sounds seem commoner from ua, uai than from ao, aoi. However, this latter may be illusory and due to the scantiness of the examples. (Jackson 1955: 52) The mention of ɔː is not relevant to the question of whether ua(i) and ao(i) have, as a regular development, fallen together, since it is found only in neose, seose ‘up’, which has been noted as having an irregular development (§3.4.6). The matter of the ‘ösound’ is discussed below. Jackson’s caveat regarding ‘the scantiness of the examples’ should be borne in mind. Jackson’s data are as follows: [eː], [ɛː] ao aoi seyr ‘carpenter’, G. saor: seːɹ (NM) keyl ‘narrow’, G. caol: keːl (EK) meayllee ‘hornless cow’, G. maolaidhe: meːli (NM) seihll ‘world’, G. saoghal: sɛːl (NM, TL) yn theihll ‘of the world’, G. an t-saogha(i)l: ən tɛːl (NM) deiney ‘men’, G. daoine: deːnʹɩ (EK) Jeheiney ‘Friday’, G. Dia hAoine: dʒɛˈheːnʹɩ (EK) keayney ‘weeping’, G. caoineadh: keːnʹɩ (EK) ua keayn ‘sea’, G. cuan: kɛdn (EK) eayn ‘lamb’, G. uan: jɛdn (EK) uai cleayshyn ‘ears’, G. clua(i)s: klɛːən (TL) [ɪː], [ɪə̯] ao keyllys ‘strait’, G. caolas: kɪːləs (EK) geay ‘wind’, G. gaoth: gɪːə (JTK, HB) aoi nuy ‘nine’, G. naoi: nɪː (NM) ua eayn ‘lamb’, G. uan: ɪːdn (JTK) creagh ‘stack’, G. cruach: krɪːax (JK) feayr ‘cold’, G. fuar: fɪːəɹ (JTK, TL) 195 Aspects of the historical phonology of Manx geayl ‘coal’, G. gual: gɪːl (TL), gɪl (JK) keayn ‘sea’, G. cuan: kɪːdn (JTK, HB, JK) uai mygeayrt ‘around’, G. má gcuairt: mə gɪːt (JK, TL) cleayshyn ‘ears’, G. clua(i)s: klɪːdn (JTK) With shortening: Gaelg, Gailck, G. Gaoidhealg: gɪlk (JK), gɪlg (NM) [uː], [uə̯] ao un ‘one’, G. aon: uːn (JTK) foyr ‘edge of blade’, G. faobhar: fuːəɹ (EK)173 ua feayr ‘cold’, G. fuar: fuːəɹ (TL), fuːr (EK) booa ‘cow’, G. bó: buə (EK) uai ooashley ‘worship’, G. uaisle: uəʃli (EK) oor ‘hour’, G. uair (but probably Middle English ‘hour’ > *úr): uːr (EK) ö [ə] ao red ‘thing’, G. réad, raod, rud: röd (EK), rɪd (TL), c’red ‘what’: krɛd (JTK) aoi geay ‘wind’, G. gaoth: göi (EK)174 ua geayl ‘coal’, G. gual: göːl (EK) geaylin ‘shoulder’, G. gualann: göːlɪʃ [sic] (EK) seose ‘up’, G. suas: söːs (NM) keayn ‘sea’, G. cuan: ködn (EK) cleaysh ‘ear’, G. cluais: klöːʃ [ɔː] ua 173 seose, neose ‘up’, G. suas, anuas: sɔːs, nʹɔːs (EK) EK’s fuːəɹ for foyr ‘edge of a blad’ (G. faobhar), would represent a development of aobh to [uː] parallel to that of naomh giving Manx noo ‘saint, holy’, nuː (JW) (HLSM II: 326). The other realizations given in HLSM (II: 176) are similar to fo̜ ːr (EKh), suggesting a development with aobha falling in with (e)abha as in lioar ‘book’ (G. leabhar) lʹo̜ ːr (NM). 174 ‘this must really be the genitive geayee = Ir. ScG. gaoithe, pronounced gɛi by [H]B in gollan geayee “swallow”’ (Jackson: 50). 196 Aspects of the historical phonology of Manx [ei] aoi nuy ‘nine’, G. naoi: nei (EK) Jackson’s öː and ɪː would suggest that ao and ua survived as distinct phones in Manx (whether or not they were distinct from one another), and had not simply fallen in with /iː/, /iә̯/, /eː/ or /ɛː/. Indeed, Jackson (36–7, 50–1) does not give any realizations of ao(i) or ua(i) with iː at all, and his ɪː does not appear as a variant of í or ia. Jackson’s ɪː is defined as a long version of short ɪ ‘as in English hit’ (11), and seems to be intended to be equivalent to Marstrander’s î (Jackson: 48). Like Marstrander, Jackson notes this sound both from ua(i) and ao(i). Two examples of a diphthongal realization of ua with [ɪ] as the primary element are given, although as Jackson notes, in the case of feayr, the schwa element could be explained by the final [r]. Jackson describes ö(ː) as follows: There is generally only one [ö-sound], a slightly retracted central ö, very poorly rounded, further back and more open than ø in French peu, and somewhat closer than œ in French pleure; though occasionally it is open enough to write œ. It is usually long, öː; when short (and sometimes when long) it represents a special development of the other vowels before (less often after) r. Marstrander heard it also on occasion before dentals, l, and sibilants…, but I did not meet this. (Jackson: 12) Jackson’s examples of [eː] and [ɛː] representing ao may represent a fronted variant of an /әː/ phoneme. This may represent an ongoing merger of ao with /eː/ or /ɛː/, perhaps to be understood as a partial loss of the fairly lexically restricted phoneme /әː/ in obsolescent Manx. That is, the terminal speakers may have been unsure which lexical items should have /әː/ and which should have /eː/, a confusion which may have been confounded by a tendency to centralize vowels before /r/ and possibly in other environments, as well as the relative smallness of the set of items with /әː/. 197 Aspects of the historical phonology of Manx 3.5.4 Wagner (1969) Wagner’s data for ao(i) and ua(i), extracted from the Manx glossary in LASID (IV: 173–88), are as follows. Tokens are marked S (south), N (north) or M (from Marstrander’s unpublished notes) (§1.5.4). [i(ː)], [iə̯] ao deyr ‘dear’, G. daor: diːr (S) y Cheyllys ‘the Sound’, G. caolas: ec y Cheyllys eɣə ˈxiːləs (M), kʹiːlʹəs (S), kʹiləs (S), kʹi.lʹəs (S) geay ‘wind’, G. gaoth: giːə (N x5), giː (N) aoi deiney ‘men’, G. daoine: diːnə (N) ua geayl ‘coal’, G. gual: giḻː (N), gʹil (S), gilː (N), gilʹː (N), gilʹ (N), gil (S) feayr ‘cold’, G. fuar: fiːə (N), fiːər (N x2), fiː (N), fʹiːə (N), fʹiːr (N), fiər (N) feayraght ‘cold’ (noun), G. fuaracht: firɑ ̣x keayn ‘sea’, G. cuan: kʹidn (S x2, N x3), kʹi.dn (S), kʹidn (S), kidn (S) geaylin ‘shoulder’, G. gualann: gilən leah ‘soon’, G. luath: liː (S), lŭiːə (M) creagh ‘stack’, G. cruach: kʹrix (S), kʹriəxən (N), kʹriəx (N), kriəx (N) uai mygeayrt ‘around’, G. magcuairt: məˈgita (N), məˈgi.ət (N), məˈgĭət (N x2, S), məˈgi.ət (N), mygeayrt-y-mysh məˈgitə moʃ (N), məˈgiːrt (M), məˈgit (N, S), məˈgiːt (N) cleaysh ‘ear’, G. clua(i)s: tliːʃ (N), kliʃ (N), kliːʒən (N) eayin ‘lambs’, G. uain: i.dn (N x2), idn (N x2) neayr’s ‘since’, G. an uair is: nĭərəs (S) geayltyn ‘shoulders’, G. *guailtean: giʟʹᴛʹʃən (S x2), ‘shoulder’ giʟʹᴛʹʃ (S) keayrt ‘time (occasion, instance)’, ny cheayrtyn ‘sometimes’, G. cuairt: kʹiətən (N), kʹiːtən (S), kʹiəðən, kʹiːt (S) ɪ, ï [ɪ(ː)] ao geay ‘wind’, G. gaoth: gɪː (S x2) eddin ‘face’, G. éadan, ScG. aodann: ïdn (S), nïðən (S) ua skeab ‘brush’, G. scuab: skɪ.ʙ (S) geayl ‘coal’, G. gual: gɪlʲ (N) 198 Aspects of the historical phonology of Manx cheayll, geayll ‘heard’, G. chuala: gɪ.l (S), kɪl (M) keayn ‘sea’, G. cuan: kɪ.dn (S) uai geayney ‘green’, G. uaine: kʹïnjə rimlagh ‘fishing-line, rope’, G. ruaimneach: rïmləx (S) y, ü ua geayl ‘coal’, G. gual: gy.l (S) uai rimlagh ‘fishing-line’, G. ruaimneach: rümlɑx (S, M) [eː], [ɛː] ao eaddagh ‘clothes’, G. éadach, ScG. aodach: eðax (S), eðɑx (N), ɛðɑx (N) eddin ‘face’, G. éadan, ScG. aodann: eðɑn (S), neðən (S) seihll ‘world’, G. saoghal: sɛːl (N), seːl (M) keyrrey ‘sheep’, G. caora: kɛrə (N x3), kɛru (N), kɛrɑx (N) tead ‘rope’, ScG. taod: tɛᴅ (N) aoi eash ‘age’, G. aois: ɛːʃ (N) keayney ‘cry’, G. caoineadh: kʹeːnjə (N, S), kʹeːŋʲə (N), kʹèːnə (N) Jerdein ‘Thursday’, G. Déardaoin: de dedn (N), dʒɑ ̣ ˈdeːdn (M) Jeheiney ‘Friday’, G. Dé hAoine: de hɛnə (N), dʒe ˈheːnjə (M) ua eayn ‘lamb’, G. uan: ɛːn (N) uai eayin ‘lambs’, G. uain: ɣẹdnʲ (M) Jelhein ‘Monday’, G. Dé Luain: de lidn (N), dʒe lʲeːdn (M) öː [ɘː] ao eaddagh ‘clothes’, G. éadach, ScG. aodach: as eaddagh zöðəx (N), öðɑx (N), nöðɑx (S) eayl ‘lime’, G. aol: ’əl (N x2) eddin ‘face’, G. éadan, ScG. aodann: öðɪnʲ seihll ‘world’, G. saoghal: töːl (S) uai cleaysh ‘ear’, G. clua(i)s: klöʃ (S), klöːʒən (S x2) 199 Aspects of the historical phonology of Manx [uː], [uə̯], [ɯː], [oː] ao geay ‘wind’, G. gaoth: go̤ i (S x2), go-i (S) (probably gaoith(e)) ua feayr ‘cold’, G. fuar: fûːṛ (S), feer feayr ‘very cold’ ˈfiːfuː (S), fuː (S), fûːr (S), fuːr (S), fuːr (M) heose ‘up’, G. thuas: hλːs (N) uai hooar ‘got, found’, G. fhuair: hooar ee huːi (N), huə̯r (S), huːə (N), huː (N), huːr (M) For ao we have two items with ə [ә] or öː [ɘː] eayl ‘lime’ (G. aol): ’əl (N, two attestations) and seihll ‘world’ (G. saoghal): töːl (S). For the short vowel in eayl, cf. Rhŷs (21). There are two southern attestations of geay ‘wind’ (G. gaoth) as gɪː, and five northern attestations of giːə, plus one of giː. For the diphthongal realization (as if gaoth > *guath), noted by Rhŷs (17) as being more of a northern feature, see §3.9.3. The remaining items include five instances of [iː], and seven of [ɛ(ː)]. These may represent loss of the /әː/ phoneme as perhaps evidenced in Jackson’s data, or freer allophonic variation in an obsolescent variety than would have been found in full native speakers of an earlier generation. Interestingly, four of the instances of [iː] are in the item keyllys ‘sound’ (G. caolas). Broderick notes realizations of this item with apparently rounded vowels y. [yˑ], ũ175 from NM and yː from HK, as well as back unrounded λː [ɯː] from HK. NM also has front high unrounded i, as in Wagner’s transcription. These high realizations could represent a high allophone of /әː/ before /l/, as noted by Rhŷs (17) in eayl ‘lime’ (G. aol), for which he records ü [y(ː)] as a variant (§3.5.1.2). For aoi Wagner gives five items, all with [e(ː)] or [ɛ(ː)], i.e. the less conservative variant noted by Rhŷs, apart from one instance as [i]. For ua we have 15 items and 45 tokens. Of these there are 29 instances of [i(ː)] or [iә̯] and 5 instances of [ɪ(ˑ)]. These 34 instances out of 45 (75.6%) would represent the expected realizations of ua following Rhŷs (§3.5.1.3). The one instance of y. in geayl ‘coal’ (G. gual) may represent the more conservative reflexes noted by Rhŷs. The 175 The vowel nasalization here (HLSM II: 504) is unexpected, especially in view of the general lack of nasalization in the Manx of the terminal speakers (§5.6); but for other examples of unhistorical vowel nasality adjacent to voiceless consonants, see Lewin (2019a: 82–9). 200 Aspects of the historical phonology of Manx difficult to interpret lŭiːə for leah ‘soon’ (G. luath), has been counted under [iː], but may also represent a more conservative realization as in Rhŷs’s lůy̆ or lüə. The remaining instances are the expected back realizations of feayr ‘cold’ (G. fuar) (also in hooar ‘got, found’, G. fuair), an instance of heose with back unrounded λː [ɯː], and an instance of [ɛː] in eayn ‘lamb’, G. uan. The [ɛː] in eayn may represent confusion between the singular and plural. For uai, several of the items given are non-prototypical in one way or another owing to phonological developments or paradigmatic analogy (keayrt, mygeayrt, cleaysh, geayltyn, §§3.9.5, 3.9.6). The high vowel [i] in Jelhein ‘Monday’ (G. Dia Luain)176 and eayin ‘lambs’ (G. uain), is more difficult to explain, unless the preocclusion plays a role; compare LASID (IV: 188) blidn or bli.dn, blein ‘year’ (G. bliadhain > Manx *bléin) for expected and also attested bʹlʹẹːnʹ (NM) (HLSM II: 34). In the case of eayn there may also be confusion between singular and plural. The vowel in kʹïnjə g(h)eayney ‘green’, G. uaine, is unexpected, but given the unexpected initial [k], the word may have been only half-remembered by the speaker; compare Rhŷs’s (24) comments on its near obsolescence. There are two instances of expected [eː] in eayin and Jelhein, although both of these are from Marstrander’s material. 3.5.5 Broderick (1986) (HLSM III) Broderick (HLSM III: 138) claims that [e(ː)] is the most common realization of ao and aoi: In L[ate] S[poken] M[anx] this [AO] is mostly found as /e(ː)/, sometimes as /i(ː)/ or /u(ː)/, occasionally as /o(ː)/; also [ö(ː)], [y(ː)], [ɯ(ː)]. AO may also represent ÉA before /t/, /d/. (HLSM III: 138) Here [in AOI] there is mostly /e(ː)/, sometimes as /i(ː)/ or /uː/. (HLSM III: 138) 176 Although the nominative form G. Luan with broad /n/ apparently existed, Cregeen’s Lhein, Lheiun (§3.4.3). 201 Aspects of the historical phonology of Manx For ua and uai he gives a number of realizations, but does not state which are more common: [UA and UAI] are found as follows: a) as /u(ː)/ or /oː/, b) as /i(ː)/ or /e(ː)/, c) as [ö(ː)], [y(ː)], [ɯ(ː)]. There may also be the diphthongs /ua/, /uә/, /iә/, /ei/. (HLSM III: 139) Broderick gives a small number of examples for each of the noted realizations, but does not attempt to quantify the frequency of their occurrence. He notes (HLSM III: 58) that [y(ː)] and ‘occasionally’ [ɯ(ː)] are sometimes attested for ao(i) and ua(i) in ‘about a dozen items’ out of the ‘about ninety or so lexical items containing in their stressed element […] (Ir./ScG.) AO(I), UA(I)’, ‘in southern Manx especially’. These are discussed in a section entitled ‘THE ALLOPHONES [y(ː)], [ɯ(ː)]’: In LSM [y(ː)] functions essentially as an allophone of /iː/ and [ɯ(ː)] as an allophone of /u:/ in limited circumstances only. In order to explain these circumstances clearly it will be necessary to refer to the Irish and Scottish counterparts of the Manx items involved. [fn.] Both [y(ː)] and [ɯ(ː)] may be found in apparent free allophonic variation in the same limited circumstances with realizations of /i/, /e(ː)/, /u/, /u(ː)/ (HLSM III: 57) It is not clear what exactly is meant here by ‘functions essentially as an allophone’. If these phones are restricted to a particular set of lexical items, which can only be defined with reference to etymology, then the most natural conclusion would seem to be that they represent a distinct phoneme or distinct phonemes (which may nonetheless be in the process of falling in with /iː/ or /eː/), rather than that they are allophones of /iː/ etc. Otherwise, we would expect to find these phones representing historical G. í, é, ú, which is not the case. Broderick (HLSM III: 60) goes on to claim that: (a) ‘in Manx […] AO(I) and UA(I) have practically completely fallen together’; (b) that [yː] and [ɯː] were perhaps more prevalent in these lexical items in the past, but equally that these phones may have existed (presumably over a long period) side by side with the more frequent variants [eː] and [iː]; 202 Aspects of the historical phonology of Manx (c) the phones [yː] and [ɯː] historically represent a phoneme equivalent to /ɯː/ in Scottish Gaelic and Donegal Irish: In Manx, as we have seen, AO(I) and UA(I) have practically completely fallen together and are noticed mostly as /eː/ or /iː/ (also with secondary shortening). We have also seen, however, that in some instances AO(I) and UA(I) appear as [y(ː)], occasionally [ɯ(ː)], thus (in the case of [ɯ(ː)]) falling in with the treatment of AO(I) in most of Scotland and parts of Donegal. These may have been more prevalent in Manx than the present evidence admits, and in theory could appear in all ninety or so words of this class. But as the Manx treatment of AO(I) and UA(I) falls in largely with the common treatment of AO(I) in Ireland, viz. /eː/ or /iː/, it does not necessarily follow that [yː] or [ɯ(ː)] were more prevalent than they are now, but that they have existed side by side with the numerically greater attested /eː/ and /iː/, particularly in southern Manx. The testimony as we have it today makes it clear that [y(ː)] and [ɯ(ː)] are evidenced only in the context of (some) AO(I), UA(I) items where both are found side by side with each other, or either one or the other is found and (in theory) could be substituted for the other. Outside this context neither is attested. The allophones [y(ː)] and [ɯ(ː)] would then be the representatives in LSM at any rate of the phoneme /ɯː/ found commonly in Scottish Gaelic (and to an extent in Donegal Irish) for AO(I). (HLSM III: 60) Broderick does not mention mid front or central (rounded or unrounded) vowels in this passage, although these are noted as reflexes of ao and ua by Rhŷs, Marstrander, and Jackson. Broderick does have a section on ‘/ә/ as [ö] or [öː]’ (HLSM III: 44–8): In certain circumstances /ә/ can be realized as [ö] or [öː] (i.e. articulated with a degree of retraction and lip-rounding). It is the result of retraction or advancing of the other vowel phonemes particularly /e/ or /eː/, especially in the environment of /r/ (even though /r/ on occasion may not be realized), but to a lesser extent in the environment of /l/, /s/ and /sʹ/, /t/ and /d/, /m/ and /n/, also /x/, /g/, /b/. (HLSM III: 44–5) [ö] would sound similar to [ә], but would tend towards [ø]. (HLSM I: 3) Several examples of items with ao(i) and ua(i) are given (cleaysh ‘ear’, G. cluais; geayl ‘coal’, G. gual; seihll ‘world’, G. saoghal; seose ‘up’, G. suas), presented as 203 Aspects of the historical phonology of Manx examples of phonemes /eː/, /iː/ retracted, but without consideration of the possibility that the central vowel may be original here (HLSM III: 46). As we have seen, Broderick claims, like Jackson, that ua(i) and ao(i) appear to have largely fallen together in Manx, and that they have at least partially fallen together with /iː/ or /eː/, although he recognizes that the existence of realizations such as [yː] and [ɯː] suggests that there may be or have been a distinct phoneme. He suggests this may have been something like the Scottish and Donegal phoneme /ɯː/, implying that he believes ua(i) had fallen in with ao(i). He suggests a parallel for this in the fact that ‘in the north [of Ireland] UA(I) can appear as /ua/ and after labials there may be unrounding, so that UA(I) may fall together with AO(I)’ (HLSM III: 60). This apparent parallel is discussed below in §3.8. It is difficult to interpret the passage ‘it does not necessarily follow that [yː] or [ɯ(ː)] were more prevalent than they are now, but that they have existed side by side with the numerically greater attested /eː/ and /iː/, particularly in southern Manx’ (HLSM III: 60). In this passage, Broderick appears to suggest the possibility that there was a stable period in the past when this old phoneme */ɯː/ was partially merged with /iː/ or /eː/, or merged in some dialects or idiolects and not in others. This scenario is justified with reference to the development of ao(i) in most Irish dialects. However, it is not clear why a similar outcome in Manx and Irish should necessarily be taken as evidence for a parallel pattern of development over a similar time-scale. Broderick comes to his conclusions based on ‘the present evidence’ (i.e. the material from the terminal speakers). It would have been useful, however, to consider other evidence, particular the orthographic evidence (both from Phillips and the later system), and especially Rhŷs, as we have done in the present study. In contrast to the scenario outlined by Broderick, Rhŷs (§3.5.1) paints a picture of nineteenth-century Manx in which (a) there is no general merging of ua and ao, but only of uai and aoi (in certain items), (b) the distinctive realizations of ao and ua (i.e. /әː/ and /ɨә̯/) are equally prevalent in the north and the south, (c) while ua seems to be the way to merging with /iә̯/ or /iː/, there is no suggestion that ao was realized as [iː] or [eː], but only as ȳ and variants thereof. In order to see whether Broderick’s 204 Aspects of the historical phonology of Manx conclusions are reasonable on the basis of his own data, a quantitative analysis of tokens from HLSM was carried out, as detailed in the following section. 3.5.5.1 A quantitative analysis of data from HLSM (II) Tables 42–45 and Chart 1 show the number of instances of different realizations of ao, aoi, ua, uai in the data in Broderick’s dictionary (HLSM II).177 All items with a known etymology from G. ao(i) or ua(i), are included, except those forming a diphthong with vocalized G. gh, dh etc. (§3.9.1). Those items with known idiosyncratic or variable developments are included (e.g. feayr ‘cold’, G. fuar; hooar ‘got, found’, G. fuair, §3.4.6).178 The wide range of phonetic realizations in Broderick’s data have been grouped into the following broader categories for the purpose of analysis. Vowel length is ignored: Grouping of vowel phones representing ao(i), ua(i) in HLSM179 [o]: o, ɔ [ɯ]: λ [ɯ], λə [ɯә̯] [u]: u, ui, uə [e]: e, ei, eə, ɛ 177 All transcribed individual instances given after the headwords in HLSM (II) were included. Different realizations from the same speaker were counted separately, and where the same transcription is noted as being from n speakers, this is counted n times. Data from the example sentences were not included, as they generally duplicate instances given in the individual item transcriptions, but sub-headings (compound words, inflected forms etc.) were included. Extracting the data from the dictionary (rather than the larger task of combing the texts in vol. 1, which, however, in any case do not include the dialogue material found in the dictionary) runs the risk of giving undue prominence to the less frequent realizations, as these are likely to be listed exhaustively with only a selection of the commonest realizations of the most frequent items. However, given the overall relatively small size of the corpus, it is likely that the data are close to being exhaustive for most items. 178 Deayrtey, ‘pour, spill’ (G. doirteadh, dortadh, dórtadh, duartan) is included under ua (see §3.9.8). Forms are excluded when it cannot be determined with certainty which class they should be assigned to (for example whether feːʃlə for feaysley ‘untie, release’, should be regarded as representing G. fuascladh or a by-form *fuaiscleadh based on the stem feayshil). Also excluded are blatant spelling pronunciations (§1.6.9.2), such as those of feoh ‘hate’ (G. fuath), as fẹːo (TC, HK) and fjoː (JW), while expected fiːə (TT) is included.178 179 This categorization may admittedly not be perfect. [ә] in particular represents a large range of symbols for various mid-central and mid-front vowels, but most commonly the ö described in the extract above (cf. also Jackson and Wagner’s use of the symbol). ɪ and ï could plausibly be included as variants of [ә], but they could also be grouped with [i]. For this reason, [ɪ] is taken as a distinct category. 205 Aspects of the historical phonology of Manx [i]: i, iə [ɪ]: ɪ, ï [ɨ] [ә]: əi, ᴇ [œ], ø, ë [ɘ] [y]: y, yə, ui Table 42. Realizations of G. ao in HLSM II y ə 11 16 11.7% 17.0% sum items: 22 ɪ 5 5.3% i 25 26.6% e 29 30.9% u 3 3.2% ɯ 4 4.3% o 1 1.1% bleayst (blaosc), cleayney (claonadh), deyr (daor), deyrey (daoradh), eayl (aol), freoagh (fraoch), geay (gaoth), geayagh (gaothach), inney-veyl (inghean mhaol), keyl (caol), Keyllys (caolas), keynnagh (caonach), keyrrey (caora), meayll (maol), seihll (saoghal), seihltagh (saoghalta), seyr (adj.), seyr (n.) (saor), seyrey (saoradh), sleayd (slaod), teayst (taos), theyreeyn (?daoraidh) sum tokens: 94 Table 43. Realizations of G. aoi in HLSM II y ə 0 7 0.0% 10.9% sum items: 15 ɪ 1 1.6% i 7 10.9% e 47 73.4% u 2 3.1% ɯ 0 0.0% o 0 0.0% deiney (daoine), deyrid (daoire), eash (aois), eoylley (aoileadh), freayney (?raonadh), geaysh (gaoisid), geayshagh (gaoisideach), Jerdein (Déardaoin), Jyheiney (Dé hAoine), keayney (caoineadh), meaish (maois), nuy (naoi), ny s’deyrey (níos daoire), skeayley (scaoileadh), Teare (Mac an tSaoir) sum tokens: 64 Table 44. Realizations of G. ua in HLSM II y ə 14 10 10.7% 7.6% sum items: 23 ɪ 10 7.6% i 69 52.7% e 8 6.1% u 11 8.4% ɯ 6 4.6% o 3 2.3% beayn (buan), beaynee (buanaidhe), cheayll / geayll (chuala), creagh (cruach), deayrtey (dórtadh, cf. duartan), eaghtyr (uachtar), eaghtyragh (uachtarach), eayn (uan), feayr (fuar), feayraght (fuaracht), feoh (fuath), geayl (gual), geaylin (gualainn), heose (thuas), keayn (cuan), leagh (luach), leah (luath), leaystey (luascadh), leaysteyder (luascadóir), neose (anuas), seose (suas), skeab (scuab), skeabey (scuabadh) sum tokens: 131 206 Aspects of the historical phonology of Manx Table 45. Realizations of G. uai in HLSM II y ə 11 10 9.6% 8.8% sum items: 12 ɪ 14 12.3% i 41 36.0% e 26 22.8% ɯ 2 1.8% u 10 8.8% o 0 0.0% cleaysh (clua(i)s), eayin (uain), feayshil (fuascail > *fuais(c)il), geayltyn (cf. ScG. guailnean), g(h)eayney (uaine), gleashaght (gluaiseacht), hooar / dooar (fuair), Jylhein (Dé Luain), keayrt (cuairt), mygeayrt (má gcuairt), ooashley (uaisle), rimlagh (ruaimneach) sum tokens: 114 Chart 1. Realizations of G. ua(i), ao(i) in data from HLSM ɪɪ uai y ә ua y ә aoi ә ao y 0.0% ɪ ɪ ɪ e i 20.0% ɪ u 40.0% y ә 50.0% ɪ u ɯ o e i 30.0% i e 60.0% u ɯ ɯ o u e i ә 10.0% u e i ɯ 70.0% 80.0% 90.0% 100.0% o These data challenge Broderick’s claim that ao(i) and ua(i) have simply fallen together. For two of the categories, there is clearly one realization which is considerably more frequent than the others. For ua, this is [i] (51.9%) (rising to 60.8% if the idiosyncratic feayr and seose / heose / neose, with their [u] realizations, are excluded). For aoi, the most frequent realization is [e] (73.4%). Ao is the most varied, with a fairly even split between [e] (30.9%) and [i] (26.6%). Ao has the highest percentage of [ә] realizations (17.0%), and [ә], [ɪ], [ɯ] and [y] taken together — which might plausibly be taken to represent variants of a phoneme /әː/ — account for 38.3% of instances of ao. 207 Aspects of the historical phonology of Manx The preponderance of [i] realizations of ua is consistent with Rhŷs’s observation of [iː, iә̯] as a reflex of ua, as the culmination of a process of fronting and unrounding, alongside more conservative, less fronted realizations. The predominance of [e] realizations of aoi is consistent with Rhŷs’s observation of [eː] alongside conservative œ̄ . The mixed results for ao may suggest the preservation of a distinct phoneme /әː/, alongside an apparent tendency (perhaps a recent and unstable one related to language obsolescence) to merge this with /eː/ or /iː/. The results for uai are also mixed. Rhŷs would lead us to expect a preponderance of [e], as with aoi, since according to him uai and aoi have merged (at least in some environments or items). However, in the data from HLSM, [i] realizations are the most frequent for uai (36.0%, 41 occurrences) with [e] in second place (22.8%, 26 occurrences). This seems to be due to the large number of occurrences of keayrt, mygeayrt (G. cuairt), cleaysh (G. cluais) and geayltyn (*guailtean) (68 occurrences, 59.6% of total for uai), which may pattern with ua rather than uai (§§3.7, 3.9.5–6). When all such idiosyncratic items are excluded from the uai category (also hooar / dooar, ooashley, rimlagh), [e] then accounts for 47.1% of instances (16) — in line with the expectations of merger between uai and aoi — and [i] for 32.4% (11). However, there are then only 6 items with a sum of 34 occurrences. 3.5.6 Front rounded realizations? A number of the descriptions discussed above either implicitly or explicitly record front rounded realizations of ao(i) and ua(i) in Manx. Rhŷs and Marstrander both make comparisons between a single Manx phone and both rounded and unrounded phones in other languages (§§3.5.11, 3.5.2.8), which makes their comments difficult to interpret. Jackson (48) casts doubt on Rhŷs’s (and Kneen’s) descriptions of front rounded vowels for ao(i), but himself records a phone ö with a degree of rounding (Jackson: 12). Broderick (HLSM III: 57) gives both [y(ː)] and [ɯ(ː)] as possible realizations of ao(i), ua(i), and these are apparently interchangeable, cf. both gyːl and gλːl [ɡɯːl] from NM for geayl ‘coal’ (G. gual). 208 Aspects of the historical phonology of Manx In listening to recordings of the last speakers for the present thesis (especially the material recorded by the Irish Folklore Commission), I did not hear any realizations which could be firmly identified as [yː] or [øː] etc., although there were plenty of centralized, unrounded realizations which could be characterized as [әː], [ɨː], [ɪː], perhaps [ɯː]. Some of these could perhaps impressionistically resemble vowels with some rounding, however. It is possible that some of the reports arise from confusion between back unrounded vowels and front rounded vowels. In an experiment, Ladefoged (1967: 133–141) gave eighteen trained phoneticians the task of listening to recordings of a number of Scottish Gaelic words including various stressed monophthongs and plotting them on a cardinal vowel diagram. While the responses were quite accurate for cross-linguistically frequently-occurring vowels such as /i(ː)/, /e(ː)/ and /u(ː)/, the phoneticians’ judgments of /ɯ(ː)/ and /ɤ(ː)/ varied greatly in degree of rounding and backness. The question of front rounded vowels in south-western Scottish Gaelic dialects raises similar difficulties; see O’Rahilly (29) and Lewin (2018: 172–4) for discussion. Since there is no clear, unambiguous evidence for front rounded vowels in Manx, the descriptions analysed above have been interpreted as referring to front-central but unrounded vowels. It remains possible, however, that there was a degree of rounding (perhaps no more than ‘very poorly rounded’, in Jackson’s [12] words) which would have served to enhance the contrast between front-central /әː/ and /ɨә̯/ on the one hand, and fully front vowels /iː/, /iә̯/, /eː/, /ɛː/, on the other. 3.6 Written evidence The two main orthographies used to write Manx are especially challenging to interpret with regard to reflexes of G. ao(i) and ua(i), but nonetheless provide valuable evidence for change in these vowels from c. 1600 onwards. 209 Aspects of the historical phonology of Manx 3.6.1 Representation of ao(i), ua(i) in Phillips 3.6.1.1 Overview In the Phillips manuscript there are dozens of orthographic sequences representing G. ao, aoi, ua and uai. These range from one to three characters, with substantial overlap between representations of the four categories. Despite this complexity, clear patterns are discernible which can be related to the phonological developments attested from the other evidence discussed in this chapter. In the following discussion, as elsewhere in the thesis, Phillips’ diacritics are disregarded (§1.6.3), and <æ>, and occasional instances of <ai>, are treated as equivalent to <e>. Taking the evidence of the initial character of the orthographic sequence, namely , <e> (inc. <æ>), <y> and (including a few instances of <o>) (Table 46, Chart 2), the following observations can be made: (a) Spellings of the <y> type, while frequent (>30%) in all four categories, are especially prevalent in the case of ao (74.6%). This presumably represents the nonmerging allophone of /әː/ described by Rhŷs (§3.5.1). (b) Spellings of the type are most prevalent in the ua category, and can be associated with back realizations of the historical /ua/ diphthong. (c) Aoi and uai have somewhat similar profiles, with <e> and <y> being the dominant representations, although <e> is more prevalent in the case of aoi. This suggests that at this early date splits were already emerging between ao and aoi, and ua and uai, respectively, with aoi and uai moving towards the merger with one another seen in the later language. The <e> and <y> spellings can be interpreted as the allophone of /әː/ described by Rhŷs which tends towards merger with /eː/ (§3.5.1.4). (d) The type is not frequent in any category (<6% for ao, aoi and uai), but is somewhat more frequent (14.9%) in the case of ua, where it can be interpreted as representing fronted reflexes of historical /ua/. 210 Aspects of the historical phonology of Manx Table 46. Spelling of G. ao(i), ua(i) in Phillips, by initial letter of orthographic sequence no. of total <e> <y> <u/o> lemmas 16 114 6 13 85 10 ao 21 101 2 52 30 17 aoi 27 174 26 4 58 52 ua 25 158 8 57 68 25 uai total % 16 21 27 25 ao aoi ua uai 114 101 174 158 5.3% 2.0% 14.9% 5.1% <e> <y> <u/o> 11.4% 74.6% 8.8% 51.5% 29.7% 16.8% 2.3% 33.3% 29.9% 36.1% 43.0% 15.8% Chart 2. Spelling of G. ao(i), ua(i) in Phillips, by initial letter of orthographic sequence 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% ao aoi ua <e> <y> uai <u,o> A glance at the tokens (§3.4) suggests that at least some of the above orthographic categories should be broken down further. Subsequent characters in orthographic sequences seem to be particular important in indicating diphthongal realizations, e.g. <ya> is especially prevalent in the ua set, <y> and <yy> in ao, <ye> and <ey> in aoi 211 Aspects of the historical phonology of Manx and uai. To investigate this further, the following finer-grained breakdown of the <e> and <y> categories has been brought to bear (Table 47): Table 47. Breakdown of orthographic categories <e> and <y>, taking into account following vowel characters <e(VV)> <e> (including <æ>, <ai>) alone or followed by one or more other vowel characters (including <ee>) <e(V)y(V)> <e> followed by one or more vowel characters, at least one of which is <y> <y(y)> <y> or <yy> <ya(V)> <ya> only, or followed by an additional vowel character <ye(V)> <ye> only, or followed by an additional vowel character other <y> <y> in combination with vowel characters other than the above, e.g. <yi> <e↔y> <e(V)y(V)> and <ye(V)> combined This breakdown is utilized in Table 48, Chart 3. Table 48. Spelling of G. ao(i) and ua(i) in Phillips ao no. of lemmas 16 total <e(VV)> <e(V)y(V)> <y(y)> <ya(V)> <ye(V)> other <y> <u/o> 114 6 5 8 68 9 7 1 10 aoi 21 101 2 36 16 10 3 11 6 17 ua 27 174 29 3 11 21 34 12 3 61 uai 25 158 8 25 32 3 14 30 21 25 <e(VV)> <e(V)y(V)> % ao no. of lemmas 16 total <y(y)> <ya(V)> <ye(V)> other <y> <u/o> 114 5.3% 4.4% 7.0% 59.6% 7.9% 6.1% 0.9% 8.8% aoi 21 101 2.0% 35.6% 15.8% 9.9% 3.0% 10.9% 5.9% 16.8% ua 27 174 16.7% 1.7% 6.3% 12.1% 19.5% 6.9% 1.7% 35.1% uai 25 158 5.1% 15.8% 20.3% 1.9% 8.9% 19.0% 13.3% 15.8% 212 Aspects of the historical phonology of Manx Chart 3. Spelling of G. ao(i), ua(i) in Phillips 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% ao <e(VV)> aoi <e(V)y(V)> <y(y)> ua <ya(V)> <ye(V)> uai other <y> <u/o> The breakdown of the data in Table 48, Chart 3 reveals the following details: (a) Orthographic sequences involving only the character <y> (single or doubled) are overwhelmingly concentrated in the ao category (59.6%, as opposed to <13% for the other categories), providing more robust evidence of monophthongal and non-merging /әː/. (b) The sequence <ya> is especially prominent in the ua category (19.5%), and less frequent (<9%) in the other categories. It is suggestive of fronted, central reflexes of historical /ua/ (i.e. [ɯә], [ɨә̯] or similar), but without monophthongization. (c) Spellings involving <e> followed by <y>, or by a sequence of vowel symbols including <y>, are more frequent in aoi (15.8%) and uai (20.3%) than in the other categories (<5%). This may be associated with /әː/ > [әː], [eː], as discussed above. (d) If, as seems reasonable, it is assumed that <e(V)y(V)> and <ye> represent similar reflexes, and their totals are combined (Chart 4), then the association between <e↔y> and aoi (26.7%), uai (39.2%) is clearer, as opposed to ao and ua (both 13.2%). 213 Aspects of the historical phonology of Manx Chart 4. Spelling of G. ao(i), ua(i) in Phillips, with <e(V)y(V)> and <ye(V)> combined 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% ao aoi <e(VV)> <e↔y> ua <y(y)> <ya(V)> uai other <y> <u/o> Note that monophthongization [ɨә̯] > [ɨː] seems to be more prevalent in certain environments or items (§3.5.1.3). Representations of ao(i) and ua(i) show a distinct pattern from those of /iː/, /ia̯ /, /eː/ and /uː/,180 the front and back vowels with which they might be expected to show (near-)merger. The data in Table 49 is from entries under A to C in Thomson’s (1953) glossary. All show very consistent (84.6% – 100%) use of the expected symbols , <e, æ> or in various combinations. Orthographic sequences based on <y, yy> are almost exclusively characteristic of ao(i) and ua(i), which also have more heterogeneous ranges of possible representations in general, as shown above. The orthographic evidence presented in this section strongly points to the conclusion that ao(i) /әː/ and ua(i) /uә̯/, /ɨә̯/, /әː/ were contrastive both with each other (with the exception of probable merger between some reflexes of aoi and uai), and with /iː/, /iә̯/, /eː/ and /uː/.181 180 Mostly representing G. í, ia, é, ú, but also new long vowels arising from fricative vocalization, as in bea /beː/ ‘life’ (G. beatha). 181 And also with /ɛː/ (G. á(i), ó(i)), which may however have been lower [aː] in Early Manx and is generally represented with variants of <a> (§2.2.4). 214 Aspects of the historical phonology of Manx Table 49. Orthographic representations of /iː/, /iə̯/, /eː/, /uː/ in Phillips (A–C, Thomson [1953] glossary) i (54), í (23), ii (3), íí (15), iy /iː/ ia (2), ía (2), íæ (2), ie (2), iy, iŷ /iә̯/ ie, îæ̂ (3), ia, éy (4), ey, êi, ée (18), ee (4), éa (4), ea (14), ê, é (5), e (5), áy, æy, æíí, /eː/ /uː/ æi (2), ǽé, ǽe, æa, ǽ (8), æ (5), á, a eu, iu (5), iú (11), iú, íu (2), îû, iúy, ôô (1), ou (3), u (28), ú (19), ŭ, ui (19), úi (5), ûi (2), ŭi, uy (18), uŷ, uy̆ , úy (7), yu total /iː/ /iә̯/ /eː/ /uː/ 96 <i(V)> 96 (100.0%) 10 <iV> 10 (100.0%) 104 <e(V)> 88 (84.6%), <iV> 14 (13.5%), <a(V)> 2 (1.9%), 132 <u(V)> 123 (95.5%), <oV> 4 (3.0%), <eu, yu> 2 (1.5%) 3.6.1.2 ua(i) in Phillips: preceding consonant conditioning and lexical diffusion Further examination of the orthographic evidence from Phillips with regard to G. ua(i) sheds light on the phonological split between back /uә̯/ and the fronted reflexes /ɨә̯/, /әː/. In the following analysis (Table 50, Chart 5), the sequences of symbols are split into three categories, based on the first symbol (as in Table 46 above), e.g. <ya> is classed under <y>. It is assumed that spellings represent the most fronted and unrounded realizations, while spellings represent the most back and rounded realizations. It is difficult to discern how <y> and <e> (including <æ>) spellings might represent differing sounds so they have been categorized together as <y/e> and assumed to represent prototypically intermediately unrounded and fronted realizations, i.e. roughly the later attested [ɨ]. It is possible that <e> spellings in some cases represent /әː/ rather than /ɨː/, however. The small number of <o> spellings are categorized together with .182 There is clear evidence of conditioning by preceding consonant. 182 <o> spellings are consistent in the case of bwoalley, bwoaill ‘hit’ (G. bualadh, buail), and this is excluded since it could perhaps represent *bóladh, *bóil, as the later spelling could be interpreted as suggesting (HLSM data on the other hand suggests /uː/, /u/) (§3.4.5). 215 Aspects of the historical phonology of Manx Table 50. Spelling of G. ua(i) in Phillips by place of articulation of preceding consonant no. of total <y/e> <u/o> lemmas vowel initial 8 84 8 75 1 labial 12 67 4 41 22 coronal 25 179 14 95 70 dorsal 8 0 53 26 27 % vowel initial labial coronal dorsal total 8 12 25 8 84 67 179 53 <y/e> 9.5% 6.0% 7.8% 49.1% <u/o> 89.3% 61.2% 53.1% 50.9% 1.2% 32.8% 39.1% 0.0% Chart 5. Spelling of G. ua(i) in Phillips by place of articulation of preceding consonant 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% vowel initial labial coronal <y/e> dorsal <u/o> For both the labial and coronal categories, <u/o> spellings represent around a third or more of instances. For the dorsal category, there are no instances of <u/o> at all. Also, there are no lexical items with preserved back /uә̯/ in Classical or Late Manx with preceding dorsal consonants. Superficially similar fronting and unrounding of ua in Ulster Irish after labials has been explained by Ó Dochartaigh (1987: 110) as dissimilation between the labial consonant and the round quality of the vowel in items such as fuar ‘cold’ (§3.8). In Manx, however, feayr ‘cold’ is among the items which may have retention of backness, 216 Aspects of the historical phonology of Manx and it might be hypothesized that the fronting represents rather dissimilation between a dorsal consonant and the back quality of the following vowel, with subsequent (incomplete) extension to other environments. This would explain why no instances of spellings occur after dorsal consonants in Phillips, and why no lexical items are found in the which there is categorical blocking of fronting after dorsal consonants. In addition, the earlier and more complete fronting of ua(i) after dorsal consonants is suggested by the much greater prevalence of spellings in this environment (49.1%) than in the others (all <10%). It should be noted, however, that the dorsal category includes relatively few tokens (53) and lemmas (8); there are over three times more tokens in the coronal category. Moving to the level of individual lexical items, it appears that a process of lexical diffusion was underway, which was less advanced in the period of the language represented by the Phillips text, judging by spellings such as lua for later leah (G. luath), luagh for later leagh (G. luach), tua for later theay (G. tuath), búan for later beayn (G. buan). None of these have /uә̯/, /uː/ in the later language, although Rhŷs’s descriptions suggest a more back realization in some varieties, which however was apparently contrastive with the <oo>, <ooa> /uː/, /uә̯/ vowels (§3.5.1.3). However, leah (G. luath), for example, is also spelled liǽ and lýa; and other items such as feayr (Phillips fiýar-, fúar, G. fuar) seem to show the later attested range of variation. Some innovating realizations are found in Phillips where the more conservative back realization is found in the later language (e.g. yasyl /ɨә̯sәl/ for later ooasle /uә̯sәl/, /wusәl/), attesting to the existence of variant forms of which it was not always the newer form which survived (§3.4.6). Similarly, Phillips’ spellings of G. fuair ‘got, found’, ScG. fhuair, d’fhuair, mostly suggest /hәːrʲ/, /dәːrʲ/, and while this realization is attested in Late Manx, the back realization was apparently more widespread (§3.8.2). For the apparent height contrast between ua = [uә̯, ɨә̯, iә̯] and uai = [әː, eː], conditioned by the broad or slender quality of the following consonant, see above (§3.6.1.1). 217 Aspects of the historical phonology of Manx 3.6.2 Representation of ao(i) and ua(i) in CM orthography In the standardized orthography of the main eighteenth-century texts, most items have one fixed spelling, making the kind of analysis given for the Phillips’ orthography (which looks for trends in a large mass of micro-variation both between and within lemmas) less feasible. There also appears to be considerably more overlap between orthographic sequences representing ao, aoi, ua, uai, and other vowels, as the following lists show.183 To avoid more sporadic or idiosyncratic spellings, only those forms attested in the Bible, and/or from both Cregeen and Kelly, are given. <eay> ao: bleayst (blaosc), cleayney (claonadh) eaynagh (aonach), eayl (aol), feayn (fao(i)n), geay (gaoth), meayl (maol), sleaydey (slaodadh), teaym (taom), t(h)eaymey (taomadh), t(h)eayst (taos) aoi: keayney (caoineadh), geaysh(t) (gaoisid), skeayl(l)ey (scaoileadh) ua: beayn (buan), beaynee (buanaidhe), cheayll, geayll (c(h)uala), deayrtey (dórtadh, *duartadh), eayn (uan), feayr (fuar), feayght (fuacht), feaysley (fuascladh), geayl (gual), geaylin (gualann), keayn (cuan), leaystey (luascadh), theay (tuath) uai: cleaysh (clua(i)s), feayshil (fuascail, *fuaisil), geayltyn (guailne, *guailtean) g(h)eayney (uaine), keayrt (cuairt), mygeayrt (má gcuairt), neayr’s (?an uair is) agh: reayrt (radharc) é: eayst ‘moon’ (éasca), falleays ‘gleam’ (EIr. folés, ScG. faileas), geayr ‘sour’ (géar), s’leayr ‘clear’ (is léir) ia: shleayst, also slheeayst, slheeas(s)id ‘thigh’ (sliasaid) other: freayll ‘keep’ (friotháladh), jeayst ‘beam’ (Eng. ‘joist’), meayn ‘ore’ (? mian, méin), skeay, also skeeah (sceith) <eayi> aoi: uai: eayil (aoil) eayin (uain) <ea(h)> ao: aoi: ua: ceaghley (claochlódh, ScG. caochladh), eash (aois), freaney ‘rage, roar’ (raoineadh) creagh (cruach), eaghtyr (uachtar), leagh (luach), leah (luath), seaghyn ‘sorrow, trouble’ (? *suathachán), skeab (scuab), sleayst (sluasaid) gleashagh(t) (gluaiseacht) earkan ‘lapwing’ (adharcán) beasagh ‘compliant’ (béasach), blean ‘flank, groin’ (bléan), breag ‘lie’ (bréag), clea ‘roof’ (cliath, cléith), crea ‘creed’ (créadh), eadaghey ‘jealousy’ (éad), eaddagh ‘clothes’ (éadach), eajee ‘abominable’ (éidigh), eam ‘call’ (éigheamh), fea ‘rest, quiet’ (féath), greasee ‘shoemaker’ (gréasaidhe), kease uai: agh: é: 183 For meanings of ao(i), ua(i) and agh items, see tables in §3.4, 3.10.1. 218 Aspects of the historical phonology of Manx /eː/: ái: ío: i: ea: ‘buttock, ham’ (céas), jea ‘yesterday’ (indé), jiulean ‘tenant farmer’ (deidhbhléan), mea ‘fat’ (méith, méath), pleat ‘talk (cf. pléadáil), rea ‘flat’ (réidh), rheam ‘province’ (réim), sheayney ‘bless’ (séanadh), trean ‘valiant’ (tréan) bea ‘life’ (beatha), clean ‘cradle’ (cliabhán), chea ‘flee’ (teitheadh), drea(i)n ‘wren’ (dreathan), feanish ‘witness, evidence’ (fiadhnaise), imnea ‘anxiety’ (imnidhe), jeadjagh (deithideach), jean ‘do, make’ (déan), lhean ‘wide’ (leathan), mean ‘middle’ (meadhón), millchea ‘mildew’ (ScG. mill-cheò), offishear ‘officer’, peeikear ‘spy’, rea ‘ram’ (reithe), soilshean ‘shine’ (soillseán) earroo ‘number’, (áireamh), gearey ‘smile, laugh’ (gáire), nearey ‘shame’ (náire). creagh ‘furrow’ (críoch) earish ‘time, weather’ (iris) feallagh ‘ones, people’ (? eallach) <eai> uai/i̯ /: s’leaie (is luaithe) agh: eairk (adharc) éi: eaishtagh ‘listen’ (éisteacht), feailley ‘feast, festival’ (féile), reaish ‘span’ (réise) other: keagh ‘madness’ (cuthach, ScG. caothach, caoch) <ey> deyll (daol), deyr (daor), deyrey (daoradh), feysht (faoiside), inney-veyl (inghean mhaol), keyl (caol), keyllys (caolas), keynnagh (caonach), keyrrey (caora), seyr (saor), seyrey (saoradh) aoi/i̯ /: fey-yerrey (faoi dheireadh) é: dangeyr ‘danger’ (dainséar), geyre ‘sharp’ (géar), rey ‘rid’ (réidh), shey (sé) ái: -eyr (-(e)óir) e.g. shiolteyr ‘sailor’ (seóltóir) eith: brey ‘birth’ (breith) eadh: fey ‘throughout’ (feadh) /eː/ livrey ‘deliver’ oi: keyll ‘wood’ (coill), rheynn ‘share, divide’ (roinn) other: veyge ‘voyage’ ao: <eyi> ao: aoi: seyir (saor) tarroo-deyill (*tarbh daoil) <eiy(gh)> aoi: keiyn (caoin) ao(i)/i̯ /: lheiy (laogh), meiygh (mao(i)th), streiyraght (sraothar) ua(i)/i̯ /: seiy (suaitheadh) /eːi̯ /: jeigh (iadhadh) /әi̯ /, /ei̯ /: eiy ‘footlock’ (iodh), eiystyr ‘halter’ (adhastar), eiyrt ‘follow, chase’ (adhart), feiyr ‘noise’ (foghar), leigh (ScG. lagh), cleiy ‘dig’ (cladh), shleiy ‘spear’ (sleagh), teiy ‘choose’ (togha) 219 Aspects of the historical phonology of Manx other: beiyn ‘animals’ (ScG. beathaichean) <eo(a),(y),(h)> ao: freoagh (fraoch) aoi: eoylley (aoileach), feoilt(agh) (faoilte) ua: beoyn (?buan), creoghys (cruadhas), feoh (fuath), heose, neose, seose (thuas, anuas, suas) agh: leodaghey (laghdaghadh) other: ben-treoghe ‘widow’ (baintreabhach), scarleod ‘scarlet’ (scarlóid), skeog ‘lock of hair’ (ciabhóg) <eo(a)i(e)> aoi/i̯ /: freoaie (fraoich) uai/i̯ /: leoaie (luaidhe), leoie (luaith), creoi (cruaidh) other: keoi ‘mad’ (cuthach, ScG. caothach, caoch), skeoigh ‘tidy’ (sciamhach) <ae> ao: gaerr, gaer (ScG. gaorr) ó /ɛː/: aeg ‘young’ (óg) ae /ɛː/: aer ‘air, sky’ (aer, EIr. aër, ScG. adhar, aighear) <ei> ao: aoi: uai: é: /eː/: éi/i̯ /: ei: 220 keirn (caorthann) deiney (daoine) lheill, lheihll (luadhail) beill ‘mouths’ (béil), beisht ‘beast’ (béist), beishteig ‘beast’ (béisteog), breid ‘veil’ (bréid), boteil ‘bottle’ (cf. Ir. buidéal), brein ‘stinking’ (bréan), -eil (verbal noun ending, -(e)áil), eilley ‘armour’ (ScG. féileadh), eisht ‘then’ (éis), er-jeid ‘on edge’ (déad), erreish ‘after’ (tar éis), garveigagh ‘roar’ (béic), jeig ‘teen’ (déag), jeir ‘tear(s)’ (déar), jeirk ‘alms’ (déirc), keiley ‘sense’ (gen.) (céille), keim ‘step, stile’ (céim), lheim ‘jump’ (léim), lheiney ‘shirt’ (léine), meill ‘lip’ (méill), meir ‘finger(s)’ (méar), my-yeish ‘in ear’ (ma dhéis), reir ‘satisfy’ (réir, riar), sheidey ‘blow’ (séideadh), sleityn ‘mountains’ (sléibhte), treigeil ‘abandon’ (tréig), veign ‘I would be’ (bhéinn, bhínn), y cheilley ‘each other’ (a chéile) ben-rein ‘queen’ (ríoghan), blein ‘year’ (bliadhain), bundeil ‘bundle’, carmeish ‘canvas’, freill ‘keep’ (friotháil), geill ‘attention’ (géill), greiney ‘sun’ (gen.) (gréine), lheid ‘such’ (leithéid), meiley ‘bowl’ (?), oaseir ‘overseer’, preis ‘press’, reill ‘rule’ (riaghladh), sheiltyn ‘think’ (saoil, síl), shirveish ‘serve, service’ (seirbhís), treisht ‘trust, hope’ jei ‘after’ (i ndéidh) beinn ‘peak’ (beinn), bheill ‘grind’ (meil, ScG. beil), creid ‘believe’ (creid), greimmey ‘grasp’ (greimeadh), greinnaghey ‘encourage’ (ScG. greannachadh, *greinn-), keiltyn ‘hide’ (ceilt), keird ‘craft’ (ceird), meinn ‘meal’ (min, mein), sheilg ‘hunt’ (seilg) Aspects of the historical phonology of Manx ai, oi: clein ‘kin’ (clainn), eirin(n)agh ‘farmer’ (aireamh), geid ‘steal’ (goid), geill ‘spring’ (gail, goil), geinnagh ‘sand’ (gainmheach), skeilt ‘cloven’ (scoilte), s’meilley ‘feebler’ (is maille), teirroo, terriu ‘bulls’ (tairbh) ai, oi + /i̯ /: deinagh ‘wearisome’ (? doighean, daighear), eirey ‘heir’ (oighre), other: eilkin ‘errand’ (?), -eig, also -age, -aag, -aig (diminutive suffix, -(e)óg), keint ‘kind’, queig ‘five’ (cúig, cóig), steillyn, steillin (ScG. stàillinn) <ei(g)h> ao: seihll (saoghal) aoi/i̯ /: dreih (draoi), mreih (mnaoi) uai: lheihll, lheill (luadhail) ua(i)/i̯ /: sleih (sluagh), treih (truagh), teigh (tuagh), veih (uaidh) éi/i̯ /: lheihys ‘healing’ (léigheas), spreih (spréidheadh) ny-yeih ‘however’ (ina dhéidh) ei/i̯ /: jeih ‘ten’ (deich), meih ‘weight’ (meidh), sneih ‘vexation’ (snighe) /әi̯ /: leih ‘forgive’ (loghadh), reih ‘choose’ (rogha) <oaie> ua(i)/i̯ /: oaie (uaigh) óigh: doaie ‘decency’ (dóigh) éi/i̯ /: oaiagh ‘perjurous’ (éitheach) oiche: cloaie ‘stone’ (gen.) (cloiche) /әi̯ /: oaie ‘face’ (aghaidh) ao(i): i(o): kirree (caoraigh, caoirigh) ching ‘sick’ (tinn), shimmey ‘many’ (is iomadh), and many others. ao: u(i): io: un (aon) muc ‘pig’ (muc), tushtey ‘understanding’ (tuigse), and many others. urley ‘eagle’ (iolar), and others. <iy> aoi: riyr (araoir, aréir) aoi/i̯ /: siyr (saothar, saoithear) /ai̯ /: criy ‘gallows’ (croich), lhiy ‘colt’ (lo(i)th), piyr ‘pair’ (ScG. paidhear), siyn ‘vessels’ (ScG. soithichean) io: er-giyn ‘following’ (iar gcionn) <uy> aoi: ua/i̯ /: iodh: iú: iu: nuy (naoi) ruy (ruadh) fuygh ‘wood’ (fiodh) shuyr ‘sister’ (siúr) juys ‘fir’ (giuthas) agh: ymmyd (adhmad) <y> 221 Aspects of the historical phonology of Manx io: <egh> ao/i̯ /: fynney ‘fur’ (fionnadh), myn ‘fine’ (mion), ynnyd ‘place’ (ionad), and many others streghernee (sraothar) Although there are no orthographic forms which are exclusive to reflexes of G. ao(i), ua(i) and agh (with the near exception of <eo>), and little to distinguish ao(i) and ua(i), the following observations about the distribution of forms can be made: (a) The representational overlap is mainly between ao(i) /әː/, ua(i) /ɨә̯/, /әː/ on the one hand and é /eː/ on the other, and to a lesser degree á /ɛː/, í /iː/, ia /iә̯/. This would appear to support the conclusion that there was a significant degree of phonetic similarity between these vowel sounds, and that ao(i) and ua(i) were fairly fronted, and mid to high, as the other evidence presented in this chapter also shows. (b) Although some of the main orthographic representations of ao(i) and ua(i), namely <eay>, <ea>, <ey>, can also represent the front mid to high vowels noted in (a), it is noteworthy that there are a number of representations which never represent ao(i), ua(i), such as <ee, eey> (/iː/), <eea, ia> (/iә̯/), <ay, ai, e_e> (/eː/), <aa, ay> (/ɛː/). This, and (c, d) below, would suggest that there were indeed phonological contrasts between reflexes of ao(i), ua(i) and these other vowels. (c) Only one representation, <eo> (including <eoy>, <eoh>, <eoa>, <eoi>), is more or less unique to ao(i) and ua(i) (although it does not help in distinguishing between the two), apart from its obviously quite distinct use for /oː/ in a few other items. To some extent the use of <eo> in the Bible orthography seems to be a recessive survival from more widespread use in less standardized versions of the CM orthography. It is used for example in CS (1707): sleoi ‘sooner’ (s’leaie, G. is luaithe), feosle ‘relieve’ (feayshil, G. fuascail); in the 1796184 edition of PC: cheoyn ‘sea’ (keayn, G. cuan), feon ‘expansive’ (feayn, G. fao(i)n), beoyn ‘eternal’ (beayn, G. buan), chleosh ‘ear’ (cleaysh, G. cluais), feoyr ‘cold’ (feayr, G. fuar), among others; and in variant 184 Possibly deriving without substantial revision from a manuscript from the first half of the eighteenth century (Max Wheeler, personal communication) (§1.3.1.1). 222 Aspects of the historical phonology of Manx spellings in Cregeen, e.g. theo ‘common people’ (theay, G. tuath), cleoyn ‘propensity’ (cleayn, G. claon, or cluain?). (d) <eay>, although also representing /eː/ in a handful of items, and /iә̯/ in a couple more, is also very strongly associated with ao(i) and ua(i). It might be wondered why, firstly, no clear way to distinguish ao(i) from ua(i) (especially ao /әː/ from ua /ɨә̯/) was adopted, and secondly, why less ambiguous representations such as <eo> (or the likes of Phillips’ <yy> and <ya>) seem to have been dispreferred and replaced by more ambiguous forms such as <ea>. To an extent, it is likely that this reflects the organic way in which the orthography evolved through the interplay of the idiosyncratic preferences of different individuals and texts. In such a process, certain distinctions and patterns might with time come to be made more clearly and consistently, but equally, there was no guarantee that the most “logical” orthographic forms would prevail. It should be remembered that accurate representation of pronunciation was not necessarily the chief concern of Manx writers (Thomson 1984: 307; Thomson and Pilgrim 1988: 4) (§§1.6.4.2, 1.66). They were native speakers of Manx and knew how the language was pronounced; their concern was with the transmission of the texts they needed to write, and with distinguishing individual lexical items (especially the many homophones and near-homophones) for the sake of semantic clarity, rather than with accurate representation of phonological contrasts which, in any case, would have had a light functional load. Another consideration is that the very complexity and fluidity of the situation with regard to the relationship between ua(i) and ao(i) may have militated against marking the contrasts too finely, since this would result in spellings suited only for certain dialects or idiolects. There may have been an impetus, whether conscious or unconscious, to develop an orthography which could encompass multiple varieties of Manx, especially in view of the collaborative process by which the Bible was 223 Aspects of the historical phonology of Manx translated by clergy originating from and residing in different parts of the island. Recall the variation attested in the following items:185 [ɨә̯] ~ [uә̯] [әː] ~ [eː] ~ [uә̯] [ɨә̯] ~ [әː] [ɨә̯] ~ [әː] ~ [eː] [ɨː[ ~ [uː] ~ [oː] feayr ‘cold’ (fuar) hooar (non-standard also heyr etc.) ‘got, found’ (fuair) geay ‘wind’ (gaoth) cleaysh ‘ear’ (cluais) heose etc. ‘up’ (thuas) The consistent retention of <eo> in heose, seose, neose may reflect the existence of a form with /oː/ (§3.4.5). Otherwise, the apparent restriction in the use of <eo> in the standard may reflect an aversion to orthographic forms which diverge too far from English norms; compare the substitution of <ai> for <ia> etc. (§1.6.5), and the problems caused by <eo> in the Manx of the terminal speakers, some of whom produce spelling pronunciations such as fẹːo, fjoː (HLSM II: 165) for feoh /fɨә̯/ ‘hate’ (G. fuath). This would also help to explain why there seems to have been no widespread attempt to adopt Phillips’ <y(y)>, <ya> etc., despite the use of <yy> in John Woods’ 1696 sermon manuscript (Lewin 2015b: 74). It is notable that there is also no clear attempt to distinguish diphthongal /ɨә̯/ from /әː/ (and the monophthongal realization [ɨː] < ua(i) which is in evidence in Late Manx), even though the other diphthongs of this type are consistently distinguished in the CM orthography from their corresponding monophthongs (<eea, ia> /iә̯/, <ee, eey, eei> /iː/; <ooa, ua> /uә̯/, <oo, ooy, ooi, u> /uː/) (§2.2.6). This may again reflect the abovementioned reluctance to distinguish too clearly between reflexes of ua(i) and ao(i), as well as the ongoing weakening of the second element of these diphthongs (*/Va̯ / > /Vә̯/ > /Vː/) throughout the attested period of Manx. 185 There was, however, evidently no way of bridging the gap between /ɨә/ and /uә/, with one variant or the other having to be chosen. 224 Aspects of the historical phonology of Manx 3.7 Instrumental data In order to investigate the realization of the Gaelic vowels ao(i) and ua(i) in the Manx of the terminal speakers, with respect to the reported mergers or near-mergers with each other and with the front vowels /iː/, /eː/ and /ɛː/ (< G. í, ia; é; á) (§3.5), an investigation of data from audio recordings of the terminal Manx speakers made by the Irish Folklore Commission in 1948 (Manx National Heritage 2003) was carried out using the software package Praat (Boersma and Weenink 2015). All instances of ao(i) and ua(i) in the native speech samples in these recordings were included with the exception of one track, and some individual instances, where the sound quality was too low or the interpretation uncertain. For the other, more frequently-occurring vowels a sample was taken for each speaker of comparable size to the datasets for ao(i) and ua(i). For the purposes of this analysis ua and uai were combined. There were relatively few tokens of uai and most of them are of the items keayrt (G. cuairt) and mygeayrt (G. má gcuairt), which appear to pattern with ua owing to the depalatalization of final */rʲtʲ/ (§3.9.6). The total number of tokens were á (193), ao (23), aoi (35), é (58), í (141), ua(i) (41). 225 Aspects of the historical phonology of Manx Chart 6. F1 tracks for front and central long vowels (corresponding to vowel height), all speakers, Irish Folklore Commission recordings í ua(i) é aoi ao á Chart 7. F2 tracks for front and central long vowels (corresponding to backness), all speakers, Irish Folklore Commission recordings í aoi á é ua(i) 226 ao Aspects of the historical phonology of Manx Chart 8. Means for each category with 95% confidence intervals, all speakers, Irish Folklore Commission recordings, showing F1 (height) and F2 (backness) 227 Aspects of the historical phonology of Manx The frequency of the first and second formants (F1 and F2), which correspond roughly to vowel height and backness respectively, were measured over ten intervals of equal duration within each vowel, and also normalized by speaker via a z-scoring procedure (that is to say, each measurement was recalculated to show the distance from the mean, measured in standard deviations), in order to improve comparability. Charts 6 and 7 show regression curves of the normalized formant measurements against time, fit using thin plate regression splines using the R package mgcv (Wood 2006), together with 95% confidence intervals. Chart 8 shows the means of the two formant measurements for each category. Although no specific statistical significance can be attached to Charts 6 to 8 in themselves, we can discern at least some qualitative patterns, including the following:  In terms of height, ua(i) clearly patterns in the vicinity of the high vowel í, while ao and aoi are of mid height in the vicinity of é, in accordance with expectations from other sources of evidence discussed in this chapter.  In terms of backness, ua(i) again patterns with í, but with some overlap with aoi and é, while ao and aoi are further back.  While there is little apparent difference in height between ao (expected /әː/) and aoi (/әː/ [ә̟ː] or /eː/), in backness they seem quite distinct, with ao being further back (in the vicinity of the lower á), which supports the expectation of an allophonic and/or phonemic split between the two categories.  á /ɛː/ is clearly distinct from é /eː/ (§2.2.3), against the claims of merger made by Broderick (HLSM III: 50). To investigate further whether statistical significance can be discerned in these apparent contrasts, pairwise comparisons were run on the differences between the midpoints (=point 5 for our purposes). T-tests were used, correcting for multiple comparisons using False Discovery Rate (FDR) (Figures 11, 12). 228 Aspects of the historical phonology of Manx Figure 11. Pairwise comparisons using t-tests with pooled standard deviation for Irish Folklore Commission data, F1 (height) á ao aoi é í ao 0.0754 — — — — aoi 2.0e-07 0.0464 — — — é 9.6e-10 0.0047 0.3020 — — í < 2e-16 6.4e-16 5.9e-13 8.4-09 — ua(i) < 2e-16 1.3e-10 2.0e-07 6.2e-05 0.2290 P-value adjustment method: FDR Figure 12. Pairwise comparisons using t-tests with pooled standard deviation for Irish Folklore Commission data, F2 (backness) á ao aoi é í ao 0.96750 — — — — aoi 4.8e-06 0.00141 — — — é 8.4e-05 0.00507 0.75133 — — í < 2e-16 9.1e-11 0.00049 0.00017 — ua(i) 9.1e-11 2.0e-05 0.23047 0.12730 0.03169 P-value adjustment method: FDR With respect to F1 (height), we find that ao and aoi are marginally different (p-value just under .05), and there is no significant difference between é and aoi or between ua(i) and í. The difference between á and ao is marginal at best, but all the other pairs are strongly different from one another, although the difference between ao and é is more marginal than the rest. With respect to F2 (backness), there is a quite marginal difference between í and ua(i), and no difference between á and ao, between é and aoi, between aoi and ua(i), and é and ua(i). The vowels í, é and á are all different from one another. Ao is different from é and aoi, but the differences are not as great as for F1. The above results show that for the terminal speakers, á is certainly contrastive with é (against Broderick’s claim, HLSM III: 50), and there is no general merger of ao(i) and ua(i), contradicting Broderick’s claim that ‘AO(I) and UA(I) have practically completely fallen together’ (HLSM III: 60). As expected from other data presented in this chapter, ua(i) shows no height difference from í, but seems to be a little different from it in backness; ao is somewhat different from é in both height and backness; and there is evidence of merger of aoi with é. Finally, there appears to be little difference between á and ao, especially in F2, although no other evidence or descriptions suggest 229 Aspects of the historical phonology of Manx merger between these vowels. It should be stressed however that the relatively low number of tokens means that these results must remain tentative. 3.8 Fronting and unrounding of ua(i) in Manx and Ulster Irish In Ulster Irish /ua/ may be fronted to [ia̯ ] or similar in certain environments (Ó Dochartaigh 1987: 110–4): A similar explanation [to that offered for fronting of /u/ before /w/, namely dissimilation] might be offered for a phonetic shift affecting a [u] vowel when it is part of the /ua/ diphthong after a labial consonant, In a number of dialects, particularly in South Donegal, though including also some examples from further north, the /u/ of the diphthong appears occasionally as [ɪ] or [i], usually accompanied by a labialised semivowel off-glide from the preceding consonant — e.g. [fʷɪәr] for fuar. (Ó Dochartaigh 1987: 110) Ó Dochartaigh endorses Ó Searcaigh’s (1925: 19) and Sommerfelt’s (1922: 31) explanation for this, namely that the fronting represents ‘a dissimilation between the initial labial and the rounded quality of the /u/, with unrounding followed by fronting to [ɪ(ː)]’ (Ó Dochartaigh 1987: 110). Further evidence for this is an intermediate form [ɯ(ː)] replacing the [u(ː)] element of the diphthong noted by Ó Searcaigh and also by Sommerfelt (1922: 145–6), who also records the complete replacement of the diphthong by long /ɯː/ in certain items. O’Rahilly’s (37) explanation for these developments is somewhat different: More important is the change, common in Ulster, of ua preceded by a labial to Iː, ʎː etc. Thus buan is pronounced in Donegal as if it were baon, viz. b(w)ʎːn or bwIːn, comparative bwIːnʹә. In the same county fuar is pronounced fwIːr or fwIːar… In these and similar words the Donegal development was, I take it, from uːΩ (= uːә, with the second element retracted) to uΩː, by shifting of length, and thence to wʎː and wIː, the originally long u of ua being finally reduced to a w offglide from the preceding labial. (O’Rahilly: 37) 230 Aspects of the historical phonology of Manx Ó Dochartaigh notes the sequence of developments suggested by O’Rahilly, but proposes a different sequence: [uːә] > [ɯːә] > [ʷɪːә] (> [ʷEː])186 (Ó Dochartaigh 1987: 111) Ó Dochartaigh’s explanation seems preferable to O’Rahilly’s, since the latter does not give an account of why the change should take place after labial consonants but not elsewhere. The apparent similarity of these developments and Manx fronting of /ua̯ / has been noted in passing by Ó Dochartaigh (1987: 112) and Broderick (HLSM III: 60). However, on closer inspection they seem to be quite distinct developments. Fronting of ua(i) in Manx may take place after any consonant, or in initial position, not just after labials, and indeed a number of the items where back /uә̯/ is or may be retained in Manx have a preceding labial consonant. The principal exceptions in Manx are given in §§3.4.5, 3.4.6. Of these, mooar and booa do not belong to the /ua̯ / class in most Gaelic dialects, and thus might be expected to behave differently in Manx. Alternatively, fronting may be blocked in them by the labial quality of the initial consonant, as in feayr and fooar. In the case of mooar the back quality of /uː/ in the comparative smoo (G. is mó) may reinforce the retention of the back quality of the vowel in mooar. It appears that the fronting of /ua̯ / in Manx either represents centralization of the [u] element to a quality closer to the second element [a] or [ә], or dissimilation between the back quality of [u] and the dorsal quality of consonants such as [k], as suggested by Phillips’ orthography (§3.6.1.2). If this is the case, this would be more or less the mirror image of the Ulster development. The Phillips data, as well as the lexical distribution of /uә̯/ and /ɨә̯/, /әː/ in the later language, indicate that fronting spread by lexical diffusion. In the case of heose etc., the motivation for the retention of back forms might be to avoid clashing with antonyms, since the regular development of both thuas ‘up’ and The development to [ʷEː] represents a realization found in East Ulster buartha (Ó Dochartaigh 1987: 111). 186 231 Aspects of the historical phonology of Manx thíos ‘down’ could give *[hiːs] in Manx.187 The forms with [oː] may represent original non-diphthongized ós (GOI: 40), or perhaps more likely later sporadic monophthongization of /uә̯/ (cf. Ó Curnáin 2007: 92–3). The retention of the back vowel in ooasle in Classical and Late Manx (but with a fronted form attested in Phillips, §§3.4.6, 3.6.1.2) may be connected with the development of an initial glide [w] (the Late Manx attested forms seem to have /wu/ rather than /uә̯/ or /uː/) (cf. ibid.: 93), or the fact that ooasle is a high-register word. 3.8.1 /ɨə̯/ ~ /uə̯/ in feayr ‘cold’ (G. fuar): register, dialect, idiolect Rhŷs (20–1) gives the following comments on the distribution and usage of the different reflexes of ua in feayr ‘cold’ (G. fuar): the uy [uә̯] of the fūyr [fuә̯r] which I have mentioned, was heard by me in Andreas in the North, also in the South, but, mostly as a slip: the person using it would quickly correct himself into füyr [fɨә̯r] or fiyr [fiә̯r] as the pronunciation considered proper to give to a stranger, but I have heard it too often for it to have been an accident, even if we had not the northern pronunciation mentioned and the parallel instance of hooar. (Rhŷs 20–1) The observation of speakers correcting themselves suggests that a perception had developed that the fronted realization of the diphthong was more correct or standard, perhaps as a result of the spelling <eay>, even though the back variant is in fact the more conservative form. The only instance of feayr in Phillips is spelled fúar, which suggests the back variant, but the derivatives fiýarghey (later feayraghey, ScG. fuarachadh) and fyaght (feayght, G. fuacht) suggest more fronted realizations. In terminal speech we have southern fuːər and fuərɑ̆x (NM), fuːrax (HK) but also ˈfiːər (TL). From the north, only front unrounded realizations are found: fiːər (TC), fiːə (DC, HB), fiːĕ̜ (HB). From this limited data we might surmise that /fuә̯r/ was the usual 187 The potential problem is exacerbated by a tendency to substitute /s/ for /ʃ/ in sheese ‘down’ (motion). Some speakers nevertheless seem to tolerate near homophones, cf. Ned Maddrell’s [siːs] ‘down’ and [sɪːs] ‘up’. 232 Aspects of the historical phonology of Manx Cregneash form (as represented by NM and HK), but that /fiә̯r/ was found further east and north and east (including in Arbory as represented by TL). However, Rhŷs’s comments suggest that the geographical distribution was more complex (see also data from Rhŷs’s notes, Broderick 2019 s.v. feayr). 3.8.2 hooar, dooar ‘got, found’ (G. f(h)uair, d’fhuair) Rhŷs (25) gives hūœr [huә̯r⁽ʲ⁾]188 as the only realization he encountered of hooar ‘got, found’, ScG. fhuair. Most of the spellings in Phillips (§3.4.6), however, as Rhŷs points out, suggest a fronted reflex: heyr, deyr (27 occurrences, ignoring diacritic variation), hæyr (14), dýyr (1). Of the remaining occurrences, duóer (1), dûêyr (1) and fóyr (1) may represent [uә̯], though this is not clear. Rhŷs (25–6) assumes the fronted reflex attested in Phillips to be obsolete, and gives a suggestion as to why the more conservative form might survive and outlive the regular development, albeit one which is difficult to interpret: If this form had been still in use it would be probably sounded hœr: it is, however, a form phonetically later than the hooar still in use. Thus it follows that the two pronunciations hūœr and hœ̄ r have been in use together, and in this instance the reason can be detected, why the older form is the one surviving. Judging from the use made by Phillips of heyr […] ‘gat,’ in the sense of ‘begat,’ I infer that association of ideas to have told against it and enabled the older form to survive, which it does as hooar. (Rhŷs 25–6) In fact the fronted development is attested in terminal speech as hęːr (HK) (HLSM II: 221), alongside forms with uː, uːə, also from HK, as well as JTK, NM and W:N. The dependent form is attested as dhere or deayr in late eighteenth-century folksong manuscripts (Thomson 1961: 22; Broderick 1981a: 118). Nevertheless, it seems that the forms with /uә̯/ were the most prevalent in the eighteenth and nineteenth centuries. ‘For unaccented œ I have nothing to show here except the case of the diphthong ū́ œ about to be mentioned as probably involved in the pronunciation of the verb hooar ‘gat, found’’ (Rhŷs: 22). The distinctive quality of the offglide may be related to the expected palatalized /rʲ/ (§4.2.1.3). 188 233 Aspects of the historical phonology of Manx It is possible that /hәːrʲ/, /dәːrʲ/ (>[heːrʲ], [deːrʲ]), was disfavoured because of the near homophony with the semantically similar haare ‘caught; reached (place)’ (ScG. tàir) /hɛːrʲ/, future dependent /dɛːrʲ/ daare (especially considering raising of /ɛː/ before /rʲ/, §2.2.3). 3.9 Other developments 3.9.1 New diphthongs: ao(i), ua(i) + vocalized fricatives The development of ao(i) and ua(i) before vocalized fricatives, especially in fronting diphthongs, deserves special discussion. It will be necessary to refer to the interaction of these with combinations involving other historical vowels. The scope of the present chapter, and considerations of space, preclude the provision of full data tables for these, but the relevant lexical items are given in the following lists. In the ensuing discussion, reference is to data to be found under the relevant headwords in HLSM (II) and Thomson (1953), unless otherwise stated. leigh (lagh) /aɣ/, /oɣ/ > /әi̯ / eiy (iodh), eiystyr (adhastar), eiyrt (adhairt), feiyr (foghar), cleigh, cleiy (cladh), fuygh (fiodh), reih (rogha), teiy (togha), leih (loghadh), oaie, oi (aghaidh) soie (suidh) /uj/ç/ > /ui̯ / > /әi̯ / or /iː/ cloie (cluiche), broie (bruich), stroie (struidh) roie (rith) /Ri/ > /rәi̯ / or /riː/ roih (righ) oie (aidhche, oidhche) /ajçә/ > /әi̯ / or /iː/189 mie (maith) /a/oj/hj/ç/ > /ai̯ / brie (braith), crie (craith), criy (croich), drine (draighean), grih (groigh), lhie (laigh(e)), lhiy (loth, *loith), Mian (*Maitheán), moidyn (maighdean), side 189 For the special development of this item in Gaelic dialects, see Ó Maolalaigh (1997: 197). 234 Aspects of the historical phonology of Manx (saighead), sie (saich, saith), siyn (saitheach, soitheach), thie (taigh), trie (traigh, troigh) jeih (deich) /ej/ɣ/ç/ > /ei̯ / lheihys (leigheas), meih (meidh), feie, feai, fey, feiy (feadh), shleiy (sleagh), eirey (eighre, oighre) jei (diaidh, déidh) (/iaj/ɣ/ >) /eːj/ɣ/ > /eːi̯ / eie (éigh(eadh)), feie (fiadh), greie (gréith), spreie (spréidh(eadh)), jeigh ((do-)iadh(adh)), (?) lhaih (léigh(eadh))190 3.9.1.1 ao(i) + /i̯ / This mostly seems to give /әːi̯ /. Table 51. ao(i) + /i̯ / Phillips lheiy y̆ i, y̆ ei, y̆ æ̆ i, y̆ oi, (pl.) yôi fey-yerrey (Bible, Cr., K.), fei(h)yerrey (SW), fyyerrey (Cr.) freoaie meiygh, meigh (K., Hymns) gleiy (Cr.), gleih (K.) ŷú, ý, ú CM pronunciation /lәːi̯ / etymology English HLSM laogh calf /fә(ː)(i̯ ) ˈjerʲә/ e̜ i TC, CC, WQ, HK, ei TC, W:N, e̜ ːi EKh, HK, ẹːi HK, ɛ-i JTK, ɛːi NM, Eːi W:N, aˌi W:S, ɛi J:HB, J:TL, J:JTK, J:EK, (pl.) Le̜ i, Le̜ iə̯n, Løiə̯n, løiə̯n TC faoi dheireadh at last heather /mәːi̯ /, ?/mәː/ fraoich (genitive) maoth /ɡlәːi̯ / glaodh glue, slime191 tender ɛːi EK, iː NM, e̜ i EKh mẹː TC, HK, (meiyghagh) møiɑ ̣x TC 190 The spelling and some attested realizations of this item may suggest assimilation to the /ɛːi̯ / (< áigh etc.) class, although the motivation for this is unclear. 191 ‘a fibre of slime or of any glutinous matter’ (Cregeen) 235 Aspects of the historical phonology of Manx mreih (Cr., K.), mree (PC 1796) nuy; nuyoo siyr streghernee (Bible, K.) streiyraght (Bible, K.), streighyr, streiyr, streighraght, streighernee (Cr.), ý, íí /mrәː̃ ĩ̯ /, ?/mrĩː/ mnaoi woman (gen.) yú, yu, ýý, (nuyoo) nú (5), nu, nýú, nyu /nәːi̯ /, /niː/ naoi; naomhadh nine; ninth eî (2), éi, ei, êî /sai̯ r/ saothar, var. saoithear, saoithir (eDIL saithir, sáithir) sraoth, sraothartach haste /strәːi̯ әr/, /strәːi̯ әrni/, /strәːi̯ әraxt/ u̜ i JW, iː NM, ɪː J:NM, ei J:EK, ᵘi TC, ai JK, ɛ-iː JTK, (nuyoo) nu̜ iu JW, nøi-u TC, niːuː NM sɑiəʳ TC, saiɛ̆ JK, sajə, sajər W:N, sɑ ̣ːʳ HK, sɑːiəʳ TT sneeze Note the /ai̯ / diphthong in the Manx reflex of G. saothar (or rather saoithear) rather than /әːi̯ / or /әi̯ /, confirmed by the spelling and the back realizations ɑi, aj etc. (HLSM); long diphthongs appear to be disfavoured in closed syllables for prosodic reasons; cf. also riyr (§§3.4.2, 3.9.4).192 The spellings of stregher-, streighyr on the other hand suggest that this item remained disyllabic and preserved /әːi̯ / in an open syllable. As in other Gaelic varieties, the language name Gaoidhealg, Gaoidhilg is irregular, and gives a monosyllable with a short vowel in Late Manx (Table 52): 192 The [ә] in some of the HLSM realizations is probably to be interpreted as a glide associated with the Late Manx realizations of coda /r/ (§4.2.3). The metre shows it is monosyllabic in PC ll. 489, 662 and Hymns 43, 130. 236 Aspects of the historical phonology of Manx Table 52. Gaoidhealg, Gaoidhilg Gaelck, Gailck, Gaelc (K.), Gaelic, Gailic, Gaelg (Cr.), Gailck (SW, Hymns, C19 newspapers); (adj.) Gaelgagh, Gailckagh etc.193 Phillips CM pronunciation etymology English HLSM (adj.) gellgah, golgkagh /ɡilʲɡʲ/, /ɡelʲɡʲ/, /ɡilʲkʲ/, /ɡelʲkʲ/ Gaoidhealg; *Gaoidhilgeach Manx language; pertaining to Manx, Manxspeaking gʹilʹkʹ JTK, HB, gilkʹ JK, gɪlk W.N, gɪLkʹ TC, gilk WQ, ge̜ lg JW, gɪlg, gilk NM, gɪlg, gelḳ W:S, gɪlk J:JK, gɪlg J:NM; (adj.) ge̜ lgɑ ̣x JW, gɪlkɑ ̣x TC194 3.9.1.2 ua(i) + /i̯ / These also mostly appear to give /әːi̯ /, but see below. For items with G. uaidh which retain back /uәi̯ / see §3.4.5. Table 53. ua(i) + /i̯ / (excluding items with synchronic /uəi̯ /) Phillips leoie yi, y̆ ei, ýei; (G. luaithreadh) liyri leoaie creoi yei (2), yoi, ei (2), oi CM pronunciation /lәːi̯ / etymology English HLSM luaith ashes øi TC, öi JW, NM, e̜ i WQ, üi CC, ɛi EC /lәːi̯ / luaidhe lead (metal) /krәːi̯ / cruaidh hard e̜ i TC, WQ, ẹːi HK, öi NM, J:JTK, ọi EC ɛːi JK, NM, ai JK, iː JTK, öi J:EK, J:JK, ɛi HB 193 The simple form Gaoidheal ‘Gael, Gaelic-speaker’ does not seem to be attested in vernacular usage in Manx; Kelly’s ‘Gael, s. a Celt, a Manxman, an Irishman, a Highlander’ is probably an antiquarian insertion. From an early date the term Manninagh ‘Manxman’ (G. Manannach) seems to have sufficed for self-identification, and there appears to be little or no tradition of ethnic identification with the Irish or Scottish Gaels prior to the Celtic revival. The adjectives Gaelgagh (*Gaoidhilgeach) and Baarlagh (*Béarlach) are used for Manx- or English-speaking, or pertaining to the respective languages. 194 Note that in the north only forms with devoiced final /kʲ/ are found, while both /ɡʲ/ and /kʲ/ are found in the south. 237 Aspects of the historical phonology of Manx treih sleih veih teigh oaie195 seiy s’leaie ruy yei, ǽig, eyi (3), éi; (truaighe) ýi, æi (2), ýei, yei, ýi, éy, eyg, ǽyi; trǽanid, tdræyid yêi, yei (3), ŷei, yi, yǽi, y̆ ei, yei ei (2) y̆ i, (pl.) téigyn ŷôi, yoi (2), yôi (4), yŏi, ýoi, y̆ oi, ŷŷi, yei (2), yéi, yæi, (pl.) yiaghyn, ýiaghyn yoi, ey̆ i, yæi, úa, y̆ i (3), yi (2), y̆ e yéi, yêi, yei (4), ýei (2), ŷæ̂ i /trәːi̯ / truagh miserable øi TC, e̜ i JC, JW /slәːi̯ / sluagh people /vәːi̯ / uaidh /tәːi̯ / tuagh from, from him, it axe /әːi̯ / uaigh, uagh grave e̜ i TC, ɛi JK, J:EK, ai JK, NM, ei W:S, öi J:TL ei, e̜ i, ei TC, ai NM e̜ i HK, EKh, ai, ɑ ̣i TC, a-i ø-i W:N ẹːi HK, ɔːi NM, öi J:NM, ɪː W:S, ᵘi TC, ɛːi HB, ei W:N, (pl.) ẹːxən, ẹiɣən HK /sәːi̯ / suaith, suaitheadh stir, mix, trouble øi TC, saːi NM, ai W:S /s lәːi̯ / is luaithe iə NM, iː HK /rәːi̯ / ruadh sooner, earlier, quicker red, ruddy ai, a-i NM, ei HK, rɛi J:EK, J:NM, J:HB, NM 3.9.1.3 Interpreting the developments of ao(i), ua(i) + /i̯ / Rhŷs’s notes on contrasts and mergers in the diphthongs /ai̯ /, /әi̯ /, /әːi̯ /, especially the relationship between the seven near-homophonic monosyllables of the shape /lV(ː)i̯ /, are complex and difficult to interpret: Open o, short…enters into a diphthong oi as in the word […] lhiy ‘a colt’ (Ir. loth ‘a colt, a filly’); […] leigh ‘law’ (Goi. lágh [sic] […] lhie ‘a lying down’ 195 Bible MSS: oaih (Numbers 19:16,18), eoi (1 Samuel 2:6), oaie (1 Kings 2:6,9, 13:22,30), oie (1 Kings 14:13), oaieaghyn (2 Chronicles 16:14), Oiee (Revelation 6:8) oaih (Revelation 20:13), oaïh (Revelation 20:14). 238 Aspects of the historical phonology of Manx (Ir. luigh, ScG. laidh); and […] leih to ‘forgive’ (O. Ir. doluigim ‘I forgive’). Ordinarily these words have much the same pronunciation, but when attention is called to it, Manx speaking people think that they make slight distinctions. As a matter of fact they never rely on those distinctions, for they resort to other means of avoiding ambiguity: thus goll dy lhie is said both of a person going to lie down in bed and of the sun setting, whereas going to law is goll gys y leigh […] literally ‘going to the law*.’ * Since the above was put in type I have taken down from the mouth of a native of Ballaugh, Mr. Thomas Callister, the following pronunciations of these words: ly̆ ́ i for (i) a calf, (2) law, (3) forgive, and lōi for (l) a colt, and (2) lying down. Here the ō represents a sound almost as open as that of a in the English word all. (Rhŷs: 9) (Rhŷs: 9) It enters into a diphthong y̆ i as in […] roie ‘a running’ (Ir. rioth, but ScG. ruith); […] lheiy ‘a calf’ (Goi. laogh), which is sometimes pronounced ly̆ i, as for example, by natives of Dalby who thereby distinguish it from leoaie ‘lead,’ pronounced by them lȳi. Here may also be mentioned a group of words with i (probably for igh or idh) forming the second element of a diphthong which I have heard variously pronounced y̆ i, ö̆ i, and even ĕi. The principal instances are the following: sleih ‘people’ ([…] Mod. Ir. sluagh […]); […] treih ‘miserable’ ([…] Mod. Ir. truagh, ScG. truagh); […] teighyn ‘hatchets,’ plural of teigh ([…] Mod. Ir. tuagh, […]), […] creoi, creoie ‘hard’ ([…] Mod. Goi. cruaidh). (Rhŷs: 11) The other diphthong into which ū enters is ūi, which I have heard in words like creoi ‘hard,’ leoie ‘ashes,’ leoaie ‘lead,’ as pronounced by natives of Ballaugh and Jurby. (Rhŷs: 11–12) The vowel ȳ enters into a diphthong ȳi in […] lheiy ‘a calf’ (Goi. laogh, […]), pronounced both lȳ ́ i (and ly̆ ́ i), but the ȳ is perceptibly longer in the pronunciation of the word now written leoaie* ‘lead’ (Ir. luaidhe […]) […]; and the pronunciation of […] leoie ‘ashes,’ (Ir. luaidh […]), is usually the same. * This is no distinction invented by a grammarian, as I learned it in questioning an aged native of Dalby. He carelessly forgot to mark sufficiently the distinction between lȳi for ‘calf’ and the longer lȳi for ‘lead,’ when he was instantly corrected by his wife, who however could not read Manx. It is right to add that y is here only an approximation, as I sometimes seem to hear the word for ‘lead’ pronounced lœ̄ i, while in the North the principal vowel unmistakably recalls the sound of the u which the word originally involved. In the parish of Ballaugh the 239 Aspects of the historical phonology of Manx word for calf is pronounced ly̆ i, and the plural lȳi, written lheiyee; but the latter is more commonly superseded by a form in -yn. (Rhŷs: 18) the perfect […] hie ‘went,’ which, in common with its Scotch equivalent chaidh, has dropped the u of the Goidelic form which appears in Irish as chuaidh. The Manx form, no less than the others, continues slender, being pronounced hắi and carefully distinguished from […] hoie ‘sat, did set’ (Ir. do shuidh […]), whose pronunciation oscillates between hy̆ ́ i and hœ̆ i. (Rhŷs: 25) Dawson says tē bĭ́u er y chāδyn [teː biu̯ er ә xaːðәn] he is swift on his feet: but he calls a foot căs [kas], and căsā́ dn [kaˈsaːᵈn]196 for a footpath. He says tŭ́ i [tui̯ ] north. Kelly distinguishes tθĕi [tᶿei̯ ] hatchet completely from tθai [tᶿai] house. Kelly says hai mī [hai miː] ‘I went’ but hei mi [hei mi] ‘I sat’, and as to his age he said tθa my ḗish trī fĭd as shaghdžḗg [tᶿa mә eːi̯ ʃ triː fid as ʃaxˈdʒeːɡ]. Quirk’s wife pointed out the distinction between lȳi [lәːi̯ ] ‘lead’ and ly̆ i [lәi̯ ] ashes. Dawson said lǖe [lɨːә] early in the sentence ha lǖe oδ y voghrə [ha lɨːә õð ә voxrә] ‘as early in the morning as &’. (John Dawson, John Kelly, Margaret Quirk, Rhŷs notebook 6: 135) Rhŷs’s comments (taken largely at face value) are summarized in Table 54. Kelly in his dictionary (1866 s.v. lheiy ‘calf’) also gives an account which suggests at least a three-way pronunciation contrast: LHEIY, a calf. This word, and leoie, or leoaie, and leigh, or leih require some practice in speaking the language to be able to pronounce them differently and distinctly. (Kelly 1866: 119) 196 For the unexpected final stress here, see §4.5.5. 240 Aspects of the historical phonology of Manx Table 54. Interpretation of Rhŷs’s descriptions of /ai̯ /, /əi̯ /, /əːi̯ / etc. expected CM vowel ai̯ Rhŷs’s descriptions ai̯ əi̯ əːi lhie (laigh) lhiy (lo(i)th) leih (logh) leigh (lagh) ‘Ordinarily…much the same pronunciation’ əi̯ ai̯ lhie (laigh) lhiy (lo(i)th) ai̯ hie (ScG. chaidh) əi̯ hoie (shuidh) roie (rith) əːi̯ lheiy (laogh) əːːi̯ leoaie (luaidhe) leoie (luaith) leoaie (luaidhe) sleih (sluagh) treih (truagh) teigh (tuagh) creoi (cruaidh) leoaie (luaidhe) leoie (luaith) creoi (cruaidh) (Thomas Collister) əi leih (logh) leigh (lagh) lheiy lheiy (laogh) ‘natives of Dalby’ (Thomas Collister; Ballaugh) əːi̯ lheiyee (laoigh) (Ballaugh) uəi̯ > əːi əi̯ , ei̯ sleih (sluagh) treih (truagh) teigh (tuagh) creoi (cruaidh) əːi̯ leoaie (luaidhe) œːi̯ uːi̯ ‘in the North’, ‘natives of Ballaugh and Jurby’ 3.9.1.4 /ai̯ / ~ /əi̯ / contrast Concerning those items which are posited here as having either the pronunciation /lai̯ / or /lәi̯ /, Rhŷs notes that ‘[o]rdinarily these words have much the same pronunciation, but when attention is called to it, Manx speaking people think that they make slight distinctions’. However, he notes from Thomas Callister of Ballaugh the historically expected distribution of lhie ‘lie’, G. laigh = /lai̯ / (apparently with a back realization) and leigh ‘law’, G. lagh, leih ‘forgive’, G. logh = /lәi̯ /. It is possible that Rhŷs failed to recognize this contrast when he was not looking for it, especially as central /әi̯ / might be difficult to distinguish from back realizations of /ai̯ /. On the other hand, a merger /әi̯ / > /ai̯ / may have been underway, and more advanced in certain speakers. There are a good deal of examples of ai, ɑi etc. in the HLSM data for the leigh (ScG. 241 Aspects of the historical phonology of Manx lagh) and soie (G. suidh(e)) sets, as well as in the lheiy (G. laogh) and leoie (G. luaith) sets, but no instances of /ai̯ / words with øi, öi etc. The jeih (G. deich) set with expected /ei̯ / also have several realizations with ai and similar in HLSM. There would thus seem to have been a tendency towards mergers in the following direction in Late Manx: /әːi/ /әi/ /әi/ /ei/ /ai/ /ei/ /ai/ The evidence of the CM orthography points to a robust contrast between /әi̯ / (G. agh) and /ai̯ / (G. aith, aigh etc.), however. The /ai̯ / items are consistently spelled <ie>, <iy>, <i_(e)>, while the /әi/ (G. agh) items are all spelled <ei(g)h>, <eiy>, <oaie>. The items with the development /ui̯ / > /әi̯ / (>/iː/) all have <oie>, as do the items /Ri/ > /rәi̯ /. The Phillips orthography has predominately <yei>, <yoi>, <yi>, <ei> etc. for the leigh set, and also for soie, soiaghey and oie and roie (for these there are also spellings , <ii> suggesting /iː/ pronunciations). The <yei> type spellings are also used for lheiy and the leoie set, with no obvious marking of the expected length contrast. As for /ai̯ /, the Phillips orthography predominantly uses <ai>, <ei>, <æi> in drine ‘thorn’, Mian ‘Matthew’, crie ‘shake’, mie ‘good’, sie ‘bad’, piyr ‘pair’, moidjyn ‘virgin’. However three items, thie ‘house’, lhie ‘lie’ and lhiy ‘colt’, mostly have <yei>, <yoi> and <yi>. Taigh is found as /tɤj/ in most Scottish Gaelic dialects.197 It is possible that in Manx, as in Scottish Gaelic, the shift /a/ > /ә/ which occurred before /ɣ/ (resulting in leigh ‘law’ /lәi̯ / etc.) also took place less consistently before /ɣʲ/. The Phillips spellings would then represent forms */tәi̯ / and */lәi̯ / which later lost ground to forms with /ai̯ /. Perhaps forms with /ai̯ / were favoured, in part, to disambiguate from otherwise homophonous /tәi̯ / teiy ‘choose’ (G. tagh) and /lәi̯ / ‘law’ (ScG. lagh), ‘forgive’ (G. logh). The spelling of lhiy ‘colt’ (G. lo(i)th), expected /lai̯ /, as lŷôi, lyôi, 197 Apart from a region on the north-west coast from Applecross to Assynt where /taj/ is found (SGDS V: 292–3). There are also areas where [te] and [toi] are found, which may be derived from /tɤj/ or /taj/. Laigh is found almost universally as /Laj/ or /La/ in Scottish Gaelic (SGDS iv: 547–8) (with both front and back and a few rounded realizations) apart from 82 and 84 which have lˋɤ̜̈ɪ̯ ç and lˋɤ̫̈ç respectively which might be interpreted as /Lɤj/. 242 Aspects of the historical phonology of Manx ly̆ oi is difficult to explain, unless these spellings represent confusion with lheiy ‘calf’ (G. laogh) = ly̆ i, ly̆ ei, ly̆ æ̆ i, ly̆ oi, pl. lyôi, or else a survival of a contrast /oi̯ ~ ai̯ / which appears to be neutralized in Classical Manx (§1.7.5). Rhŷs (25) describes a clear distinction between /ai̯ / and /әi̯ / resulting from historical /ui̯ / in the soie class. 3.9.1.5 lheiy ‘calf’, G. laogh Rhŷs reports both long lȳ ́ i [lәːi̯ ] and short ly̆ ́ i [lәi̯ ] realizations of lheiy ‘calf’ (G. laogh). Thomas Callister of Ballaugh is reported as using ly̆ ́ i [lәi̯ ] for ‘calf’, expected /lәːi/, as well as ‘law’ (G. lagh), and ‘forgive’ (G. logh(adh)), expected /lәi̯ /. Rhŷs also describes a length contrast in Ballaugh between the singular lheiy and plural lheiyee, G. laoigh, and in Dalby he notes short ly̆ i [lәi̯ ] lheiy ‘calf’ contrasting with long lȳi [lәːi̯ ] leoaie ‘lead’ (G. luaidhe). Rhŷs also seems to imply a three-way contrast between [lәi̯ ] leigh ‘law’ and leih ‘forgive’, lȳi [lәːi̯ ] lheiy ‘calf’, and ‘the longer lȳi’ ?[lәːːi̯ ] leoaie ‘lead’. This confusing picture possibly represents instability in the length contrast in diphthongs (and perhaps in vowels more generally), and/or Rhŷs’s failure to perceive the contrast consistently (for example he reports a short diphthong for expected /ɛːi̯ / (5) and a long diphthong for expected /ũĩ̯ / (46)). On the other hand, certain explanations for the observed facts may be suggested. The development of a length contrast between singular and plural may represent an attempt to keep these distinct once vocalization of both /ɣ/ and /ɣʲ/ to /j/ had rendered them otherwise identical. The longer plural form could also represent bisyllabic /lәːi̯ .i/, with the final /j/ reanalysed as the plural suffix /i/ (G. -igh). This may also be the analysis represented by the plural spelling of the biblical orthographic lheiyee, if this is not merely an orthographic device to mark the semantic difference between homophones. Some of the original translators of the Bible at any rate did not use this -ee termination, nor make a clear orthographic distinction between singular and plural, since in the Bible manuscripts we have plural ley, lheiy (1 Kings 12:32) and lheiy (1 Samuel 14:32), with the standard spelling inserted by the editor. It is possible also that the claimed extra-long pronunciation of lheiyee represents a spelling pronunciation 243 Aspects of the historical phonology of Manx from the Biblical form, especially since, as Rhŷs notes, this was ‘more commonly superseded by a form in -yn’ (18). A plural form with termination -aghyn is attested in the manuscript of 1 Samuel 6:7. Pressure to disambiguate /lәːi̯ / lheiy ‘calf’ from leoaie ‘lead’ (G. luaidhe) and leoie ‘ashes’ (G. luaith), once the latter two had also come to be pronounced /lәːi̯ /, might also explain the shortening of lheiy ‘calf’ (G. laogh) that Rhŷs notes from some speakers. The influence of the short diphthong in lhiy /lai̯ / ‘colt’ can also not be ruled out. Furthermore, bearing in mind the apparent survival of triphthongal /uәi̯ / in gruaie ‘cheek’, G. gruaidh etc. in CM (§3.4.5), it is possible that leoaie and leoie contained three vowel qualities until a later date (whether /uәi̯ / or /ɨәi̯ /), whereas lheiy would have been diphthongal, and open to vowel-shortening, for a considerably longer period. 3.9.1.6 leoie etc., G. ua(i) + /i̯ / Rhŷs reports a long diphthong ȳi /әːi̯ / in leoaie ‘lead’ (G. luaidhe) and leoie ‘ashes’ (G. luaith). However, he notes (Rhŷs: 11) short y̆ i, ö̆ i /әi/ or ĕi /ei̯ / in a series of other items with G. ua(i)gh/dh which we would expect to have a similar development: sleih ‘people’ (G. sluagh), treih ‘miserable’ (G. truagh), teigh ‘axe’ (G. tuagh) and creoi ‘hard’ (G. cruaidh). It is possible that Rhŷs did not perceive the length of the diphthong in these items since (apart from teigh) they do not form minimal pairs with items with short /әi̯ /, and so he would not have been listening out for a contrast. Rhŷs (11–2) also notes a diphthong ūi /uːi̯ / in creoi, leoie and leoaie ‘as pronounced by natives of Ballaugh and Jurby’, reporting (18) that ‘in the North the principal vowel unmistakably recalls the sound of the u which the word originally involved’. This pronunciation is not recorded in Rhŷs’s fieldnotes, however, and one wonders whether this description represents a post hoc analysis based on these items’ etymology. On the other hand, it is interesting that Rhŷs notes only these three items here, with their distinctive spelling involving <eo>, and does not mention the items with historical uagh which have the spelling <ei(g)h>, which renders them indistinguishable, as far as the orthography is concerned, from the the leigh class (G. 244 Aspects of the historical phonology of Manx agh etc., expected /әi̯ /). The spelling of creoi, leoie and leoaie possibly represents a conservative pronunciation of these particular items. Another potential factor here is that for two of these items there could have been analogical pressure to retain the conservative vowel reflex in the form of the derived nouns creoghys ‘hardness, hardship’ (G. cruas, cruadhas, cruachás), creoghan ‘gadfly, harsh creditor’ (G. cruadhachán) (Kelly) and the genitive leoh ‘of ashes’ (G. luatha) (Cregeen). On the other hand, the spellings leoie and leoaie could simply be intended to distinguish them from the near homophones leigh ‘law’ (G. lagh), and leih ‘forgive’ (G. logh). CC’s198 pronunciation lüi of leoie ‘ashes’ perhaps represents Rhŷs’s ūi. According to Broderick’s vowel chart (HLSM I: 1), ü is a high back unrounded vowel and is ‘similar to λ and could very well be interchanged for it’ (ibid.: 2).199 (Broderick’s λ = [ɯ], although he also uses the latter symbol, e.g. HLSM III: 60.) The transcription lüi should thus be understood as [lɯi̯ ]. If Rhŷs’s comment that ‘the principal vowel unmistakably recalls the sound of the u’ is taken to refer to height and backness, but not necessarily rounding, this may agree with the existence of a realization of G. ua as [ɯː] or [ɯә̯] (Rhŷs’s ů(y) and ü(y)). As discussed above, there is no clear evidence in the Phillips orthography for pronunciations of the leoie (G. luaith) set as /uәi̯ /, or of a clear differentiation between these items and the leigh (ScG. lagh) and lheiy (G. laogh) sets. It would seem that the pronunciation of G. ua(i)dh/gh/th as /әːi̯ / goes back to the early seventeenth century, or else that the Phillips orthography fails to distinguish between /ɯәi̯ / or /ɯːi̯ /, /әːi̯ /, and /әi̯ /, which, if distinct, must nonetheless have sounded similar (especially to an L2 198 Presumably Caesar Cashen of Peel (b. c. 1859), although he is not included in Broderick’s list of speakers (HLSM I: xxvii–xxviii). Cashen was literate and held Manx classes in Peel in the early twentieth century (Broderick 1999: 75), so there is perhaps a risk that his realizations represent spelling pronunciations. 199 The use of this character probably follows Wagner (LASID I: xvii), for whom ü is ‘a slightly rounded variety of ï… ü and ø are much less rounded than German ü, ö, although they give a similar acoustic impression’. Wagner’s ï is a mid-high central unrounded vowel. 245 Aspects of the historical phonology of Manx speaker, if the orthography is attributed to Phillips himself), as well as representing a contrast with little functional load. 3.9.1.7 /əi̯ /, /əːi̯ /, (/ɯːi̯ /) > /ei̯ / Many of the realizations of the leigh, soie, roie, lheiy and leoie sets in HLSM have transcriptions representing [ei̯ ] or similar. This may represent a reduced phoneme inventory (i.e. the loss or avoidance of sounds not found in English) among the terminal speakers, but Rhŷs’s comment on members of the leoie sets that he ‘heard variously pronounced y̆ i, ö̆ i, and even ĕi’ suggests that this merger was underway among some of his informants. The CM spellings <eih, eiy, eigh> used to represent /әi̯ / and /әːi̯ / may suggest that this tendency towards merger with /e(ː)i/ was already underway in the eighteenth century. They are more consistently distinguished in the Phillips orthography, with /ә(ː)i̯ / being represented by <yei, yoi, yi, ei> while /e(ː)i̯ / is usually <ei, æi>. The merger /ә(ː)i/ > /e(ː)i/ may be seen as part of the wider merger of aoi, uai /әː/ > /eː/ before slender consonants reported by Rhŷs (§3.5.1.4). 3.9.1.8 /əi̯ /, /əːi̯ /, (/ɯːi̯ /) > /iː/ There is also a tendency to monophthongize /ә(ː)i̯ / to /iː/, which seems to be lexically and dialectally conditioned. In particular it is more frequent in the south (cf. HLSM I: 162). Monophthongization is not found in the leigh category with original broad /ɣ/. The Phillips orthography apparently records both pronunciations. It is attested from southern speakers in the soie (G. suidh(e)) set, alongside the diphthongal pronunciation. Phillips’ spellings sigi, siggi for imperative plural soie-jee (Ir. suidhigí) apparently represent the /iː/ pronunciation alongside the more frequent spellings with <yi, ei, yei>. In this set /iː/ could arise directly from fronting of /u/ before a slender consonant (§2.1.8), or via /әi̯ /. In the roie set /i/ is original, and realizations with /iː/ are attested from southern speakers alongside diphthongal realizations. In the lheiy set there are only diphthongal realizations, apart from nuy ‘nine’ (G. naoi), where /niː/ is attested from southern speakers alongside diphthongal realizations. The 246 Aspects of the historical phonology of Manx spellings myníí (Ph.) and mree (PC 1796) for mreih (G. mnaoi), genitive of ben ‘woman’,200 perhaps also represents a pronunciation with /ĩː/. In the leoie set we have mostly diphthongal realizations from both north and south, but kriː for creoi ‘hard’ (G. cruaidh) from the northern speaker JTK, ɪː (Wagner, southern informant) for oaie ‘grave’ (G. uaigh), and sliə, slʹiː, from the southern speakers NM and HK respectively, for comparative s’leaie ‘sooner, -est’ (G. is luaithe), although this could represent the uninflected form leah (G. luath). In this set /iː/ may represent the general fronting and unrounding of G. ua(i). 3.9.1.9 ao(i) + /u̯ / This gives both diphthongal and monophthongal forms. Table 55. ao(i) + /u̯ / Phillips cheu CM pronunciation /tʃe(ː)u̯ / etymology English HLSM taobh > *téabh side øu JW, TC, au JK, NM, TL, u TC, JTK, EC /krәːu̯ / craobh øu JW, TT ǽu, êa, eú, eu fôêr, fŷior ? /eːun⁽ʲ⁾әs/, > /juːn⁽ʲ⁾әs/ /foːr/, ?/fuː.әr/ aoibhneas, ScG. éibhneas faobhar bunch of shrub joy ú (11), u (8) /nuː/ naomh cheu (13), chéêŷf, chǽf, chǽf, chæf, cháyf, chǽyf, chæyf, chêyf201 crouw eunys foyr noo edge of blade holy; saint juːnəs JW, TC, jũnəsɑx TT fo̜ ːr EKh, fu̜ ːəṛ J:EK, ə ˈnɔːr WːS, ən ˈo̜ ːə EL, (foyragh) foːrɑ ̣x HK uː JW, HK 200 = G. dative form mnaoi rather than the historical genitive mná (Thomson 1995: 142). Note unexpected maintenance of /v/ here, as if the form were *taoibh, or rather *téibh (slender bh, mh are often retained in Phillips, written <f(f)>, where later they are vocalized to /u̯ /, as was the case already in Phillips with their broad equivalents). Given that the noun is feminine according to Cregeen, this could be a dative form, perhaps under the influence of lá(i)mh. 201 247 Aspects of the historical phonology of Manx sou /sәːu̯ / saobh vextion, distress; (prefix) foolish etc. 3.9.2 /eː/ > /əː/ in Scottish Gaelic and Manx A number of items in Scottish Gaelic show ao (northern /ɯː/, southern /әː/) for OIr. é /eː/ before broad consonants (O’Rahilly: 32; Dillon 1953: 322–3; Ó Maolalaigh 2007: 226), especially in initial position as in aodach ‘clothes’ (Ir. éadach), aodann ‘face’ (Ir. éadan), aog ‘death’ (Ir. éag), aogasg ‘countenance’ (Ir. éagcosc), aotrom ‘light’ (Ir. éadrom); also taod ‘halter, rope’ (Ir. téad), faod ‘may, can’ (Ir. féad), maodal ‘stomach, paunch’ (Ir. méadal) and others. This change is not consistent in Scottish Gaelic; some words which one might expect to be affected do not have reflexes in ao at all, such as eud(ach) ‘jealousy, zeal’. Others, including eug / aog, and the adjectives with the negative prefix eu-, ao- such as eu-coltach, ao-coltach, are variable. Even the items which appear to have ao universally in the spoken language may have written variants with eu, as in eudach ‘clothes’, which are presumably literary and influenced by the Classical Irish standard. O’Rahilly (32) claims that this change is ‘unmistakeable proof that the Ωː or ʎː of current ScG. could have developed from Eː, for it has a number of words in which an historic ē followed by a non-palatal consonant has acquired the value of ao, after first passing through E:’. O’Rahilly’s Eː seems to be intended to represent a retracted [eː] or fronted [әː], while ʎː represents [ɯː]. O’Rahilly is probably right: it can easily be seen how an interchange between /eː/ and /әː/ could take place when these were similar sounds, adjacent in the vowel space. It is much less likely that /eː/ would synchronically interchange with /ɯː/. Since aodach etc. is found throughout Scotland (and apparently the Isle of Man), it presumably represents an early change that affected the whole of the Scottish Gaelic area before the vocalization of adh/gh and the raising and backing of /әː/ to /ɯː/ in the northern dialects. 248 Aspects of the historical phonology of Manx Dillon (1953) suggests that the existence of doublets with ao and eu in Scottish Gaelic is an example of ‘semantic distribution’ whereby different phonological reflexes of the same item gain different meanings; Ó Maolalaigh (1997: 114, 200, 227, 514, 634) refers to this as ‘homophonic lexical split’. Dillon (1953: 323) gives a number of examples, including taod ‘halter’ : teud ‘harpstring’, faodail ‘waif’ : eudail ‘treasure, darling’, and saod ‘good condition or humour’ : seud ‘way’. In all of these doublets, the form preserving /eː/ can be seen as the higher-register or literary form, although eudail has come into popular use as a term of endearment (m’ eudail, m’ iadail).202 Ó Maolalaigh (2007) makes the following observations concerning the change é > ao: it is unlikely to be purely coincidental that the change é > ao in Scottish Gaelic is confined to words with initial éa- and f-, s-, t-. The development in the case of féad > faod clearly represents a subclass of initial éa- given the lenition product of f-. Lenition of both s- and t- yields h which is phonemically neutral with regard to palatalisation. The retraction of original é in certain words with initial f-, s- and t- in Scottish Gaelic may first have occurred in lenited variants fh-, sh-, th-, and the loss of initial palatal quality may have been due to hypercorrection in much the same way that non-palatal initial s- and t- in some Irish dialects have been replaced by palatal s-, and t- in certain CiCʹ sequences, e.g. saoil > síl, tuit > tit etc. (Ó Maolalaigh 2007: 226) There are also one or two cases of /әː/ > /eː/, notably aoibhinn (O.Ir. oíbind) which is found as éibhinn in Scottish Gaelic (alongside the historic form), and apparently in Manx, if eunys /juːnәs/ ‘joy’ (HLSM II: 154) derives from éibhneas rather than aoibhneas.203 Similarly, the case of Ir. saoil > síl is paralleled in Manx by the form sheill, sheiltyn (apparently = *séil) (§5.5.1). There is some evidence in Manx of /eː/ > /әː/ in some of the same items as in Scottish Gaelic. Rhŷs (17) notes ȳ [әː] as the most frequent realization of the initial vowel in eddin ‘face’, Ir. éadan, suggesting a form parallel to Scottish aodann. However he notes ē [eː] ‘once or twice in the South’. Spellings with <y> in Phillips, such as yddyn, 202 Homophony also offers a possible motivation for the initial change /eː/ > /әː/, distinguishing the preexisting homophones eudach > aodach ‘clothes’ from eudach ‘jealousy’. Interestingly, the latter item may itself have undergone a subsequent homophonic split between eud /eːd/ ‘jealousy’ and iadach /iadәx/ ‘zeal’ (Dwelly s.v. eud; Dillon 1953: 323). 203 Breatnach (1994: 233) notes a Middle Irish spelling ébind. 249 Aspects of the historical phonology of Manx ydyn, occur eight times, and would seem to suggest the /әː/ pronunciation, while there are nine occurrences with <e>, <æ>, <é>, <ey> which might imply /eː/. Eaddagh, eaddeeyn ‘clothes’, Ir. éadach, ScG. aodach, is consistently spelt with <y> in Phillips (12 occurrences). Lhuyd’s collector writes adyn for eddin (Ifans and Thomson 1980: 137). Tedd (G. téad, ScG. taod) is attested in Phillips as tæddyn, teddyn, and tydyn; the spelling tedd is found in Isaiah 5:18, while Cregeen has both tedd and tead.204 Phillips only has <e> in eddrym (G. éadrom, ScG. aotrom) (three occurrences, all with single following <d>). The Classical Manx spellings, and some of those in Phillips, appear to suggest short vowels in these words. From terminal speech only short ø (TC) and ɪ (NM) are attested for eddin (HLSM II: 141). The former especially suggests a pronunciation of the ao rather than the é type. For eaddagh there is ï, eː (both JW), e (JTK, NM) and ę (TT), while for eddrym there is ę (HK), e (JTK) and ɛ (W:N). Tead / tedd is attested with ę (HK, JW), eː (JTK), ɛ (W:N). The initial /t/ rather than /tʃ/ confirms that the Manx form represents taod rather than téad, however, possibly with vowel shortening. For the other items it is quite possible that historical éadan, éadach, éadrom co-existed in Manx with the Scottish-style development to initial ao. For shortening of stressed long vowels in polysyllables, see §551; shortening in tead, tedd possibly originates in the plural form teddyn. The Manx form fod, foddee ‘can’ (G. féad, ScG. faod) apparently has an irregular development to short /o/, confirmed by Phillips where forms of this verb are consistently spelt with <o>, sometimes with following <dd>. However, in Late Manx, NM has realizations with ɪ, ɛ, ə, apparently representing (shortened) féad or faod, alongside a form with ɔ, which agrees with the Manx spelling. Other speakers only have forms with /o/. Similarly, G. féileacán ‘butterfly’ gives Manx foillycan (Cr.), folican (K.), fɔːljəkən (W:S, HLSM II: 173) presumably via a form *faoileacán. Finally, Cregeen’s spelling of word meydlagh, meyhdlagh ‘heavy and slow in moving on account of size; unwieldy’ (Ir. méadlach, ScG. maodalach) is ambiguous and could represent either /eː/ or /әː/ (cf. §3.6.2 for <ey>). 204 G. téad with initial /tʲ/ would give something like *chead in Manx orthography. 250 Aspects of the historical phonology of Manx 3.9.3 geay ‘wind’, G. gaoth For the noun geay ‘wind’ (G. gaoth), as well as a regular development gȳ [ɡәː], Rhŷs (17) notes diphthongal pronunciations (probably not a ‘dissyllable’): Here also belongs one of the pronunciations of the Manx word for wind, namely that which I should represent as gȳ, to be heard in the South of the Island, and to be equated with the Irish and Scotch gaoth ‘wind’; but it is superseded in the North (and sometimes in the South too) by a form which I should almost represent as a dissyllable gǖә or gœ̄ ә. (Rhŷs: 17) Phillips’ spellings (gya, gýa, gýæ, gye, gua, gúa, also gyei, gýæi) appear mostly to represent the diphthongal variant. This diphthong is otherwise noted by Rhŷs as reflexes of G. ua (§3.5.1.3), and it may be that the northern (and occasionally southern) pronunciation of gaoth represents a form *guath on the pattern of leah ‘soon’ (G. luath), theay ‘people’ (G. tuath), and feoh ‘hate’ (G. fuath), or is a back formation from the genitive (or dative) form geayee /ɡәːi̯ / (G. gaoith(e)), since /әːi̯ / can represent both uai/i̯ / and aoi/i̯ / (§3.9.1). Strachan has gů, which may represent the monophthongal form, or monophthongized form of the /ɡɨә̯/ (Strachan 1897), while Lhuyd’s collector gives gèe (Ifans and Thomson 1980: 133) which probably represents the /ɡәː/ form. The Terminal Manx forms are as follows: from the south we have gλːə, gʹeːə, gʹeː (NM), ən ɣλː (HK), giː (TCr), go̤ i, go-i (Wagner), and from the north guiː (TC), gʹiːə (JK), gɪːə (J:JTK, HB). These forms are somewhat difficult to interpret, given that monophthongal forms may represent either historical ao, or monophthongization of ua, and Wagner’s southern forms seem to represent oblique gaoith(e). There are four occurrences of geay in PC in rhyming position, all rhyming with leah (G. luath), which suggests the /ɡɨә̯/ realization. 251 Aspects of the historical phonology of Manx 3.9.4 riyr ‘last night’, G. aréir, araoir This is irráir in Early Irish, which Ó Briain (1923: 318) derives from fár ‘sunrise, dawn’ (eDIL s.v. fáir), giving approximately the meaning ‘the time before the dawn’. For the irregular change ái /aː/ > éi /eː/, Ó Briain suggests the analogy of indé ‘yesterday’. The G. variant form with aoi (ScG. a-raoir) would then be an example of the interchange between é and ao (§3.9.2). The spelling and attested realizations of this item in Manx show a diphthong (e.g. rɑ ̣iər NM, rɑ ̣ːiər JW, HLSM II: 369). The development, or maintenance, of this diphthong is unexpected; one might tentatively suggest the influence of oie ‘night’ (G. oidhche; northern Manx /әi̯ /) which often precedes this word. 3.9.5 cleaysh ‘ear’, G. clua(i)s etc. G. cluas ‘ear’ is found with a generalized dative form cleaysh (cluais) as nominative in Classical and Late Manx. Rhŷs (24) reports that this was realized as clůsh [klɯːʃ], clüsh [klɨːʃ], plural clůshyn [ˈklɯːʃәn], clüshyn [ˈklɨːʃәn] in the north, while clœsh [klәːʃәn], plural clœshyn [ˈklәːʃәn] was ‘usual in the South, and may be heard as far North as Kirk Michael’. Forms which appear to indicate forms with both [s] and [ʃ] (i.e. G. nominative cluas and dative cluais) are found in Phillips (e.g. kluas, kluash, klyesh, pl. klúasyn, klyasyn, klúashyn, klýæshyn), although to some extent <s> and <sh> are used interchangeably for both [s] and [ʃ] in Phillips (Rhŷs: 155–6; Thomson 1953: 10; Lewin 2015b: 78; Wheeler 2019: 4–5), so that the phones intended by the author or the scribe are not entirely certain. An apparent survival of cluas is found in the appendix to the earliest (1707) printed book in Manx (PSD: 19), in the passage Te fosley nyn gleays ‘He is opening their Ear’, which may perhaps represent a fossilized genitive plural after the verbal noun, or else simply the original nominative singular form translating the English singular. Rhŷs’s northern forms apparently show the usual reflex of ua as /ɨә̯/, /ɨː/ rather than that of uai as /әː/, notwithstanding following slender [ʃ]. This could be a lexical 252 Aspects of the historical phonology of Manx exception, or a result of the analogy of cluas before this form disappeared from the language. Cf. Rhŷs’s (24–5) comment: ‘[t]his will serve to illustrate the difficulty of distinguishing in Manx between changes purely phonetic and others which are declensional with their roots in analogy’. The southern form would thus show the usual development of uai to /әː/. From Terminal Manx we have from the north kliʃ, kliʒən (TC), klöːʃ, klɪːdn (J:JTK), klɪʒʹən, klɪːʒən (HB), and from the south kleːʃ (HK), klʹẹːʃ (JW), kleːïʃ (EKh), klẹːʃən (NM), kleːiə̯ (SK), klɛːən (J:TL), klöːʒən (W:S) (HLSM II: 83). From the north, then, we mostly have high front unrounded vowels, which could represent the most progressive reflex of ua (/uә̯/ > /ɯә̯/ > /ɨә̯/ > /iә̯/) in Rhŷs’s description, thus supporting the conclusion that the vowel in the northern reflex of cluais behaves as if followed by a broad consonant. The southern reflexes mostly have front mid unrounded vowels, following the regular development of uai, falling in with aoi and giving /әː/ or /eː/. The form noted by Wagner from the south with öː may be equivalent to Rhŷs’s œ̄ , although Jackson also gives ö for a northern speaker. A similar case may be geayltyn ‘shoulders’ (G. guailne, but Manx *guailtean), which on the evidence from HLSM appears to have /ɨә̯/ rather than /әː/, presumably on the basis of paradigmatic analogy from the singular geaylin (G. gualainn) with broad l. 3.9.6 keayrt ‘time, occasion’, G. cuairt Judging by the HLSM forms and the spellings with <iy> etc. in Phillips, this item, and the derived preposition mygeayrt ‘around’ (G. ma gcuairt), have mostly /ɨә̯/ rather than /әː/, presumably because final /rʲtʲ/ is regularly depalatalized in Manx (Jackson: 82; HLSM III: 86). 253 Aspects of the historical phonology of Manx 3.9.7 seaghyn ‘sorrow, affliction, trouble’ Table 56. seaghyn Phillips seaghyn ya (6), ýa (6), eia (2) CM pronunciation ?/sɨә̯xәn/, /sәːxәn/ etymology English HLSM ?sua(i)th- grief, sorrow, affliction sẹːəxən, sᵘiːɑ ̣ən TC, siːən W:N, siːɣən HK, (part.) sẹːxənətʹ, sᵘiːxənitʹ TC, siːɣənətʹ JW Seaghyn ‘affliction, sorrow, grief, trouble, agitation’ (Cregeen), ‘distress, trouble’ (Kelly) appears to have ao(i) or ua(i), but its etymology is not entirely clear. Broderick (HLSM II: 388) suggests the following derivation: Prob[ably] a deriv[ative] of OIr. sóïd [sic], later suaithid, vn. OIr. sóud [sic], later súathad ‘turn, stir, agitate’, the simplex giving Mx. seiy [‘stirring, mixing’] qv.: w[ith] (a)chan. But v. also Ir. saochán, sub saobhán in Dinn[een]. (HLSM II: 388) Thomson (1953: 302) suggests ‘cf Seiy, and Ir saoth’. For figurative senses of soïd, sua(i)th205 etc. relating to mental states, cf. ‘upset; s[uaithim] mé féin, I exercise myself’ (Dinneen, s.v. suaithim), ‘Tá a aigne suaite, his mind is fuddled’ (Ó Dónaill, s.v. suaith); also eDIL (s.v. soud, [e]) ‘turn, mood, experience, behaviour’. The predominant high realizations in HLSM (also Rhŷs notebooks, Broderick 2019 s.v. seaghyn) would point to ua, as would Phillips’ spellings in <ya>, <ýa> (although these could also possibly represent a trisyllabic realization /sәː.әxәn/). However Thomas, Kermode’s sȳɣyn [sәːɣәn] (Strachan 1897), TC’s forms with ẹː (if not examples of the spelling pronunciations to which this speaker was prone), and Phillips’ seiaghyn might point to /әː/, /әːi̯ /. Possibly these forms represent by-forms with suathand suaith- respectively (cf. Ir. stem suaith, vn. suathadh). 205 These verbs are not equated in eDIL or LEIA. Thomson (1953: 303, s.v. seiy) seems to connect them, ‘Ir sódh, suidhim, also suathadh’. 254 Aspects of the historical phonology of Manx Broderick is probably right to see the ending -achán in this item, as in gweeaghyn ‘curse’ (G. guidheachán), nieeaghyn ‘washing’ (G. nigheachán). Note that Phillips has one instance of -an in the final syllable (otherwise -yn), which might represent older /an/ (for interchange of /an/ and /әn/ in these words, see §4.4.7.3). Derivations from saoth-, saobh- etc. seem less plausible, since the forms with high vowels would then be difficult to account for. 3.9.8 deayrtey ‘pour’, G. dort, doirt, dórt, duart Deayrtey, ‘pour, spill’, Ir. doirteadh, dortadh, dórtadh, ScG. dòrtadh, OIr. do-fortai, do-foirti, -dortai, dórtad, is included under ua in this chapter (§3.4.3), because it is clear from the Phillips and later spellings, and from the realizations in HLSM, that the Manx should be regarded as a form *duartadh. Compare Ir. duartan ‘downpour, a torrential shower; calamity’ (Dinneen), dúartan ‘downpour, shower’ (eDIL). 3.9.9 smooinaghtyn ‘think’, G. smuain, smaoin O’Rahilly (38) notes that ‘[i]n smaoin < smuain the aoi is universal in current Irish, and is also common in Sc[ottish]; here the aoi seems to have been taken over from the verb saoil’. In Manx, this appears in the verb smooinee, verbal noun smooinaghtyn (in Phillips mostly smún-, with or without the diacritic, and sometimes smúin, smuyn), which apparently represents *smúin, or else the general tendency to monophthongize /uә̯/, /iә̯/ in Manx (§2.2.6), although this would be very early in Phillips. Smooinaghtyn is recorded with u or uː from JK, HB, TC, NM and TT, although Wagner transcribes the vowel as ï (HLSM II: 419), which might plausibly represent smaoin, or shortening (cf. §5.5.1) of the vowel to *smuin. The lack of fronting and unrounding in the Manx realizations may be another example of the lack of fronting after a labial consonant, as in feayr ‘cold’ (G. fuar) etc. (§§3.6.1.2, 3.8). It is also worth noting that the form saoil does not exist in Manx, being represented by sheill (*séil, 255 Aspects of the historical phonology of Manx cf. Ir. síl) (§§3.9.2, 5.5.1). However, there is possibly analogical influence from cooinaghtyn ‘remember, memory’ (G. cuimhne), which has /uː/ in Manx (HLSM II: 92). 3.9.10 Manx derivatives of G. draoi, druadh The spellings of derivatives of G. draoi ‘druid’ (OIr. druí, druï), gen. druadh in Manx are complex. There are two occurrences in the Bible: fer-druiagh ‘charmer’ (Deuteronomy 18:11) and druaiaghtee ‘enchanters’ (Jeremiah 27:9). The latter would appear to point quite clearly to a derivative *drua(i)dh(e)acht(ach) (for usual G. draoidheacht), with preservation of back /uәi̯ /, as in dwoaie, gruaie, twoaie (§3.4.5). Cregeen, whether deliberately or not, simplifies this spelling to druaight ‘a Druid’, druaightagh ‘Druidic’, ‘a Druid’ (citing the Jeremiah passage), druaightys ‘Druidism, enchantment’, and also gives an alternative abstract derivation drualtys, driualtys. If genuine, all of these suggest /uә(i̯ )/. Kelly in the earlier manuscript of his dictionary (MNHL MS 1477) gives druaiaghtagh with a reference to Jeremiah interlined between druaie ‘a Charmer, wizard, Druid, G. Draoi’ and druiagh ‘Pertaining to a Charmer, enchanting’ with the example cloagey druiagh. In the later MS 1045–47 he has druiagh, druiaghtagh, druiaght, with the Jeremiah passage cited with the spelling druiaghtee rather than Bible druaiaghtee; this is also the case in the 1866 printed version. For the simple form, G. draoi, Kelly has drui, pl. ny druee or darui (druï, ny druee, daruï MS 1045–47) with a long paragraph on the history and etymology of the term ‘druid’, citing both G. forms draoi and druadh. This entry does not appear in MS 1477 and is probably a later insertion inspired by Kelly’s antiquarian etymologizing tendencies (§1.6.8.1). Kelly also has druai ‘a dwarf, a pigmy; a sorcerer, an enchanter. (Ir. droich)’ (MS 1477 ‘druai p. -yn. A dwarf’). The reference is apparently to ScG. troich, droich ‘dwarf’, Ir. troch ‘wretch’, but this would give */drai̯ / in Manx (cf. criy /krai̯ / ‘gallows’, G. croich, §1.7.5). This druai might be equated with Manx dreih, dreigh ‘wretch’, which according to Thomson (1998: 122) represents G. draoi ‘druid’ 256 Aspects of the historical phonology of Manx with pejoration of meaning, but Kelly has a separate entry (s.v. dreigh) for this and does not equate the two. These forms do not occur in HLSM, but Rhŷs notes a form of the abstract noun from one of his informants: drȫghagh or dreghagh ‘enchantment’ (cŭr ayd fo ghrḗghagh ‘to put them under enchantment’ – I cannot discover druiaght anywhere except in the spoken language.) (Tom Kermode, Rhŷs notebook 6, original deletion) This form appears to be a little garbled, but points to a realization /әː/ or /eː/, whether from draoi- or from a fronted variant of drua(i)dh-.206 All in all, the evidence is not entirely clear but suggests that by-forms with /druәi̯ -/ (<drua(i)dh) and /drәːi̯ / (<draoi or drua(i)dh) may have existed side by side. It is also possible that the Manx spellings are influenced by English ‘druid’ and/or the Gaelic forms. 3.9.11 Unstressed and shortened ao(i), ua(i) Ao(i) and ua(i) are found shortened in initial syllables through stress shift (§5.1) or sporadic stressed long vowel shortening in polysyllables (§5.5.1). They are also shortened in post-tonic position, such as cassid ‘accusation’ (G. casaoid), and fynneraght ‘coolness, breeze’ (G. fionnuaracht), but this is not discussed here, since the possible results in this position are more limited (§5.1.6). Stressed or pre-tonic unstressed shortened vowels generally retain (broadly) their original quality (§§5.1.3, 5.5.1), but this is problematic in the case of ao(i) /әː/ and ua(i) /ɨә̯/, as they do not historically have short counterparts in the vowel inventory. (However, see §3.9.2 for the possible development of short stressed /ˈә/ in certain items.) 206 Cregeen also has ‘fo ghruaight, s. under druidism or inchantment [sic]’. The fossilized lenition after fo here and in Kermode’s realization would suggest this is a genuine, established phrase (cf. Broderick 1984a: 166). 257 Aspects of the historical phonology of Manx Rhŷs (18) notes short y̆ [ә] as the usual realization of ao in freoagh ‘heather’ (G. fraoch), although he states that he has ‘occasionally’ heard it long, but only long forms are given in HLSM (II: 174). He also notes the short sound in un ‘one’ (G. aon), which is also indicated by the spelling, although some long forms are given in HLSM (II: 468– 9). For data on these, see §§3.4.1, 3.4.2; for fod ‘can’ (G. féad, ScG. faod), see §3.9.2. The short vowel in Gailck, Gaelg, /ɡilʲkʲ/ etc., ‘Manx Gaelic’ also seems to be long established; irregular developments of this word are found throughout Gaelic dialects (§3.9.1.1). Similar vocalism is found in the shortened stressed vowel in kirree ‘sheep’ (G. cao(i)r(a)igh), where harmony with final /i/ might perhaps play a role. A problematic word is foillan ‘gull’ (G. faoileán) (also with -óg). The heavy-heavy structure of this word would lead us to expect stress shift (§5.1), as indeed seems to be the case in Kelly’s form fooilleig, unless a form with a shortened vowel in the first syllable developed at an early date, or the Manx form represents G. faoileann with the analogical development /әN/ > /an/ found in a number of other items (§4.4.7.3). Some of the forms in HLSM (II: 173) have long [oː], which may represent a case of the ‘occasional interchange of ao with other vowels’ noted by O’Rahilly (34–5), e.g. ScG. ònar alongside aonar ‘alone’, cf also the semantic split failt ‘welcome’ (G. fáilte), feoilt ‘generous’ (G. faoilte). Toshiagh-jioarey (Cr.), toshiaght-joarrey (K.) ‘coroner’,207 apparently G. taoiseach deóra (Thomson 1988: 141), with shortened vowel /o/, may similarly represent the variant tóiseach (cf. Ó Maolalaigh 2008a: 213– 4), or else result from the weak stress in the initial element of the compound. Table 57. Pre-tonic ao(i), ua(i) Phillips caghlaa y (10) CM pronunciation /kaˈxlɛː/ carrane a /kaˈrɛːn/ cuarán sandal /dai̯ ˈɛːn/ ? daoi + án (HLSM II: 121) worm dhiane (Cr, K., Bible), 207 etymology English HLSM c(l)aochlódh, ScG. caochladh change, difference ɔ NM, ɑ̜ NM, HK, JK, EK, a JW, HB, ə JK ə SK, J.EK, a SK, krɛː̃ nən TC, kre̜ ːnən JTK dɑ ̣iˈẹːᵈn HK, (pl.) døˈiniə̯n (sic) TC A Manx court official whose role is quite different from that of coroners in the United Kingdom (see <https://www.courts.im/court-information/coroners/>). 258 Aspects of the historical phonology of Manx daiane, gaiane (K.) farrane feoghaig (Cr.), fughage (K.) freoaghane (Cr.), freeoaghane (K.) leaystane sonnaase treicknane (Cr.), tratneayn (K.) unnane, annane (CS), annanjeig (Cr.), annan (Mona’s Herald 22.12.1840); unnaneys y (2) u, o (2), ou a (26), y, aý /faˈrɛːn/ fuarán spring, fountain periwinkle (shellfish) /fuˈxɛːɡ/ faochóg /frә(ː)ˈxɛːn/ fraochán bilberry /lɨә̯ˈstɛːn/ /soˈnɛːs/ luascán saobhnós /trɨkʲˈnʲeːn/ tuairgnín swing arrogance, riot beetle, mallet /uˈnɛːn/, /aˈnɛːn/, /ˈanan/ aonán one; unity a JN ə W:?N o̜ TC anan JK, JTK, o̤ ˈneːn TK, (unnane-jeig) ɑ ̣nɑ ̣n dʹʒeg JK, anan ˈdʹʒeg JK, arəndʹʒe̜ g Fa, arən ˈdʹʒe̜ g Co The spellings of items such as freaoghane, leaystane clearly show an awareness of their etymology, and possibly a corresponding pronunciation (at least in careful speech), although it is less easy to account for Cregeen’s feoghaig with <eo>, which apparently shows an awareness of the original vowel in the initial unstressed syllable, despite the non-attestation of the corresponding simplex form (G. faocha, ScG. faoch). Compare also with retained back realizations boirane ‘clamorous fellow’ (Cr.), ‘gidd[iness]’ (K. s.v. kione-y-lhei) (G. buaidhreán), bwoail(l)teen ‘mallet’ (G. buailtín). 3.9.11.1 Pre-tonic ao, ua > /w/ Where the shortened vowel is followed synchronically by a long vowel without an intervening consonant or glide, it may be reduced to non-syllabic /w/ (cf. also bwaane, bwaag /bwɛːn/, /bwɛːɡ/ < bothán, bothóg): 259 Aspects of the historical phonology of Manx Table 58. Pre-tonic ao, ua > /w/ Phillips mwane whaaley208 fóaly, fualy CM pronunciation /mwɛːn/ etymology English maothán /xwɛːlә/ fuagháil embryo, foetus sew HLSM hwe̜ ːle̜ HB, kwɛːlə TT, kwe̜ ːlɑ̜, kwe̜ ːlə, k′eːlɑ̜, k′o̜ ːl NM, kwe̜ ːl′ə HK, fʷɛːlə WːN 3.9.11.2 unnane, annan ‘one’, G. aonán Cox’s (2011: 277; 2013: 271) derivation of Manx unnane, annan, ScG. aonan from OIr. a n-óen is implausible, as this would give /-әːn/ (*-eayn or the like) not /-ɛːn/ in Manx, and, as shown elsewhere in this chapter, there is little evidence of these vowels being confused.210 It seems much simpler to derive both the Scottish and Manx forms from G. aonán, with northern Manx annan being easily explainable as a form arising under weak postlexical or phrasal stress in e.g. unnane-jeig /ˌәnɛːn ˈdʒeɡ/ ‘eleven’ (G. aonán déag) > annan-jeig /ˌanan ˈdʒeɡ/, as in other cases discussed in §5.1.1.5, and later generalized. It is perhaps no accident that Cregeen has this spelling only in a form of aonán déag, and that it is also only attested in HLSM (II: 9, 469) in this phrase. The phrasal stress in dagh annan ‘each one’ (G. gach aonán) (Mona’s Herald 22.12.1840) is uncertain but perhaps represents /ˌdax aˈnɛːn/ > /dax ˈanan/. 208 For the development of this item see Thomson (1981: 142–3), and compare foast ‘yet, still’ (G. fós, ScG. fhathast), realized as hwɛːs, hwoːs etc. by some speakers (HLSM II: 170), and hooar for older fooar ‘got, found’ (G. fuair, ScG. fhuair) (Thomson 1995: 121). The stages of the development are apparently /ˈfua̯ ɣaːlәɣ/ > /fua̯ ˈɣɛːlә/ > /fuә̯ˈɛːlә/ > /ˈfwɛːlә/ > /ˈxwɛːlә/ (> [ˈkwɛːlә]), with dissimilation between the labials /f/ and /w/. 210 The stress shift in a n-óen > ScG. aonan, Manx annan is also unlikely, and the n in a n-óen might be expected to be fortis, giving *aonnan. 260 Aspects of the historical phonology of Manx Chapter 4 Sonorant consonants and associated developments 4.1 Introduction Earlier Gaelic varieties are generally considered to have had a system of four-way sonorant contrasts in rhotics, coronal nasals and laterals: In Old Irish there was probably a contrast between laminal dental and apical alveolar coronals, each of which could be palatalized or velarized, giving four possibilities […] these possibilities for the laterals would be l̪ ˠ, lˠ, l̪ ʲ, lʲ. (Ladefoged et al. 1998: 14) Reduction of these to either three or two contrasts is widespread in Gaelic dialects (Ó Maolalaigh 1997: 57–65), and no dialects retain four rhotics (§4.2.4). In the Late Manx of the terminal speakers, the following sonorant consonant phonemes (nasal stops, laterals and rhotics) are securely attested (cf. HLSM III: 2, 106–17): /m n nʲ l lʲ r ŋ ŋʲ/ In addition, /rʲ/ seems to have been marginally retained by some speakers, and some may have had a three-way lateral contrast /l̪ ~ l̠ ~ lʲ/. There is evidence of a more complex system, with greater retention of fortis and lenis contrasts in Early Manx, and perhaps later. In general, the Manx orthographies do not show fortis-lenis contrasts clearly (i.e. there is no equivalent of the Irish-Scottish use of single v. double consonants medially and finally), and do not always mark the broad-slender contrast either. For example, medial <ll> may represent all four original G. laterals /L Lʲ l lʲ/ in the following items: ollan /olan/ ‘wool’, G. olann /olәN/ balley /balʲә/ ‘town’, G. baile /balʲә/ bollag /bolaɡ/ ‘skull’, G. ballóg /baLoːɡ/ keylley /kelʲә/ ‘forest’ gen., G. caille /kaLʲә/ 261 Aspects of the historical phonology of Manx The double <ll> here serves to mark the preceding short vowel, as per English spelling conventions. Palatalization is, however, often marked by a preceding or following : caill /kalʲ/ ‘lose’, G. caill tooilley /tulʲә/ ‘more’, G. tuilleadh thooilley /tulʲә/ ‘flood’, G. tuile doillee /dolʲi/ ‘difficult’, G. doiligh Because of the opaqueness and inconsistency of the orthographies, our analysis of the development of these consonants must rely heavily on phonetic descriptions of Late Manx (although these are not always easy to interpret, and date from a period when the system had already been significantly simplified), and on evidence of collateral historical changes within or connected with the sonorant system (e.g. development of vowels before historically fortis sonorants, and the development /Nʲ/ > /ŋʲ/ which is reflected in the orthography). 4.2 Rhotics In the Manx of the terminal speakers, and to some extent that of Rhŷs’s informants, the rhotics are notable for significant loss in coda position, change in quality, and alterations to preceding vowels, apparently reflecting parallel developments in English. However, it is evident that loss or weakening of rhotics, and the changes in vowel quality, are a late development, since they are not shown in the standard orthography of eighteenth and nineteenth century Manx. Early loss of rhotics is only found in certain limited environments. 4.2.1 /r ~ rʲ/ Twentieth-century descriptions of the Manx of the terminal speakers mostly report merger of /rʲ/ and /r/ as non-palatalized tap or approximant [r] with only a few traces of palatalized [rʲ]. However, Rhŷs and Strachan report more robust maintenance of this 262 Aspects of the historical phonology of Manx contrast, and orthographic evidence suggests it is maintained at least in some positions in Classical and pre-terminal Late Manx. 4.2.1.1 /rʲ/ in Classical Manx There are representations in the eighteenth-century orthography which clearly indicate medial and final /rʲ/, including the following pairs: ooir /uːrʲ/ ‘earth’, G. úir oor /uːr/ ‘hour’211 coair212 /koːrʲ/ ‘near’, G. i gcóir (but coar ‘kind, decent’, G. cóir) coayr /koːr/ ‘odd’, G. corr coirrey /korʲә/ ‘kettle, caldron, furnace’, G. coire correy /korә/ ‘sowing’ (gen.), G. cor, cur laair (Cr.) /lɛːrʲ/ ‘mare’, G. láir (but laayr K.) laare /lɛːr/ ‘floor’, G. lár See also: erriu /erʲu/ ‘on you’, G. oirbh (but erroo ‘ploughman’, also /erʲu/,213 G. aireamh) merriu /merʲu/ ‘dead’ pl., G. mairbh terriu, teirroo /terʲu/ ‘bulls’, G. tairbh luirg /lurʲɡʲ/ ‘staffs’, G. luirg fuirree /furʲi/ ‘remain, stay, wait’, G. fuirigh oirr /orʲ/ ‘edge’, G. oir mooir, muir /murʲ/ ‘sea’, G. muir buirroogh /buˈrʲuːx/ ‘roar’, G. búirfeadhach A diaeresis apparently represents palatalization in the following morphological alteration: 211 Probably a later borrowing of the English / French word than G. uair < Latin hora, which however appears to be present in neayr’s ‘since’ (§3.4.4). 212 But coayr in Ezekiel 44:25. 213 /rʲ/ confirmed by Rhŷs (notebook 6: 115). 263 Aspects of the historical phonology of Manx jeir /dʒeːr/ ‘tear’, G. déar jeïr /dʒeːrʲ/ ‘tears’ pl., G. déir Cf. mair /meːrʲ/ ‘finger’ (G. méar), and meir /meːrʲ/ (pl.) (G. méir). However, very often there is no clear indication of /rʲ/. As shown by erroo and coar above, the desire to keep homophones orthographically distinct often seems to trump phonological accuracy in Manx orthography (§1.6.4.2). arrey /arʲә/ ‘watch’, G. aire (cf. carrey /karә/ ‘friend’, G. cara; marrey /marә/ ‘sea’ gen., G. mara) earroo /erʲu/ ‘number’, G. áireamh obbyr /obәrʲ/ ‘work’, G. obair s’lhiurey /lʲiurʲә/ ‘longer’, G. is libhre cheer /tʃiːrʲ/ ‘country’, G. tír (cf. feer /fiːr/ ‘true, very’, G. fíor) Note also seyr ‘free’ and seyir (also seiyr) ‘carpenter(s)’ in the Bible (G. saor), where the two spellings might have been expected to distinguish singular /sәːr/ and plural /sәːrʲ/ (G. saor, saoir), but appear to be used to distinguish the adjective from the noun.214 Whether such ambiguous spellings suggest incipient merger of /r/ and /rʲ/ is unclear, since broad and slender contrasts in other consonants are frequently obscured in the orthography as well. 4.2.1.2 /rʲ/ in clusters While /rʲ/ is reasonably well evidenced in the orthography in medial and final position as an individual consonant, it is less clear to what extent it was found in clusters such as br, cr, gr etc. While spellings such as cliaghtey ‘custom, practice’ (G. cleachtadh) clearly attest to initial /klʲ/, we do not find e.g. *criagh for cragh ‘plunder, prey, disaster’ (G. creach) so it is unclear whether this represents /krax/ or /krʲax/. The initial clusters /tr/ and /dr/, at least, tend to be non-palatalized throughout in most Scottish dialects (e.g. SGDS III: 238–41 dream, V: 386–7 tric), and this may be represented by <thr> in Manx three ‘three’ (G. trí), where the /t/ is certainly non-palatalized (HLSM 214 Originally a single lemma (eDIL, s.v. 2 saer). 264 Aspects of the historical phonology of Manx II: 459). On the other hand, there is some evidence for initial /(s)krʲ/ in Rhŷs’s notebooks, e.g. skřiu [skrʲiu̯ ] screeu ‘write’ (G. scríobh) (Margaret Cowley, Rhŷs notebook 6: 66).215 We may compare Ó Maolalaigh’s (1997: 67–8) comments on the weak development of palatalization in certain initial clusters in Scottish Gaelic, which he sees as an archaism as contrasted with Irish. Given that Manx seems to go with Scottish Gaelic in having either lost or not developed palatalized labials (§1.7.6), it might be guessed that it would show similar reflexes to Scottish Gaelic in initial clusters. In view of the lack of certainty in this matter, non-palatalized /r/ is assumed in all initial /(C)Cr/ clusters in this thesis. 4.2.1.3 /r ~ rʲ/ in pre-terminal Late Manx Rhŷs expresses considerable uncertainty about the articulation of Manx rhotics. He describes the main realization of broad r (G. /R, r/) as follows: I am bound to speak with the utmost diffidence of the Manx r’s, as I have but a very imperfect idea how they are produced. In most words initial r in Manx produces on my ear the effect of English r, but I greatly doubt that it is formed in the same way. At any rate, it seems highly probable that the r which is associated with broad vowels is approximately an [a]mbidental r. This I would write ρ […] I should say that the part of the tongue made to vibrate lies on a lower level than in the case of English r, and that the edge of the tongue is brought somewhat closer to the edge of the upper teeth instead of recurved, as is done when pronouncing the English liquid. (Rhŷs: 147–8) Rhŷs (148–9) also distinguishes an ‘[a]lveolar r […] a sound produced as nearly as possible like the English r. He admits he has ‘no certain instances’ but ‘should guess that we have it in’ ayr ‘father’ (G. athair), fer ‘man’ (G. fear). Slender /rʲ/ would be 215 If there was a contrast /kr ~ krʲ/ in initial position, then the form krᵊiː cree ‘heart’ noted by Jackson (121) would seem to represent G. croidhe rather than ScG. cridhe. 265 Aspects of the historical phonology of Manx expected in athair (though see below). It is not clear what the significance of this claimed variant is. Rhŷs (149–50) gives a description of a palatalized (‘mouillé’) rhotic. He gives clear affirmation of its presence in medial and final position (although with few examples), and also claims with less certainty to have noted it in initial position: By this [R Mouillé] I mean an r pronounced analogously to ñ [nʲ] and ɫ [lʲ], that is to say, a palatalized r. It may be represented as r,̃ and it stands mainly for Aryan [Indo-European] r associated with a slender vowel or i̯ ; but first of all I wish to state how far I hear it is a distinct kind of r from the ρ already discussed. Initially I have sometimes thought I noticed a crispness or sharpness which argued as r ̃ rather than a ρ, as for instance in words like […] ree ‘a king’ (Med. Ir. rí […]); but I have never felt certain about it, and no help is to be got from the semivowel yod which has been sunk in most places where one would expect it, as in riu ‘to you’ (Med. Ir. frib, Mod. Ir. ribh) and roo ‘to them’ (Med. Ir. friu, Mod. Ir. riu), which are both pronounced rū̃ [rʲuː], possibly rū [ruː]. Medially this crispness which I have mentioned has often struck me as quite unmistakable and amounting now and then almost to the sharpness of a Welsh or Italian r, as for example in words like marish ‘with, together with’ […] Finally, the difference between r ̃ and ρ or r is rendered still more certain by a sort of a parasitic whisper, which if reinforced would make a sound approaching ch in the German word ‘ich’. This has attracted my attention especially in the case of the word for ‘gold’, namely […] airh, which seems derived not from the Latin aurum but from its genitive auri […], as it is pronounced ā ̣r;̃ [ɛːrʲˢ] or ēr;̃ [eːrʲˢ], if I may use ; for this kind of whisper. (Rhŷs: 149–50) Regarding the reported final /rʲ/ in airh, a more plausible explanation is suggested by the noun’s feminine gender in Manx, as in airh vuigh as palchey j’ee ‘yellow gold and plenty of it’ (Cregeen) (G. *óir bhuidhe agus pailte dí), airh ghlen ‘pure gold’ (Bible) (G. *óir ghlan), in which case óir could represent generalization of an oblique form. Compare also the genitive or adjectival form airhey, which perhaps originates in a reanalysis of the adjective órdha ‘golden’ as feminine genitive *óire. The noun is masculine in other Gaelic varieties, but was neuter in Old Irish (eDIL s.v. ór). In contrast to /r/, slender /rʲ/ is noted as Rhŷs as not being subject to deletion in final 266 Aspects of the historical phonology of Manx or preconsonantal position:216 This r ̃ is not liable to be elided or assimilated, though we have an apparent exception in the word […] lajer ‘strong,’ pronounced lā́ žy ̣ ̆ r [lɛːʒәr] or lā́ žy ̣ ̆ [lɛːʒә] ([…] Mod. Goi. láidir); this is, however, not so much perhaps a matter of phonology as of declension, just as if the Irish word were to be changed from láidir to láidear. (Rhŷs: 50) Rhŷs’s notebooks shed further light on the matter. He distinguishes between r and ř as well as r̊ (‘= English r’, Rhŷs notebook 6: 77). It later becomes clear that ř means palatalized [rʲ] (r͂ in Rhŷs: 1894), although in the earliest diary entries (from 1888) it appears medially and finally in a number of positions where broad /r/ would be expected. Later, it is used more consistently for expected [rʲ], as well as in some words in initial position (§4.2.1.5), although some unexpected uses still appear, such as scarrey ‘divide’ (G. scaradh) (Rhŷs notebook 6: 82). The change in Rhŷs’s usage suggests a gradual attuning of the ear to the contrast, which presumably was relatively unsalient. Explicit comments are as follows, including consideration of minimal pairs: He pronounces aynrit [sic: aanrit] as y̆ ́ nřiti [әnrʲitʲ]217 with a very slight pal[atalized] r and so does his wife (John Joughin, Rhŷs notebook 6: 58) He pronounces airh [‘gold’, G. ó(i)r] ēṛ [eːr] almost īr [iːr] but ēr [e̞ ːr] ‘father’ [ayr, G. athair] or ā ̣r [ɛːr], which? (Thomas Collister, Rhŷs notebook 6: 169) aer [‘air’, G. aer], ayr [‘father’ G. athair], airh [‘gold’, G. ó(i)r] all pronounced the same he thought: I doubt it. (William J. Caine, Rhŷs notebook 7: 198) Although his transcription of ooir gys ooir ‘earth to earth’ (G. úir) as ū(ř) dy ū(ř) might suggest otherwise (Richard Qualtrough, Rhŷs notebook 6: 14). At this point (12.09.1888) Rhŷs seems not to have noticed palatalization in /rʲ/ (if present), and ř seems to represent a weakened (?) coda [r]. 217 G. anart, anairt > ?ainrit. 216 267 Aspects of the historical phonology of Manx aer [G. aer] & ayr [G. athair]218 both pr[onounced] ā ̣r [ɛːr] but airh pr[onounced] differently, ā ̣r ̃ [ɛːrʲ] (Thomas Collister, Rhŷs notebook 7: 199) In the following the higher vowel realization of /ɛː/ ( > [eː]) preceding /rʲ/ is noted (§2.2.3), but not the palatalized rhotic itself: λẹr [l̪ er] ‘a mare’ with ẹ [e(ː)] as in airh ‘gold’ and nearey ‘shame’ (Thomas Collister, Rhŷs notebook 6: 174) Strachan (1897: 55) also notes the contrast /r ~ rʲ/, although he expresses similar uncertainty to Rhŷs: Broad and slender r have been distinguished by r and r͂ , but I doubt if I have always distinguished them aright. In χrī [xriː] heart, and rī [riː] arm, a broad r stands before a slender vowel. A Manxman with whom I talked distinguished r͂ ī [rʲiː] king from rī [riː] arm. (Strachan 1897: 55) Again this suggests a maintained contrast, but of relatively little phonetic salience. 4.2.1.4 /r ~ rʲ/ in terminal Late Manx Marstrander (56) notes two rhotic phones, but for the most part his ‘løst artikulert bakre r’ (‘loosely articulated back r’) ʅ can be understood as an allophone occurring preconsonantally in codas. He does not note a palatalized [rʲ], although he refers to raising of /ɛː/ (G. ái) ‘foran gammelt palatalt r’ (‘before old slender r’) (§2.2.3). However, he does note a contrast between ʅ in e.g. corp ‘body’ and r in kirp ‘bodies’ (G. corp, cuirp), which would appear to represent a trace of the /r ~ rʲ/ contrast noted by Rhŷs (§4.2.1.1), in that /r/ is more liable to weakening or elision than /rʲ/. 218 A reflex of athair with final broad /r/ may reflect the historical genitive athar (Manx ayrey /ɛːrә/, cf. Ir. variant athara), or reflect syncopation in the plural as in ScG. athraichean (Manx ayraghyn). 268 Aspects of the historical phonology of Manx Jackson (117–8), like Marstrander, notes two predominant rhotic phones which represent positional allophones rather than the historical Gaelic contrasts. He does not note anything corresponding to the contrast noted by Marstrander in corp ~ kirp. In general there are only two r-sounds, and their occurrence is independent of whether they were originally “broad” or “slender”, lenited or non-lenited, double or single, in dental groups or not. Indeed it is clear that the second variety is simply a recent weakening of the first so that the elaborate Gaelic system of r-sounds was really reduced in principle to one […]. The first is a one-flap alveolar r, the second is a quite weak alveolar fricative ɹ. The first occurs initially, intervocally, and after consonants; the second before consonants and finally. (Jackson: 117–8) Jackson adduces evidence that the weakened variant is a recent development, noting a comment by one of his informants to this effect: However, it is very significant that [Eleanor] K[arran] and [Thomas] L[eece] sometimes use final r where others have ɹ; and on mooar “big” L[eece] commented to me that ɹ here is wrong and is a modern corruption. (Jackson: 118) There may be centralization of vowels preceding coda /r/ (Jackson: 119). Jackson (118) also notes palatalized /rʲ/ in initial position in a limited set of items, as also noted by Rhŷs and Broderick (§4.2.1.5). Broderick’s (HLSM III: 17–8, 107, 113–7) descriptions are similar to Jackson’s. 4.2.1.5 Initial /rʲ/ Jackson (118–19) and Broderick (HLSM III: 114) note /rʲ/ in a limited set of items: There is also rarely a palatalized rʹ, which is a weak alveolar buzz similar to the “slender” r of Modern Irish. This occurs, initially only, precisely where initial rʹ occurs in Ir. and ScG., namely in lenited position, but only as a fixed characteristic in a few words in which it is also stereotyped in ScG. Gaelic. It represents therefore original lenited “slender” R-. For this initial rʹ some Manx speakers use, however, r. (Jackson: 118) 269 Aspects of the historical phonology of Manx The items noted by Jackson and Broderick are the prepositional forms riu (G. ribh) and roo ‘to them’ (G. riu) and the temporal adverb rieau ‘ever’ (G. riamh), and also the nouns rio ‘frost’ (G. reódh) and red ‘thing’ (G. réad, rud). Rhŷs (149–50) records a somewhat wider distribution for initial /rʲ/, including in ree ‘king’, although his comments are tentative: ree ‘a king’ (Med. Ir. rí […]); but I have never felt certain about it, and no help is to be got from the semivowel yod which has been sunk in most places where one would expect it, as in riu ‘to you’ (Med. Ir. frib, Mod. Ir. ribh) and roo ‘to them’ (Med. Ir. friu, Mod. Ir. riu), which are both pronounced rū̃ [rʲuː], possibly rū [ruː]. (Rhŷs: 149–50) A realization perceived as [z] may also perhaps be interpreted as /rʲ/: I once thought I heard rhyt ‘to thee’ (Ir. riot) pronounced zy̆ t, but the man reading to be quickly corrected himself: this was also in the North. (Rhŷs: 149) In his notebooks, Rhŷs is ‘bothered’ by initial [rʲ] as he cannot explain it: řen [rʲen] ‘did’, řå f rå ‘too’ [rɔː] very guttural but řo [rʲoː] frost – both [‘did’ and ‘frost’] seem to have ř [rʲ] but why? rolā́ g̣ [roˈlɛːɡ] ‘a star’, ruggitch [ruɡitᶴ] ‘born’. I am bothered by these words as they all seem to have ř [rʲ] whereas I should have expected rå, ruggit, & rolæg to have r̊ [r, ɹ] (= English r). (William Mylrea, Rhŷs notebook 6: 77) Jy sær̊ n [dʒәˈsæɹn] [‘Saturday’], řī [rʲiː] ‘king’ ben rēᵈn [ben reːᵈn] ‘queen’, mažə r̊ ḗu [maʒә ˈɹeːu̯ ] ‘oar’ řo “rhew” [Welsh, ‘frost’], but he seemed to sound řṓish [rʲoː(i̯ )ʃ] [‘before’] and not r̊ ṓish [ɹoː(i̯ )ʃ], but he said ruggitsh [ruɡitʃ] ‘born’ right enough. I am inclined to think that initially ř [rʲ] is ousting r̊ [r, ɹ] as an initial, just as ɫ [lʲ] tends to do l.219 (Daniel Kelly, Rhŷs notebook 6: 82) Initial /rʲ/ is well-attested in lenited position in Irish dialects (Ó Murchú 1986). Ó Murchú (1986: 22) explains this as overgeneralization of /rʲ/ as the lenited form of both 219 It is unclear what the comparison with the laterals refers to. 270 Aspects of the historical phonology of Manx historically broad and slender initial rhotics after the falling together of unlenited /R/ and /Rʲ/. In the items noted by Rhŷs, ren ‘did, made’ (G. rinne) is a preterite verbal form where lenition would be expected; the participle ruggit ‘born’ (a new formation in Manx) can be explained as generalization of lenited r from the preterite rug ‘bore, was born’ (G. rug, rugadh). Lenited roish (G. roimh) may represent analogy with forms of rish ‘to, with, by’. The spread of [rʲ] to other items noted by Rhŷs, including presumably in non-lenited positions, may perhaps represent a ‘last gasp’ phenomenon (cf. Barras 2018) in the face of the trend towards merger of /r/ and /rʲ/, and the rapid shifts in articulation of the rhotics in Late Manx. As noted by Jackson (119), Cregeen’s (vii) comment and spelling rhed seems to indicate /rʲ/ in red ‘thing’. This realization is recorded from two of Jackson’s informants. Jackson claims that ‘[t]here seems no reason for the rʹ in this case’; but it is easy enough to derive it from the lenited form in frequent collocations such as dy chooilley red ‘everything’ (G. gach uile rud). The spelling rio ‘frost, ice’ (G. reódh) may represent the initial /rʲ/ discussed here and noted in this item by Rhŷs, Jackson and Broderick, or perhaps it represents maintenance into Classical Manx of a diphthongal realization of G. eó, as in certain Scottish dialects (e.g. Borgstrøm 1940: 148). It is much less likely to represent survival of original */Rʲ/, given the evidence of early loss of this phoneme throughout Gaelic dialects. 4.2.2 Early loss of /r/, /rʲ/ As noted above, <r, rr> generally appears in the Manx orthography where a rhotic would be historically expected, and any loss seems to be associated with the obsolescent stage of language shift. The exceptions are as follows: 271 Aspects of the historical phonology of Manx 4.2.2.1 /rs, rʃ/ > /s/ (?/ʂ/) The rhotics are regularly deleted in medial and final -rs- in native and established loanwords, although /r/ may be retained or reintroduced by analogy (e.g. coorse, persoon, see below). According to Jackson (125–6), the s in such items is ‘cacuminal’ (retroflex) or alveolar, as also in Irish and Scottish dialects (with or without deletion of r), as opposed to the usual broad /s/ which is described as dental. Since the orthography (both Phillips and CM) shows this as <s(s)>, it appears that this was not interpreted as a cluster /rs/, which would suggest the necessity of positing a distinct phoneme /ʂ/ or /s̱ /. Broderick also notes this alveolar s (HLSM III: 119). A near minimal triplet would be: messyn ‘fruits’, ScG. measan; eshyn ‘he’ G. eisean, essyn ‘doorjamb’ ScG. ursainn /mesәn eʃәn eʂәn/. The following items have rs > s(s): as ‘said’, G. arsa claasagh ‘harp’, G. cláirseach custey ‘cursed’ essyn ‘doorjamb’, G. ursann fess, fesst ‘spindle’, G. fearsaid foster ‘forester’ pesson ‘parson’ possan ‘parcel of sheep’, Eng. ‘portion’ wistad ‘worsted’ (fabric) /r/ is retained, restored or introduced in the following: coorse ‘course’, G. cúrsa, but Ph. kuys (Thomson 1995: 132) erskyn ‘above’, G. os cionn (Thomson 1981: 50) ersooyl ‘away’, G. ar siubhal (Thomson 1981: 22) persoon ‘person’ 4.2.2.2 Other cases of early rhotic deletion G. urchar, EIr. aurchor (§2.1.10) gives Phillips orghyr /or(ә)xәr/, but CM orraghey /orәxә/ ‘bow-shot’, with deletion of the second /r/. The appearance of adjacent /r/ and /s/ across the word boundary in the collocation orraghey sidey ‘arrow shot’ (G. urchar saighde), as in Genesis 21:16, may explain the development (§4.2.2.1). and/or the 272 Aspects of the historical phonology of Manx influence of the shape of dorraghey ‘dark’ (G. dorcha) and verbal nouns in -aghey (ScG. -achadh). There is simplification of the cluster /RLʲ/ to non-palatalized /l/ in G. magairle, Manx maggle ‘testicle’ (with syncopated plural /maɡlәn/, HLSM II: 286). The only item of similar shape is coyrle ‘advice’ (G. comhairle), but here the cluster is preserved (ko̜ ːṛlʹ etc., HLSM II: 98), perhaps because it synchronically constitutes the coda of a stressed syllable /kõːrlʲ/. There may also be analogous simplification of /RN/ to /n/ in Manx abane ‘ankle’ (EIr. odbrann, ScG. adhbrann) if this is derived from earlier *adhbarn (Ó Maolalaigh forthcoming b) (§3.4.7.4). 4.2.3 Weakening and loss of /r/ Late Manx The weakening and loss of coda /r/ is widely attested in the material from the terminal speakers (Marstrander: 56; Jackson 118–24; HLSM III: 113–5). Jackson (118) attributes this to English influence:220 Further, ɹ is very often dropped altogether or much reduced, and in addition it is apt to develop before it after a vowel, or substitute for itself, an ə. Both these features have notable analogues in standard Southern English, and also, significantly, in the English of the Isle of Man. (Jackson: 118) Following these descriptions, this weakening and associated developments can be summarized as follows: (a) Weakening of coda /r, rʲ/ to an alveolar approximant [ɹ]. (b) Insertion of [ә] before /r, rʲ/. (c) Loss of /r, rʲ/. (d) [ә] in place of coda /r, rʲ/. (e) Centralization of vowels before coda /r, rʲ/, with or without retention of rhotic. 220 Although see §3.3.2.1. 273 Aspects of the historical phonology of Manx In addition, there are occasional cases of intrusive [r], as in non-rhotic varieties of English, in the speech of the terminal speakers (cf. HLSM II: 169). The following example is from Ned Maddrell: ha ˈrau mi ˈruː eːvɪlʹ ˈlai ɑ̜ rəðə cha row mee rieau abyl lhaih eh [r] edyr ‘I was never able to read it [Manx] at all’ (HLSM II: 267) In Rhŷs’s informants these developments are not so advanced (for example, they apparently do not affect /rʲ/, §4.2.1.3), but their incipient presence is noted: y̆ [ә] is a favourite a vowel before the broad r in Manx as the same vowel sound is before r in English words (Rhŷs: 16) This ρ [r] is assimilated very readily to other consonants, as in […] jiarg and pronounced džy̆ g [dʒәɡ] ‘red’ (Goi. dearg), and this is especially the case before consonants of the ambidental group, as in ard ‘high’, pronounced y̆ δ [әd̪ ]. […] This is not the only point of similarity between the Manx ρ and English r; for like the latter it is wont to be preceded by a furtive y [ә] […] In some cases this ρ like English r disappears as in […] feer ‘true’ (Goi. fíor […]), which in such phrases as […] feer veg ‘very small,’ is frequently pronounced fī́y̆ veg [fiːә veɡ] or fī veg [fiː veɡ]. (Rhŷs: 148) It has already been noted that these developments must be recent, as they are not shown in the eighteenth-century orthography, where r always appeared where historically expected apart from in certain limited circumstances (§4.2.2), and Jackson provides evidence that the change was underway in the lifetime of his informants. Given the very close congruence between these developments and those of the English dialects of the Isle of Man, and the sociolinguistic situation of language shift, it is highly likely that the changes observed reflect language contact. Rhotics have been noted to be particularly liable to change via contact; compare, for example, the spread of uvular rhotics through several adjacent western European languages (Trudgill 1974). For English influence on Gaelic rhotics, we may compare the replacement of 274 Aspects of the historical phonology of Manx /r/ and /rʲ/ by an English alveolar approximant [ɹ] in the Irish of younger ‘posttraditional’ speakers (Ó Curnáin 2012: 290, 292). Coda rhotics were until recently widespread in north-west England, the area with which Manx English has the closest correspondences (Barry 1984), and are still retained in a shrinking area of Lancashire (Wells 1982: 367; Barras 2018: 364–5). Traces of rhoticity were recorded in traditional rural Manx English in the midtwentieth century (Barry 1984: 174–5), which ‘points to a change from rhoticity to non-rhoticity, around the end of the nineteenth century’ (Hamer 2007: 173). We may note the pronunciation of ‘November’ with a clearly audible final approximant [ɹ] by the native Manx speaker John Nelson (1839/40–1910)221 in a wax cylinder recording from 1906.222 This [ɹ] appears to be his realization of /r/ (and /rʲ/?) in codas, as in Hiarn ‘Lord’ (G. a Thighearna), ort ‘on you’ (G. ort), cur ‘put’ (G. cu(i)r), danjeyr ‘danger’, whereas a flap [ɾ] is used elsewhere, including in final position when immediately followed by an initial vowel (e.g. ayr ain ‘our father’, G. athair againn).223 It appears that coda /r/ followed a parallel trajectory in Manx English and Manx Gaelic in the nineteenth century, i.e. [r, ɾ] > [(ә)ɹ] > [ә] > Ø. It is possible that the occasional instances of ‘strongly trilled’ realizations in final position noted by Broderick (HLSM III: 18) represent hypercorrection in reaction to these developments. Onset rhotics in conservative Manx English may resemble those noted for Manx /r/: usually either an alveolar trill [r] or tap [ɾ] […] This variant is now becoming restricted to the speech of older adults, particularly males (Hamer 2007: 172) 221 See HLSM I: 320–3; Broderick 2018a: 141–2. ‘Manx language archive recording: Part of the Evening Prayer in Manx read by John Nelson.’ Manx National Heritage. <https://youtu.be/6rwnK0ZeB9Y> [accessed 12.02.2019]. 223 The sound quality is not good enough to be sure whether or not any subtle contrast between /r/ and /rʲ/ is maintained in Nelson’s Manx. 222 275 Aspects of the historical phonology of Manx 4.2.4 Fortis /R/ /R/ and */Rʲ/ have fallen together224 in all Gaelic dialects (Jackson: 117; Greene 1977: 159; Ó Murchú 1986: 21; 1989b; Ó Maolalaigh 1997: 53–7),225 leaving a maximal ternary system of rhotic consonant phonemes in conservative dialects, and reduction to one or two phonemes in others. There is no synchronic evidence of a rhotic fortislenis contrast in Late Manx in medial or final position (Marstrander: 56; Jackson: 117– 8; HLSM III: 107), although see above (§4.2.1.5) for traces of initial lenition in the form of a palatalization contrast. However, the occasional trilled realizations described by Broderick (HLSM III: 18) may represent a non-phonemic trace of earlier /R/, as noted in Iorras Aithneach Irish (Ó Curnáin 2007: 228–34). There is no orthographic evidence for a /R ~ r/ contrast, although that is not to say that such a contrast did not exist in earlier periods, as there is no obvious way it would be encoded in the orthographies;226 compare the fortis-lenis contrast in laterals and coronal nasals, which must have existed at least in Philips’s period if not later, but is not indicated orthographically (see below). As in other dialects, there is lengthening of certain short vowels before historical /R/, e.g. baare /bɛːr/ ‘top’ (G. barr) (§4.6.1). There may also be diphthongization of /i(ː)/ in a handful of items after initial /R/ (Jackson: 121), namely roie ‘run’ (G. rith, ScG. ruith), roih ‘arm’ (G. righ), and also Ph. ry̆ i, also ríí etc., CM ree ‘king’ (G. rí). 4.2.5 Other realizations of rhotics Rhŷs (149) notes that the medial cluster /nr/ may be articulated as [nz] or [ndr]. The former may be a northern development (see also HLSM III: 18). Broderick (HLSM III: 17–8) also notes sporadic realization of intervocalic [r] as [ð] in mairagh ‘tomorrow’, (G. amáireach, amárach). Possibly gooddin (Cregeen) for Kelly’s gurrin ‘pimple’ (G. 224 Ó Maolalaigh (1997: 56–7) explores the possibility that */Rʲ/ never developed in the first place in some dialects. 225 Although a slender rhotic may be restored in paradigms by analogy (Ó Curnáin 2007: 222). 226 Cf. CM arran ‘bread’ (G. arán), carragh ‘scabby’ (G. carrach). The <rr> here indicates the shortness of the preceding vowel and it is not possible to be sure whether there was any contrast in the rhotics. 276 Aspects of the historical phonology of Manx goirín) (§4.4.7.2) reflects this realization. This is perhaps a realization of /rʲ/ specifically. 4.3 Laterals 4.3.1 Introduction For the most part, descriptions of the laterals in Late Manx suggest loss of the fortislenis contrast, with a reduction to a binary palatalization contrast (shown below). However, some of the descriptions point to a more complex picture than this. l L lʲ Lʲ L lʲ 4.3.2 Lateral contrasts in Late Manx Rhŷs (145–7) notes ‘ambidental’, ‘alveolar’ and ‘mouillé’ (palatalized) laterals in Manx in the following items: ‘ambidental’: laue, ‘hand’, G. lámh lheiy ‘calf’, G. laogh moal ‘slow, feeble, bad’, G. mall ‘alveolar’: injil ‘low’, G. íseal vel ‘is’, G. an bhfuil cummal ‘hold’, G. congbháil, ScG. cumail elley ‘other’, G. eile ainle ‘angel’, G. aingeal inney-veyl ‘maidservant’, G. inghean mhaol ‘mouillé’:227 lheie ‘melt’, G. leaghadh er-lheh ‘apart’, G. ar leith cliaghtey ‘to be wont’, G. cleachtadh fliaghey ‘rain’, G. fleachadh 227 Also as a prosthetic consonant in initial /bj/ > [blʲ], /fj/ > [flʲ]. 277 Aspects of the historical phonology of Manx glione ‘valley’, G. gleann fuill ‘blood’, G. fuil sooill ‘eye’, G. súil lhiastyn ‘owe’, G. dleastanas The correspondence of the distribution of Rhŷs’s dental and alveolar laterals with the historical contrasts is not immediately obvious. He emphasizes the strongly dental character of λ [l̪ ]: I write it [the ‘ambidental’ lateral] λ […] one is apt, while undoing the contact with the teeth, to produce a slight sound of ð, especially at the end of a word. Thus the combination ūλ is pronounced almost as if written ūλð [uːl̪ ð], and forms the Manx word for an apple […] ooyl (Goi. ubhal228 […]) (Rhŷs: 145) In his notebooks, Rhŷs refers to the dental lateral as an ‘old-fashioned’ feature particularly noticeable in certain speakers: the Mull of Galloway he usually call[s] yn Vṍulᶞ.229 He had an old fashion[ed] pronunciation probably for he pronounced the word for apple ūδlᶞ. (John Boyd, Rhŷs notebook 6: 166) Visited Police Constable Caley […]: he is a native of Jurby and […] remembers understanding Manx better than English, though he now prefers saying or relating things in English […] He says […] coulᵟ [koul̪ ] for coll hazel his broad ls sound very hollow and different from lⁱ [lʲ] or English l generally. I noticed this in his ūlᵟ [uːl̪ ] for ooyl ‘apple’ as contrasted with ūlⁱ [uːlʲ] from fuill ‘blood’ (William Caley, Rhŷs notebook 6: 58–9) 228 The dental lateral is original (eDIL s.v. uball, ubull) although later spellings with -ll and -l are found and the usual modern Scottish spelling is ubhal (Ó Buachalla 1988: 42). 229 i.e. maol. The vowel of John Boyd’s yn Vṍulᶞ [әn ˈvõu̯ l̪ d̪ ] (Rhŷs notebook 6: 166) may reflect the influence of the English form ‘Mull of Galloway’. Compare also ‘Mull Hill’ (older also ‘Mule Hill’, and contemprary local pronunciation [mjuːl]) near Cregneash in Rushen, apparently also from maol (PNIM VI: 463). It is possible that the labial resulted in a degree of rounding of the vowel, at least perceptually. 278 Aspects of the historical phonology of Manx Strachan (1897: 55) mentions a similar set of three laterals to that described by Rhŷs:230 In connexion with original broad vowels l has the thick sound found also in Irish and Scotch Gaelic, which Rhŷs describes and represents by λ. Where l seemed to me to be mouillé it has been represented by ł. This sound is clearest when it stands at the beginning of a word and is followed by a broad vowel. In some cases it was hard to decide whether l was mouillé or was simply an alveolar l. (Strachan 1897: 55) Notably, Strachan (1897: 55) contradicts Rhŷs’s (146) comment on elley ‘other’ (G. eile): elley ‘other,’ pronounced ĕ́ ḷ ẹ [ele], not ĕ́ ɫ̣ i̯ ә [elʲjә] or ĕ́ ɫ̣ ә [elʲә] (Rhŷs: 146) In ełe ‘other’ l seemed distinctly mouillé. (Strachan 1897: 55) Like Rhŷs and Strachan, Marstrander (55–6) reports three laterals, although some speakers have only two: Woodworth, Kelly og Crebbin (Four Roads) skjelner bare mellem et velart og palatalt l ɫ, begge artikulert på alveolene og brukt henholdsvis foran bakre og fremre vokaler. Derimot kjenner Christian, Taggart, Quane, Quale (Castletown) og Crebbin (Bradda Village) foruten alveolart l også et interdentalt l foran bakre vokaler. Dette l som kan betegnes λ er åpenbart den gamle fortis, men den har i mansk vunnet et større utbredelse enn i irsk. Det heter ikke alene λē, λag, kʹīλ, tәuλ, pәuλ, men også kλәun, bλē, fλәuʹnәs, koʹλbaχ, dʹžīdλ; det heter šīλ og šīl, foλt og folt, men alltid eʹnal og -al i alle infinitiver; det er intet fremlydsskifte λ : l. Mansken tilstreper åpenbart (som sydirsken) å innskrenke l-lydens antall til to: et velart l foran bakre og et palatalt ɫ foran fremre vokaler. Av de to velare l-lyd, avvek det interdentale λ betydelig skarpere fra det palatale (alveolare) ɫ enn det (likeledes alveolare) l; det var bare rimelig at det blev valgt og grep ut over sit oprinnelige område. [Woodworth, Kelly and Crebbin (Four Roads) differentiate only between a broad and slender [l] [lʲ], both articulated on the alveolar ridge and before back and front vowels respectively. However, Christian, Taggart, Quane, Quale (Castletown) and Crebbin (Bradda Village) have in addition to alveolar l also 230 Although in the transcribed text λ does not appear. 279 Aspects of the historical phonology of Manx an interdental l before back vowels. This l which can be transcribed λ [l̪ ] is evidently the old fortis, but it has in Manx gained a wider distribution than in Irish. We have not only [l̪ ɛː], [l̪ aɡ], [kʲiːl̪ ], [tәul̪ ], [pәul̪ ], but also [kl̪ әun], [bl̪ ɛː], [fl̪ әunәs], [kol̪ bax], [dʒiːdl̪ ]; we have [ʃiːl̪ ] and [ʃiːl], [fol̪ t] and [folt], but always [enal] and -[al] in all infinitives;231 there is no initial mutation [l̪ ] : [l]. Manx is clearly striving (like southern Irish) to restrict the number of l-sounds to two: a broad [l] before back and a slender [lʲ] before front vowels. Of the two broad lsounds, the interdental [l̪ ] was significantly more sharply differentiated from the slender (alveolar) [lʲ] than from the (likewise alveolar) [l]; it was only natural that it was chosen and expanded beyond its original environment.] (Marstrander: 55–6) Jackson (107–11) and Broderick (HLSM III: 107) record only a two-way phonemic distinction, although both refer to traces of the earlier fortis-lenis contrast: Some writers note the occurrence of strong forms, both broad and slender, i.e. of ʟ, ɴ, ʟʹ and ɴʹ, but I never heard them and doubt that they now exist, whatever ɴʹ may have been until recent times. [fn.] Marstrander denies the survival of the strong forms as such, but sees a trace of ʟ in the interdental variety he noted from some of his speakers for “broad” l […], it evidently had no phonemic significance and bore no relation to the old system. I never heard this among my speakers, with whom “broad” l is alveolar. (Jackson: 107) In L[ate] S[poken] M[anx] /l/ and /n/ have lost their original velar quality, though there are still some traces of it in /l/, viz. [ɫ]; this varies freely with [l] and has no phonemic significance. (HLSM III: 107) However, Jackson (110–11) refers to the apparently greater frequency of plain sonorants representing original lenis slender /lʲ, nʲ/ as evidence of a survival of the fortis-lenis contrast until ‘recent times’: Internally and finally lʹ, nʹ as well as l (*n ? and l, n), stand for original “slender” strong ll, nn (or l, n in a dental group); and also for original weak single “slender” l and n. However, it is notable that l (and l, n) are somewhat rarer in the case of the originally strong sounds than in the case of the originally 231 laa ‘day’ (G. lá), lhag ‘hollow’ or ‘weak’ (G. lag), keeayll ‘sense’ (G. ciall), towl ‘hole’ (G. toll), poyll ‘pool’ (G. poll), cloan ‘children’ (clann), blaa ‘flower’ (G. bláth), flaunys ‘heaven’ (G. flaitheamhnas), colbagh ‘heifer’ (G. colpthach), jeeill ‘harm’ (G. díoghail), sheel ‘seed’ (G. síol), folt ‘hair’ (G. folt), ennal ‘breath’ (G. anál), -al verbal noun ending (G. -áil). 280 Aspects of the historical phonology of Manx weak sounds, and it may be that a system of internal and final “slender” l and n = lʹ and nʹ (older ʟʹ, ɴʹ) versus internal and final “slender” l and n = l and n (older lʹ and nʹ) survived late and only broke down in recent times. (Jackson: 110–11) 4.3.2.1 Merger of /L/ and /l/ The ‘old-fashioned’ ‘ambidental’ or ‘interdental’ lateral noted from some speakers by Rhŷs and Marstrander clearly represents the merger of the dental and alveolar broad laterals, with the result being dental, as in Scottish dialects where this merger has occurred (Wentworth 2002; Musil 2017: 11; Ó Maolalaigh forthcoming a: 321) and for the most part in Irish.232 This is acknowledged explicitly by Marstrander (56), and is clear in the appearance of Rhŷs’s strongly dental λð, i.e. [l̪ ᶞ], in G. maol, where alveolar /l/ would historically be expected. In some dialects or speakers this has further merged with the alveolar /l̠ / discussed below. That in Early Manx /L/ and /l/ remained contrastive in medial and final position is shown by the different treatment of the stressed vowel /a/ before historical G. l and ll: there is categorical development of /a/ > /o/ in e.g. mollaght ‘curse’ (G. mallacht), whereas backing and rounding is only incipient in e.g. thalloo ‘land’ (G. talamh) (§2.1.1.2). The merger /L, l/ > /L/ must therefore have taken place between the seventeenth and nineteenth centuries. 4.3.2.2 Alveolar /ḻ/ Slender /Lʲ/ and /lʲ/ had apparently also merged as a single palatalized lateral in most environments by the Late Manx period, according to the fieldwork sources discussed above (cf. Ó Maolalaigh [forthcoming a: 333–7] for similar developments in Scottish Gaelic). However, we must also account for a third lateral, a plain alveolar [l̠ ]. This is 232 In Irish merger to a dental lateral is reported by Ó Cuív (1944: 46, 48), de Bhaldraithe (1945: 40–1), Breatnach (1947: 47), Mhac an Fhailigh (1968: 41), Ní Chasaide (1979), Ó Sé (2000: 17–8). However, Ó Curnáin (2007: 200–1) notes alveolar [l] representing both historical /L/ and /l/ in the majority of Iorras Aithneach speakers who have the merger. 281 Aspects of the historical phonology of Manx most robustly attested in the suffixes -al, both verbal (G. -áil) and adjectival (G. -amhail), and, according to Marstrander, apparently in nouns such as ennal ‘breath’ (G. anál) as well. In data from HLSM II, [lʲ] is usual in stressed reflexes of -áil and -amhail, but never occurs in the unstressed reflexes, except where this is fused into a monosyllabic form through fricative vocalization (goaill /ɡoːlʲ/ ‘take’, G. gabháil and jeeill /dʒiːlʲ/ ‘harm’, G. díoghbháil), or secondary lenition (two instances of credjal ‘believe’; see also the development of the verbal noun suffix -(a)in, §4.4.7.1). On the other hand, it appears that dental [l̪ ] does not occur in the -al ending either. Alveolar [l̠ ] also occurs consistently in the dependent present tense of the substantive verb vel, cha nel (Ir. fuil, ScG. eil, etc.) (HLSM I: 75–7, II: 66, 472). Rhŷs (146) emphasizes that [ḻ] rather than [lʲ] is present in elley ‘other’ (G. eile), although in the HLSM data, [lʲ] is found in three out of the four instances given, and this is corroborated by Strachan (1897: 55). Other cases of alveolar [ḻ] in Rhŷs are from historical broad /l/ and /L/ and presumably are from speakers who do not have the dental lateral. We must therefore presume a phoneme /ḻ/ with a restricted distribution among speakers with a ternary lateral system. We can therefore sketch an evolution of the lateral system as such, with Late Manx speakers having either system (2) or system (3): (1) l L lʲ Lʲ (2) (3) l̪ ḻ lʲ ḻ lʲ This Manx development is remarkable, since merger of broad and slender laterals is otherwise largely unknown in Gaelic dialects,233 although in the coronal nasals the parallel development /nʲ/ > /n/ is widespread in Scottish Gaelic (Ó Maolalaigh 2001: 16). For a somewhat similar situation, we may compare the ‘limited l v. ʟ opposition’ possibly present in some speakers of Iorras Aithneach Irish resulting from retention of ‘an alveolar articulation of morphologically depalatalised lʹ […] e.g. verbs in -áil -ɑːlʹ > future -álfaidh -ɑːlə’ (Ó Curnáin 2007: 201; cf. S. Ó Murchú 1989: 80). Phonemes 233 Ó Maolalaigh (forthcoming a: 329–30) gives a few examples of velarized laterals in Scottish Gaelic dialects in place of expected /lʲ/ or /Lʲ/, but specific etymological or analogical grounds (e.g. historically attested by-forms with /L/ or /l/, e.g. brollach / broilleach) can be suggested for most of them. 282 Aspects of the historical phonology of Manx with such limited distributions are likely to be inherently unstable and liable to be reduced. In relation to the development of unstressed -al it should be noted that post-tonic unstressed */alʲ/ and */anʲ/ seem not to occur at all in the phonology (§4.4.7.3).234 A possible source for generalization of /ḻ/ (=the historical broad lenis lateral /l/) in adjectives would be derivatives such as gennallys ‘gladness’ (G. *geanamhlas), spreading to gennal ‘glad’ (G. geanamhail).235 On the other hand, Late Manx /ḻ/ in three-lateral idiolects (in -al, vel and perhaps elley) can be seen as representing historical slender lenis /lʲ/, preserved in certain stereotyped circumstances when other instances of lenis /lʲ/ had merged with /Lʲ/. Loss of secondary articulations is perhaps inherently more likely in final position than in initial or medial position, and in unstressed syllables (cf. verbal noun -yn /әn/, G. -(a)in, §4.4.7.1) — although this is not necessarily expected to be the case with sonorants, where the acoustic cue to the secondary articulation is salient during the consonant itself, rather than simply during formant transitions in adjacent vowels as with stops. That [l] in -al is not simply an allophone of /l̪ / in unstressed position is shown by the appearance of Rhŷs’s lᵟ in some instances of injil ‘low’ (G. íseal), and the names Maghal ‘Maughold’ and Cranstal (Rhŷs notebook 6: 165), which would give a contrast between unstressed /al̪ / and /aḻ/). Some apparent instances of /ḻ/ for expected /lʲ/ may be the result of environment, as millish ‘sweet’ (G. milis) (Rhŷs notebook 6: 162; HLSM II: 299) where palatalization on- and off-glides might not be easily distinguishable from the adjacent high vowels in [mɪlʲɪʃ]. 234 Although there is one possible instance of /anʲ/ (§4.4.7.3). There is orthographic evidence for maintenance of this alternation in e.g. the doubly-suffixed eadolagh ‘jealous’ (G. *éadamhlach) vs. adjectives in simple -oil (G. -amhail) (but cradoilagh ‘mocking’ < G. cnáid; gerjoilagh ‘joyous’ < G. gairdeach), and in verbal nouns in -ail, derivative -alys, -alagh, e.g. kiarail ‘intend, care’ /kʲaˈrɛːlʲ/, kiaralagh ‘careful’ /kʲaˈrɛːlax/. 235 283 Aspects of the historical phonology of Manx 4.3.2.3 Initial lenition of laterals There is little or no evidence as to whether and to what degree lenition of initial laterals may have been maintained in Manx. The variant spellings lesh and liesh for G. le(is) ‘with; with him, it’ given by Cregeen (s.v. liesh) may suggest coexistence of originally lenited and unlenited forms of this preposition. i.e. /lʲeʃ/ < */Lʲeʃ/ and /ḻeʃ/ < */lʲeʃ/. However, in the Bible liesh is restricted to s’liesh, by-liesh, my-liesh in the sense ‘own’ (e.g. y vooinjer by-liesh eh ‘the owners thereof’, Luke 19. 33), and therefore maybe an orthographic attempt to differentiate senses, or reflect the fact that the cluster -s l- in is leis would be expected to have a fortis lateral after the sibilant, i.e. /sLʲ/. 4.3.3 Lateral contrasts in Early Manx There is evidence for a maintenance of fortis / lenis contrasts /L ~ Lʲ ~ l ~ lʲ/ in the Phillips orthography. For example the spelling <all> is used for both G. /aL/ and /al/, without indication of the later development /aL/ > /o:l/ (although the diacritic in iáll may represent a degree of vowel lengthening, i.e. [jaːL]). The fact that these lexical sets later diverge in their development (/al/ > /al/; /aL/ > /oːL/ > /oːl/) entails that the consonants were contrastive in Phillips’ time, prior to transphonologization into a contrast of the preceding vowel. boayl ‘place’, G. ball. Ph. ball, boll gial ‘white’, G. geal. Ph. gall giall ‘promise’, G. geall. Ph. iall (2), iáll, jall gioal ‘pledge’, G. geall. Ph. giall It is less easy to demonstrate the persistence of the /Lʲ ~ lʲ/ contrast in e.g. keeill ‘church’ (G. cill) and mill ‘honey’ (G. mil), given that only lengthening of certain vowels occurs before /Lʲ/, rather than a change in vowel quality. The spellings of keeill with <í> may indicate lengthening: keeill ‘church’, G. cill. Ph. kíll (2), kííll, kill mill ‘honey’, G. mil. Ph. mill (2), mil 284 Aspects of the historical phonology of Manx However, given the evidence for merger of broad /L/ and /l/ in 19th century Manx (§4.3.2.1) and the fact that this has been shown cross-dialectally to occur prior to merger of the slender laterals (Ó Maolalaigh 1997: 59), it follows that, if it is the case that /L/ and /l/ were contrastive around 1600, it is likely that /Lʲ/ and /lʲ/ remained distinct at this period, too, as /Nʲ/ and /nʲ/ certainly did (§4.4.6). 4.4 Coronal nasals 4.4.1 Introduction The development of the coronal nasals is somewhat more complex than that of the laterals, but there is also more orthographic evidence, especially for the development of /Nʲ/ and its interaction with original /ŋʲ/ (< */ŋʲɡʲ/). As with the laterals, there is some evidence which suggests an alveolar /ṉ/ with a limited distribution alongside dental /ṉ/ and palatalized /nʲ/, although the evidence is less clear than in the case of the laterals. 4.4.2 Contrasts in coronal nasals in Late Manx Rhŷs (133–5) notes three phones corresponding to the Gaelic coronal nasals (leaving aside [ŋ] and [ŋʲ]), namely dental (‘ambidental’) ν [n̪ ], alveolar n [n̠ ] and palatal or palatalized (‘mouillé’) ñ [nʲ]. The ‘ambidental’ nasal is noted in noa ‘new’ (G. nuadh) and kione ‘head’ (G. ceann), where fortis dental /N/ is historically expected, and also in lane ‘full’ (G. lán), where the lenis alveolar /n/ would be expected. Rhŷs (133) also notes [n̪ ] in Manx English ‘no’. The ‘mouillé’ nasal appears for G. fortis and lenis /Nʲ ~ nʲ/, in e.g. niart ‘strength’ (G. neart), veign ‘I would be’ (G. bheinn, bhínn), thallooin ‘earth’ (gen.) (G. talmhain), blein ‘year’ (G. bliadhain), yn irriney ‘the truth’ (G. an fhírinne), er n’yannoo ‘have, has done, made’ (G. iar ndéanamh), yn aspick ‘the bishop’ (G. an easpaig) yñ i̯ aspick [әˈnʲaspik] (Rhŷs: 134–5). 285 Aspects of the historical phonology of Manx Rhŷs’s alveolar nasal, which ‘is the sound of English and Welsh236 n; but it occurs oftener in those languages than in Manx’ (ibid.: 134), has the most difficult distribution to explain. Rhŷs also notes alveolar [n̠ ] from original /N/ in nearey ‘shame’ (G. náire), bannit ‘blessed’ (G. beannaighthe), sheelnaue ‘mankind’ (G. síol nÁdhaimh), bwoirryn ‘female’ (G. boireann), and original /n/ in inneen ‘daughter’ (G. inghean) (with reference to the final n), feeyn ‘wine’ (G. fíon), grian ‘sun’ (G. grian). He records variation between [ṉ] and [nʲ] in the dependent copula form (cha) nee (copula, G. an é, chan é etc.) nī, hă nī́ [(ha) ˈn̠ iː], but also hă ñi̯ ē, hă ñi̯ ē̃ [ha ˈnʲ⁽ẽ⁾ː] ‘heard […] at Cregneish’, and also notes [ṉ] in nee ‘will do’ (G. (do) (gh)ní) (§4.4.4). Strachan (1897: 55) notes only two coronal nasals, alveolar n [ṉ] and ‘mouillé’ ñ [nʲ]. He refers explicitly to ‘Rhŷs’s second (alveolar) n’: Rhŷs’s second (alveolar) n seemed to me to be sounded in mennick [‘often’, G. meinic], činnә [chengey ‘tongue’, G. teanga], jinnaχ [jinnagh ‘would do’, G. déanadh, deineadh], jinnu [jannoo ‘do’, G. déanamh], Inid [Innyd, ‘Ash Wednesday’, G. Inid], perhaps henk [haink ‘came’, G. tháinig], though there it may have been a little mouillé, hannik [honnick ‘saw’, G. chonaic], finnish [feanish ‘witness’, G. fiadhnaise]. But the two sounds were not always easy to distinguish. (Strachan 1897: 55) Marstrander (57), Jackson (107) and Broderick (HLSM III: 14–5, 106–7) note a two- way distinction between [n] and [nʲ], with no trace of the fortis-lenis contrast apart from diphthongization before historical */N/, and /ŋʲ/ from */Nʲ/. None of them mention the dental [n̪ ] noted by Rhŷs. 236 Presumably southern Welsh, as northern Welsh /n/ is often strongly dental (Ball and Williams 2001: 63). 286 Aspects of the historical phonology of Manx 4.4.3 Evidence for /nʲ/ > /ṉ/ As discussed above, Jackson (110–1) suggests that plain [ḻ] and [ṉ] occur more frequently for the original lenis /lʲ/ and /nʲ/ than for fortis /Lʲ/ and /Nʲ/, and that this may be evidence for late survival of the fortis-lenis contrast. Data from Broderick’s dictionary (HLSM II) was collated to test Jackson’s hypothesis (Table 59, Chart 9). Table 59. Incidence of palatalized and non-palatalized realizations of Gaelic laterals and coronal nasals in data from HLSM II237 l, n lʲ, nʲ no. tokens lʲ 35.6% (165) 64.4% (299) 464 Lʲ 31.7% (53) 68.3% (114) 167 nʲ 45.9% (206) 54.1% (243) 449 Nʲ 31.6% (49) 68.4% (106) 155 Chart 9. Incidence of palatalized and non-palatalized realizations of Gaelic laterals and coronal nasals in data from HLSM II 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% lʲ Lʲ nʲ l, n Nʲ lʲ, nʲ 237 All transcribed individual items given after the headwords in HLSM (II) were included, from the main dictionary, the place-names and the addenda. Different transcriptions from the same speaker were counted separately, and where the same transcription is noted as being from n speakers, this is counted n times. Data from the example sentences were generally not included, as they generally duplicate instances given in the individual item transcriptions, but sub-headings (compound words, inflected forms etc.) were included. Items ending in unstressed -al (G. -áil, -amhail), which generally have unpalatalized [l] were excluded, as were items in -yn (G. -(a)in) except geddyn, feddyn. Only intervocalic medial and postvocalic final sonorants were investigated. The main transcriptions in HLSM are n, nʹ, l, lʹ. Other occasional transcriptions include e.g. Nʹ, nʹj, nj, nʲ, ᵈʹn counted under [nʲ], N counted under [n], and a similar range of variants for the laterals. Realizations with ŋ(ʹ) were not counted in rheynn ‘divide’ (G. roinn) and Nherin ‘Ireland’ (G. Éirinn), which also have forms with [n] or [nʲ], and items with only ŋ(ʹ) were excluded entirely (e.g. ching ‘sick’, G. tinn). 287 Aspects of the historical phonology of Manx Overall, these data show the considerable extent of depalatalization in the Manx of the terminal speakers, as noted by Broderick and Jackson – in all four cases over 30% of tokens show non-palatalized forms. The percentages show there is little difference between the figures for reflexes of historical */lʲ/ and */Lʲ/. However, for */nʲ/ the percentage of tokens showing nonpalatalization is considerably higher than for */Nʲ/. Moreover, a number of */nʲ/ items (with at least 5 tokens) which have exclusively non-palatalized realizations, namely the following:238 anney ‘commandment’, G. aithne (6 tokens) ennym ‘name’, G. ainm (8 tokens) accan ‘complaint’, G. acaoine (9 tokens) hene ‘self’, G. féin (6 tokens) imman ‘drive’, G. iomáin (5 tokens) mwannal ‘neck’, G. muineál (9 tokens) shen ‘that’, G. sin (12 tokens) There are no such items for */Nʲ/, */lʲ/, */Lʲ/, apart from skillin ‘shilling’ (G. scilling, scillinn, 5 tokens). Another pair of related items which come close is enn(ey) ‘recognition’ (G. aithne), ennaghtyn ‘feel, perceive’ (ScG. aithneachdainn), which taken together have 7 tokens with [n], and 1 with [nʲ]. Both anney and ennaghtyn are spelled in Phillips exclusively with forms lacking an explicit indication of palatalization (anny, an(n)aghyn; e(a)naghtyn). Accan and imman will be discussed below (§4.4.7.3). G. sin has numerous variant forms, including ones with final non-palatalized /n/ (eDIL). It might be suggested that /nʲ/ was depalatalized in anney and enney (both G. aithne) owing to its adjacency to non-palatalized /h/ (of which there is no trace in the attested period of Manx, here or in any non-initial position), while the non-palatalization in ennym may be connected to its adjacency to non-palatalized /m/ (prior to svarabhakti, or in syncopated forms pl. enmyn ‘names’, verb enmys, enmaghey ‘to name’). 238 Spain ‘spoon’ might be included here (6 tokens with [n]), on the basis of ScG. spàin, Ir. variant form spáin (Ó Dónaill) (< Norse spánn, or Scots spane), but it is possible we have a form *spán here. 288 Aspects of the historical phonology of Manx On the other hand the adjacency to preceding front vowels in these items, and in shen, hene (see also shin ‘we’, G. sinn, below) may be a contributory factor, as in Scottish Gaelic (Ó Maolalaigh 1997: 63–4). One or both of these factors (adjacency to a historically plain consonant, and preceding front vowel) might explain other possible cases of depalatalization, such as rhenniagh ‘fern’ (G. raithneach, 5 [n], 1 [nʲ]), eunys ‘joy’ (G. aoibhneas, éibhneas, 4 [n]) and lhiennoo ‘infants’ (G. lein(i)bh, 2 [n]). In the last of these the spelling is fairly firm evidence for depalatalized [n], since otherwise *lhienniu might be expected (cf. thenniu ‘thaw’, ScG. taineamh; terriu ‘bulls’, G. tairbh). In the case of anney ‘commandment’ (G. aithne), Cregeen’s rather opaque comment ‘sounded Ahney’ suggests something noteworthy about the pronunciation. The only similar comment of Cregeen’s is for ennagh ‘some, certain’ (G. éigin(each)), for which we find the comment ‘pronounced Ehnnagh’. No explanation is given for the value of this <hn(n)>, and Wheeler (2018) suggests in the case of ennagh ‘i.e. with slender /nʲ/’. However, it is more likely, in view of the other evidence presented here, that Cregeen is referring to depalatalization, i.e. a change [nʲ] > [ṉ], which would be of restricted distribution and thus of note. In HLSM II, ennagh has 3 tokens with [n] and 1 with [nʲ]. For the development of G. ui in mwannal ‘neck’ (G. muineál), see §2.1.8. There may also be depalatalization after /iː/. The CM spelling sheeyney ‘stretch, reach’ (G. síneadh) (HLSM II: [n] 1), with <y> which typically indicates a following broad or plain consonant (rather than e.g. *sheeiney), may indicate a depalatalized form. The opposite development appears to have taken place in speeiney ‘peel, strip’ (G. spíonadh) (HLSM II: 1 [n], 3 [nʲ]).239 The spelling variants meen, meein (both found in the Bible) ‘fine, soft, tender; dear’ (G. mín) may point to palatalized and depalatalized by-forms (1 [n], 4 [nʲ], HLSM II). There is further evidence in HLSM (II) of confusion between historical /iːnʲ/ (ín etc.) and /iːn/ (íon etc.): 239 From spíon ‘thorn’ (< Latin spina) (eDIL s.v. spín), cf. ScG. variant spìn (Dwelly). 289 Aspects of the historical phonology of Manx Historical /iːnʲ/: bwilleen ‘loaf’ (G. builbhín) 5 [n], 6 [nʲ] lhemeen ‘moth’ (G. leaghman > -ín) 1 [n] meen, meein ‘fine, gentle’ (G. mín) 1 [n], 4 [nʲ] reen, s’reen ‘tough’ (G. righin, is righne) 2 [n], 2 [nʲ] saveen ‘slumber’ (G. sáimhín) 1 [n], 4 [nʲ] sheeyney ‘stretch, reach’ (G. síneadh) 1 [n] shilleen ‘slug’ (G. seilchide > -ín) 1 [n] Trilleen ‘Pleiades’ (G. Tréidín) 1 [n] Historical /iːn/: inneen ‘daughter’, girl’ (G. inghean) 2 [n], 6 [nʲ] berreen ‘cake’ (G. bairghean) 1 [n] (initial stress) cooilleeney ‘fulfil’ (G. coimhlíonadh) 1 [n] lhieeney ‘fill’ (G. líonadh) 4 [n] lieen ‘net’ (G. líon) 4 [n], 1 [nʲ] speeiney ‘peel, strip’ (G. spíonadh) 1 [n], 3 [nʲ] whilleen ‘as many’ (G. a choimhlíon) 1 [n] (initial stress) There are also instances of the plural termination -eenyn with [nʲ] (4, all TC), which probably involves -anna- etymologically (§5.1.4), but [n] in raanteenys ‘surety’ (TC). In addition, Rhŷs (Broderick 2019) has examples of cooilleeney and whilleen with [nʲ]. 4.4.4 /N/ and /n/ > [n̪ ], [n̠ ] As discussed above (§4.4.2), Rhŷs reports both a dental and alveolar nasal. The distribution of these does not correspond with that of historical G. */N/ and */n/, however; for example, Rhŷs reports dental [n̪ ] in lane ‘full’ (G. lán), but alveolar [n̠ ] in bannit ‘blessed’ (G. beannaighthe). Rhŷs (134) attempts to explain the occurrence of [n̠ ] in nearey ‘shame’ (G. náire) ‘pronounced nē ̣rә’ [ṉeːrә] and inneen ‘daughter’ (G. inghean) ‘pronounced ñi̯ īñ ’ [nʲĩːn̠ ] as being motivated by the synchronic quality of an adjacent vowel. In some of the instances mentioned, it is to be noticed that the [alveolar] n occurs where the other Goidelic dialects postulate an ambidental nasal 290 Aspects of the historical phonology of Manx associated with a broad vowel241 whereas in Manx that vowel has been narrowed as in nearey, and inneen, so that here at least alveolar n appears as a compromise between ν [n̪ ] and ñ [n]. (Rhŷs: 134) However, it seems more likely that we have here a merger of /N/ and /n/, with the two realizations used indiscriminately, perhaps varying allophonically, or according to speaker and dialect. It is also possible that the two merged as dental [n̪ ], with subsequent change to an alveolar realization. The influence of English alveolar [n̠ ] may be relevant. The development would thus be similar to that of */L/ and */l/, except that the dental lateral [l̪ ] seems to have survived longer, being reported by both Rhŷs and Marstrander (§4.3.2). 4.4.5 /ṉ/ representing fossilized initial lenition A trace of initial lenition of nasals is seen in the appearance of alveolar /n/ in nee ‘will do’ (G. do ghní), as opposed to nhee ‘thing’ (ní), and niee ‘wash’ (G. nighe) (Rhŷs: 134): nee ‘will or shall do’, sounded nĩ [ṉĩː] with an alveolar n […] [fn.] From the last three instances [nee (copula), cha nee, nee ‘will do’], all pronounced with n, must be distinguished niee to ‘wash’ (Ir. nighe […]), and nhee [‘thing’] (Ir. nídh) […]. The latter two are pronounced identically, ñī̃ [nʲĩː]. Four of these words occur in the following sentences: Nee oos [sic] nagh jean eh niee son nhee erbee. Cha nee. ‘Is it thou that wilt not wash it for anything in the world? No.’ (Rhŷs: 134) However, nee (G. do ghní) may also be found with [nʲ]: nee ‘he will do’ (with Eng. n), but neem is ñĭm [nʲim] ‘I will do’ (William J. Caine, Rhŷs notebook 7: 198) 241 This is correct with regard to initial n in náire /Naːrʲә/, but not final n in inghean /inʲɣʲәn/; Rhŷs does not seem to have fully understood the distribution of the Gaelic fortis-lenis contrasts. 291 Aspects of the historical phonology of Manx Rhŷs does not report any cases of dental [n̪ ] representing historical lenis /nʲ/. As discussed above, /nʲ/ appears to be depalatalized in a limited set of items. It is possible therefore that some speakers into the Late Manx period had a limited three-way phonemic contrast, similar to that apparently found in the laterals. Possible contrastive examples would be the following: /nʲiː/ niee ‘wash’, G. nigh(e); nhee ‘thing’, G. ní /n̠ iː/ nee ‘will do’, G. do ghní /n̪ iː/ (also /n̪ әi̯ /) nuy ‘nine’, G. naoi /banʲә/ bainney ‘milk’, G. bainne /an̠ ә/ anney ‘commandment’, G. aithne /ban̪ әtʲ/ bannit ‘blessed’, G. beannaighthe As suggested above with regard to the laterals, the defective distributions of these three-sonorant systems in Manx may have contributed to their reduction to binary contrasts. 4.4.6 Reflexes of G. /Nʲ/ The most notable survival of the fortis and lenis contrasts in Classical and Late Manx is the realization of historical /Nʲ/ as a palatalized velar nasal /ŋʲ/ (Rhŷs: 135–6; Marstrander: 57; Jackson: 111; HLSM III: 110). As noted by Rhŷs, this development appears to be only incipient in seventeenth-century Manx, judging by Phillips’ orthography: The combination of n mouillé and i is liable to be changed into a palatal ŋ as in [Ph.] shuinyn, now shinyn ‘we, us’ (Goi. sinne) which is very commonly pronounced shiŋyn [ʃiŋәn] (perhaps shiŋi̯ yn [ʃiŋʲәn]); and I conclude that the same n mouillé, rather than the ordinary n, was the starting point of the change illustrated by such words as kiínn, kiin, kin, now written king and pronounced kĭŋ̣ [kɪŋ] ‘heads’ (Goi, cinn […]); chiinn, chinn, now written çhing ‘sick, ailing’ (Goi. tinn); and piín, now ping ‘a penny’ (Ir. píghin, ScG. peighinn […]); also reyng, rǽyng, reng, rǽyn, now rheynn ‘did divide (Ir. doroinn, ScG. roinn). Here, as in væing, the form favoured by the scribe of the Phillips Prayerbook has not prevailed, and other instances parallel to væing, as contrasted with veign [‘I should be’], occur commonly enough, as for instance at [Moore and Rhŷs 1895] p. 537, where we have near one another ætlieing, now etlin ‘I should 292 Aspects of the historical phonology of Manx flee’, ’urring ‘I should remain’ (Ir. fhuirighinn): the other pronunciation is evidenced in the same passages by raghein ‘I should go’ (Goi. rachainn), and ienin, now yinnin ‘I should make’ (Goi. dhéanainn). (Rhŷs: 135–6) Rhŷs (136–7) also remarks on the reverse development, /ŋʲ/ > /nʲ/, with the notable example of chengey ‘tongue’ (G. teanga): In some instances ŋi̯ [ŋʲ] is changed into ñi̯ [nʲ], thus reversing, as it were, a change to which attention has just been called under the nasal mouillé. Take the following examples: [Ph.] luinie, now lhuingey, genitive of lhong (Ir. long, genitive luinge […]) ‘a ship,’ in the phrase kall luinie, now coayl-lhuingey ‘shipwreck,’ where at the present day the pronunciation with ŋ is the only one to be heard in the Island […]; and [Ph.] chiange, now çhengey ‘tongue,’ […] which, however, occurs also written my hinnge. The O[ld] Ir[ish] was tenge, genitive plural tengad (Med[ieval] Ir[ish] tenga, Mod[ern] Goi[delic] teanga), but in spite of the ng the Manx pronunciation is mainly tsẽ ñi̯ ә [tʃenʲә], identical wholly with that of chene, now written çhenney ‘fire’ (Goi. teine […]), which has in consequence become obsolete as the ordinary word for fire in Modern Manx. It is right, however, to say that the pronunciation of çhengey ‘tongue’ with ŋ is by no means extinct, as I have heard it regularly used at Cregneish in the South; but Phillips’ spelling hinnge seems to suggest the other and more prevalent pronunciation. (Rhŷs: 136–7) The development of fortis /Nʲ/ to /ŋʲ/, with no such change in original lenis /nʲ/, and the lack of indication of the /ŋʲ/ realization in Phillips’ orthography, shows that there must have been a contrast /Nʲ ~ nʲ/ in Early Manx, and perhaps later. See for example the following near minimal pair: /nʲ/ > /nʲ/ kynney ‘kindred, tribe’, G. cineadh,242 Ph. kieny (3), kiĕny (2), kiney (2), kĭy̆ ny, kyne, kyney /Nʲ/ > /ŋʲ/ chingey ‘sick’ (pl.), G. tinne, Ph. chinny The development /Nʲ/ to /ŋʲ/ is almost categorical after in /i/ in fully stressed syllables (Table 60). 242 The CM reflex with <nn> /nʲ/ rather than <ng> /ŋʲ/ shows that this represents the historical G. form cineadh (eDIL s.v. ciniud), as opposed to ScG. cinneadh, which may be influenced by cinneadh ‘growing, growth’, and/or reflect the ScG. split /nʲ/ > /Nʲ/ or /n/ (Ó Maolalaigh 2001: 16); cf. Borgstrøm (1941: 38). 293 Aspects of the historical phonology of Manx Table 60. inn /ˈiNʲ/ > /iŋʲ/, /inʲ/ item Phillips bing; bingys bín; binniys, binnys (2), binnis, bínnys bing ching; chingys chinn (3), chin̄ (2), chinny; chinnys CM pronunciation /biŋʲ/ etymology English binn; binneas sweet-sounding, melodious; melody /biŋʲ/ ScG. binn jury /tʃiŋʲ/, /tʃiŋʲәs/ tinn; tinneas sick; sickness ingan /iŋʲan/ inneóin anvil innagh /inʲax/243 inneach woof (of cloth) /kʲiŋʲ/ cinn head (gen., pl.) lhing /lʲiŋʲ/ linn period, age, time lhingey /lʲiŋʲә/ linne shuin (7), shin /ʃin/244 sinn cavity between rocks in river we, us shuiniyn (4) /ʃiŋʲәn/ sinne we, us (emph.) /ʃlʲiŋʲan/ slinneán shoulder-blade king kinn (4), kiin shin 245 shinyn shlingan Apart from shin ‘we’ (G. sinn) (see below §4.4.6.2 for discussion), the only exception is innagh ‘woof’ (G. inneach). Here the development of a velar nasal is perhaps blocked by the presence of the following velar fricative; we may compare the blocking of the change /sk/ > /st/ by a following dental stop (Lewin 2015b: 72). Two other items have variants with /iŋʲ/ from -uinn and -ainn: Table 61. Other instances of /Nʲ/ > /ŋʲ/ item Phillips cruin, cring, cruing (Cr.) krŭin, krunn (2), kruinn, kruin, krŭyn, kruyn (2), krynn, krun reyn, rǽyn (2), reynn, renn, ren̄ , rêŷng, reygn, ræing, reng, ren, reyng, ræyng, ræyn, (impv. pl.) renji, reynnigi, (fut. 1sg.) renniym rheynn, ring (Cr.) CM pronunciation /krunʲ/, /kriŋʲ/ etymology English cruinn round, close, compact /reːnʲ/, /riŋʲ/246 rainn, roinn division, divide ïnɑ ̣x TC, HK (HLSM II: 225). ʃin, ʃɪn, ʃen (HLSM II: 403). 245 Likewise ainyn /iŋʲәn/ ‘at us’ (G. againne), dooinyn /diŋʲәn/ ‘to us’ (G. dúinne) etc. (HLSM I: 61). 246 re̜ idʹnʹ TT, riŋ JW, raiᵈn, reːn TM, rødʹᵊnᵊʹ TC (HLSM II: 366–7). 243 244 294 Aspects of the historical phonology of Manx 4.4.6.1 /Nʲ/ in final unstressed syllables Spellings with <ng>, apparently indicating [ŋʲ], are also found in final unstressed syllables, both in Phillips and eighteenth-century sources, although the standard Classical Manx orthography mostly favours -in, and the Late Manx fieldwork data generally has [inʲ, in], with one instance of [ŋ]. In Phillips, as noted by Rhŷs (see above), the conditional/imperfect 1sg. ending G. -(a)inn can be spelled with <ng> (Thomson 1953: 50): ætlieing ‘I would fly’, G. eitlighinn urring ‘I would remain’, G. fhuirighinn væing ‘I would be’, G. bheinn but: gy niinshiyn ‘I would tell’, G. insinn vêîn (3), vêînsh, væynsh ‘I would be’, G. bheinn(se) leshshin, lessyn ‘I should’, G. dleasienīsh, ienin (2) ‘I would do’, G. déanainn etc. rağh̆ ein ‘I would go’, G. rachainn, raghainn odin ‘I could’, G. fhéadainn ioinsh ‘I would get’, G. gheibhinnse The place-names Mannin ‘Isle of Man’ and Lunnin ‘London’ appear as Mannyng, Lunnyng in certain eighteenth century texts, notably on title pages such as that of Matthew’s Gospel of 1748 and Yn fer-raauee Creestee (1763) which is prentyt ayns Lunnyng ‘printed in London’ and the Epistles and Revelation (1767), prentyt ayns Mannyng ‘printed in the Isle of Man’. In HLSM there is one instance of Nherin ‘Ireland’ (G. Eirinn) with ŋʹ, alongside six instances of [nʲ]. The spelling conning ‘rabbit’, found in the Bible and manuscript, also suggests [ŋʲ] in a form representing coinín > *coinin > *coininn, with fortis /Nʲ/ owing to MacNeill’s Law (see e.g. Ó Buachalla 1988).247 The reverse change is attested in chengey ‘tongue’ (G. teanga). The regular development of this seems to be attested in Phillips chiange etc. (i.e. /tʃaŋә/,with the 247 We may tentatively conclude that Manx follows Irish dialects in observing MacNeill’s Law, as opposed to Scottish Gaelic (Ó Buachalla 1988: 42), given that no evidence has come to light of Nalbin ‘Scotland’, (G. Albain, but ScG. Albainn), with [ŋʲ], as opposed to Mannin (G. Manainn), Lunnin (ScG. Lunnainn) and Nherin (G. Éirinn). 295 Aspects of the historical phonology of Manx usual development of ea > /a/) (Rhŷs 136–7), whereas the forms with /enʲ/ in the later language must represent a by-form *teinge (cf. EIr. forms teng, teing, ting, eDIL, LEIA s.v. tengae). As noted by Rhŷs (136), the same change is apparently evident in Ph. luinie, later lhuingey, genitive of lhong ‘ship’ (G. long). The later restoration of /ŋʲ/ here may be explained by analogy with the nominative form, or adoption of a pronunciation with /iŋʲ/, rather than /uŋʲ ~ unʲ/ (cf. cruinn, cring above). 248 These developments point to towards merger of original /Nʲ/ and /ŋʲ/ (the latter originally restricted to a small number of items, from earlier [ŋʲɡʲ]), with a [ŋʲ] realization of the merged phone following /i/, and [Nʲ ~ nʲ] elsewhere. The later predominance of /nʲ/ rather than /ŋʲ/ in final unstressed syllables may represent a general change /ŋʲ/ > /nʲ/, /n̠ / in this position, also when the /ŋʲ/ is original, as in farling ‘farthing’ (G. feóirling), LM faːrlʹən HK, faːʳlən JW (HLSM II: 158), and aghin ‘petition’ (G. athchuinge), although in the latter case the orthography indicates /Nʲ/ or /nʲ/ as early as Phillips (Thomson 1953: 147), likewise in Ph. farsyn ‘manifest’ (G. farsaing) (ibid.: 211), an item not found in later texts.249 Compare unstressed -(a)ing in Scottish Gaelic, which is realized as -(a)inn /әNʲ/ in most dialects (SGDS III: 346– 7), and similarly in many Irish dialects (LASID I: 183).250 248 The original velar nasals for reasons of space are not discussed separately here, but are assumed to have generally followed the widespread Irish development /ŋɡ/ > /ŋ/, /ŋʲɡʲ/ > /ŋʲ/ (Quiggin 1906: 106– 8; Ó Cuív 1944: 120; Breatnach 1947: 141; de Búrca 1958: 132; Wagner 1959: 31–2; Henderson 1974: 146–7; Ó hUiginn 1994: 561; Ua Súilleabháin 1994: 489; GOI: 94). Retention of /ŋɡ/, /ŋʲɡʲ/ is the rule in most Scottish Gaelic varieties (e.g. SGDS III: 336–7, IV: 266–7; v: 146–7, 322–3), and is found also in certain Irish dialects (de Bhaldraithe 1945: 39; Ó Sé 2000: 17–8; Ó Curnáin 2007: 198). In Manx there may also be retention of /ŋɡ/ in certain items such as bangan ‘branch’ (G. beangán), baŋgan (J:NM), but also bãŋan (TT), bẽ̜ŋan (JW) (HLSM II: 23). See Jackson (105–6), HLSM (III: 105–6). 249 Ph. aghyin, aghein etc. This item is often confused with accan ‘complaint’ (§4.4.7.3) in Early and Classical Manx texts (Thomson 1981: 122), and the form aːɣan (JW) (HLSM II: 6, s.v. aghin) may in fact represent accan. 250 And also with loss of the nasal as -(a)ig, cf. Dillon (1962: 579). 296 Aspects of the historical phonology of Manx 4.4.6.2 Other developments of /Nʲ/ sliennoo ‘surname, to surname’, G. sloinneadh, slonnadh; sloinneamhain The spelling of this in Phillips slonniu, slonniú suggests the expected development /sloNʲu/. The Classical Manx spelling sliennoo however, and Late Manx transcriptions ʃlʹe̜ nu TC, ʃlʹĩnu TT, sliennooit ‘surnamed’ ʃlʹe̜ nuə̯tʹ TC (HLSM II: 415), point to an irregular development to a form /slʲen̠ u/, perhaps influenced by /en̠ әm/ ennym ‘name’ (G. ainm), and lhiennoo ‘children’ (G. leinbh). The non-palatalized [n] may go back to the by-form slonnadh (eDIL s.v. slondud, sloinded), however. shin ‘we’, G. sinn In contrast to the emphatic form shinyn /ʃiŋʲәn/, there is no evidence of the simple pronoun shin being realized as */ʃiŋʲ/. This is likely because pronouns are usually lightly stressed. Indeed, transcriptions in HLSM (II: 403) show a form which is consistently non-palatalized, suggesting that this item has fallen in with historical lenis /nʲ/ and follows those items such as hene, anney with the development /nʲ/ > /n̠ / discussed above (§4.4.3). 4.4.7 Unstressed final /an/, /ən/, /ənʲ/ Final unstressed syllables of the shape /Vn(ʲ)/ are generally written -yn, -in and -an in the Classical Manx orthography. The diachronic derivation and phonological signification of these orthographical terminations are somewhat complex and merit detailed discussion. Table 62. Summary of developments of post-tonic unstressed /Vn⁽ʲ⁾/ orthographical pronunciation principal origins form -yn /әn/, occasionally -(e)an, -(e)ann, -(a)in /әnʲ/ -in /әnʲ/, /әn/ -ín, -ean, -eann, -(a)in, -(a)inn -an /an/ (marginally -(e)án (/-éan), -(e)ann /anʲ/?) 297 Aspects of the historical phonology of Manx There are three particular complexities which require consideration:  -yn seems predominantly to represent /әn/, but this includes a large number of items, mainly verbal nouns, where other Gaelic dialects have Ir. -(a)in(t), ScG. -(a)inn (§4.4.7.1).  -an is robustly attested as representing /an/, but this includes not only reflexes of -án but also a number of items where /әn/ (< G. -(e)ann) would be expected (§4.4.7.3).  -in represents mainly /әnʲ/ [inʲ], but also a fair number of items with /әn/, especially after slender consonants (§4.4.7.2); a further illustration of the ambiguity of representation of palatality in the Manx orthography (§1.6.4.6). In more detail the three orthographical terminations represent the following synchronic and diachronic forms: -yn /әn/  the regular plural suffix (ScG. -(e)an), and complex variants thereof, -(a)ghyn, -(i)nyn, -eeyn, -(t)eenyn, e.g. muckyn ‘pigs’ (G. muc), meeaghyn ‘months’ (G. mí), joughinyn ‘drinks’ (G. deoch), glionteenyn ‘valleys’ (G. gleann).  verbal noun suffix (G. -(a)in) and complex -aghtyn /axtәn/, (G. -(e)achtain), e.g. clashtyn ‘hear’ (G. claistin), ennaghtyn ‘feel’ (ScG. aithneachdainn).  various items in G. -(e)an, -(e)ann, e.g. moidyn ‘maiden, virgin’ (G. maighdean); fyrryn ‘male’ (G. fireann).  certain items with expected G. -(a)in(n), and where it may not be clear whether Manx has /әn/ or /әnʲ/, e.g. cossyn ‘win, earn’, stem of cosney (G. cosain, cosnadh), geddyn, feddyn ‘get, find’ (ScG. faotainn) (some realizations with final [nʲ] in HLSM), screeuyn ‘writing, letter’ (G. usu. scríbhinn, but original nominative scríbheann, EIr. scríbend).  a handful of items with expected G. -án, e.g. er-shaghryn ‘astray’ (G. ar seachrán), nieeaghyn ‘washing’ (G. nigheachán). -in /әnʲ/, /әn/  /әnʲ/, unstressed reflex of diminutive suffix -ín, e.g. drillin ‘spark’ (G. drithlín), caillin ‘girl’ (G. cailín), kibbin ‘peg, stake’ (G. cipín), gurrin ‘pimple’ (G. 298 Aspects of the historical phonology of Manx goirín).  /әnʲ/, a small number of verbal nouns in G. -(a)in which appear to retain final /nʲ/, e.g. fakin ‘seeing’ (G. faicsin, ScG. faicinn), toilchin ‘deserve’ (G. toilleamhain, ScG. toilltinn).  /әnʲ/, other items in original -(a)in, e.g. shiaghtin ‘week’ (G. seachtmhain), Nalbin ‘Scotland’ (G. Albain).  /әnʲ/, items in original -(a)inn, -(a)ing, e.g. cuishlin ‘vein’ (G. cuislinn), Mannin ‘Isle of Man’ (G. Manainn) (see also §4.4.6.1), yllin, uhllin ‘stackyard’ (G. iothlainn), skillin ‘shilling’ (G. scilling, scillinn).  /Cʲәn/, /Cʲәnʲ/, orthographic indicating preceding slender consonant; it is not always clear whether or not n is palatalized, e.g. claiggin ‘skull’ (G. claigeann), egin ‘compulsion’ (G. éigean, -in, ScG. éiginn), mwyllin ‘mill’ (G. muileann).  /Cәn/, no obvious rationale for orthographic ; e.g. cheddin ‘same’ (G. céadna > *céadan), eddin ‘face’ (G. éadan). -an  /an/, diminutives and other nouns in original -án (inc. < EIr. -én), e.g. beggan ‘little’ (G. beagán), quaillan ‘pup, whelp, cub’ (G. cuileán), arran ‘bread’ (G. arán).  /an/, unstressed reflex of other original terminations of the form /Vːnʲ/, with depalatalization, e.g. imman ‘driving’ (G. iomáin), accan ‘complain’ (G. acaoine), ingan ‘anvil’ (G. inneóin), follan ‘wholesome’ (G. folláin).  /an/, various items in original -(e)an, -(e)ann, assimilated to the /an/ class rather than showing expected */әn/, e.g. crackan ‘skin’ (G. craiceann), ollan ‘wool’ (G. olann), astan ‘eel’ (G. eascann), doghan ‘disease’ (G. dochann) (Table 64). 4.4.7.1 verbal nouns in -yn, -in The verbal noun ending G. -(a)in (Mod. Ir. often -(a)int; ScG. -(a)inn), originally the dative form of verbal nouns in EIr. -(i)u, e.g. aicsiu ‘seeing’ > (f)aicsin (Stüber 1997: 299 Aspects of the historical phonology of Manx 231–2), usually appears in Classical Manx orthography as -yn, and judging from the Late Manx phonetic data is overwhelmingly realized with non-palatalized [n]. The main exception is fakin ‘see’ (G. faicsin, ScG. faicinn), which is consistently spelled as such in the printed texts (also fackin), even where other verbal nouns regularly have -yn, and for which there is ample evidence of [nʲ] (HLSM [n] 5, [nʲ] 3, also Rhŷs [Broderick 2019 s.v. fakin]). Another item where the -in spelling survives even in the most orthographically standardized texts is toilchin ‘deserve’ (ScG. toilltinn), but this is not attested in HLSM. Some earlier and less-standardized eighteenth-century texts have more widespread use of -in, but it is unclear whether this represents an earlier /әnʲ/ or simply less discriminate use of <y> and .251 Out of 19 verbal nouns in -yn, -in in HLSM only two are solidly attested with [nʲ], namely fakin and feddyn, geddyn ‘get, find’ (Sc faotainn) ([n] 5, [nʲ] 8).252 This may have to do with the fact that these are frequently-occurring irregular verbs; in the case of fakin at least it may also be connected with the contraction to a monosyllabic form [fainʲ] (Rhŷs: 120, 122). It seems that word-final changes in palatality are more common in unstressed syllables that in stressed monosyllables. Depalatalization of this ending is also found in certain Scottish dialects (SGDS II: 416– 7, cluinntinn; Borgstrøm 1940: 68; Oftedal 1956: 252). Whether this is a secondary development from /Nʲ/, or a reflection of the split development of G. lenis /nʲ/ to both /Nʲ/ and /n/ in Scottish dialects (Ó Maolalaigh 2001: 16), is unclear. In Manx the development was perhaps supported by the analogy of verbal nouns with original /n/ (including those with G. -achán > Manx /axәn/, see below), and those possessing related forms with historical /n/, e.g. lhiastyn ‘owe’, noun lhiastynys ‘debt’ (G. dleastanas). Depalatalization is also found in a couple of other items which are not verbal nouns, including Boaldyn ‘May-day’ (G. Beal(l)taine, ScG. Bealltainn; HLSM [n] 3) and 251 Cf. plurals foilchin ‘faults’ (later foiljyn) and noijin ‘enemies’ (noidjyn), where the ending was certainly /әn/, in CS (Lewin and Wheeler 2017). 252 There is also one attestation of [nʲ] in bentyn ‘touch, belong’ (ScG. beantainn), alongside three with [n]. 300 Aspects of the historical phonology of Manx perhaps screeuyn ‘writing, letter’ (G. scríbhinn), which may have been interpreted as plural ‘writings’ (although it has itself a plural screeunyn), and/or associated with the verbal nouns in -yn, unless it represents original nominative scríbheann, EIr. scríbend (HLSM [n] 3, but pl. skru:nʹən NM). 4.4.7.2 -in Where -in represents the unstressed reflex of G. diminutive -ín, there appears to be a tendency towards depalatalization in Late Manx, although the evidence is sparse (Table 63). There are enough cases of [nʲ] from different speakers and items to suggest this is not a general merger with /әn/, however. Table 63. Unstressed reflexes of G. -ín etymology English HLSM caillin CM pronunciation /kalʲәnʲ/ cailín girl inʹ HK gibbin /ɡibәnʲ/ goibín sand-eel iN JTK gurrin (K.), gooddin (Cr.) kibbin /ɡurʲәnʲ/ goirín pimple ənʹ TT, ən JW253 /kʲibәnʲ/ cipín peg, stake ən HK, J:JK perkin /perʲkʲәnʲ/ ScG. poircean? porpoise ən TK, JW skurrin /skurәnʲ/ ScG. sgurran? rump, back inʹ TC Similarly original -(a)in in items such as shiaghtin ‘week’ (G. seachtmhain) have several depalatalized realizations (7 in HLSM II: 400) but are also found with palatalization (tʹʃaːxtənʹ HK, saxtinʹ EKh). The same is true of those with original -(a)inn (§4.4.6.1), such as aalin ‘beautiful’ (HLSM II: 1 [n] 3, HK; [nʲ] 3, JW, NM, TT). In some cases where a final broad -n would be historically expected, there is robust attestation of [nʲ], which can be readily explained as generalization of a feminine oblique case form, as in dorrin ‘storm’ (G. doineann) (inʹ, ənʹ TC, HK, JW, HLSM II: 253 The latter is under gurran in HLSM (II: 215), but given there are no attested forms with a clear vowel [a], and no orthographic forms with <an>, it is likely that this is simply a non-palatalized realization of -in (< G. -ín). 301 Aspects of the historical phonology of Manx 128),254 and probably egin ‘force’ (G. éigean ‘ā, f.; later also o, m.’ eDIL s.v. éicen; ScG. éiginn f.) which is feminine according to Cregeen (ənʹ HK, ən JW, HLSM II: 149, s.v. er-egin).255 In other cases -in appears to represent palatalization of the preceding consonant, but not the -n, as in mwyllin ‘mill’ (m., G. muileann) (ən JK, W:N, J:EK, JTK, but one instance of inʹ TC, HLSM II: ), cadjin ‘common’ (G. coitcheann) (ən TC, TT), claig(g)in ‘skull’ (G. cloigeann, ScG. claigeann) (klɑgən TC, HLSM II: 82).256 Certain other items have -in following a broad consonant, and apparently with broad [n], for no obvious reason; including cheddin ‘same’ (G. céadna > *céadan) and myrgeddin ‘likewise’ (G. mar an gcéadna); eddin ‘face’ (G. éadan, ScG. aodann);257 sheeabin ‘soap’ (ScG. siabann). When further endings are added, in can be especially ambiguous, as in Manninagh ‘Manx(man)’: here the orthographic -in presumably represents the final /nʲ/ (historical /Nʲ/) in Mannin ‘Isle of Man’ (G. Manainn) (§4.4.6.1), but is carried over into the derivative, which however has only [n] (HLSM II: 289) and probably represents G. Manannach.258 4.4.7.3 -an Transcriptions in HLSM show that the clear vowel /a/ is generally well-preserved in -an (< G. -án), although reduced realizations with [ә] occur, and the CM orthography has a consistent one-to-one correspondence between /an/ and <an>, as 254 According to Cregeen this noun is masculine (no evidence either way has come to light in texts), but it is feminine in other Gaelic varieties (eDIL s.v. doinenn, Dinneen, Ó Dónaill s.v. doineann, Dwelly s.v. doineann, doireann). 255 In shegin dou ‘I must’, also frequently spelled sheign, the form is frequently reduced to ʃein, sain etc., with the n frequently depalatalized, probably as a result of consistently preceding /d/, although one palatalized form is attested in HLSM (II: 395), alongside 7 with [n]. 256 But possibly with feminine declension (‘o, n. and [m.] Later also ā, [f.]’ eDIL s.v. cloicenn). Cregeen does not give a gender for this item; there is one case of lenition in na’n chlaigin ‘than the skull’ (2 Kings 9. 35), but na’n here is possibly treated as regular preposition + article, causing lenition. 257 This noun is given as feminine by Cregeen, a designation which is supported by some agreement evidence (Wheeler 2017: 24), so an oblique form *éadain could be posited (cf. airh ‘gold’, f. =*óir, §4.2.1.3); but the data from HLSM (II: 141) and Rhŷs (Broderick 2019 s.v. eddin) has only nonpalatalized [n]. Phillips has mostly -yn (14 instances), one instance of -in, but the spellings of the plural ydyniyn and edyniyn may be equivalent the ScG. plural aodainnean. 258 As in Éireannach ‘Irish’, Albanach ‘Scottish’ etc.; ScG. Manainneach (found alongside Manannach) is probably a modern reformation. 302 Aspects of the historical phonology of Manx shown in the examples in Table 64. There are also items where G. -an, -ann, -aing (> Manx */әn/) would be expected, but where -an /an/ is clearly attested (Table 65). Table 64. Some examples of Manx -an < G. -án Phillips arran an (5) CM pronunciation /aran/ etymology English HLSM arán bread an HK, NM, aᵈn EL ɑ ̣n WQ, HK, EKh ɑ ̣n JW, an JK beggan an (3) /beɡan/ beagán little ellan (pl.) ellanyn (2) /elʲan/, /alʲan/ oileán, EIr. ailén island partan /partan/ partán crab roddan /rodan/ ScG. rodan rat sporran /sporan/ sparán, sporán purse thurran /turan/ ? torrán, túr + án haystack ɑ ̣n HK, JW, an EL, ɑn TT CM etymology English HLSM /astan/ eascann eel /krak⁽ʲ⁾an/ (§2.1.6) /doxan/ croiceann skin dochann disease ən NM, an TK, ən WQ ən TC, JK, JTK ɑn TC, ɑ ̣n WQ an JW, ɑ ̣n TT, ɑn TC /foxan/ fochann /iŋan/ EIr. ingen (ar méraib)259 young corn in blade issue, young, offspring (of an JW, JTK, NM, JW, ɑ ̣n JK, TC, ɑn JK an EC, an NM, pl. paːrtərən, pɑ ̣ːtərən NM ɑ̃ ṇ JW, ɑ̆n W:S, ən J:EK, ɑ ̣n JW, an NM an JW, ɑ ̣n HK Table 65. Unhistorical -an Phillips astan crackan yn (2) doghan foghan ingan yn 259 That is to say, a legal term ingen ar méraib ‘nail on fingers’ metaphorically denoting (possibly with reference to a manner of calculating kinship by use of the hands [McLeod 2000]) ‘descendants beyond the son of the great great grandson, i.e. collateral kin beyond the third cousins’ (eDIL s.v. 2 ingen; Patterson 1990: 138–40; McLeod 2000: 6–8), sometimes found without ar méraib (eDIL). This would agree with Kelly’s etymology ‘Ir. ionga’. Judging from the standard dictionaries (Dinneen, Ó Dónaill, Dwelly), which lack this sense of ionga, it has not survived in the other modern Gaelic languages. In Manx it appears to have undergone a semantic shift from referring to distant progeny of humans to referring primarily to the young of animals; note e.g. the differing translations of ‘fruit’ in Deuteronomy 30:9: As nee’n Chiarn oo y vishaghey ayns ooilley obbraghyn dty laue, ayns sluight dty chorp, ayns ingan dty vaase, as ayns dty vess hallooin son dty vie ‘And the Lord thy God will make thee plenteous in every work of thine hand, in the fruit of thy body, and in the fruit of thy cattle, and in the fruit of thy land, for good’, where ingan refers to animal young but sluight (G. sliocht) to human 303 Aspects of the historical phonology of Manx animals) lieckan (pl.) liettanyn loaghtyn (Bible, Cr.), loghtan (K.) oalan (K.), oalyn, olan (PNIM III: 123) ollaghan ollan sollan ayn, án ən, ɑ ̣n TC, ɑn HK an JK /lʲekʲan/ leiceann cheek /loxtәn/, /loxtan/ ScG. lachdann brown, tawny /oːlan/, /oːlәn/ G. abhlann, cf. ScG. abhlan wafer /olәxan/ treadle of spade /olan/ ealchaing, ScG. ealachainn olann wool an JK /solan/ salann salt aN NM, an NM, JK, JTK, ən NM, TC ɑn JW, W:N ɑ̜n HK Cases of unhistorical clear vowels in final syllables are attested in Scottish dialects, e.g. oisean /ɔʃan/ (Ó Maolalaigh 2001: 9) for historical oisinn ‘corner’, and with other final consonants, e.g. galar ‘disease’, iodhal ‘idol’ etc. Ó Maolalaigh explains cases such as oisean as resulting from ‘truailliú agus analach […] leis an deireadh díspeagtha -ean/-ein’.260 Similar analogical substitutions are also attested in Irish (Ó Curnáin 2007: 124–7). This likely explains the Manx cases given above. It is perhaps no accident that several of the items given above refer to small animals (astan, ingan; cf. braddan ‘salmon’, scaddan ‘herring’, partan ‘crab’ etc.), parts of offspring. For the ending -an, distinguishing this item from yngyn ‘nail’, see the discussion in the present section. One might otherwise suspect that ingan is to be equated with oikan ‘infant, baby’ (Cr. also inkan, oinkan) (G. *naoidheacán, Ph. pl. nikanyn, ikanyn), where variants with a medial nasal consonant (‘on the south side of the island’, Cregeen s.v. oikan) can be explained by rhinoglottophilia (§5.6) and/or the influence of the original initial n- (Lewin 2019a: 87). Ingan would then represent voicing of the stop in the medial cluster /ŋk/ > /ŋɡ/. Telling against this interpretation is the fact that ingan ‘young of animals’ (often in an uncount sense) and oikan ‘human infant’ are clearly distinguished in the texts and the dictionaries; moreover, the rhinoglottophilia account requires voiceless /k/ (although secondary voicing is not necessarily precluded). However, the form chied oingyn maaish ‘firstling that cometh of a beast’ (i.e. ‘first young of cattle’) in the manuscript of Exodus 13ː12 (printed chied ingan maaish) may attest to confusion or fluctuation between these two items and also a reflex of ionga with historical /-әn/ rather than /-an/ (actually the manuscript appears to have oingan with y written over a, and then the whole word rewritten oingyn in the original hand, without subsequent emendation within the manuscript). 260 ‘contamination and analogy with the diminutive ending -ean/-ein’. 304 Aspects of the historical phonology of Manx larger wholes (crackan, lieckan, foghan, ollaghan), or substances often divided into small parts or portions (ollan, sollan, foghan). With the exception of ollaghan and perhaps doghan, all of them also refer to concrete and natural entities (or in the case of loghtan, attributes). We may contrast three items showing the opposite development, historical G. -án > Manx -yn, all of which are more abstract concepts and may possibly have been influenced by the class of verbal nouns in -yn (possibly also seaghyn ‘sorrow, affliction’, §3.9.7).261 Table 66. Unhistorical -yn < G. -án gweeaghyn nieeaghyn, niaghyn ershagh(y)ryn; shaghrynys; shaghrynagh Phillips CM etymology English HLSM yn (2) /ɡwiːaxәn/ guidheachán curse ən TC, JW /nʲiːaxәn/ nigheachán washing /erʲ ʃax(ә)rәn/ seachrán astray; confusion; stray ən TC, W:N, W:S ən TC, W:N, JW, EKh, TT an (5), yn (2), án Some of the data from Phillips (<yn> in foghan, crackan; <an, án> in shaghryn) indicate that these developments were not complete in Early Manx. The confusion between these terminations was possibly aided by the breakdown of the /N ~ n/ contrast (in -ann, -án), which may post-date Phillips (§4.4.4). There are also a few items in final /an/ deriving from terminations with historical */nʲ/. For the most part, this is depalatalized in the Manx realizations. Table 67. -an < G. /Vːnʲ(ə)/ Phillips accan an (5), æ̆ in CM pronunciation /akan/ follan ayn /folan/ imman an (5) /iman/ ingan 261 /iŋʲan/ etymology English HLSM acaoine, ScG. acain complain, complaint folláin, ScG. fallain iomáin, ScG. iomain inneóin, ScG. innean, innein wholesome ən TC, NM, JTK, W:S, aːn JW ɑ ̣n TC ɑn TC, ɑ ̣n JW drive ɑn, ən TC anvil ən TC, aNʹ JK, ɑn W:N Nieeaghyn may indeed function as a verbal noun (HLSM II: 324), alongside niee (G. nighe). 305 Aspects of the historical phonology of Manx On the whole these items seem to have been assimilated in Manx to the /an/ class. Since there are, in addition, no known plurals or genitives of these at any period with /anʲ/, one might posit a constraint against post-tonic unstressed */anʲ/ in Manx (cf. the apparent constraint against post-tonic unstressed */alʲ/, §4.3.2.2), although against this we have the isolated example of ingan ‘anvil’ from JK with Nʹ,262 and Phillips’ spelling with <æ̆ in> might be taken as representing final /nʲ/. Suffixed plurals such as bradanan ‘salmon’ (rather than slenderized bradain) are typical of eastern Scottish dialects, according to Ó Maolalaigh (2003b: 158), who notes that there is some evidence of non-inflection of -án in Early Irish (GOI: 178; Carney 1964: 155), as well as occurrence of nominative plurals in -ána, representing spread of accusative plural forms (see eDIL s.v. bratán). Ó Maolalaigh posits that plurals in -ain did in fact develop in Eastern Scottish Gaelic, but that the contrast -an and -ain was neutralized by the extensive merger of original broad and slender lenis /n, nʲ/ in these dialects. As discussed above (§4.4.3), depalatalization of original lenis /nʲ/ does not seem to be so widespread in Manx, but is found in certain environments, including after /a/ in anney ‘commandment’ (G. aithne), so it is possible that there was depalatalization in final unstressed /anʲ/. We should also note the general decline of final slenderization in nominal inflection in Manx, especially in the genitive singular, with only a handful of examples remaining, such as baaish, genitive of baase ‘death’ (G. bás).263 4.4.7.4 Stressed reflexes of G. -éan, -án As observed by Ó Maolalaigh (2001: 31), Manx overall patterns with Irish rather than Scottish Gaelic in the development of the diminutive suffix -ín (Manx -een, -in), which Ó Maolalaigh suggests may derive from an earlier suffix *-éin. In Scottish Gaelic, on 262 John Kneen, the Gaaue (‘blacksmith’) (Broderick 2018a: 131–3), who might be expected to have been familiar with this word. On the other hand, Kneen gives the impression to me of being a less careful speaker than some of the other HLSM informants, and the level of confusion and hypercorrection in the terminal speakers means that individual instances of (non-)palatalization cannot be taken as conclusive (Broderick 1999: 85–90). 263 There are, however, more cases of retention of slenderization in plurals, including /n/ > /nʲ/ as in eayn ‘lamb’, pl. eayin (G. uan, uain); but cf. raunyn ‘seals’ (Lamentation 4. 3; Cregeen; HLSM II: 361) for G. rón, pl. róin. 306 Aspects of the historical phonology of Manx the other hand, there is no raising of the vowel in this suffix and it falls in with reflexes of -án, é(a)n. In one respect, however, Manx may preserve an archaism not otherwise preserved in Gaelic dialects, namely the apparent survival of a termination /ˈeːn/ in certain items, which in Irish falls in with -án (e.g. EIr. ailén ‘island’ > Mod. Ir. oileán). Table 68. Manx -ean /eːn/ < G. -éan (Mod.Ir. -eán) Phillips binjean (Cr., K.) cheinjean /tʃenˈdʒeːn/ jeeigean (Cr.) jiulean soilshean CM pronunciation /binʲˈdʒeːn/ /dʒi(ː)ɡʲeːn/ /dʒuˈlʲeːn/ áyn, éyn, eyn, en /solʲˈʃeːn/ etymology English EIr. bintén, < G. binid teinteán, EIr. *teintén G. díog, díg, EIr. *dígén deibhleán, EIr. deidblén soillseán, EIr. *soillsén curds and whey bonfire HLSM rill small tenant farmer264 shine265 ẹːᵈn JW, WQ eːᵈn SK, eːᵊn NM, eːn J:EK, ẹːn TC There is a similar orthographical contrast in the verbal noun suffix -ail (mostly after broad consonants), -eil (mostly following slender consonants) (G. -áil) and the diminutive -age, -aig, -aag (after broad consonants), -eig (after slender consonants) (G. -óg > Manx *-ág). Here forms *-éil,266 *-éag may conceivably have developed by analogy with -án, -én. 264 See Thomson (1988: 142), Wheeler (2018 s.v. jiulean). The use of this as a verbal noun ‘to shine’ seems to be peculiar to Manx, and appears to be a semantic development of the diminutive soillseán ‘a torch, a taper, a ray of light’ (Dinneen) found in other dialects. 266 Note however Ó Cuív (1980: 127): ‘It is noteworthy that in some of the verbs the vn. ending follows a palatalised consonant. Hence the process of analogy has given a morpheme -(e)áil. [fn.] This may be contrasted with the termination -eáil (gen. of -eál) and related elements, reflexes of earlier -éil, -él etc. (< *-ethl- etc.). However, there is no doubt that a morpheme -él (with oblique form -éil) was used in the formation of verbal (or abstract) nouns for some centuries from the late Mid[dle] Ir[ish] period on.’ See further ibid.: 134. 265 307 Aspects of the historical phonology of Manx 4.4.8 Summary of developments in coronal and velar nasals Figure 13 summarizes the developments in the system of coronal and velar nasals underway in Early and Classical Manx (1–2) and the point reached in Late Manx (2– 3), based on the discussion above. Figure 13. Summary of developments of coronal and velar nasals (1) n N nʲ Nʲ ŋʲ ŋ (2) n̪ /ṉ nʲ (Nʲ) ŋʲ ŋ (3) ṉ nʲ ŋʲ ŋ As discussed above, some aspects of these developments, especially the distribution of dental [n̪ ] and alveolar [ṉ], remain somewhat unclear, as we are largely reliant on the descriptions of Rhŷs, which only give a limited snapshot of the range of dialectal, idiolectal, lexical and diachronic variation and change which may have existed. 4.5 Preocclusion 4.5.1 Introduction Preocclusion of stressed final nasals and laterals is one of the best-known features of Manx phonology and has attracted a certain amount of scholarly attention (§§4.5.2, 4.5.3), although there has been no consensus on its characteristics or origins. In writing the phenomenon is only attested in certain folksong manuscripts in idiosyncratic orthographies from the nineteenth century, although there is reason to believe it developed significantly earlier than this. In the speech of the terminal speakers as represented for example in the Irish Folklore Commission recordings (Manx National Heritage 2003), preocclusion is very frequent with some speakers (such as HB, NM), especially with final [n], [nʲ], and rare and/or very weak in other speakers (such as JK, JTK). It is usually quite weak and often 308 Aspects of the historical phonology of Manx difficult to hear, and frequently absent entirely, even in speakers who often have it. It seems to be particularly prone to absence under weak phrasal or sentence stress and in rapid speech. Preocclusion appears to vary freely with lengthening of the sonorant (often with a shortened vowel), lengthening of the vowel (with the sonorant being short) and occasionally “postocclusion” (with [l]), all of which can be seen as strategies to enhance syllable weight (§4.5.5). Some examples are given in the spectrograms below (Figures 14a–g), which were generated in Praat (Boersma and Weenink 2015). The numbers refer to the disc and track in Manx National Heritage (2003). Figure 14a. [ʃeᵈn] shen ‘that’ (G. sin) with preocclusion, HB (1:14) ʃ ᵈ e n Figure 14b. [ɡʲeˑᵇmˑ] geam ‘calling’ (G. éigheamh) with preocclusion, HB (1:14) ɡʲ eː ᵇ m 309 Aspects of the historical phonology of Manx Figure 14c. [xɨᵈn] keayn ‘sea’ (G. cuan) with preocclusion, NM (2:9) x ɨː ᵈ n Figure 14d. [veᵈn] ben ‘wife’ (G. bean) with preocclusion, NM (2:10) v ᵈ e n Figure 14e. [ʃuˑlːᵈ] shooyl ‘walk’ (G. siubhal) with “postocclusion”, NM (2:10) ʃ 310 uˑ lː ᵈ Aspects of the historical phonology of Manx Figure 14f. [eʳsuˑ(ᵈ)lː] ersooyl ‘away’ (G. ar siubhal) with shortening of vowel and lengthening of sonorant, NM (2:19). The presence of preocclusion is doubtful. eʳ s uˑ lː Figure 14g. [tʃiŋʲː] ching ‘sick’ (G. tinn), NM with lengthened sonorant but no audible preocclusion, NM (2:9) t ʃ i ŋʲ 4.5.1.1 Cross-linguistic typology Cross-linguistically preocclusion or pre-stopping does not seem to be an especially common development. It is found within northern Europe in North Germanic (Icelandic, Faroese, certain Norwegian dialects) (Sandøy 2005, Røsstad 2011), Sámi (Sammallahti 1998) and Cornish (Chaudhri 2007). The distribution, realization and phonological function of preocclusion in these languages are quite different from one another, but they all seem to develop from historical long or geminate sonorants and/or sonorant clusters. This and other features have been argued to provide evidence for a northern European sprachbund (e.g. Wagner 1964). Iosad (2016b) argues on 311 Aspects of the historical phonology of Manx chronological and historical grounds that direct contact influence of these languages on each other in respect of preocclusion and other features is implausible, but that more fundamental structural similarities between these languages — which may themselves reflect older language contact (cf. Salmons 1992) — ‘conspire to encourage the repeated genesis of shared features’ (Iosad 2016b: 15). Outside northern Europe, pre-stopping (as it is conventionally known in this context) is particularly prevalent in Australian Aboriginal languages (Ladefoged and Maddieson 1996: 128–9; Loakes et al. 2008) and Austronesian languages (Jardine et al. 2015). The origin of Australian pre-stopping seems to be different to the northern European phenomena, and has been argued to be a strategy to preserve place of articulation distinctions in languages which typically have an unusually large number of places of articulation and few manners of articulation (Butcher and Loakes 2008; Loakes et al. 2008: 90). 4.5.2 Descriptions The first detailed description of Manx preocclusion is that of Rhŷs (143–4): I must mention a phenomenon of considerable importance in the present pronunciation of Manx. It consists in prefixing to a final nasal the corresponding voiced mute. Thus […] trome ‘heavy’ (Med. Ir. tromm, Mod. Ir. and ScG. trom […]) is pronounced in a way which sometimes strikes one as being τrŏ́ um [t̪ roum] and sometimes τrọ̆bm [t̪ roᵇm] or τrŭbm [t̪ ruᵇm] with a sort of precarious b; and similarly with other words such as […] kione ‘head’ (Goi. ceann) which becomes ki̯ õuν [kʲõun̪ ] or ki̯ ō̃δn [kʲõːd̪ n], while the plural […] king, is pronounced sometimes kĭƽŋ [kiᶢʲŋʲ]; bleïn ‘a year,’ becomes blĭdñ [bliᵈnʲ] and […] lhong ‘a ship,’ becomes λŏgƞ [l̪ oᶢŋ] or λŭgƞ [l̪ uᶢŋ]. The same thing happened now and then with rν [rn̪ ] as in […] oarn ‘barley,’ pronounced orδν [ord̪ n̪ ] […] (Goi. eórna); and with rn, pronounced r̥ dñ [rᵈnʲ], as in cuirn or keirn ‘the rowan or mountain ash’ ([…] Mod. Ir. caorthainn […]). This modification began before the orthography of Phillips’ translation had been fixed upon, as one would otherwise have expected tromm, for example, or trom, rather than troum, tróym, or trúm. In all the cases mentioned the vowel was short and the nasal consonant as in tromm was long, so to say, so that metrically speaking u̯ m or bm is an equivalent for mm. So it is needless to say that the neatest cases of this phenomenon happen to be all accented final syllables, and those which have been here enumerated ended, etymologically speaking, in a mixed equivalent for mm, νν, nn, ŋŋ, or ƞƞ. But (2) the same thing has happened, probably later, 312 Aspects of the historical phonology of Manx where the nasal consonant was short but preceded by a long vowel, and here the reinforcement of the consonantal element took place, metrically speaking, at the expense of the vowel: at any rate this may be supposed to be the tendency. Thus though […] thallooin ‘earth’s’ ([…] Med. Ir. talam, genitive talman) retains the length of the vowel of its final syllable after that syllable is modified, so that the word sounds τaλū̃ dñ [t̪ aˈl̪ ũːᵈnʲ] with the stress on the last syllable, and […] bane ‘white’ (Goi bán) is also pronounced with its a not perceptibly shortened in the South, but in the northern half of the Island the pronunciation in by̆ δν [bәd̪ n̪ ] with the vowel as short as may be. […] it should be remarked that the less distinctly one hears the parasitic consonant the less is the quantity of the vowel tampered with. (Rhŷs: 142–3) Rhŷs’s discussion of the topic is notable for his suggestion as to the origins of Manx preocclusion (§§4.5.4.1, 4.5.5) and comments on its synchronic behaviour as an active prosodic or metrical phenomenon (§§4.5.5.2, 4.5.5.3); for the evidence provided that the phenomenon may be lexically conditioned;268 and for evidence of variation between idiolects and dialects. He notes preocclusion in most of the environments it occurs, including before labial, coronal and velar nasals and in rhotic-nasal clusters, and shows an intuitive understanding of the relationship between preocclusion and vowel and sonorant length. However, it is notable that he does not mention preocclusion with laterals, although there is evidence of this in his notebooks: He pronounced ooyl ‘apple’ mostly ūδλ [uːd̪ l̪ ] sometimes ū́ λδ [uːl̪ d̪ ], but in that case the δ was fainter: the pronunciation ūδλ I have heard of before as the habitual pron[unciation] of an old man in the neighbourhood of Ramsey. (John Stephen, Rhŷs notebook 7: 196) Rhŷs’s notes also contain some comments on idiolectal variation in preocclusion: this man Brew had a constant tendency to pronounce every final n as dn (James Samuel Brew, Rhŷs notebook 6: 73) 268 For example Rhŷs (143) claims that preocclusion occurs more often in Jelhein ‘Monday’ (G. Dé Luain) and Jardain ‘Thursday’ (G. Déardaoin) than in Jecrean ‘Wednesday’ (G. Dé Céadaoin) (Rhŷs: 143). However, preocclusion is well-attested in Jecrean in Rhŷs’s notes: in notebook 6 (p. 152) in a comparative table of items three out of four speakers have dñ [ᵈnʲ]. 313 Aspects of the historical phonology of Manx Strachan (1897: 55) notes preocclusion only before n, ‘sometimes […] quite distinct, sometimes barely audible, and records that he ‘seemed sometimes to hear the same sound when English was spoken, e.g. agädn for again’. Marstrander (58) describes preocclusion before both nasals and laterals, and also notices it in Manx English in stūdl ‘stool’, spūdn ‘spoon’, stībm ‘steam’. He notes that preocclusion occurs irrespective of sonorant quantity, although it is unclear whether this is a synchronic or a diachronic observation: Utviklingen synes ikke å ha noe med konsonantens kvantitet å gjøre. Den forklares heller ikke ved en forsinket åpning av ganeseilet, da den jo også foreligger ved l. [The development seems not to have to do with the consonant quantity. Nor is it explained by a delayed opening of the velum, since it also occurs with l.] (Marstrander: 58) Jackson (113–4) notes preocclusion in nasals only; he notes Rhŷs’s and Marstrander’s descriptions but states he did not encounter preocclusion with laterals or rhotic-lateral clusters himself. He notes preocclusion in Manx English with final /n/ only, as in ‘seen = siᵈn and the like’. Before -n or -nn of either quality when final in stressed monosyllables there has very commonly developed in Manx a kind of fugitive unexploded d. What seems to happen is that in producing the n the occlusion begins just before the velum is lowered, so that the sound is denasalized at the beginning. I write ᵈn for this. It is most certainly not a glottal stop, as it has been called. (Jackson: 113) Jackson notes a number of examples where he only heard preoccluded nasals, such as bɛᵈn ben ‘woman’ (G. bean), fiːᵈn ‘wine’ (G. fíon), driːᵇm dreeym ‘back’ (G. druim), ɔːᵈn oarn ‘barley’ (G. eórna). However, he also notes some items for which he heard both preoccluded and non-preoccluded forms, e.g. tʰrɒᵇm, tʰroːᵇm, tʰroːm trome ‘heavy’ (G. trom), eeym ‘butter’ (G. im), and an item with only non-preoccluded forms: ‘in ching “sick”, […] with original -nn, I heard only ŋʹ’ (Jackson: 115). Wagner comments briefly on Manx preocclusion, noting similar developments ‘in Cornish, West Norse, Lapp, as well as in some Siberian languages’ (Wagner 1956: 314 Aspects of the historical phonology of Manx 109) and suggesting this is evidence of ‘a certain North Eurasian ‘Sprachlandschaft’’ (he also notes preocclusion in Dublin English, which he attributes to ‘West Nordic’ influence). According to Wagner, preocclusion is restricted to southern Manx: Im modernen Manx scheinen diese Formen auf die südlichen Dialekte beschränkt zu sein, während das Nord-Manxische Formen aufweist, die mit entsprechenden schott[isch]-gäl[ischen] Formen verwandt sind. Karte 89 meines LASI, welche die Manx-Formen für ir. gann „scarce‟ illustriert, gibt eine Form ɡauә̯n für den nördlichen Dialekt und eine Form ɡoːdn für den südlichen. [In Modern Manx these forms seem to be restricted to the southern dialects, while Northern Manx shows forms which are related to the corresponding Scottish Gaelic forms. Map 89 of my LASID, which illustrates the Manx forms for Irish gann ‘scarce’, gives a form gauә̯n for the northern dialect and a form ɡoːdn for the southern one.] (Wagner 1964: 293) Notwithstanding this claim, plentiful evidence is found of preocclusion in the north in the other accounts discussed here. Indeed, in Wagner’s northern form ɡauə̯n, the otherwise suspicious [ә] may well represent weak preocclusion. However, it was noted above (§4.5.1) that two of Wagner’s three northern informants (Broderick 1999: 71), JK and JJK, mostly have very weak or absent preocclusion, which may explain Wagner’s claim. Unlike Jackson, Wagner (1956: 109) does note preocclusion with laterals, giving the example of ʃuːdl shooyl ‘walking’ (G. siubhal). Broderick (HLSM III: 28–9) introduces preocclusion as follows: In L[ate] S[poken] M[anx] there can occur usually in stressed monosyllables (but also in stressed final syllables of disyllables and stressed medial syllables followed by a short monosyllabic unstressed suffix — whether the stressed syllable be long or short) ending in a nasal or lateral a development known as preocclusion. That is to say, that just prior to the articulation of the nasal or lateral the corresponding (voiced) stop is realized, but with nasal or lateral relase, i.e. [bᴺ] before /m/, [dᴺ] before /n/, [ɡᴺ] before /ŋ/, [dᴸ] before /l/. (HLSM III: 28–9) Broderick’s is the only primary description to note preocclusion in medial positions. He gives two examples of this (HLSM III: 29): 315 Aspects of the historical phonology of Manx brynnagh ‘flattering’ /bre[dʹᴺ]nʹax/ lieenyn ‘nets’ /lʹi[dᴺ]nәn/ (HLSM III: 29) According to HLSM (II: 49, 277) the realization bre̜ dʹnʹɑ ̣x brynnagh ‘flattering, comely’ is from JW, found alongside bre̜ nʹərɑ ̣x brynneragh ‘act of flattering’ (ScG. brionnal, brionnalachd), while lʹiᵈnən lieenyn ‘nets’ (G. líon) is from NM, who has lʹiᵈnʹ, lʹi.ᵈn, lʹiːᵈn in the singular. A further instance is dʹʒɔᵈnɑ ̣x joanagh ‘dusty’ (G. deann) (JW), cf. joan ‘dust’ dʹʒɔᵈn (JW) (HLSM II: 238). Broderick suggests that the preoccluded sonorants are probably to be analysed as allophones in free variation with their non-preoccluded equivalents: It is my view that […] though in a given set of circumstances preocclusion can take place, nevertheless reflexes containing no preocclusion (in most cases) also occur, thus indicating that preocclusion plays no role whatever in the context of meaning and import. That is to say, that the preocclusive forms [bm], [dn], [ɡŋ], [dl] are special realizations of the corresponding phonemes (/m/, /n/ /N/, /l/), and in this regard I would view preocclusion in LSM as having allophonic rather than phonemic status. (HLSM III: 31) Williams (1994b: 714) comes to the same conclusion. However, Broderick also claims there is some evidence of incipient separation, including syllabification, and thus phonologization of the preocclusive stop: it may be noticed that in his phonetic corpus of LSM Marstrander sometimes renders the preocclusive dental as a spirant, viz. /[ð]n/, which suggests that it was becoming separated from its homorganic nasal and the whole unit was developing into [ðən], as in [bẹːðən] ‘boats’ [baatyn, ScG. bàtaichean]. Indeed he sometimes writes as if the unit had already developed a centralized vowel, e.g. (without spirantization of the dental) [ʃedən] ‘that’, usu. [ʃen] [shen, G. sin]. That is to say, that the dental was now being released orally instead of nasally, i.e. as a separate segment. In other words a process of phonemicization was taking (or had taken) place. (HLSM III: 31–2) Broderick (HLSM III: 32–4) uses a comparison with English syllabic nasals in e.g. ‘button’ [bʌtn̩ ] and evidence from verse to argue that a monosyllable containing 316 Aspects of the historical phonology of Manx preocclusion can be considered bisyllabic, although it is not clear why it cannot be considered one long syllable in the terms employed. The vowel in [ʃedn̩ ] is short, and bearing in mind that in Manx a long syllable has the value of two short syllables, the short syllable here is, therefore, made up by the preocclusive element plus the nasal plosion. We can see the same in [dʹʒidn̩ ] ‘eager’ […] which occupies a position of stress and therefore (in metrical terms) can have a long or two short syllables. In this instance the vowel is short, indicating that two short syllables are required to make up the quantity. The short vowel contributes to the first short syllable; the second is therefore made up by the preocclusive element plus the nasal plosive. That is to say, that (in Manx verse terms at any rate) preocclusion renders an additional syllable to the word (here a stressed monosyllable) so affected. […] It is my view that the same applies in ordinary speech, i.e. that preocclusion renders a (stressed) monosyllabe [sic] disyllabic, and a disyllable trisyllabic. (HLSM III: 33–4) This argument does not seem to stand up to scrutiny. Even if some kind of exaggerated articulation were found in verse which could be interpreted, perceptually at least, as suggestive of an additional syllable (for which Broderick does not present evidence, although see the written data, §4.5.3), there is no reason to think this would be relevant to ‘ordinary speech’ where preocclusion seems in fact to have been fairly faint on the whole, and often absent or only barely perceptible. The English syllabic sonorants do not seem pertinent to the discussion, given the optional presence of a vowel [ә] as Broderick himself notes (HLSM III: 32), and the fact that words such as ‘button’ count without doubt as bisyllabic in all circumstances. Broderick (HLSM I: 162–3) also comments on dialect differences in preocclusion: An intrusive d can also appear before final -l, and a g before final -ng. These features seem to be peculiar to the South.269 (HLSM I: 162) 269 Broderick gives an exception loᶢŋ from TC, but explains this by noting ‘[h]is father came from Lonan (ie. on the southside) from whom he likely inherited any southern forms in his speech’ (HLSM I: 162). Another counter-example is kʹiᵈl (W:N) keeill ‘church’ (G. cill) (HLSM II: 245). 317 Aspects of the historical phonology of Manx The intrusive b [before m] is common to both areas, but from the limited evidence available it seems to be more absent in the South. (HLSM I: 163) It has been noticed that there is significant variation and disagreement between the descriptions of Manx preocclusion in previous scholarship. The following claims have been made, and shown here to be inaccurate or incomplete:    some descriptions do not note preocclusion with laterals (Rhŷs, Strachan, Jackson) preocclusion is claimed to be restricted to certain dialects (Wagner) preocclusion is claimed to be syllabic (Broderick) In part at least these discrepancies between descriptions likely reflect the relative lack of salience of non-contrastive preocclusion, which has been noted in a cross-linguistic context: Butcher and Loakes (2008) note that non-contrastive pre-stopped realizations are difficult to perceive auditorily, even by field researchers experienced in working with the languages in question. Our research anecdotally supports this observation. Members of our research team found non-contrastive lateral prestopping in Kaytetye difficult to perceive, but perceiving contrastive nasal prestopping was unproblematic. (Harvey at al. 2015: 246) It is likely that Manx preocclusion has always been non-contrastive insomuch as even when it was restricted to stressed final fortis sonorants (§4.5.5), it would not have been the only, nor necessarily the primary, cue for the contrasts, which also involved differences of place and manner of articulation, and length. 4.5.3 Written evidence A notable feature of Manx preocclusion is the lack of written evidence for it. Even though it was certainly prevalent throughout the island by the early nineteenth century at the latest, going on the evidence of Rhŷs and Strachan, and quite possibly centuries earlier than this, it is never represented in either of the two main orthographies, and is 318 Aspects of the historical phonology of Manx rare in texts with non-standard orthographies. It is, however, indicated in certain nineteenth-century folksong manuscripts in non-standard orthographies. Indication of preocclusion is especially prevalent in a song manuscript Manx National Heritage Library MD 900 MS 08307, edited by Broderick (2015). This manuscript is of uncertain provenance, but was most likely compiled between 1830 and 1840 (Broderick 1984a: 157). Preocclusion is represented in the manuscript as <dn>, <dyn>, <din>, <bm>, <bym>. There are at least270 81 instances of indicated preocclusion in the text of MS 08307. There are also many occurrences of eligible items with no indication of preocclusion (e.g. dhoan, dhon, wooan, woan, aun ‘brown’, G. donn, alongside odn, woadn, woadyn), and there are no cases of representation of preocclusion with the velar nasals or laterals. Table 69. Representation of preocclusion in MNHL MD 900 MS 08307 (ed. Broderick 2015) spelling in MS CM orthography etymology English kiodn, kiodyn koodn, ?choadyn skoadn creedn seidn, seidyn lhedn, laydn frowdn dhowdn vlowdn odn, woadn, woadyn foadn kione coon s’goan, s’coan creen shegin lane frown dowin blieaun dhoan, dhone foyn (K.), foain (Cr.)271 boyn skynn spain ceann cumhang is gann críon is éigean lán domhain bleaghan donn fonn head, end narrow hardly ripe, withered must full, many frown deep milking brown sward, ground no. of occurrences 7 2 1 1 5 2 1 1 1 3 1 bonn scian ScG. spàin heel knife spoon 1 1 1 voadn skidn, skydyn speidn 270 In a few cases the sense of the passage is unclear and so the reading of the word is not entirely certain; these are indicated with a question mark in the table below. A handful of other possible instances were so uncertain that they were omitted entirely. 271 Moore (1896: 70–1) interprets this as foaidyn ‘sods’, but the metre clearly requires a monosyllable with preocclusion, rhyming with boyn ‘heel’, and it is probably to be equated with Kelly’s foyn ‘the grass or ground underfoot, earth’s mantle or covering’, Cregeen’s foain ‘the sward, the green grassy surface of the earth or ground; fo-ain, (under us)’, which is evidently G. fonn ‘base, foundation, soil, ground, land, territory’ (cf. Ó Dónaill, Dinneen, eDIL). Kelly’s spelling is probably more accurate; Cregeen’s is unlikely to represent palatalization here, but rather reflects his predilection for inserting unwarranted (§1.6.5), and in this case he is probably influenced by his proposed etymology, and perhaps the preceding entry foaid ‘sod’ (G. fó(i)d). The stanza does not occur in the version of the song given in Broderick (1980–2: 11–3) from the Clucas Collection. In John Nelson’s reading of Moore’s version he pronounces singular foaid ‘sod’ [foːdʲ] (Trebitsch and Remmer 2003: disc 2, track 19). 319 Aspects of the historical phonology of Manx yoadn Jeleidn, Jeleidyn lhoodyn, lodyn, glodyn keedyn, keadyn, keeadyn, keedyn, keddyn peidyn shidyn greedyn bleadyn, vleadyn theidyn, huidin veadyn feedyn pleadyn ghloadyn, chlaudyn, chloodyn, chloadyn, cloadyn headyn, peedyn lheedyn voadyn veedyn vowdyn chodyn yeeadyn streidyn roadyn leadyn reidyn lhedyn graibm joan Jelhein glione deann Dia Luain gleann dust Monday valley, glen 1 2 2 keayn cuan sea 11 pian sheiltyn, shein green blein thoin mean feeyn plain cloan pian saoiltin pain think green year bottom middle wine plain children 1 1 1 4 2 1 2 1 5 hene, pene lieen moain ?mee(i)n ?bouin chionn eayin stroin raun ?lhean ?rheynn y Lhane greme féin líon móin mín boghainn teann uain sró(i)n rón leathan roinn *lán < Norse lón greim self linen turf, peat tender waist fast, tight lambs nose seal wide divide the Lhen272 bite, morsel; grasp, grip back call jump, leap before me 1 1 1 1 1 1 1 1 1 1 1 1 1 dreebm, gheeabm gaibm leabm, leabym roabym dreeym eam lheim roym druim éigheamh léim romham bliadhain tó(i)n meadhón fíon clann 2 1 2 1 It is noteworthy that almost all instances of indicated preocclusion in this text are in stressed final syllables with a synchronically long273 vowel or diphthong — whether from an original long vowel or diphthong, as lane /lɛːn/ (G. lán), fricative vocalization, as lhean /lʲeːn/ (G. leathan), blieaun /blʲau̯ nʲ/ (G. bleaghan), or original short vowel + fortis sonorant, as kione S /kʲoːn/, N /kʲau̯ n/ (G. ceann). The only possible exceptions to this are skynn /skʲin/ ‘knife’, which seems to have a short vowel in Manx, although it has a diphthong scian in other Gaelic varieties (originally bisyllabic sciän, eDIL; cf. 272 Brook forming the boundary between the parishes of Jurby and Andreas (cf. Rhŷs: 143); for etymology see PNIM (III: 146) and Marstrander (231–4). 273 I.e. underlyingly long; disregarding optional vowel shortening which is a result of preocclusion itself, and assumed to be non-categorical. 320 Aspects of the historical phonology of Manx ScG. dat. sgithinn), and one or two items with original fortis sonorants which can have short vowels in Classical and Late Manx (e.g. chionn S /tʃon/, N /tʃau̯ n/, G. teann). There are no cases of indicated preocclusion with a short vowel + historically lenis sonorant, such as ben ‘woman’ (G. bean), although this item appears frequently in the text. The implications of this will be discussed below (§§4.5.4.3, 4.5.5.2). Preocclusion is also sometimes indicated in songs transcribed by John Clague (1842– 1908), and edited by Gilchrist (1925) and Broderick (2018e). Examples include hedyn for henn ‘old’ (G. sean) (Broderick 2018e: 32), a rowdle for er-rouyl ‘mad, keen’ (etymology uncertain) (ibid: 38), kiddlyn for ?keayn ‘sea’ (G. cuan) (Gilchrist 1925: 213), cheady[n] for keayn (ibid.: 214), sheedyn for sheean ‘noise’ (G. sian) (ibid.: 214). Broderick (2018e: 32) argues that spellings such as hedyn provide evidence for a bisyllabic realization of preocclusion: Originally monosyllabic, preocclusion developed into a disyllable during the course of the 19th-century. This particular example was seemingly heard by Clague as disyllabic. […] Clague evidently did not know much, or any, Manx at all when he first started collecting material (a point also noted by Gilchrist [1925]: ix), and so took down the text as he heard it. (Broderick 2018e: 32) In later texts by which time his knowledge of Manx has increased, representation of preocclusion is not in evidence in Clague’s texts, and the orthography is closer to the standard. Gilchrist (1925: ix–x) comments explicitly on preocclusion in the Clague collection, and in addition to noting its status as a native development of Manx, remarks on its appearance in the singing of ‘old sailors of English nationality’: One point, however, may be noted. Prof. Strachan [1897: 55] speaks of a “d” sound some-times heard before final “n” of a word, as in “chea(d)yn”=sea. I am informed that “b” is sometimes similarly heard before “m,” as in “Tho(b)m=Tom. The same peculiarity used to be found in the singing of old sailors of English nationality. Captain Whall [1913: 43] calls it a “regulation pronunciation which has quite gone out.” He gives a verse of “The Female Smuggler” to illustrate it, which begins: “O come list a-whidle adnd you soodn shadll hear,” and in this instance of intruded sounds it should be noticed that they are not necessitated by any extra syllabic notes in the tune. W. Clark Russell gives similar examples of this sailor mannerism in singing. (Gilchrist 1925: ix–x, original italics) 321 Aspects of the historical phonology of Manx Although spellings such as hedyn, kiodyn etc. in these manuscripts could be taken to suggest a syllabic pronunciation with an intervening schwa, or else a syllabic sonorant as in Broderick’s example of English ‘button’ (above), it is also possible that any perception of syllabicity comes from the perspective of English itself. We know that Clague at least was not a native speaker of Manx and may have only had a limited command of the language when he began collecting folksongs. It is unclear who wrote MS 08307, but this collector may have been from a similar background — at any rate the orthography employed might suggest an unfamiliarity with conventional Manx literacy. It should be noted that if preocclusion were indeed full syllabified, such that the preocclusive stops were analysed as intervocalic stop segments, we might expect indication in MS 08307 of secondary lenition with originally preocclusive stops. With original intervocalic stops we find e.g. ovvyr [ovәrʲ, oβәrʲ] for obbyr ‘work’ (G. obair), bathyn [bɛːðәn] for baatyn ‘boats’, (ScG. bàtaichean). However, we do not find e.g. *leavym for leabym (lheim ‘jump’, G. léim) (however, cf. Broderick’s comments on Marstrander’s transcriptions, §4.5.2). That preocclusion was especially exaggerated in singing for metrical reasons, such that it might be perceived as syllabic, is also possible, but given the complete lack of recordings of Manx traditional singing we can only speculate on this. There certainly seems to be no evidence of this in ordinary speech, and so little basis for Broderick’s claims regarding syllabicity. So far as is known, there are no cases of indication of preocclusion in the often highly non-standard orthographies of the carval manuscripts and the writings of Edward Faragher. This may be evidence that preocclusion was not particularly salient to native Manx speakers. 4.5.4 The origins of Manx preocclusion: previous hypotheses Four hypotheses have been proposed by scholars for the origins of Manx preocclusion. Other commentaries are purely synchronically descriptive (§4.5.2) and do not deal extensively with questions of diachronic development. 322 Aspects of the historical phonology of Manx 4.5.4.1 Rhŷs (1894) Rhŷs (142–3) (see above §4.5.2 for full quotation) suggests that preocclusion began as a reflex of geminate sonorants (i.e. [mm] > [u̯ m] > [ᵇm]), and later spread to the items with original lenis sonorants. He remarks that it can be understood as a further development of realizations more widespread in Gaelic dialects (Rhŷs: 143–4): if one might venture to relegate to a second place the extreme form of the Manx modification [i.e. with preocclusion], treating it as a development of the stage represented by τrŏu̯ m [t̪ roum], for example, in the case of trome, one would find that it ranges itself with a dialectic peculiarity of the Gaelic of the South of Ireland. Thus τrŏu̯ m is the pronunciation actually current not only in Manx but also in a great part of the South of Ireland. (Rhŷs: 143–4) It is argued below (§4.5.5) that Rhŷs’s analysis is substantially correct. 4.5.4.2 Chaudhri (2007) In his thesis on Cornish historical phonology, Chaudhri (2007: 39–44) includes a discussion of Manx preocclusion as a point of comparison with the analogous development in Cornish. He notes that, unlike in Cornish, Manx preocclusion does not occur in medial position (but see §4.5.4.3), and affects both original short and long sonorants. He suggests that the length contrast in the sonorants had already disappeared by the time of preocclusion in Manx: The reason given by Jackson [113–5] that pre-occlusion in Manx does not occur in non-final stressed syllables is that the affected consonant must be in absolute final position in the word. This may be because Manx, unlike Cornish, did not retain the phonemic oppositions /nn/–/n/, /mm/–/m/ by the time of pre-occlusion (whether or not this had earlier been the case) and the appearance of the long varieties of /n/ [nn] and /m/ [mm] was determined only by final position in a stressed syllable. (Chaudhri 2007: 39–40) 323 Aspects of the historical phonology of Manx The crucial difference is that Cornish must have retained the long phonemes /nn/ and /mm/ at least until the time of pre-occlusion. It did not therefore rely upon a process of gemination based on position. The parallel of Manx and Cornish pre-occlusion is by no means a direct one. (Chaudhri 2007: 44) Chaudhri thus posits a new gemination in Manx by which all stressed final nasal and lateral sonorants (all at this point short, whether or not they had earlier been geminate or non-geminate) were lengthened (i.e. (re-)geminated), as a precursor to preocclusion. With regard to eeym ‘butter’ (G. im) and kione ‘head’ (G. ceann), Chaudhri (2007: 40) comments that the relationship between preocclusion and vowel lengthening is not clear: It is not clear whether this lengthening of the vowel is associated with preocclusion in this word or, if not, which change came first. (Chaudhri 2007: 40) In the case of words containing an original short vowel, it seems most likely that any lengthening of the vowel occurred earlier than pre-occlusion. (Chaudhri 2007: 42) Chaudhri (2007: 43) considers that preocclusion in Manx is determined only by position and has nothing to do with the quality (fortis or lenis; tenseness in Chaudhri’s terminology) of the sonorant. In any case, pre-occlusion in Manx happened in words containing an original short vowel irrespective of whether the consonant was originally single or double. It happened equally in words containing an original long vowel. Whether or not a long vowel was originally long or short, it seems that there may possibly have been a tendency to shorten long vowels to compensate for the increasing tenseness of the following consonant, as Jackson suggested. Rhŷs considered that pre-occlusion occurred first after short vowels and later spread to syllables containing long ones; this appears to fit with his and Jackson’s hypothesis well […] The logical consequence of these observations is that pre-occlusion in Manx had no phonemic motivation but was instead determined only by word final position in a stressed syllable. This seems to be a good explanation for its comparatively wider operation, although it does not entirely explain why the additional changes /l/ [l] > [dl] and /ŋ/ [ŋ] > [gŋ] occurred only in Manx. It may 324 Aspects of the historical phonology of Manx have served to emphasise the long nature of a syllable where the inherited vowels tended to be shortened, although the evidence is equivocal. (Chaudhri 2007: 43) Chaudhri’s invocation of Rhŷs overlooks the fact that the latter explicitly links the genesis of preocclusion with original fortis sonorants (see §4.5.4.1). Chaudhri (2007: 55) argues that stressed position in itself favours the development of preocclusion. Although the following comment is made in relation to Cornish, combined with the above remarks on Manx we may infer that Chaudhri thinks stressed position in and of itself is sufficient to motivate preocclusion in Manx, given that he supposes that, unlike in Cornish, there was no longer any fortis-lenis, tense-lax or geminate-non-geminate contrast in Manx at the time of the development of preocclusion: It is reasonable to suppose on a general basis that phonemes are articulated with additional tenseness in a stressed syllable and moreover that they may receive heavier articulation when the vocalic element of the syllable is relatively short and the consonantal element is relatively long. This is a good description of the environment in which pre-occlusion is in fact observed. (Chaudhri 2007: 55) As discussed above, Chaudhri’s (2007: 42) suggestion that ‘[i]n the case of words containing an original short vowel, it seems most likely that any lengthening of the vowel occurred earlier than pre-occlusion’ seems to imply an earlier stage characterized by loss of gemination and lengthening of the short vowel, presumably by compensatory lengthening, followed by new gemination and subsequent preocclusion. This seems to be based on an overly simplistic view of compensatory lengthening / diphthongization and preocclusion as binary alternatives. It is possible that preocclusion and vocalic lengthening / diphthongization arose more or less simultaneously; it is normal that multiple cues for a phonological contrast should exist side by side, and that diachronic changes should involve gradual shifts in the prominence of different cues. Even in the most conservative Gaelic dialects which retain long sonorants, the vowel may be somewhat lengthened also, as noted by Jones (2010: 61) in Jura Gaelic: 325 Aspects of the historical phonology of Manx The consonants /ʟ/, /lʹ/, /ɴ/, /nʹ/ and /ʀ/ are given by Holmer (1938: 68) as occurring in lengthened form, represented orthographically by doubling. Holmer gives such words as ceann, mall and barrachd with these consonants denoted as long and the vowel immediately preceding them short. He contrasts this with the forms familiar in northern dialects where the consonant is short with the preceding vowel undergoing diphthongisation. In the data I have gathered there does in fact appear to be some lengthening of the vowel preceding the ‘doubled’ forms of /ʟ/, /lʹ/, /ɴ/, /nʹ/ and /m/ where this occurs in a monosyllable or in the stressed syllable of a polysyllabic word where the /ʟ/, /lʹ/, /ɴ/, /nʹ/ or /m/ forms a homorganic cluster with a following consonant as, for example in beanntan [bi̯ ɑˑɴˑd̬ ən] (‘mountains’). The syllable is long with partial lengthening of the vowel and lengthening of the consonant as well. I mark this phonetically as a half long vowel followed by a half long consonant. (Jones 2010: 61) As noted by Jones (2010: 62), this is implicit in the medieval Gaelic grammarians’ concept of ‘middle quantity’ (síneadh meadhónach) (Greene 1952), and the occasional marking of vowel length in such items as far back as the Old Irish period (GOI: 32): Original short vowels are sometimes marked long when followed in the same syllable by unlenited m, 1, n, r […]. Accordingly they must have at least sounded longer than the normal short vowel. Most, though not all, of them are long in the modern dialects also. (GOI: 32) The examples given [in GOI] are of the type ránn, tróm, báll, […]. Now all these […] still have a short vowel in many of the modern dialects, e.g. Donegal, where the usual treatment is short vowel plus long consonant. That is what the traditional spelling points to and the type from which the forms found in the other dialects […] are logical developments. There is of course no reason to believe that the vowels of these syllables sounded longer than the normal short vowels; it was the syllable itself which was half-long and therefore occasionally marked long. The syllable ferr was felt to be longer than fer, but not as long as fér. (Greene 1952: 212–3) If vowel and consonant length can co-exist simultaneously, and preocclusion is a development of the latter, then there is no obstacle to the initial restriction of preocclusion to long sonorants (and sonorant clusters). Chaudhri does accept that preocclusion may have spread from one environment to another (ibid.: 43), namely 326 Aspects of the historical phonology of Manx ‘pre-occlusion occurred first after short vowels and later spread to syllables containing long ones’. In explaining why preocclusion does not occur medially, Chaudhri suggests that Manx ‘did not retain the phonemic oppositions /nn/–/n/, /mm/–/m/ by the time of preocclusion’ (since otherwise we might expect medial G. -nn- etc. to give medial [dn] as in Cornish). However, it is quite unremarkable for the fortis sonorants to develop differently in medial and final position in Gaelic dialects. In Manx itself there is typically lengthening or diphthongization, and sometimes modification of quality (rounding) before coda fortis sonorants (e.g. G. ceann ‘head’ > Manx kione /kʲoːn/, /kʲau̯ n/, phonetically [kʲo(ː)ᵈn], [kʲau̯ ᵈn]), but before medial fortis sonorants there is only modification of quality (e.g. G. ceannach ‘buy’ > Manx kionnagh /kʲonax/). In Gaelic dialects in general we can identify at least four stages, from the most conservative to the most innovative: (1) Geminate sonorants retained both medially and finally, with no categorical vowel lengthening. Donegal dialects (Quiggin 1906: 77–8, 122; Wagner 1959: 17–26; Henderson 1974: 139–44), e.g. kʹaɴːɪ ceannuighthe ‘bought’ (LASID IV: 143, point 83). (2) Geminate sonorants retained finally but shortened medially, with no categorical vowel lengthening or diphthongization. Southern Scottish dialects (Holmer 1957: 87;274 Holmer 1962a: 21–4, 27–30; Ó Murchú 1989a: 107–10; Jones 2010: 62–3, 74–5), e.g. ceann ‘head’ kʹɛn‵ˑ, ceannaich ‘buy’ kʹɛ’n‵ɪ̇ç (SGDS II: 326, 336, point 53, Islay).275 274 Holmer’s (1957: 87) descriptions of Arran Gaelic seem to suggest optional retention of intervocalic fortis sonorant length, perhaps with morphological conditioning. 275 Holmer (1938: 81) tentatively suggests that medial nn may be lengthened in Islay as transcribes it and ll as such, e.g. ə-nɔlːᵘikj Nollaig ‘Christmas’, k῾jɛnːiç ceannaich ‘buy’ (ibid.: 137, 197). Jones (2010: 74) casts some doubt on Holmer’s descriptions, and gives forms such as kʹɑʔɴiç ceannaich. One wonders whether Holmer perceived sequences [ʔN] as a long sonorant; in my experience glottalization is often quite weak in Islay Gaelic. On the other hand, Ó Maolalaigh (2014b: 12) suggests that ‘Holmer’s description reflects the speech of older conservative speakers and that he ignored or failed to hear glottalisation in the speech of other speakers’. In general, Holmer’s discussion of this topic seems somewhat confused; so he claims that ‘[i]n Islay, Gigha, and certain parts of Skye, no difference is heard between the lenited and non-lenited n-sounds’. For what may be regarded as an intermediate stage 327 Aspects of the historical phonology of Manx (3) Geminate sonorants shortened in all positions but (all or some) fortis-lenis contrasts retained through place of articulation (dental v. alveolar) and secondary articulation (velarization, palatalization); there may be vowel lengthening or diphthongization before coda fortis sonorants. Connacht (de Bhaldraithe 1945: 106–11; de Búrca 1958: 131–3; Mhac an Fhailigh 1968: 160–3; Ó Curnáin 2007: 210–22, 234–7), Clare (Holmer 1962b: 38–42, 55–6), northern Scottish dialects (Borgstrøm 1937: 90–95, 111–19; 1940: 38–46, 65– 72, 142–8, 159–65; 1941: 24–29, 35–41, 77–82, 95–99; Oftedal 1956: 87–93, 120–29). (4) Geminate sonorants shortened in all positions, fortis-lenis contrast in sonorants entirely lost; there may be vowel lengthening or diphthongization before original coda fortis sonorants.276 Most of Munster (Ó Cuív 1944: 119– 22; Breatnach 1947: 140–3; Ó Sé 2000: 17–18), Late Manx. These developments may be represented schematically as follows:277 (1) (2) (3) (4) ceannach /kʲaNːax/ /kʲaNax/ /kʲaNax/ /kʲanax/ ceann /kʲaNː/ /kʲaNː/ /kʲәu̯ N/ /kʲәu̯ n/ It is quite plausible that Manx was at stage (2) at the point when preocclusion first developed, and that original fortis sonorants were still geminate at this point (contrary to Chaudhri’s claim) and thus liable to be affected by the initial development of preocclusion, whereas medial fortis sonorants had already been shortened, and so were unaffected by preocclusion, unlike in Cornish. between (2) and (3) in the Gaelic of Colonsay (situated between the southern area typified by Islay and the more northerly dialect area typified by diphthongization in items like ceann), see Scouller (2017: 76). 276 Except for /ŋʲ/ as a reflex of /Nʲ/, see above. 277 The details are somewhat simplified, especially as regards vowel quality. 328 Aspects of the historical phonology of Manx 4.5.4.3 Ó Maolalaigh (2014b) Ó Maolalaigh (2014b) briefly considers Manx preocclusion in a paper on ‘glottal and related features’ in the Gaelic languages. In unpublished lecture notes, he tentatively suggests that preocclusion resembles the phenomena of glottalization, h-insertion, devoicing and gemination in other dialects. He notes that Manx preocclusion can shorten a preceding long vowel, and proposes that ‘preocclusion following long vowels may be a secondary development’, implying that preocclusion began in stressed monosyllables with original short vowels. It seems that pre-occlusion has the affect of shortening a preceding long vowel, which is reminiscent of the shortening of vowels before geminates in Donegal Irish. The development of pre-occlusion following long vowels may be a secondary development. […] My suggestion is that pre-occlusion in Manx may be yet another reflex of glottalisation in the Gaelic languages. Phonetically speaking, the preglottalisation of sonorants is similar to preoccluded sonorants or prestopped sonorants in the occlusion or closing of the oral cavity. They are acoustically very similar, it seems to me. (Ó Maolalaigh 2014b: 22–3) Wagner (1956: 109) similarly saw the origins of preocclusion in glottalization: The occlusive element of the sonores (ᵗn), as well as the pre-aspiration of the tenues must arise from a glottal stop. (Wagner 1956: 109) Ó Maolalaigh (2014b: 23) lays considerably weight on the two examples of medial preocclusion from HLSM cited above (brynnagh and lieenyn) (§4.5.2), suggesting that ‘[p]erhaps it was once more common intervocalically but has been lost’, without, however, suggesting a mechanism or motivation for this loss, or for its retention in these items. 329 Aspects of the historical phonology of Manx The environments in which pre-occlusion occurs is [sic] very similar to that of glottalisation and gemination in ScG and Irish, i.e. it occurs in word final position and intervocalically following a short vowel. Unfortunately, I have only two examples of this in intervocalic position. Perhaps it was once more common intervocalically but has been lost. (Ó Maolalaigh 2014b: 23) It appears to be implicit in the following discussion from Ó Maolalaigh’s (2014b: 24) conclusion that he considers the environment of short vowel + lenis (‘light’) sonorant to be the prototypical environment for preocclusion, from which it subsequently spread to other environments (short vowel + fortis sonorant, long vowel + lenis sonorant): The joint evidence from Holmer and Wagner suggest that glottalisation may have occurred originally only with the light single sonorants. The absence of glottalisation with tense sonorants originally, can be related to the fact that glottalisation is not associated with heavy syllables. We have seen that in the case of syllables with long vowels and epenthetic or svarabhaktic environments. We can extend that to syllables containing geminate tense double sonorants too, although there seems to have been fluctuation between tense geminates and tense non-geminates in intervocalic position. Given the presence of glottalisation with heavy sonorants nowadays in ScG, it seems that gemination spread to these once they were reduced to non-geminate consonants. Perhaps the spread of glottalisation itself was a catalyst in the reduction of long tense sonorants – just as we have seen in the case of the shortening of stressed long vowels before geminate and pre-occlusive stops. The most conducive environment for glottalisation, gemination and preocclusion is a preceding short vowel. Indeed, we have seen that gemination in Irish and pre-occlusion in Manx can have the effect of shortening preceding long vowels. (Ó Maolalaigh 2014b: 24) Ó Maolalaigh (2014b: 27) further presents the following reconstruction of preocclusion as a later stage in a series of developments of glottalization. This reconstruction is predicated on the same (possibly unsound) assumption made by Chaudhri that degemination in e.g. cam, cill occurred prior to the development of preocclusion (§4.5.4.2). 330 Aspects of the historical phonology of Manx (5) Glottalisation of stops may result in pre-occlusion: cam kamm > kam > kʹaᵇm, cill kʲiʟʲː > kʲiʟʲ > kʲiᵈl (Ó Maolalaigh 2014b: 27) Ó Maolalaigh’s hypothesis that Manx preocclusion developed out of glottalization in similar environments to those found with the latter phenomenon in southern Scottish Gaelic dialects supposes that it was originally prevalent word medially. However, there is no evidence for this. As Ó Maolalaigh notes, only two possible cases of medial preocclusion occur in HLSM (II: 49, 277, III: 29). NM’s plural lʹiᵈnən lieenyn ‘nets’ (G. líon) may be influenced by preocclusion in the singular, or indeed by the /nt/ cluster of the historical irregular plural lieenteenyn (Bible, Cregeen); similarly JW’s joanagh dʹʒɔᵈnɑ ̣x ‘dusty’ probably reflects the monosyllabic stem joan. JW’s bre̜ dʹnʹɑ ̣x ‘flattering, comely’ and bre̜ nʹərɑ ̣x ‘act of flattering’ (the latter without preocclusion, it should be noted) are anomalous in other ways, as palatalization is not expected here (if the etymology brionnal is correct). It seems more likely that medial preocclusion here is a speech error (perhaps influenced by a semantically related item such as taitnyssagh ‘pleasant’, medial /tʲnʲ/?), than that medial preocclusion was once widespread before its unmotivated loss. Indeed, if preocclusion ever had developed medially after short vowels, there would be motivation for retaining it in the interests of increased syllable weight, as with glottalization and gemination etc. in other dialects. In addition, medial preocclusion would be expected to be more prone to being reanalysed as medial clusters /d.n/ etc., with syllable boundary, given the pre-existing phonotactics of the language; in which case they would be unlikely to subsequently disappear. As mentioned above, Ó Maolalaigh’s (2014b: 27) reconstruction of the development of preocclusion with original fortis sonorants apparently suggests loss of gemination prior to the development of preocclusion. As discussed above in relation to Chaudhri’s hypothesis, there is no reason to suppose this, and it will be argued that there is good typological reason rather to suppose that preocclusion developed from original final geminate sonorants (§4.5.5). The development of an oral stop from a glottal segment (buccalization), on the other hand, is reported to be typologically very rare (Trask 1995; La Voie 1996: 304; Hall 2009: 150–1). 331 Aspects of the historical phonology of Manx If the evidence of MNHL MD 900 MS 08307 discussed above (§4.5.3), is taken to suggest that preocclusion, after developing in final geminate sonorants, spread first to long vowel + lenis sonorants, and only subsequently to short vowels + lenis sonorants (see §4.5.5.2), then the environments in which preocclusion originates and is initially favoured are quite the opposite of those in which glottalization and the other features discussed by Ó Maolalaigh (2014b: 14, 27) are most prevalent. Nevertheless, it may be argued that preocclusion ended up serving prosodic ends similar to those of glottalization and related phenomena (§4.5.5). 4.5.4.4 Broderick (2018e) Similarly to Rhŷs (§4.5.4.1), Broderick (2018e: 13) in a brief comment suggests that preocclusion began with the original fortis sonorants, although he does not mention a link with the length of the original (geminate) fortis sonorants. Preocclusion became quite prevalent in L[ate] M[anx] whereby original fortis /L/, /N/, /M/, /Ŋ/, in losing their fortis quality, would be preceded by the corresponding stop, viz. /dl/, /dn/, /bm/, /gŋ/ to differentiate them from their lenis counterparts. (Broderick 2018e: 32) It seems more likely that the development of preocclusion precedes the loss of the fortis-lenis contrasts (§4.5.5), rather than compensating for it as suggested by Broderick. 4.5.5 The origin and spread of preocclusion 4.5.5.1 Typological and phonetic considerations Given the typological comparisons with other northern European languages mentioned above (§4.5.1.1), it seems that the most likely origin for preocclusion in Manx would be, as Rhŷs supposed, as a development of the original long sonorants /mː/, /Nː/, /Nʲː/, /ŋː/, /ŋʲː/, /Lː/, /Lʲː/ and the clusters /RN/, /RNʲ/, /RL/, /RLʲ/. 332 Aspects of the historical phonology of Manx As far as the nasals are concerned, the initial phonetic cause of preocclusion would be misalignment between oral closure and opening of the velum, resulting in an interval where the oral occlusion has been made but the velum is still closed. The longer the sonorant is, the more time there is for this to occur and be perceived and conventionalized. In languages which have long nasal sonorants but without consistent preocclusion, preocclusion may nevertheless occur sporadically.278 An intrusive oral stop is also phonetically natural in the rhotic + nasal/lateral clusters where there is a transition from a purely sonorant consonant with no coronal contact to a nasal stop or lateral with complete or partial coronal closure.279 From this perspective the development is perhaps less natural in the long laterals /Lː, Lʲː/, as observed by Chaudhri (2007: 54): The nasals [n] and [m] possess close oral counterparts [d] and [b], whereas the articulation of the lateral [l] is relatively further removed from that of [d] than is true of [n]. The exact phonetic realisation of /r/ and /rr/ in Middle Cornish may have been flapped [r], trilled or tapped [R] or retroflex [ɹ]. However, all of these sounds would be articulated even less closely to a plosive counterpart such as [d]. This may explain why /nn/ and /mm/ were inherently more likely to be pre-ploded as [ᵈn] and [ᵇm] than /ll/ and /rr/. It is likely that the further type of pre-occlusion [ll] > [ᵈl] occurred in Manx, but not in Cornish, because the phonetic motivation for these changes was sufficiently greater that the articulatory distance between [l] and [d] could be overcome. (Chaudhri 2007: 54) Marstrander (58) takes the difference between nasals and laterals in this regard as evidence against oral-velar misalignment as an explanation for the initial development, but his analysis confounds different periods of the development, and overlooks the possibility that preocclusion could spread analogically from nn and rl to ll: Utviklingen synes ikke å ha noe med konsonantens kvantitet å gjøre. Den forklares heller ikke ved en forsinket åpning av ganeseilet, da den jo også foreligger ved l. 278 Pavel Iosad, personal communication, has noted this in Welsh honni ‘claim’. Compare the development of medial ɴr > ɴḍr, ɴṭr and ʟ > ʟḍr, ʟṭr in Cois Fhairrge Irish (de Bhaldraithe 1945: 36–7), and Manx maynrey ‘happy’ (G. méanar) /meːnrә/ > [meːndrә] etc. (§4.2.5) (Rhŷs: 149; HLSM III: 18). 279 333 Aspects of the historical phonology of Manx [The development does not seem to have to do with the quantity of the consonant. Nor is it explained by a delayed opening of the velum, since it also occurs with l.] (Marstrander: 58) The fact that preocclusion with laterals is not noticed by some scholars (see §4.5.2), and seems to be less prevalent in general with laterals than nasals, may be evidence that the development was not so well-established with laterals. 4.5.5.2 Generalization and reanalysis of preocclusion From the long sonorants and sonorant clusters, preocclusion would have spread to original short or lenis sonorants, perhaps in association with the mergers between fortis and lenis sonorants which seem to have taken place between the seventeenth and nineteenth centuries (§§4.3, 5.4). Unfortunately, given that preocclusion is not found written until after this spread had already taken place, there can be no firm evidence for this supposition. However, the evidence of MNHL MD 900 MS 08307 may provide a clue. The fact that preocclusion is not found in items of the ben category (original short vowel + original lenis sonorant) in this manuscript may represent an intermediate stage, where preocclusion has spread to the lane (G. lán) category (original long vowel + original fortis) but not to the ben category. Possibly preocclusion was reanalysed as a marker of long (i.e. bimoraic) or heavy syllables — that is, vowel length/diphthongality, sonorant length and preocclusion became interchangeable and co-existing markers of syllable weight. Compare Iosad (2016b: 13), who comments briefly that moraicity is ‘possibly’ associated with Manx preocclusion; and also Chaudhri’s (2007: 43) comment that preocclusion ‘may have served to emphasise the long nature of a syllable where the inherited vowels tended to be shortened’. From here, preocclusion would finally spread to the ben category. This final stage may have been encouraged by the analogy of the small number of monosyllables with original final long sonorants in which the vowel had not been lengthened, such as /ka[ᵇ]m/ cam ‘bent’ (G. cam), /tʃi[ᶢ]ŋʲ/ ching ‘sick’ (G. tinn) (§§4.6.1.3, 4.6.1.12), including those in which short vowels had been retained or 334 Aspects of the historical phonology of Manx restored by paradigmatic analogy, such as /tʃo[ᵈ]n/ chionn ‘tight, fast’ (G. teann) (§4.6.4), as well as anomalous items such as skynn ‘knife’ (G. scian) which may have had a long vowel or diphthong earlier (§4.5.3). The pressure of a ‘bimoraic norm’ in Gaelic (cf. Iosad 2016b) may further motivate the development of /be[ᵈ]n/ etc. Sporadic gemination of consonants after short vowels in Manx (HLSM III: 27–8), as in other Gaelic dialects, may be a further manifestation of this tendency (Ó Maolalaigh 2014b: 21). As shown in the spectrograms above (§4.5.1), preocclusion seems to be interchangeable with sonorant lengthening, and occasionally “postocclusion” (cf. Rhŷs’s comment cited above on ooyl ‘apple’, G. ubhal, ‘mostly ūδλ [uːd̪ l̪ ] sometimes ū́ λδ [uːl̪ d̪ ]’, notebook 7: 176), and these can all be seen as realizations of the same synchronic phenomenon. 4.5.5.3 Preocclusion as a synchronic prosodic process There is evidence that preocclusion in Late Manx is synchronically a prosodic process which is implemented after other processes. Hence it may be found in polysyllables with unstressed final syllables where these are optionally reduced to monosyllabic realizations via secondary lenition of medial fricatives, as in jeeaghyn ‘looking’ (G. déachain): /dʒiːxәn/ > [dʒiːxәn], [dʒiːɣәn], [dʒiː.әn], [dʒiːᵈn] dʹʒiᵈn HB, SK, dʹʒiəᵈn JK, dʹʒiːən JTK, dʹʒiən JK, NM, dʹʒiɣən, dʹʒiːgʹən NM, dʹʒiːɣən JW(HLSM II: 229–30) This may apparently be lexicalized, as in shegin da ‘must’ (G. is éigean do), often spelled sheign in eighteenth- and nineteenth-century texts, for monosyllabic [ʃeinʲ], [ʃeiᵈnʲ] etc: seiᵈʹnʹ HK, saiᵈn NM, baiᵈn NM, ʃein TC, ʃiːn JK, sain JTK, EL, se̜ in JW, sein EKh, böin RC, brain [sic] W:S (HLSM II: 28, 395–6) Compare er-egin ‘by force’ (G. ar éigean) with retention of the bisyllabic realization and no preocclusion: e̜ r ˈẹːɣənʹ (HK), e̜ r ˈẹːɣən (JW) (HLSM II: 149). 335 Aspects of the historical phonology of Manx Anomalous preocclusion in normally unstressed final syllables may be further evidence of preocclusion as a live synchronic prosodic process, as in arran ‘bread’ (G. arán) (HLSM II: 11), cassan ‘path’ (G. cosán) (Rhŷs notebook 6: 133). With cassan in particular (expected /ˈkasan/ [ˈkazan], [ˈkaðan]) it seems possible that the informant was deliberately stressing the final syllable to draw attention to the contrast (in vowel quality) with cassyn ‘feet’ (/ˈkasәn/ [ˈkazәn], [ˈkaðәn]) in response to Rhŷs’s questioning: Dawson says tē bĭ́u er y chāδyn [teː biu̯ er ә xaːðәn] he is swift on his feet: but he calls a foot căs [kas], and căsā́ dn [kaˈsaːᵈn] for a footpath. (John Dawson, Rhŷs notebook 6: 133) 4.6 Vowel lengthening and diphthongization before coda fortis sonorants 4.6.1 In monosyllables In certain short vowel + fortis sonorant combinations in stressed monosyllables, the vowel may be lengthened or diphthongized, as found also in southern Irish and northern Scottish dialects (O’Rahilly: 49–52) (cf. §4.5.4.2). The evidence of Phillips shows that these developments before m, nn, ll were only incipient in the early seventeenth century, but they are complete by the time of the eighteenth-century texts. In some combinations there is a clear dialectal split between southern monophthongal realizations and northern diphthongal realizations (§4.6.1.34). The development of long vowels before final rr is common to all Gaelic dialects (O’Rahilly: 50), and as far as is known no dialects preserve a long rhotic */Rː/. The vowel /ɛː/ in the Manx development of -(e)arr shows it to be of some antiquity (§2.1.3.1). Thus the following four developments must have taken place in the order shown: (1) (2) (3) (4) 336 /e/ ea > /a/ (§2.1.3) /aRː/ > /aːR/ /aː/ > /ɛː/ (§2.2.2) /ah#/ > /aː#/ Aspects of the historical phonology of Manx 4.6.1.1 -ionn /iN/ > /in/, /uːn/ Table 70. -ionn Phillips er-giyn CM pronunciation /erʲ ˈɡʲin/ erskyn /erʲ ˈskʲin/ erskyn (3), aerskyn, erskin, er an skyn-syn,m er an skynsyn, er anskyn, er an skinn, er-yskyn, er yskynn fynn (K.); fyn-ruy (Cr.); Fyn (NBHR: 169) lhune /lʲuːn/ etymology English iar gcionn next, following 280 ós cionn above fionn; fionnruadh; Fionn white; (fyn-ruy) ‘having brown hair or fur’ (Cr.); Finn (MacCool) lionn, leann beer The short vowel in erskyn and er-giyn (G. cionn) may result from weak stress (cf. Ó Curnáin 2007: 211), although this would not be the case in the rare item fynn, also name Fyn; here high register and low frequency may explain the conservative realization. 4.6.1.2 -im /im/ > /iːm/ Table 71. -im Phillips eeym CM pronunciation /iːm/ etymology English im butter 4.6.1.3 -inn /iNʲ/ > /iŋʲ/ (/in/) See §4.4.6. All items have retention of short /i/. 280 See Thomson 1981: 50. 337 Aspects of the historical phonology of Manx 4.6.1.4 -ill /iLʲ/ > /iːlʲ/, /ilʲ/ Table 72. -ill Phillips keeill kíll (2), kill, kííll m(h)ill CM pronunciation /kʲiːlʲ/ etymology English cill church /milʲ/ mill, milleadh spoil CM pronunciation /fam/ etymology English feam stem of seaweed 4.6.1.5 -eam /em/ > /am/ Table 73. -eam Phillips fam Compare famlagh, famyragh ‘seaweed’ (G. feamnach). 4.6.1.6 -eim /em⁽ʲ⁾/ > /em/, /im/, /eːm/ (?) Table 74. -eim Phillips crem (Cr.), cremagh (adj.); cremeyder ‘fault-finder’ greme (Bible, Cr.), greim (Bible, Cr., K.), greimm (Bible) 338 grým, grym̄ , gryem, grĕy̆ m CM pronunciation ?/krem/ etymology English creim sore, ailment /ɡrem/, /ɡrim/ ?/ɡreːm/ greim bite; hold, grasp; stitch (in side) HLSM (n., vn.?) grɪmː NM, grɪm J:TL, grẹːᵇm TT, greᵇm JK, (vn.) grɪmal NM, griːmə TT, grĩmə TC, grãmə HK, (pret.) grẽm ̣ TC, grim TT, (part.) grĩmətʹ JW Aspects of the historical phonology of Manx From the spellings and HLSM realizations there appear to have been multiple by-forms of greme etc. with short and long vowels,281 and their lectal and paradigmatic distribution is not entirely clear. It may be significant that in the Bible the spelling ghreimm (suggesting a short vowel?) is restricted to the stem of the verb greimmey ‘seize’, whereas the noun is spelt greme.282 Compare also certain items in G. -éim, where there appears to be longstanding interchange with -eim(m); cf. ʎem, ʎemˑ etc. leum in south-western Argyll (SGDS IV: 246), and shortening in Teelin Irish bʹemʹ béim, kʹemʹ céim, ʟʹemʹ léim etc. (Wagner 1959: 12). Table 75. -éim Phillips keim kemm lheim liem, lĭĕm, liæ̆ m, (pret.) læ̆ ym, (fut. 1sg.) liemmym CM pronunciation /kʲe(ː)m/ /lʲe(ː)m/ etymology English HLSM céim, E.Ir. céimm, ceim etc. step, stile léim jump, leap, hop køm, kʹẹːm TC, kẹᵇm HK, keːm J:EK, (pl.) kømən kẹːmən, kʹẹːmən TC lʹeᵇm JK, JTK, J:EK, (vn. lheimey) lʹimə HB, (lheimyragh, ey) lʹe̜ mərɑ ̣x, lʹẽːmərɑ ̣xən, lʹemərə TC 4.6.1.7 -eann /eN/ > /oːn/, /au̯ n/, (/uːn/, /on/) Table 76. -eann Phillips chionn glione chiann, chiàn, chĭăn CM pronunciation S /tʃon/ N /tʃau̯ n/ etymology English HLSM teann tight, fast S /ɡlʲoːn/ N /ɡlʲau̯ n/ gleann valley S tʹʃɑ̜ᵈn TK, tʹʃɔᵈn NM; N tʹʃaun, tʹʃauN JTK S lʹɔᵈn EKh, lʹeN, lʹɑ̜ᵈN, 281 Although forms such as grɪmː (NM) could represent secondary shortening with (weak) preocclusion, i.e. [ɡriᵇm] (§§4.5.1, 4.5.5.2). 282 There are a number of other spellings of the noun in the Bible MSS, including greim, greym, gream. 339 Aspects of the historical phonology of Manx joan jan, jàn, jann S /dʒoːn/ N /dʒau̯ n/ deann dust kione kian (10), kiann, kian̄ S /kʲoːn/ N /kʲau̯ n/ ceann head, end 283 glen283 NM, lʹoᵈn, l̠ ɔᵈn, lʹɑ̜ᵈn J:EK, ɣʹlʹɔᵈn, glʹɔᵈn, lʹöᵈn, lʹɔᵈn TT, lʹöᵈn JW, lʹɔᵈn HK, glaᵘn TK, glen SK; N glʹøun, glʹöun TC, glʹaun, gle̜ n, glʹeN, glʹɛᵘn, glaᵘn, glɛ̜n, glʹen JK, JTK, lʹaun HB, (Peel) glʹɔᵈn, glʹo̜ ːn WQ, glo̤ ᵈn EC S dʹʒɔᵈn JW N dʹʒaun JK, dʹʒøun TC S kʹɔːn, kʹọːn, kʹo̜ ːn, kʹoⁱᵈnʹ NM, kʹo̜ ːᵈn EKh, kʹoːᵈn JW, HK, kʹɔ̜ːn SK, kʹoːn TL, kʹʲɔᵈn J:TL, kʹʲo.ᵈn J:EK, kʹʲoːn, kʹʲoːᵈn W:S; N kʹøun, kʹöun TC, kʹauN, kʹɛuN, kʹɔⁱᵈnʹ, kʹọᵈn JK, kʹaun JTK, HB, kʹɛun J:HB, J:JK, kʹŏun J:JK, kʹʲoun, kʹɑũn W:N, kjəun Fa, kʹoːn Myl Forms with short /e/ such as glen (NM) may be pretonic forms in place-names, and/or influenced by English ‘glen’, as found in English forms of Manx place-names, e.g. Glen Chass, Glen Auldyn, etc. (see HLSM II: 502–3). 340 Aspects of the historical phonology of Manx 4.6.1.8 -einn /eNʲ/ > ?/eːnʲ/ Table 77. -einn Phillips beinn CM pronunciation ?/beːnʲ/ etymology English HLSM beinn peak, top, pinnacle beːNʹ JK HLSM 4.6.1.9 -eall /eL/ > /oːl/, (/ol/) Table 78. -eall Phillips etymology English broill (Cr.) CM pronunciation /broːl/ breall? part of tool284 foall (K., PC) /foːl/ feall deceit gioal giall /ɡʲoːl/ geall pledge gʹoːlʹ TC giall gial, giall, gall, iall (2), iáll, jall /ɡʲoːl/, ?/ɡʲal/ geall promise (pret.) gʹɑl TC, but vn. gʹo̜ ːldən JW, gʹʲɔːldən J:EK lhuss-nymoal-moirrey (B.), lus ny moyl Moirrey (Cr.), luss-nymoal-moirree (K.) moll, molley /lus nә moːl285 morʲә/ ScG. lus nam meall móra /mol/ poll, pohll, polley (Cr., K.) /pol/ meall, mealladh peall, pealladh deceive mat, stick together, entangle 284 ‘the part of a tool that bruises down by being hammered on, as on the upper end of a jumper, a chisel, or the point of a rivet. There is no corresponding word in English’ (Cregeen). If the etymology is correct, this is presumably a further example of Cregeen’s extraneous use of (§1.6.5), perhaps under the influence of unrelated brooillagh ‘crumbs, fragments’ (Cregeen) (G. brúireach). 285 Rhŷs notes short mol, mụl (Thomas Collister, Rhŷs notebook 7: 198), presumably owing to postlexical phrasal stress. 341 Aspects of the historical phonology of Manx 4.6.1.10 -earr /eR/ > /ɛːr/ Table 79. -earr Phillips baare; baarey giare; giarey kiare share, nhare, ba(a)re CM pronunciation /bɛːr/; /bɛːrә/ etymology English bearr; bearradh gearr; gearradh cearr shave, cut hair left is fearr better, best CM pronunciation /ʃaŋ/ etymology English seang /streŋ/ sreang lean, emaciated string CM pronunciation /am/ etymology English am ‘stature, size, puberty’ (Cr.), ‘time, period, season, era’ (K.), disposition, fit state (SW)287 /kam/ cam /mam/ Ir. mám, ScG. mam, E.Ir. mám, maam giarr (2), giar, iar kiar /ɡʲɛːr/; /ɡʲɛːrә/ sháyr (3), shayr (3), niarr, bâŷr, b’áyr /ʃɛːr/, /nʲɛːr/, /bɛːr/ /kʲɛːr/ short; cut 4.6.1.11 -eang /eŋɡ/ > /eŋ/, /aŋ/ Table 80. -eang Phillips shang streng stréin, (pl.) strengyn 4.6.1.12 -am /am/ > /am/ Table 81. -am Phillips amm (Cr., SW), am (K.) cam, camm mam 287 kamm handful; boil, blain A rare word in Manx, apparently with specialized semantic developments. The usual word for ‘time’ is traa (G. tráth). 342 Aspects of the historical phonology of Manx 4.6.1.13 -ann /aN/ > /oːn/, /au̯ n/, (/uːn/) Table 82. -ann Phillips CM pronunciation S /oːn/, N /uːn/ etymology English HLSM ann in him, it; there S ɔːn, ɔːᵈn, o̜ ːn NM; N uːn JTK, JK, u.ᵈn HB S klo̜ ːᵈn HK, klo̜ ːn, klo̜ ᵈn NM, kloːᵈn JW, klɔːᵈn TL S krɔᵈn NM, krɑdn HK; N kro̜ n TC S go̜ ːᵈn, gɔːᵈn, sko̜ ᵈn NM, skoːᵈn JW, skɔːᵈn EL; N gøũn, skøun TC, gãuə̯n W:N ayn ayn, áyn cloan klaun (3), kláun (4), klaún S /kloːn/ N /klau̯ n/ clann children croan, cron (Cr.) kran̄ , kran, krón S /kroːn/ N /krau̯ n/ crann mast, pole, tree goan, goaun (both Bible), scoan skaun S /ɡoːn/ N /ɡau̯ n/ gann; is gann scarce; hardly Northern /uːn/ is presumably a secondary development under weak stress of */au̯ n/. 4.6.1.14 -ainn /aNʲ/ > /eːnʲ/, /ai̯ nʲ/ Table 83. -ainn Phillips bine (pl.) beinyn, banniyn, bainyn clein rheynn, ring (Cr.) reyn, rǽyn (2), reynn, renn, ren̄ , rêŷng, reygn, ræing, reng, ren, reyng, CM pronunciation /bainʲ/ etymology English HLSM bainne, boinne drop baiᵈn NM, bɑ̜iᵈnʹ, bɑ ̣in JW, bɑ ̣ːiᵈʹnʹ HK, bɑdʹnʹ W:S, (pl.) bɑ ̣:inʹən HK ? /kleːnʲ/ clainn, cloinn (dat. clann) rainn, roinn kin /reːnʲ/, /riŋʲ/ (see also fn. 288) divide, division, share re̜ idʹnʹ TT, riŋ JW, raiᵈn, reːn TM, rødʹᵊnᵊʹ TC 343 Aspects of the historical phonology of Manx ræyng, ræyn, (impv. pl.) renji, reynnigi, (fut. 1sg.) renniym 4.6.1.15 -all /aL/ > /oːl/, (?/au̯ l/) Table 84. -all Phillips CM pronunciation /boːl/ etymology English HLSM ball place bɔːl JK, bo̜ ːl JTK, bɔ̜ːl HK, boːl JW, bᵘo̜ lː HB ko̜ ːl NM boayl ball, boll coayl /koːl/ call lose doal kall (5), káll, kal, kàl, kiall dall (3) /doːl/ dall blind hoal, noal háll, nall /hoːl/, /noːl/ thall, anall over moal mall (2) /moːl/, ?/mau̯ l/ mall poor, feeble, slow 344 doːl JW, do̜ ːl HK, JK, HB, NM noːl JW, əˈnɔːl, əˈno̜ ːl EL, nɔːl, əˈno̜ ːl TC, nɑ̜ːl W:S mo̜ ːl TC, JW, mɔ̜ːl, mɔːl NM, mọːl JTK, mɔːl HB, J:EK, mau̜ l J:JK, mo̤ l W:S, moːl W:S Aspects of the historical phonology of Manx 4.6.1.16 -aill /aLʲ/ > /alʲ/, /elʲ/ Table 85. -aill Phillips etymology English HLSM gail, gháil CM pronunciation /kalʲ/ caill lose kalʹ JW keyll kæil, kélliy ?/kelʲ/ caill, coill, ScG. coille wood, forest saill sáil /salʲ/ saill fat, brawn, grease, blubber kølʹ TC, ke̜ lʹ HK, kiːl, kyːl, kiːlʹ NM, kɪːlʹ J:NM, xɛlʹ J:JK, kidlʹ W:S, (pl.) ke̜ lʹən HK sɛːlʹ TC caill 288 For doail ‘blind’ (n.pl.) (G. daill, doill) see §4.6.3. 4.6.1.17 -arr /aR/ > /ɛːr/ Table 86. -arr Phillips baare báyr caayr (K.), cayr (Cr.) faare farr, na fáyrsyn CM pronunciation /bɛːr/ etymology English barr /kɛːr/ carr /fɛːr/ i bhfarradh289 top, tip, point, crop wagon, coach, car, vehicle etc. near 288 The spelling with <ey> and some of the attested realizations of this item in HLSM may suggest a form with assimilation to the /әː/ set (<G. ao(i)), from an earlier (?allophonic) short [ә] realization – as also perhaps in rheynn ‘divide, share’ (G. rainn, roinn) (and other items?) (§4.6.1.14). 289 The loss of -adh here is unusual (§5.2). It possibly results from reanalysis of the combination of this preposition with the definite article, i.e. i bhfarradh an /faRә әN/ > /faRә#N/ > /faR#әN/ > /faR/ > /faːR/ /fɛːr/. This would mean that loss of final unstressed /ɣ/ (</ð/) was early, preceding lengthening in monosyllabic /aR/ and /aː/ > /ɛː/. The etymology of farradh (EIr. arrad), is obscure, although it probably involves the prefix ar- (LEIA s.v. arrad). 345 Aspects of the historical phonology of Manx 4.6.1.18 -a(i)ng /aŋɡ/ > /aŋ/ Table 87. -a(i)ng Phillips stang CM pronunciation /staŋ/ etymology English sta(i)ng wooden horse, stock, whipping post 4.6.1.19 -om /om/ > /oːm/, ?/au̯ m/ Table 88. -om Phillips etymology English HLSM croym CM pronunciation /kroːm/ crom S (impv.) kro̜ ːm TT lhome /loːm/ lom v. stoop, incline; adj. stooped bare trome /troːm/290 trom heavy S loːm JW, (lhome-leear) lo̜ ːm ˈliːr TT N (lhomelane) lum leᵈn, lo̜ m leᵈn JK, (lhomelomarcan) loːm ˈlo̜ məkən S troːᵇm, troːm NM, troːm JW, tro̜ ːm EKh, tʰroːm J:EK, tʰroːᵇm J:TL N trobm JK, trubm W:N, tʰrɑ̜ᵇm J:JTK Rhŷs (143) notes an apparently diphthongal realization of this word as τrŏ́ um [t̪ rou̯ m], alongside τrŏ ̣bm [t̪ roᵇm], τrŭbm [t̪ ruᵇm], but does not comment on its dialectal distribution. From his comments τrŏ́ um might also be interpreted as weak preocclusion or geminate [mː], however (§4.5.2). Rhŷs does not mention these forms in his discussion of the dialectal contrast between diphthongal and nondiphthongal realization of items in -(e)ann (Rhŷs: 160). 290 346 Aspects of the historical phonology of Manx 4.6.1.20 -oim /om⁽ʲ⁾/ > /em/ Table 89. -oim Phillips breim CM pronunciation /brem/, ?/brim/ etymology English HLSM broim fart brɪm NM, (breimeragh) brɪmərax NM, brïmərɑx W:N, vre̜ mərɑ ̣x TC etymology English HLSM bonn heel, sole /doːn/ donn brown, bay /droːn/ (K. droun = /drau̯ n/?) dronn hump; humpback /ton/, /to:n/, /tau̯ n/ tonn wave S bo̜ ᵈn TT, bᵘuᵈn, boᵈn NM; N bøun TC, (pl.) bøunən TC, baunən JK S do̜ ːᵈn TT, dɔːn J:EK, J:TL; N dõ̜ ːn TC, doːᵈn S droːᵈn JW, dro̜ ːᵈn TT, (adj.) dro̜ ːnɑx TT N droːᵈn HK (adj.) droːnɑ ̣x TC to̜ n TC, (pl.) to̜ nən TC 4.6.1.21 -onn /oN/ > /oːn/, /au̯ n/, (/on/) Table 90. -onn Phillips boyn, (pl.) boynyn, boynnyn (both Bible) (pl.) bŏynyn doyn (Bible), dhoan, dhone (both Cr.) drone, droyn (Cr.), dron, droun (K.) tonn, toayn, touyn (Hymns) tóyn, (pl.) tonnyn (3) CM pronunciation S /boːn/ N /bau̯ n/ 347 Aspects of the historical phonology of Manx 4.6.1.22 -oinn /oNʲ/ > /eːnʲ/ Table 91. -oinn Phillips breinn brú (2), bru, breyn (2), brey̆ n, brenn (3), brêin, brein, (gen.) brónniy, (pl. breniyn) CM pronunciation /breːnʲ/ etymology English HLSM broinn (brú) womb vrẹːnʹ, (pl.) brẹːnʹən TC etymology English HLSM coll hazel N køul, køəl TC S /poːl/ N /pau̯ l/? poll pool S /toːl/ N /tau̯ l/ toll hole S poːl HK, (unstressed) pu.l, po̤ l, pə NM, pəul M S toːl JW, NM, to̜ ːl EKh, töul Mrs EKh toːl W:S; N toul W:N CM pronunciation /tolʲ/ etymology English HLSM toill, tuill deserve toilliu tɔlʹu, to̜ lʹu TC, tilʹu EKh 4.6.1.23 -oll /oL/ > /oːl/, /au̯ l/ Table 92. -oll Phillips couyl (Bible), coll, cohll (Cr.), coull (K.) poyll towl toll, tóyll CM pronunciation S /koːl/ N /kau̯ l/ 4.6.1.24 -oill /oLʲ/ > /olʲ/ Table 93. -oill Phillips toill 348 toilliu tolliu 2, tollíu, tolliú, toiliu, toilchin toilchyn (3), toilchiyn Aspects of the historical phonology of Manx 4.6.1.25 -orr /oR/ > /oːr/ Table 94. -orr Phillips coayr (Bible, Cr.); coair, coyr, coir (Cr.); coair, corr (K.) coar, coayr (Bible, Cr.) Ph. torr̀ torr̀ CM pronunciation /koːr/ etymology English corr odd (number) /koːr/ corr */toːr/291 torr crane, heron, bittern heap CM pronunciation /loŋ/ etymology English long ship CM pronunciation /kum/ etymology English cum, cumadh congbháil, ScG. cùm, cumail form, shape HLSM kɔːr W:S, kɔː JK, ko̜ ːʳ TT 4.6.1.26 -ong /oŋɡ/ > /oŋ/ Table 95. -ong Phillips lhong long (5) 4.6.1.27 -um /um/ > /um/ Table 96. -um Phillips cum, cummey cum, cummal kúm̄ , gumm, ghumm, ghum̄ (2), ghúmm (2), ghum /kum/ hold; dwell 291 This item appears not to survive in CM and LM, except in the diminutive form thorran /toran/ ‘dungheap’ (G. torrán). 349 Aspects of the historical phonology of Manx 4.6.1.28 -uim /umʲ/ > /iːm/ Table 97. -uim Phillips dreeym drym, dry̆ im, dryim, (pl.) drymmyn, drimmyn sym CM pronunciation /driːm/ etymology English druim back /sim/ suim sum CM pronunciation /nuːn/ etymology English anunn over 4.6.1.29 -unn /uN/ > /uːn/ Table 98. -unn Phillips noon núnn 4.6.1.30 -uinn /uNʲ/ > /unʲ/, /inʲ/, /iŋʲ/ Table 99. -uinn Phillips cluinn, clynn cruin(n), (Bible, Cr., K.); cring, cruing (Cr.) cruinn 292 klŭyn, klŭy̆ n, kluỳn, kluin, kluíín, gluyn (2), gluinn, gluin krŭin, krunn, kruinn, kruin, krŭyn, kruyn, krynn, krunn, krun CM pronunciation /klunʲ/, /klinʲ/ etymology English cluinn hear /krunʲ/, /kriŋʲ/ cruinn round /krunʲ/, /kreːnʲ/? crann (pl.) masts These forms seem to represent crainn rather than cruinn. 350 HLSM krẹːᵈnʹ HK, krẹːᵈnʹᵊ NM292 Aspects of the historical phonology of Manx 4.6.1.31 -uill /uLʲ/ > /uːlʲ/ Table 100. -uill Phillips buill puill puill tuill CM pronunciation /bulʲ/ etymology English HLSM ball (pl.) places buⁱlʹ NM, bo̜ ⁱlʹ HK, bo̤ lʹ JW, TC /pulʲ/ poll (pl.) pools /tulʲ/ toll (pl.) holes tûːl W:N 4.6.1.32 -uing /uŋʲɡʲ/ > /wiŋʲ/ Table 101. -uing Phillips quing quing (3), kuing mwing CM pronunciation /kwiŋʲ/ etymology English cuing yoke /mwiŋʲ/ mong, muing mane 4.6.1.33 Summary Table 102 summarizes the above developments of historical short vowels + fortis sonorants, showing the environments in which the vowel remains short, becomes long or diphthongal, and where both developments are found. Table 102. Predominant CM/LM reflexes of stressed short vowels before final fortis sonorants L Lʲ R ŋ ŋʲ m N Nʲ i iː i, uː i — iː, i — — — e a oː, au̯ eː oː, o — ɛː e, a — a a oː, au̯ eː oː a, e ɛː a — o oː oː, o, au̯ eː oː, au̯ — oː o — u u, iː uː u, i — u — — i long vowel or diphthong long or short vowel — no instances 351 Aspects of the historical phonology of Manx 4.6.1.34 Dialectal variation There are clear north-south dialectal splits between in some of the vowel + fortis sonorant combinations (Rhŷs: 160; HLSM I: 161), as shown in Table 103: Table 103. Dialectal splits in stressed short vowel + final fortis sonorant sequences -eann -ann -onn -all -oll south /oːn/ /oːn/ /oːn/, /on/ /oːl/ /oːl/ north /au̯ n/ /au̯ n/ /au̯ n/, /oːn/. /on/ /oːl/, (/au̯ l/) /au̯ l/ Notice that -all and -oll are differentiated in the north but not in the south; otherwise (e)a and o fall together before nn and ll. Northern varieties also show diphthongs in other items where there are long monophthongs in the south, as doo S /duː/, N /dau̯ / ‘black’ (G. dubh), ooh S /uː/, N /au̯ / ‘egg’ (G. ugh), jiu S /dʒuː/, N /dʒau̯ / ‘today’ (G. i ndiu) (Rhŷs: 161; HLSM II: 121, 238, 341). In the standard Classical Manx orthography (i.e. in the Bible) the orthography generally, although not exclusively, represents the southern / monophthongal forms (Table 104). In this connection we may note the southern origin of some of the key figures in the standardization of the Manx orthography, such as Philip Moore and John Kelly (both of Douglas), although further research on the orthographic variants in various texts (such as the Bible translation manuscripts and sermons) is needed in order to understand more fully how this southern bias came about. Table 104. Representation of dialect variants in Bible orthography Bible spelling etymology English chionn glione joan kione ayn cloan teann gleann deann ceann ann clann tight, fast valley dust head, end in him, in it, there children 352 dialectal correspondence south south south south ? (see §1.6.3) south Aspects of the historical phonology of Manx croan goan, goaun, s’coan boyn doyn tonn couyl poyll towl crann gann bonn donn tonn coll poll toll mast, pole, tree scarce, hardly heel, sole brown wave hazel pool hole doo ooh jiu dubh ogh, ugh, ubh i ndiu, ScG. an-diugh black egg today south south / north south south both? north south ? (spelling could indicate both) south south south 4.6.2 Voiced homorganic rhotic clusters We mostly found lengthening of short stressed vowels preceding voiced homorganic rhotic clusters rd, rn, rl, as is widespread in other Gaelic dialects (O’Rahilly: 50), although apparently retention of short vowels with /i/ and /u/. Table 105. Vowel developments before rd, rn, rl eard, eird keird Phillips CM pronunciation etymology English HLSM duyne na kerdȷý /kʲeːrd/, /kʲeːrdʒә/ ceard, ceird craft /kʲeːrdi/ ceardcha, ScG. ceàrdach smithy kʹe̜ ːṛd NM, kʹed, kʹöd, köːd JK, köːd HB, (ferkeirdey, -ee) fɛṛ kʹödi J:NM, fer köʳdʹʒə EKh kʹe̜ ʳdi TC, kʹeːdi, kʹeṛdi JK, kʹaːṛdi, kʹe̜ ːᵊrdi NM, (pl.) kʹeːᵊdiə̯n JK, kʹɛːdiə̯n NM /meːrdәrax/, /meːrdәrәs/ meirdreach fornicator; fornication kairdee (Cr.), caardagh (K.) eird maarderagh; maarderys fer mardruoil, mardrys (5) 353 Aspects of the historical phonology of Manx ard ard aird ard, ardjyn gerjagh(ey) syrjey, yrjid ord boayrd oard, oayrd (Bible, Cr.) oardaghey uird buird ayrd (38), áyrd (6), ard (9), árd (3), (pl.) ardy, árdy /aːrd/ ard high aːd J:EK, NM, Eːd J:CW, öːd J:TL, öːʳd TC, ɑːd NM arjyn (5), árjyn /aːrd/, /aːrdʒәn/ aird direction, region gyrjaghy (7), gyrjaghe, gyrjaghey (4), gýrjaghey, gyrji (2), gyrjitt syrje (3), na sýrje, nás ýrje, sirrje, sýyrje, b’ýrje; (n.) ýrje /ɡeːrdʒaxә/, ?/ɡәːrdʒaxә/ gairdeach comfort, joy öːṛd, aːʳd HK, örd, ïʳd TCr, ɑ ̣ːd TM (gerjys) göːdʹʒɪs AK /sirdʒә/, /irdʒәdʲ/293 airde higher; height öːʳdʹʒədʹ TT, ördʹʒədʹ JW, HK, ïʳdʹʒədʲ, jørdʹʒəd TC boyrd (3), bóyrd (2), boỳrd, board, bord (2) ordyn /boːrd/ bord table, board /oːrd/ ord ord- (18), órd- /oːrdaxә/ ordaghadh (sledge) hammer order böːʳḍ W:N, vɔ̆ːṛd JK, bᵘu̜ ːṛd NM, bo̜ ːrd TC oːʳd, (pl.) oːʳdən HK buyrd /buːrd/ buird tables, boards earrann part, portion *earn < earrann, ighearn, iarann ayrn ayrn (3), áyrn /aːrn/ (2), arn, árn, æarn, (pl.) arnyn (3) 293 aːṛn NM, HK, aːʳn JW, aːn TM, NM, (pl.) aːʳnən TC, In syrjey, yrjid we seem to have the morphophonemic alternation /a/ > /i/, as often in Manx (§2.1.9), with <y> representing short /i/. The medial cluster /rʲdʲ/ apparently has depalatalization of the rhotic (>/rdʒ/), with preceding /i/ treated like io-, with retracted allophone, and even the development of prosthetic glide [j] in TC’s jørdʹʒəd. In the case of gerjaghey (G. gairdeach), we would expect /a/ or /e/ (with lengthening) rather than raising to /i/, but Phillips consistent <y, ý> suggests some retraction before depalatalized /r/; perhaps there is merger with ao /әː/, so gaird- > *gáird > *gaoird-. 354 Aspects of the historical phonology of Manx chiarn (ayrniagh) ɑ ̣ːrniɑ ̣x, aːrniə̯n TC tʹʃaːʳn JW, tʹʃɑ ̣ːṛn NM, (voc.) xʹjaːn J:EK chiarn (7), chiárn, chíarn, charn, (pl.) chiarnyn, chárnyn iarn (2), iærn /tʃaːrn/ tighearna lord /jaːrn/ iarann, EIr. iarn iron jɑ ̣ːrn TC, jaːᵈn JK, HK, jaːn JK, W:N, nʹaːᵈn HK, ɡʹaːn JTK arn, *arn < arraing, arthain carn /kaːrn/ carn cairn, heap kaːrnən TC er-mayrn er mayrn (3), er m’ayrn /erʲ maːrn/ ar marthain left, remaining tayrn tayrn (3), tăy̆ rn, tarn, darniym, tarngi, tarni, harn, hayrn, háyrn /taːrn/ tarraing pull, draw e̜ r ˈmɑ ̣ːᵈn HK, e̜ r ˈmaːʳn JW, ɛ̆ ˈmɑːᵈn NM, e̜ r ˈmaːʳn TC taːn NM, TK, HB, taːᵈn JTK, tɑːrən W:S airn arn /aːrnʲ/ airne sloe barnagh /baːrnʲax/ bairneach limpet, flitter böːʳnɑ ̣x HK /kaːrn/ corn /kuːrn/ corn horn, trumpet can kaːrn, (pl.) kaːrnən TC ku̜ ːṛn NM döːᵈn JK, do̜ ːrən, do̜ ːʳn TC, (pl.) do̜ ːṛnən NM skɔːṛnɑx, sko̜ ːṛnɑx NM, skɔːrnəx W:S, sko̜ ːrnɑ ̣x TC yiarn (see also §5.3) orn cayrn karn, gharn curn doarn, doarney dornyn (2), dorny, er nan orny, goyrn /doːrn/, /doːrnә/ dorn, dornadh fist scoarnagh skorniagh294 (2), skornagh, storinagh (sic) /skoːrnax/ scornach, scoirneach throat /soːrn/ sorn fire-place in kiln sorn, surn (Cr.) 294 = by-form G. scoirneach? 355 Aspects of the historical phonology of Manx oirn moyrn taarnagh uirn curnaght moyrn (2), môŷrn (2); mórniagh tarniagh (3), tárniagh, (gen.) tarni, (pl.) tarniaghyn; (vn.) tarnaghy /moːrnʲ/, /moːrnʲax/ muirn, moirn pride; proud mɔːnax JK /taːrnʲax/ toirneach thunder tøːʳnɑ ̣x, (gen.) taːrni TC, töːʳnax JW, HK, toːnjə [sic] W:S kurnagh (3), kurnah, kurnaght, kurnyght /ku(ː)rnʲaxt/ ? cruithneacht > *cuirneacht wheat køʳnɑ ̣x TC, ko̜ ːnax, köːṛnax, korənʹax, kɔ̆ːṛnʹax JTK, köːnax JK /duːrnʲ/ duirn fist (pl., gen.) arlu (5), arlú (5); iarlaghy (2), arlyghey, arlii, arluíít, arlaghy, arlyghy, arliaghy, arlaghey /aːrlu/ earlamh, EIr. aurlam ready; prepare, cook ɛːlu, öːṛlu NM, öʳlətʹ HK merliagh (5), merliygh, (pl.) merli (4) /meːrlʲax/ meirleach thief mɛːlʹɑx, möʳlʹɑx W:N, mo̜ ːlʹɑ ̣x TC /ɡoːrlә/ galar disease /oːrlax/ orlach, ordlach inch /doːrlʲәʃ/ doirling gap in hedge durn (Bible), duirn (Cr., K.) earl aarloo; aarlaghey eirl maarliagh orl gorley (Cr. K.), goarley (Cr.), gallar (Cr.) oarlagh oirl doarlish 356 dorlys, dorlysyn do̜ ːʳləʃ, döːʳləʃ JW, döːⁱliʃ, Aspects of the historical phonology of Manx dɔilɪʃ JK, doːliʃ JTK, do̜ ːlïʃ HK *url < *iorl < iolar burley urley yllrée, ýlrey uirl murlhin (Cr.), moorlin (K.) /burlә/ biolar, biolra cress børlə TC /urlә/ iolar eagle öːʳlə, (pl.) öːʳlɑ ̣xən HK /murlʲәnʲ/ Ir. muirleóg, murlóg, ScG. mùrlag, mùrlainn hamper mŏːlʹin JTK Note the indication of vowel length in e.g. boayrd /boːrd/ ‘table, board’ (G. bord) in contrast to rhotic + unvoiced consonant clusters, such as ort /ort/ ‘on you’ (G. ort), fort /fort/ ‘ability’ (G. feart?), gort /ɡort/ ‘stale’ (G. goirt), which contrast with long /oː/ in coyrt ‘give, put, send’ (G. tabhairt), loayrt /loːrt/ ‘speak’ (G. labhairt), roayrt ‘springtide’ (G. rabharta). 4.6.3 Other medial clusters Lengthening and diphthongization may also take place before other, mainly homorganic, clusters involving sonorants, although it is not nearly as prevalent as in certain other Gaelic dialects (e.g. Ó Curnáin 2007: 210–22, 234–7). Table 106. Vowel developments before other medial clusters Phillips etymology English oonlaghey CM pronunciation /uːnlaxә/ ion(n)ladh wash Boaldyn /boːldәn⁽ʲ⁾/ Beal(l)taine May Day /boːndә/ Ir. banda, ScG. bannd band (of iron etc.) /boːndәrә/ banaltra nurse boandey boandyrey, boandyrys boyndy (‘bondservant’) baintri (‘nourish’) HLSM bɔːldən JK, bɑ̜ːldən TM, vɔːldɪnʹ NM bo̜ undəʳ TC, bo̜ ːndə TT bøundər, bo̜ ːndər, vo̜ ːndər TC, bɑːndərəs W:N, bo̤ nðrəs NM, 357 Aspects of the historical phonology of Manx bɔ̜ːndərəs HB, bõ̜ ːndərə W:?N stoandey /stoːndә/ 295 ScG. stannd standish blunt, stammering relish, ‘kitchen’297 moandagh /moːndax/ man(n)tach aunlyn, oanlyn, oalyn296 (Cr.), awnlin (K.) oanluckey /oːnlәn⁽ʲ⁾/, /au̯ nlәn⁽ʲ⁾/ annlann anlaky, anlaghy, anlyky (2), anliky, (part.) ánlikit /oːnlәkә/ annlacadh, adhlacadh, EIr. adnacul bury; burial, funeral (moddey oaldey) madiállty (2), (pl.) madiallty (2) /oːldә/ allta, allaidh wild; (moddey oaldey) wolf moaldey /moːldә/ ScG. màlda? poor, mean foalsey /foːlsә/ fallsa false oalsum (Cr.), ousym (K.) /oːlsәm/, /au̯ (l)sәm/ Norse halsband > *allsam connlach tie on cattle straw cunnradh, connradh exchange, barter oaldey coonlagh kunlygh /kuːnlax/ coonrey kúnre, kúnrey, kúynre /kuːnrә/ 295 øunlənʹ TC o̜ ːnləkə, õ̜ ːnləkə TC, oːnlɪkə JTK, õ̜ ːⁿləkɑ̜ NM, (part.) oːnlɪkətʹ JTK, ɔː̃ ⁿlət, oːnlɪkʹ JK, õːⁿləkɪtʹ NM, oːnlikʹət, uːnləkʹət, ũːŋɡlə W:N fo̜ ːlsə, fo̜ ːlzə HK kuːnlax JK, JTK, TL But cf. sondagh ‘covetous’ (G. san(n)tach), apparently with short /o/; and fondagh ‘sufficient’ (G. foghantach), fo̜ ndɑ ̣x TC (HLSM II: 175) where /oː/ might be expected. The tendency to shorten stressed long vowels in polysyllables (§5.5.1) may explain these forms. 296 ‘Though the former of these [oanlyn] may, perhaps, be the best orthography, yet see aunlyn’ (Cregeen s.v. oanlyn, oalyn). Cregeen’s preference for a spelling clearly indicating monophthongal /oː/ probably reflects his southern dialect. 297 ‘relish or moisture that is taken with bread, potatoes, &c’ (Cregeen), ‘a kitchen, any kind of food eaten with bread, as butter, cheese, milk’ (Kelly). 358 Aspects of the historical phonology of Manx Note the diphthongal realizations in some of these items, pointing to the same dialectal variation as found in monosyllables (§4.6.1.34). 4.6.4 Paradigmatic uniformity Historically, the restriction of lengthening and diphthongization to stressed positions preceding coda fortis sonorants resulted in alternations such as the following between monosyllabic radical forms with a lengthened vowel, and morphologically complex derivatives or inflections in which the original short vowel was maintained: glione ‘valley’ S /ɡlʲoːn/, N /ɡlʲau̯ n/ (G. gleann) gen. glionney /ɡlʲonә/ (G. gleanna) However, both lengthened and unlengthened forms may spread analogically, as found in some Irish dialects (Ó Sé 2000: 42; Ó Curnáin 2007: 212–3; Ó Direáin 2015: 43–4, 46). For example, the adjective giare /ɡʲɛːr/ ‘short’ (G. gearr), with a long vowel, shows the expected alternation with plural giarrey /ɡʲarә/ (G. gearra), with a short vowel; but the verb giarey /ɡʲɛːrә/, giare /ɡʲɛːr/ ‘cut’ (G. vn. gearradh, stem gearr) apparently has generalization of the lengthened form (Thomson 1998: 86); similarly in baarey /bɛːrә/ ‘shave’ (G. bearradh).298 On the other hand, tonn /ton/ ‘wave’ (G. tonn) and chionn /tʃon/ ‘tight, fast’ (teann) appear to show generalization (or maintenance) of the short vowel on the pattern of plural tonnyn /tonәn/ ‘waves’, plural adjective / verbal noun chionney /tʃonә/ ‘press’ (G. teanna; teannadh). In these cases spellings suggesting expected /toːn, tau̯ n/ and /tʃoːn, tʃau̯ n/ are also found. Further examples are as follows: Generalization of short vowel: mill /milʲ/ ‘spoil’, vn. milley /milʲә/ (G. mill, milleadh), rather than */miːlʲ/ (cf. keeill /kʲiːlʲ/ ‘church’, G. cill) moll /mol/ ‘deceive’, vn. molley /molә/ (G. meall, mealladh), rather than */moːl/, similarly poll, polley ‘matt, entangle’ (G. pealladh) Forms from HLSM (II: 195–6): vn. gʹaːrə, gʹaːru, gʹaːro JK, gʹeːrə HB; gaːrəsmɑ ̣d JW giarrysmayd ‘we will cut’. 298 359 Aspects of the historical phonology of Manx Generalization of long vowel: baare /bɛːr/ ‘shave’ (G. bearr) > vn. baarey /bɛːrә/ (G. bearradh), rather than */barә/ eeym /iːm/ ‘butter’ (G. im) > eeym(m)ey (Cr.) ?/iːmә/, rather than */imә/ lhome /loːm/ ‘bare’ (G. lom) > lhoamey /loːmә/, lommey /lomә/ (both Cr.), but lomman /loman/ ‘scorching wind’ (G. lomán), lhommyrt /lomәrt/ ‘shear’ (G. lomairt)299 trome /troːm/ ‘heavy’ > tromey /troːmә/ pl., also trommey /tromә/ (Psalm 12:5) doal /doːl/ ‘blind’ (G. dall) > doaley pl. (G. dalla), doalley300 ?/doːlә/ ‘to blind’ (Exodus 23:8, 32:32–33) (G. dalladh), but dolley /dolә/ ‘to blind, blot’ (Bible), dollan /dolan/ ‘fan’ (G. dallán) croym /kroːm/ ‘bend, stoop’ > croymmey vn. (G. cromadh) /kroːmә/ (o̜ ː TC, TT, oː TC, HLSM II: 109) 299 See also spellings loamrey (Cr. , Bible), lomrey (Cr.) ‘fleece’ (G. lomradh) ? = /lomrә/ or /loːmrә/. It is possible that <oa> here simply represents vowel quality, or recalls the spelling of the stem doal, rather than representing length, i.e. /dolә/ rather than /doːlә/. 300 360 Aspects of the historical phonology of Manx Chapter 5 Suprasegmental phonology 5.1 Stress In earlier periods lexical stress in Gaelic fell on the first syllable,301 and vowels in any syllable, whether stressed or not, could be either long or short (GOI: 27, 31; O’Rahilly: 83–5; Ó Sé 1989: 148). This remains the case in some Irish varieties today, notably in most Connacht dialects (Ó Sé 1989: 148; Green 1997: 93).302 However, long vowels in unstressed syllables are cross-linguistically dispreferred. This mismatch between stress and syllable weight has long been recognized by Gaelic scholars such as O’Rahilly (84–5): Now in a language with strong stress, like Irish, words containing an unstressed long syllable […] are more or less in a state of unstable equilibrium. Little will be required to upset the equilibrium, which once upset, will only be restored when one or other of two opposing tendencies has taken effect. Either […] the long unstressed syllable will be shortened, or else the stress will be attracted to the long vowel. (O’Rahilly: 84–5) This intuition is captured by the ‘Weight-to-Stress Principle’ (Prince 1990), cited by Green (1997) in his analysis of developments in Gaelic prosody, which states ‘[i]f heavy, then stressed’ and contraposed, ‘[i]f unstressed, then light’ (Prince 1990: 358). Connacht Irish dialects continue to tolerate a violation of the Weight-to-Stress Principle, while in general (and leaving aside certain details), Munster dialects have shifted stress to non-initial heavy syllables, while Ulster and Scottish varieties have shortened non-initial long vowels and retained initial stress. The situation in Manx is particularly complex in that both solutions are found: some words have stress-shift, while others have retention of initial stress with shortening of 301 Disregarding elements such as the prefixes of deuterotonic verb forms in Early Irish (GOI: 27). Ó Sé (1989: 166–8) discusses evidence that Old Irish stress was considerably less prominent than in the modern dialects; as a result the Weight-to-Stress Principle may have been less relevant in this period. 302 There is evidence that Connacht dialects may have had non-initial stress in the past (O’Rahilly: 99– 100; Ó Sé 1984; 1989: 159–60). 361 Aspects of the historical phonology of Manx unstressed long vowel. This is referred to by Thomson (1960: 122) and Broderick (HLSM III: 149) as ‘the problem of stress in Manx’. It has been observed since O’Rahilly (114) that these two developments are conditioned by the weight of the initial syllable: there is a strong tendency towards stress shift in items with historical long or ‘half-long’ vowels (including diphthongs), in the initial syllable, while items with short vowels in the initial syllable are likely to show initial stress with vowel shortening (see also Jackson: 20; HLSM III: 148–9; Green 1997: 90–1; Ó Sé 1991). This is well illustrated by the pair of items /muˈrɛːn/ mooarane ‘many, much’ (G. mórán), with stress shift, and /ˈbeɡan/ beggan ‘little, few’ (G. beagán). In addition, many Anglo-Norman borrowings (such as shirveish ‘service’, G. seirbhís; resoon ‘reason’, G. réasún) have final stress, which may be original. Ó Sé (1991) notes that these tendencies are not categorical, citing counterexamples with stress shift despite light initial syllables (§5.1.5.2), and suggests that the patterns observed represent the results of lexical diffusion. Ó Sé (1991: 162) also notes that ‘[w]ords in which a short vowel is followed by a cluster of sonorant consonant + voiced consonant (e.g. ordóg ‘thumb’) have tended to be treated in Manx like words with a long vowel in the first syllable (e.g. fág(bh)áil ‘leaving’)’. In general, stress is transparently represented in the Classical Manx orthography. For example, the suffixes which have both stressed and unstressed reflexes are clearly distinguished: G.: -án -ín -óg -amhail -áil -óir Stressed: -ane /ˈɛːn/ -een /ˈiːn/ -age, -aig, -eig, -aag /ˈɛːɡ/ -oil /ˈoːlʲ/ -ail, -eil /ɛːlʲ/ -eyr /ɛːrʲ/ Unstressed: -an /an/ -in /әnʲ/ -ag /aɡ/ -al /al/ -al /al/ -er /erʲ/ Cregeen indicates stress on most headwords in his dictionary, but as Wheeler (2018: ii) notes, ‘[t]here is a considerable degree of inconsistency in Cregeen’s marking of stress’, and stress is not infrequently marked in an unexpected position, usually on the initial syllable in words where the phonological structure and spelling, and recorded 362 Aspects of the historical phonology of Manx fieldwork data, show forward stress. It is possible that some of this variability ‘may correspond to real variation in pronunciation’ (ibid.: ii), perhaps including (postlexical) stress retraction (§5.1.3), or that some unexpected stress markings are printing errors. Because of this uncertainty, Cregeen’s marking of stress is not used as evidence in the present discussion. 5.1.1 Previous accounts of stress shift and long vowel shortening 5.1.1.1 O’Rahilly (1932) O’Rahilly (114) claims that non-initial stress was originally introduced into Manx (as in southern Irish dialects) with Anglo-Norman loanwords. He observes that ‘all such borrowings have a long vowel in the second syllable, and all, or practically all of them have (or had) in the first syllable either a long vowel, or else a ‘half-long’ vowel’. According to O’Rahilly, the addition of these items to the Manx lexicon motivated the analogical shifting of stress in native items (including earlier borrowings) of the same (heavy-heavy) pattern. Subsequently, ‘another phonetic law, by which long unstressed vowels were shortened, came into operation’ and ‘[t]he long terminations of those words which had escaped the accent-shift were accordingly shortened, as in Scottish Gaelic’. Thirdly, ‘after the above changes had established themselves’ (O’Rahilly: 115), vocalization of fricatives could create new long vowels ‘in hitherto unstressed syllables’, and ‘the word in its new shape was brought into conformity with the stresssystem of the language’. Although not stated explicitly, it is implied in O’Rahilly’s account that the first syllable conditioning was no longer operational at the stage when medial fricatives were vocalized — since otherwise the cited examples annoon ‘weak’ (G. anbhfann), shilleeid ‘slug’ (G. seilchide) etc. would have given something like */ˈanan/, */ˈʃelʲәdʲ(ә)/ rather than attested /aˈnuːn/, /ʃiˈlʲiːdʲ/ — and that by this period, post-tonic unstressed long vowels were no longer permitted in Manx phonology (‘the stress-system of the language’). 363 Aspects of the historical phonology of Manx It will be argued below that much of O’Rahilly’s account is plausible, although internal phonetic factors may explain the stress shift in heavy-heavy better than the alleged impact of the Anglo-Norman borrowings, which may, however, have been a contributory factor. In addition, it is not necessary to invoke, as O’Rahilly (117) does, ‘the influence of Scottish Gaelic’ to account for post-tonic shortening. 5.1.1.2 Jackson (1955) Jackson (20) notes the first syllable conditioning, which he calls ‘remarkable’. However, he does not offer an explanation of the phenomenon, or discuss the ordering of the changes, although he references O’Rahilly’s discussion. He also notes (76–7) that long vowels arising from vocalization of fricatives may attract stress, but does not discuss the ordering of this development in relation to other processes. 5.1.1.3 Ó Baoill (1980) Ó Baoill (1980: 102) makes brief mention of forward stress in Manx in relation to the lack of epenthesis in items such as colmane ‘dove’ (G. colmán). He argues that stress shift precedes epenthesis, and draws conclusions about dating as follows: It is a well known fact about the stress rule of Munster Irish that it must apply after the application of the rule of epenthesis in words like feargach and Luimneach. In this case the epenthesis rule blocks the application of the stress rule. I would suggest that the same procedure applies in the case of epenthesis in words like colmane and carnane in Manx […] We may conclude from these relevant facts from Munster Irish and Manx that epenthesis occurs after stressed syllables only. If this ordering of rules is correct, then the stress rules, which place stress […] on -ane in words like colmane , carnane and on -aag in faasaag in Manx, is very old indeed, and may go back at least to the Old Irish period and perhaps even further. (Ó Baoill 1980: 102) However, it is shown below (§5.4) that Ó Baoill’s arguments regarding Manx data are incomplete and that no conclusion can be drawn from internal evidence about the relative ordering of epenthesis and stress shift. In any case, Manx and Irish developments are not necessarily parallel, either structurally or chronologically. 364 Aspects of the historical phonology of Manx 5.1.1.4 Broderick (1986) Broderick303 (HLSM III: 151–3) proposes an alternative explanation for forward stress, namely that it was a result of shortening of the vowel in the initial syllable: The contention is that, rather than the first syllable in those words containing forward stress being shortened as a result of the stress-shift, the opposite is the case; that is, that the stress is advanced because of the shortening of the initial syllable. It is also a noticeable feature of Manx that stressed long vowels in monosyllables are usually, or can be shortened [examples given including eeast (G. iasc), moain (G. móin); slane (G. slán)] […] It is […] suggested that the proclivity of Manx to shorten stressed long vowels in monosyllables spread to the initial (original) stressed long vowel vowel in words of the faagail type […] causing it to shorten […] as a result of which the stress shifted to the long second syllable […] In the case of those words containing initial stress on an original short vowel, but whose second syllable, once long, has been shortened, as in beggan, thunnag etc., it may be asked why then did the stress not shift to the long vowel when it was long. The answer, perhaps, is that the long vowel here had already become shortened at the time of the stress-shift in disyllables of the faagail type. (HLSM III: 151–2, original emphasis) There are several difficulties with Broderick’s account. Ó Sé (1991: 172) identifies one of them: I am reluctant to follow Broderick’s […] explanation of the stress shift as resulting from shortening of stressed long vowels. It is indeed the case that Late Spoken Manx showed (often variable) shortening of stressed long vowels in monosyllables, as in slane (slen) [sic] ‘whole’ (Ir. slán), eeast [jis. jiːs] ‘fish’ (Ir. iasc), but I am not convinced that this development is old enough to have contributed to the stress shift, giving e.g. ˈfágáil > ˈfăgáil > faˈgáil. (Ó Sé 1991: 172) In fact the eighteenth-century orthography clearly and consistently indicates long vowels or diphthongs in moain, slane, eeast etc., so the variable shortening observed in Late Manx must be a recent development, and, as Broderick himself notes (HLSM III: 151), it may be associated with preocclusion in items such as slane. Vowel shortening is longer established, and shown orthographically, in polysyllabic items 303 According to Broderick (HLSM conclusion’. III: 151), Heinrich Wagner ‘told me he had come to a similar 365 Aspects of the historical phonology of Manx such as firriney ‘truth’ (G. fírinne), currym ‘duty’ (G. cúram), and in certain categories of items such as frequently-occurring verb forms (§5.5.1). However, stressed vowel shortening is not sufficiently consistent to be considered a viable explanation for the regular pattern of stress shift observed in original heavy-heavy items, nor is there any evidence that it is old enough. We might note also that Late Manx also has a countervailing ‘proclivity’ to lengthen short stressed vowels (§5.5.2). Furthermore, Broderick offers no explanation for why unstressed vowel shortening occurred in the light-heavy category (beggan etc.) without affecting heavy-heavy items (faagail etc.). His only argument for the earliness of the shortening is that suffixes with original /aː/ and /oː/ have /a/ in their shortened Manx form, which, according to Broderick (HLSM III: 153), ‘indicates that shortening of the second syllable took place in words of this type before OIr. Á (and Ó), became /eː/ [i.e. /ɛː/] in Manx’. The latter claim is also made by Jackson (20). However, it is uncertain how old the fronting and raising of á is (§2.2.2). Phillips (and indeed the later orthography) represent this primarily as <a, aa> etc., which may have indicated something closer to [æː, aː] in earlier periods. This less fronted realization seems to have survived even into the Late Manx of certain speakers or dialects (HLSM III: 123). In any case, it does not follow that the quality of the shortened /a/ must reflect the quality of the earlier long vowel. In both Manx and Scottish Gaelic shortened /a/ (in ScG. also /ɛ/) represents a variety of original vowels (cf. Ó Maolalaigh 2001: 8). There also seems to have been a constraint in Manx forbidding any vowel except /ә/ and /a/304 in unstressed closed syllables.305 Broderick (HLSM III: 153) claims that ‘[t]he addition of the Anglo-Norman/Middle English loanwords with forward stress would fit into this pattern, and perhaps help to establish it, but would follow rather than establish the stress-shift’. Broderick mentions the items such as jarrood (G. dearmhad) with secondary long vowels from fricative vocalization, but does not discuss how they fit into the chronology of the other developments, except to note that they pre-date Phillips (1610). 304 And perhaps /e/ in -er (G. -óir). Compare Ó Dochartaigh’s (1987: 57) discussion of the relationship between vowel shortening and vowel quality reduction. 305 366 Aspects of the historical phonology of Manx 5.1.1.5 Ó Sé (1991) Ó Sé (1991)306 agrees with Broderick in proposing that unstressed vowel shortening occurred before stress shift, but for different reasons. He notes that his ‘outline is of course speculative, but it is designed to take account of geographical and social factors, as well as using such insights as can be gained from quantification’. Ó Sé dates unstressed vowel shortening to the period between the ninth and thirteenth century when Man was politically, as well as geographically, close to western Scotland and Ulster: It is reasonable to assume that Manx speech shared in linguistic changes taking place in the vicinity, and that unstressed vowel shortening arrived as part of a process of spread throughout the region; [Ó] Dochartaigh (1987: 34) shows it to have diffused across Ulster from northeast to southwest and it seems reasonable to regard it as a Scottish innovation which gradually spread south, into Ulster and Man, over a period of some centuries. (Ó Sé 1991: 167) For the relationship between the length of the stressed and unstressed syllable, Ó Sé (1991: 168) cites the case of Achill Irish (Ó Dochartaigh 1978; 1987: 32–4), which has been observed to show the opposite correlation to that found in Manx: 306 Ó Sé’s is the most detailed treatment of the topic to date, being based on a quantitive analysis of items from HLSM II. Some of Ó Sé’s analysis of individual items is faulty, however, as the following examples show (the list is not exhaustive). The items folliaght ‘secret’ (G. folaigheacht) and markiagh ‘ride’ (G. marcaigheacht) should not be included in Ó Sé’s (1991: 177) ‘contracted’ category, as there is no evidence that -aigheacht ever gave */iːxt/ in Manx as it did in Irish. Rather the treatment is as in Scottish Gaelic, with retention of a syllable boundary /i.axt/ (Non-coalescence of the vowels here is probably due to the retention of /a/ before /x/, whereas -(a)idh/ghe(a)- elsewhere gives /iː/ which can attract stress, as in Creesteenyn ‘Christians’ /kri(ː)ˈstiːnәn/ (*Críostaidheannan > *Críostaíonnan), but not Creesteeaght ‘Christianity; the eucharist’ /ˈkriːsti.axt/, G. Críostaidheacht). Caaig ‘jay’ (Ir. cabhóg, ScG. cathag) and scaan ‘ghost’ (G. scáthán) are included under vowel shortening rather than stress shift, but owing to coalescence of the two syllables the results of the two treatments would be formally identical, so these items should have been excluded (Ó Sé 1991: 178). Tanroag(an) ‘scallop’ is included under stress shift (Ó Sé 1991: 179), but in fact the basic form is roagan (Cregeen), with an obscure first element tan- (HLSM II: 442); it seems that we in fact have vowel shortening in the suffix -án, perhaps from *rothacán based on roth ‘wheel’? Thassane ‘hiss’ is included as an exception (stress shift despite initial heavy syllable), on the basis of Ir. tasán, but cf. ScG. tàsan (Dwelly). Seaghyn ‘sorrow’ is explained as *saochán, and thus an exception of the opposite kind, but may rather be derived from suaith- (§3.9.7), and is probably a regular development of an original trisyllable (§5.1.5.5). 367 Aspects of the historical phonology of Manx words of the beagán ‘few’ type retain their second vowel long whereas the mórán ‘many’ type shorten it […] Manx shows another way for shortening to spread, by an assimilative rather than a dissimilative relationship between the two syllables. (Ó Sé 1991: 168) Vowel length assimilation, however, is not a known phonological process. An alternative explanation for the Achill development would be that stress shift occurred in light-heavy items in order to resolve the violation of the Weight-to-Stress Principle, followed by stress retraction as in other Connacht dialects (O’Rahilly: 99–100; Ó Sé 1984; Ó Sé 1989: 159–60), while in heavy-heavy items there was no stress shift, but shortening of the unstressed long vowel. As for Manx, assuming that Ó Sé is right that vowel shortening in light-heavy items preceded stress shift in heavy-heavy ones, Green’s (1997) account, discussed below, provides a better explanation. Ó Sé (1991: 169) claims that vowel shortening spread through the lexis by creating doublets, affecting first light-heavy items and later beginning to affect those in the heavy-heavy category: By assuming short-term variability in the operation of unstressed vowel shortening we may find an explanation for the doublets, and for the fact that they involve only words with a heavy initial syllable, the shape which we associate with stress shift. I propose that unstressed vowel shortening spread by creating doublets, so arán > (arán ~ arăn). As it diffused through the lexicon over time the earliest words to have been affected […] lost their original forms (so only arăn survives) but the more recently affected would retain a double form (ˈdornóg ~ ˈdornŏg). Those doublets which survived in Late Spoken Manx (caraig, dornaig etc.) come from this overlap between the two sound changes. (Ó Sé 1991: 169) However, the restriction of doublets to original heavy-heavy items can be otherwise explained. Firstly, initial stress remains the most frequent pattern in Manx, so there is always potential analogical pressure to assimilate to this pattern (especially, perhaps, with less frequent or poorly remembered words). Secondly, some of the doublets noted by Ó Sé (1991: 177–8) contain transparently analysable morphemes which could easily be restored (e.g. dornaig ‘handle’ = doarn ‘fist’ + diminutive; eeasteyr 368 Aspects of the historical phonology of Manx ‘fisherman’ = eeast ‘fish’ + agentive suffix), with the unstressed version of the suffix substituted. Thirdly, postlexical stress retraction in Irish dialects with forward stress is a welldocumented phenomenon (Ó Sé 1989: 151; 2000: 52–55; Iosad 2013: 70–1) and could explain some of these cases. Indeed, Ó Sé (1989: 156) himself is careful to note this possibility when discussing the evidence of Lavin (1957) on stress in an East Mayo Irish dialect. This appears to be the case in some of the instances from HLSM: for example, in yn dornaig y skynn shoh [sic] ‘the handle of the knife’ ən ˈdörne̜ g ə skin ʃo̜ ː (HLSM II: 128), dornaig would be expected to have a secondary postlexical stress within the genitive phrase. The same speaker (TC) also has expected stress in the same word: dörˈnɛːg, dørˈnɛːg. Orthographic evidence is decidedly against the co-existence of by-forms with initial stress: spellings such as faasaag, caraig, eeasteyr, faagail all clearly show final syllables with long vowels, and there are no known orthographic variants such as *faasag, *carrag, *eeaster, *faagal. Moreover, if the variants with initial stress were original, we would expect them to preserve the original quality and length of the vowel in the first syllable. However, the form of caraig ‘beetle’ (G. ciaróg) with initial stress given in HLSM (II: 59) is kʹarɔg (alongside finally stressed kʹəˈrɛːg, both W:S), whereas, if a form with the original initial stressed syllable had survived without undergoing stress shift and initial syllable vowel reduction, we would expect a form *keearag, CM */ˈkʲiә̯raɡ/, LM */ˈkʲiːraɡ/ (cf. ScG. ciarag). It therefore seems clear that forms such as kʹarɔg represent secondary, most likely postlexical, restressing of the initial syllable, rather than the result of an inchoate spread of Ó Sé’s proposed lexical diffusion of post-tonic vowel shortening into the heavy-heavy category. According to Ó Sé (1991: 170), vowel shortening was halted by external sociolinguistic factors: The social cause for unstressed vowel shortening being discontinued will have been the collapse of the Kingdom of Man and the Isles in 1266 A.D., providing a “catastrophic overturn of the social values that are helping the change proceed 369 Aspects of the historical phonology of Manx in an orderly fashion” (Coates 1987: 194). Manx was subsequently cut off from linguistic changes affecting other areas of Gaelic speech. (Ó Sé 1991: 170) It is plausible that this historical turning-point would prevent new changes taking place in other Gaelic dialects from diffusing into Manx. It is unclear, however, why an ongoing change which was supposedly already well-established in the language would be halted in this way. On the contrary, given that unstressed long vowels appear to have been preserved longer in high-register literary varieties used or patronized by Gaelic-speaking elites, it is more likely that the removal of such elites would accelerate, rather than halt, the change (cf. O’Rahilly: 105; Ó Dochartaigh 1978: 332). Ó Sé (1991: 171) claims that ‘stress shift in Manx will postdate the thirteenth century, and will therefore have coincided with increasing contact with English (containing a large Romance adstratum by this stage)’. He suggests that ‘there is good reason to believe unstressed vowel shortening to be an internal development in Gaelic’ but that ‘it is not improbable that language contact did play a role in the stress shift’. Regarding the ‘contracted words’ with original medial fricatives, Ó Sé (1991: 169–70) claims that they behave similarly to items with original long vowels: This study casts some new light on the history in Manx of words like bunadhas, geanamhail, peacamhail, although only 25 such words occur in the sample. The fact that these contracted words do not pattern very differently from those with original long vowels suggests that vowel shortening did not greatly precede contraction. Some of them are attested in seventeenth century Manx with forms which have not survived […] Phillips’ translation of the Book of Common Prayer (c. 1610) has gan(n)oil, gniartuoil for gennal, niartal […]. It is not impossible that most or all of these words were subject to variability of form at that time. (Ó Sé 1991: 169–70) However, it is not the case that the ‘contracted words’ pattern similarly to those with original long vowels; on the contrary, they are not subject to initial syllable weight conditioning, generally showing forward stress regardless of the weight of the preceding syllable (bunnys, G. bunadhas, and the adjectives in -oil, -al, G. -amhail, 370 Aspects of the historical phonology of Manx are special cases, §5.1.4). Ó Sé presumably failed to perceive this owing to the smallness of his sample. 5.1.1.6 Green (1997) Green (1997: 92) explains first syllable conditioning as follows: At an early date, Manx promoted TROQ and demoted MAX(µ) […] in the primary grammar (made up of native words), this had the effect of shortening the long vowel in bégaːn […]; but there was no effect on (H H) words like bóːkaːn […], because they did not violate TROQ. (Green 1997: 92) This relates to ‘trochaic quantity’ (Prince 1990: 359; Hayes 1995: 79–85), the observation that in trochaic languages light-light (and heavy-heavy) trochees are better formed than light-heavy ones, which provides a motivation for reduction of lightheavy words to light-light without affecting the heavy-heavy category. Green (1997: 92) assumes that Anglo-Norman borrowings were prespecified for final stress. He claims that stress shift occurred once vowel shortening in the beagán type had occurred, and was prompted by the analogy of the Anglo-Norman items: Later, as happened in [Scottish] Gaelic, old (Ĺ H) words like bʹégan were reinterpreted as underlyingly (Ĺ L). Once this happened, the only (L H) words in the language were the end-stressed Anglo-Norman words like bodʹéːl. At this point, the Cw clusters […] received an epenthetic ә, and later, the sequence әwә contracted to uː. (78) CwV > Cәwә > Cuː dʹárwad > d′árәwәd > d′áruːd ‘forgetting’ [jarrood, G. dearmhad] The new (L H) words like d′áruːd took over the forward stress of the AngloNorman words, becoming d′arúːd and the like. Also, the native (H H) words like bóːkaːn took over the Anglo-Norman stress pattern, becoming boːkáːn. (Green 1997: 92) Green (1997: 90–3) claims that items with historical medial fricatives are treated differently according to whether the fricative was originally intervocalic (e.g. 371 Aspects of the historical phonology of Manx geanamhail, Manx gennal) or followed a consonant (e.g. dearmhad). A wider set of evidence casts doubt on this, however (§5.1.4). Green’s explanation for first syllable conditioning by appealing to cross-linguistic metrical preferences is phonologically plausible. A possible objection could be raised in relation to the fact that in Late Manx all unstressed historical long vowels are apparently shortened, including those in original initial syllables (O’Rahilly: 115; Jackson: 20; HLSM III: 148). Green’s account would require two long vowel shortenings, first affecting light-heavy items (e.g. arán /ˈaraːn/ > /ˈaran/, Manx arran ‘bread’) and later affecting initial syllables after stress shift (e.g. fuarán /ˈfuә̯raːn/ > /fuә̯ˈraːn/ > /faˈrɛːn/ > [fәˈrɛːn], [frɛːn], Manx farrane ‘fountain, spring’). It might be argued that it would be more economical to posit only one vowel shortening, following stress shift. However, there is some evidence that initial syllables in stress-shifted items are treated differently to post-tonic syllables, and that vowel shortening may be a more recent development in the former environment (§5.1.3). 5.1.2 An explanation for stress shift targeting heavy-heavy items A possible explanation for early stress shift targeting only heavy-heavy items may be found in the phenomenon of peak delay, whereby the pitch peak moves further to the right in longer words than in shorter words. As Iosad (2016a: 82) notes, ‘[i]n languages with peak delay, longer domains are associated with a later placement of the tonal peak; hence, disyllabic words would have associated their peaks further to the right compared to monosyllabic words’. One might reasonably suppose that longer syllables, and thus overall word length, would also result in greater peak delay, and thus that heavy-heavy items would show pitch peak further to the right that light-heavy items. There is also potential for a mutually-reinforcing relationship in that pitch rises are better perceived on longer vowels, and longer vowels are perceived as having rises. These factors could result in reanalysis of initially-stressed heavy-heavy words as having primary stress on the second syllable. 372 Aspects of the historical phonology of Manx Both this account, whereby stress shift preceded post-tonic vowel shortening, and Green’s proposal discussed above, which posits the reverse, seem plausible. Since both changes, and also vocalization of medial fricatives, took place before the seventeenth century and therefore before the beginning of the Manx literary tradition, it may not be possible to reach a firm conclusion on this matter. 5.1.3 Initial syllable shortening in items with forward stress Pretonic vowels in old heavy-heavy words are noted as short in descriptions of Late Manx (O’Rahilly: 115; Jackson: 20; HLSM III: 148), and generally transcribed as such in HLSM etc., in contrast to the situation in Munster Irish where such vowels retain their length (e.g. Ó Sé 2000: 46–55), but the spelling evidence presents a mixed picture. Morphologically transparent forms with stressed endings usually retain the long vowel spelling of the stem, but this may be merely orthographic: faagail ‘leaving’, G. fág(bh)áil. fɛ̆ˈge̜ ːlʹ NM, HB, fəˈgaːlʹ, faˈgaːlʹ JK307 (HLSM II: 154) faag ‘leave’, G. fág, (impv.) fɛːg TT, JW, faagit (part.) fɛːgətʹ HK, fe̜ ːgɪtʹ NM, fɛːgitʹᶴ AK, faːgɪtʹ JTK (HLSM II: 154) mooarane (mórán) muˈrɛːᵈn JW (HLSM II: 305) mooar ‘big, great’, G. mór muːr, muːə TC, muːə JK, NM, muː TK, TL, moːər NM, mu̜ ːr J:TL, mu̜ ːṛ, mu̜ ːəṛ J:EK, J:JTK, J:HB, muː W:S, pl. muːrə TC, NM (HLSM II: 305) Items where the first syllable is no longer a recognizable independent element tend to be spelt without indication of vowel length, as in the following examples. Notice that original vowel quality tends to be indicated308 (although with some interchange between e.g. /u/ and /o/, /a/ and /e/; and /ia/, /ua/ > /a/; /әː/ > /a/, /o/, /u/): Trinaid ‘Trinity’, G. Tríonóid smarage ‘ember’, G. sméaróid saveen ‘slumber’, G. sáimhín Also with stress retraction ˈfe̜ gɛlʹ, ˈfʹɛːɣal (JTK). Some deviations from this can be explained by the consonantal environment, e.g. rollage ‘star’ from G. réaltóg, where initial [rˠ] results in a back quality of the following vowel, represented by <o> (cf. roie ‘run’, G. rith; roih ‘arm’, G. righ) (cf. §4.2.4). 307 308 373 Aspects of the historical phonology of Manx cronnane ‘purring’, G. crónán suggane ‘straw rope’, G. súgán sonnaase ‘arrogance’, G. saobhnós (§3.9.11) caraig ‘beetle’, G. ciaróg carrane ‘sandle’, G. cuarán However there are other items such as faasaag ‘beard’ (G. féasóg), peeaghane ‘hoarseness’ (G. piachán), among others, where the spelling could reasonably be interpreted to allow for maintenance of length or diphthongization in the initial syllable, perhaps in careful speech. G. féas, *piach have no attested independent existence in Manx. There are also examples of transparent formations with the semiproductive suffix -oil /ˈoːlʲ/ (G. -(e)amhail, Ir. -(i)úil, ScG. -ail, -eil), such as gloyroil ‘glorious’ (G. *glóireamhail, cf. Ir. glórmhar, ScG. glòrmhor), which might reasonably be expected to retain some length (and perhaps secondary stress?), i.e. [ɡlo(ː)ˈroːlʲ]. Items such as thousane ‘thousand’ (tøuˈzẹːn EKh, to̜ uˈze̜ ːn NM, touˈzaːn JK, HLSM II: 450), sidoor ‘soldier’ (G. saighdiúir), lourane ‘leper’ (G. lobhrán), boirane ‘troublesome person’ (G. buaidhreán), in which a diphthong is, or may be, retained in the initial unstressed syllable, show that bimoraic syllables were tolerated in this position, although they are not found post-tonically.309 The evidence presented here shows that pretonic original long vowels and diphthongs (a) retained their quality in the Classical and Late Manx periods to a sufficient degree to be recognized in orthography, and (b) may, in some cases at least, have (optionally) retained their duration. This is in marked contrast to the treatment of post-tonic closed syllables, where all long vowels are reduced either to /a/ or /ә/ (and /e/?), the only vowels which can occur in this position. This differential treatment between pretonic and post-tonic original long vowels might point towards the two distinct shortenings required by Green’s hypothesis (§5.1.1.6), one preceding stress shift and the other following it. 309 However, these seem to have often been reduced to monosyllables and even schwa in production, e.g. sidoor, səˈduːr TC (HLSM II: 407), sidžūryn [siˈdʒuːrәn] (Rhŷs notebook 6: 58). Phillips’ spellings appear to represent both diphthongal and monophthongal realizations: sajúr, sêidjúr, (pl.) sajúryn, sajuryn (3), sejúryn, seidjúryn. 374 Aspects of the historical phonology of Manx On the other hand, the peak delay account of stress shift discussed above (§5.1.2), whereby ˈHH words became HˈH or LˈH could perhaps involve an intermediate stage with secondary stress retained — or at least some form of prominence — on the initial syllable, i.e. ˌHˈH or ˌLˈH, which might explain retention of vowel quality and length in the initial syllable. At any rate, it is not implausible that there should be some difference in treatment between vowel shortening in a syllable originally stressed, and in a syllable which had never borne stress. In addition, paradigmatic uniformity in alternations such as faag ~ faagail would provide a motivation for retention of vowel quality and length in the stem when unstressed. This motivation does not exist with regard to post-tonic original long vowels: although certain suffixes have developed stressed and unstressed by-forms such as -án > -an /an/, -ane /ˈɛːn/, there is no alternation between these in a single paradigm. There is some evidence of complete loss of the pre-tonic unstressed vowel (Rhŷs: 15, 21, 43; O’Rahilly: 115), as found more widely in Irish dialects (Ó Sé 1984). In a handful of items loss of the unstressed vowel is shown in the orthography. plaase ‘palace’, G. pálás praase (HLSM II: 354), puddase (Cr.), potase (K.) ‘potato’, cf. Ir. práta streipe (Cr.) ‘stirrup’, G. stíoróip (O’Rahilly: 115) farrane ‘spring, fountain’, G. farrane, fy̆ rā ̣n [fәˈrɛːn], frā ̣n [frɛːn] (Rhŷs: 15) carrane ‘sandle’, G. cuarán, cy̆ rā́ ṇ [kәˈrɛːn], crā ̣n [krɛːn] (Rhŷs: 15, 21), kʹəreːᵈn, kʹaˈreːnən SK, kəˈrɛːn J:EK, krɛː̃ nən TC, kre̜ ːnən JTK (HLSM II: 60) Mylecharaine, surname, G. Mac Giolla Chiaráin …ch(y̆ )rā̃ ̣ñ [x(әˈ)rɛː̃ nʲ] (Rhŷs: 15, 43), only without syncope in HLSM (II: 490): moləkɑˈreːn TT, ˈmɑlikaˈrɛːᵈn, mɑ ̣ləkɑ ̣ˈrɛːᵈn JW 5.1.4 Long vowels arising from vocalization of fricatives It is clear that first syllable conditioning does not apply to the category of items where long vowels in non-initial syllables result from Ó Sé’s (1991) ‘contracted words’, since in most cases these have forward stress regardless of the weight of the initial syllable. 375 Aspects of the historical phonology of Manx The general development of such items may be outlined as follows, with svarabhakti assumed in original clusters such as /rv/, /rʲç/: G. dearmhad > Manx jarrood ‘forget’ /ˈdʲer(ә)vәd/ > [ˈdʲarәwәd] > /dʒaˈruːd/ G. airchis > Manx erreeish ‘compassion’ /ˈarʲ(ә)çәʃ/ > [ˈarʲәjәʃ] > /eˈrʲiːʃ/ The following is an exhaustive list of ‘contracted’ items with forward stress the etymology of which can be securely determined:310 Historical light311 initial syllable: annoon /aˈnuːn/ ‘weak’, G. anbhfann anugh (Ph.) /iˈnuːx/ ‘timely’, G. ionbhadhach arroogh, erroogh /aˈruːx/ ‘chimb’, G. *earrbhach, ScG. earrach, EIr. errbu (eDIL) berreen /beˈrʲiːn/ ‘cake’, G. bairghean karús, karus, karrýus, karryûs (Ph.) /kaˈruːs/ ‘Lent’, CM Cargys /ˈkarɡәs/, G. Carghas312 erreeish /eˈrʲiːʃ/ ‘compassion’, G. airchis elúyn (Ph.) /eˈlʲuːnʲ/ ‘nurture’, G. aileamhain (CM ellyn; §5.1.4.1) ferroogh /feˈruːx/ ‘eyelid’, G. forbhrú etc.313 inneen /iˈnʲiːn/ ‘daughter, girl’, G. inghean jarrood /dʒaˈruːd/ ‘forget’, G. dearmhad kiannoort /kʲaˈnuːrt/ ‘governor’, G. ceannphort kynoauin (K.) ?/kʲiˈnʲoːnʲ/ ‘fate’, G. cinneamhain muinneel /muˈnʲiːlʲ/ ‘sleeve’, G. muinchille parús, parus (Ph.) /paˈruːs/ ‘paradise’, G. parrthas, parrdhas314 310 Pandoogh (Cr.), pantoogh (K.) ‘pant’ probably has /-uːx/ on the pattern of buirroogh, mhinnoogh, strinnoogh. Khennoogh (Cr.) ‘carping, cavilling’ might be similarly explained as a derivative of G. cáin. Ladoose (K.) ‘thrift, industry, economy’ may correspond to Ir. ládas ‘self-will, obstinacy’ with /uːs/ on the pattern of tarroose (PSD: 15) ‘industry’, from tarroogh. An item of similar shape, khyrloghe, translating ‘brokenhanded’ (Leviticus 21. 19), ‘unsound, carious’ (Cr.), ‘benumbed with cold, torpid’ (K., s.v. kyrloghe) is probably a compound containing lámhach (first element corr-, cearr-, crith-?). 311 It might be pointed out that the svarabhakti assumed in earlier forms of these words would give an initial heavy syllable (cf. calmane, G. colmán), i.e. [ˈ[dʲarәv]әd] . But it is unclear how the weight conferred by this could be retained once the svarabhakti vowel + fricative had coalesced as a new long vowel [dʲaruːd]. 312 See fn. 94. 313 See O’Rahilly 1942b: 216–7. 314 CM pargys, pargeiys (Thomson 1995: 115). 376 Aspects of the historical phonology of Manx shilleeid (Cr.), shelleed, shilleed (K.), /ʃeˈlʲiːdʲ/ ‘slug, snail’, G. seilchide strinnoogh /striˈnuːx/ ‘snore’, G. srannfach, srannfadhach tarroogh /taˈruːx/ ‘thrifty, industrious’, G. tarbhach, tairbheach thallooin /taˈluːnʲ/ ‘earth, land’ (gen.), G. talmhain Historical heavy initial syllable: buirroogh / buˈrʲuːx/ ‘roar’, G. búirfeadhach brooightooil /bruxˈtuːlʲ/ ‘belch’, G. brúchtghail carnoain (Cr.) /kʲarˈnoːn/, kiornane (K.), ‘large beetle’, G. cearnabhán etc.315 fegooish /feˈɡuːʃ/ ‘without’, G. (f)éagmhais jymmoose /dʒiˈmuːs/ ‘wrath’, G. díomdha + as imnea /imˈnʲeː/ ‘concern, anxiety’, G. imnidhe etc. jeanúgh, jeanugh (Ph.) /dʒeˈnuːx/ ‘maker’ G. déanmhach, déanmhaidh + ach? kegeesh /keˈɡʲiːʃ/ ‘fortnight’, G. cóicthigheas lannoon /laˈnuːnʲ/ ‘twin’, G. lánamhain mennuigh (K.), mhinoogh (Cr.) /meˈnuːx/ ‘yawn’, G. méanfach, méanfadhach sheeloghe /ʃiˈloːx/ ‘generation’, G. síolbhach smooirooil (Cr.) /smuˈruːlʲ/ ‘smile, smirk, titter’, Ir. smúraíl (ÓD) Similar developments can also be seen in Manx surnames (e.g. Kerruish /kʲeˈruːʃ/, G. Mac Fhearghais, Quilliam 1989: 65) and place-names (e.g. Barrule /baˈruːl/, Norse varða-fjall, vörðu-fjall, PNIM III: 307, IV: 62). There is also a stressed plural termination -(t)eenyn which can be traced to medial fricative vocalization: Creestee /ˈkriːsti/ ‘Christian’, G. Críostaidhe Creesteenyn /kriˈstiːnәn/ ‘Christians’, G. Críostaidhe + anna + an annym /ˈanәm/ ‘soul’, G. anam anmeenyn /anˈmiːnәn/, G. anam + ?adha + anna + an The reconstruction is conjectural however, and there is likely to have been a degree of reanalysis and restructuring, cf. the complex array of plural suffixes and extensions and combinations thereof found in Connacht Irish dialects (e.g. Ó Curnáin 2007: 659– 315 Cregeen’s form points first to shortening of -án to /an/, as regularly in trisyllables, i.e. */ˈkʲarnәvan/, followed by vocalization and stress shift to /kʲarˈnoːn/. Kelly’s form suggests either a contracted form *cearnán, with stress shift as in other heavy-heavy items, or modification of the termination /ˈoːn/ to /ˈɛːn/ by analogy with other diminutives in -ane. Both forms are attested in HLSM (II: 60): koˈnɛːn, kʹŏˈnã ̣ːn TC, kəˈnoːn W:S. 377 Aspects of the historical phonology of Manx 881). This termination is found in the following nouns (all in Cregeen and/or the Bible): annym ‘soul’, G. anam, pl. anmeenyn Creestee ‘Christian’, G. Críostaidhe, pl. Creesteenyn gioal ‘pledge, mortgage’, G. geall, pl. gioalteenyn glione ‘valley’, G. gleann, pl. glionteenyn jaghee ‘tithe’, G. deachmhadh, pl. jagheenyn keeill ‘church’, G. cill, pl. kialteenyn (G. ceall-) lieen ‘net’, G. líon, pl. lieenteenyn raane ‘surety’, G. ráth, ScG. ràthan, pl. raanteenyn According to Cregeen, some of these have a variant plural ending –(t)eeyn, which presumably represents unstressed /-ti.әn/. Note also other derivatives such as raanteenys ‘suretiship’, jagheenys ‘to tithe’. A similar formation is naboonyn /naˈbuːnәn/ ‘neighbours’ (also nabooyn),316 sg. naboo /ˈnɛːbu/ (ScG. nàbaidh), also naboonys ‘neighbourhood’. Finally, there is the adjective-forming suffix G. -amhail, which has two reflexes in Manx, stressed -oil /ˈoːlʲ/ and unstressed -al /al/. The following list is restricted to words appearing in the Bible: -oil (also -o(i)lagh) historical heavy initial syllable: baasoil ‘deadly’ (G. bás), breeoil ‘powerful, effectual’ (G. brígh), craidoilagh ‘mocking’ (G. cnáid), eadolagh ‘jealous’ (G. éad + amhail + ach), feohdoil ‘hateful, abominable’ (G. fuath), floaoil ‘fluent’ (Eng. flow), foayroil ‘favourable’ (G. fábhar), gerjoil ‘joyful’, gerjoilagh ‘comfortable’ (G. gaird-), gloyroil ‘glorious’ (G. glóir), graihoil ‘loving’ (G. grádh), graysoil ‘gracious’ (G. grás), laaoil ‘daily’ (G. lá), reeoil ‘royal’ (G. rí), sayntoilagh ‘covetous’ (G. sainnt), schleioil ‘skilful’ (G. ?gleo), sheeoil ‘peaceful’ (G. síth), slayntoil ‘wholesome’ (G. sláint), staydoil ‘stately’ (G. stáid), trocoil ‘merciful’ (G. trócaire) historical light initial syllable: boggoil (G. bogadh), chymmoil ‘compassionate, pitying’ (G. time), creeoil ‘hearty, courageous’ (G. croidhe), ennoil ‘beloved’ (G. TC’s form ˈnaːbunən must be a blend of the two variants; whereas the same speaker has the expected forward stress in nɑˈbuːnəs ‘neighbourhood’ (HLSM II: 318). 316 378 Aspects of the historical phonology of Manx aithne), messoil ‘fruitful, fertile’ (G. meas), paittoil ‘pestilential’ (G. pait), peccoil ‘sinful’ (G. peaca), raahoil (G. rath) polysyllabic stems: onnoroil ‘honourable’ (G. onóir), spyrrydoil ‘spiritual’ (G. spiorad) -al historical heavy initial syllable: booisal ‘thankful, pleasing’ (G. buidheachas), cairal ‘just, righteous’ (G. cóir), kenjal ‘kind’ (Eng., cf. Ir. cineálta), pleasal ‘pleasing’, pooaral ‘powerful’ historical light initial syllable: aghtal ‘clever, capable’ (G. acht), baghtal ‘clear, evident’ (G. beacht), blaystal ‘tasty’ (G. blasta), costal(agh) ‘costly’, cronnal ‘evident, conspicuous, famous’ (G. cron), dunnal ‘courageous, valiant’ (G. duine), gennal ‘merry, glad’ (G. gean), meshtal ‘drunk’ (G. meisce), niartal ‘mighty’ (G. neart) The unstressed form -al is not found in Phillips, where booisal, dunnal, gennal, meshtal, niartal all appear with stressed -oil (Thomson 1953: 33–4).317 Although there may be some tendency towards -oil with heavy initial syllables and -al with light initial syllables, there is no consistent pattern, and the evidence of Phillips seems to suggest forward stress with -oil across the board as with other ‘contracted’ words, with forms such as niartal being later developments. The division between -oil and -al may have developed by analogy with the reflexes of the verbal noun ending -áil, stressed -ail, -eil /ˈɛːlʲ/, unstressed -al /al/, where the long vowel is original. It may be significant that the adjectives with -al can in general be characterized as more everyday, registerneutral lexis, whereas many of those with -oil appear to belong to a more literary higher register, including terms of religious significance such as peccoil, gloyroil, feohdoil etc., and several (gloyroil, onnoroil) appear to be new formations in Manx unattested in other Gaelic varieties (glórmhar is Ir./ScG. for ‘glorious’, for example). Onnoroil and spyrrydoil have stress on a third syllable, whereas there is usually shortening in such cases (§5.1.5.5). It appears that -oil and -al remained as semi-productive suffixes 317 But Ph. duynalys ‘humanity’, CM dunnallys, alongside adjective duinôîl; Ph. gannylys, ganlys ‘gladness’, CM gennallys, alongside adjective ganoil etc. 379 Aspects of the historical phonology of Manx in Manx and therefore these items cannot be taken as evidence of the general pattern in words containing secondary long vowels from vocalized fricatives. Green (1997: 90–3) claims that items with historical medial fricatives are treated differently according to whether the fricative was originally intervocalic (e.g. geanamhail, Manx gennal) or was part of a consonant cluster (e.g. dearmhad). The former type, in this account, developed an unstressed long vowel (>*/ˈɡʲenoːlʲ/) at an early stage, and thus were treated like words with historical post-tonic long vowels. This implies that such items would be subject to initial syllable conditioning, and that the initial stress in gennal is explained by the lightness of the initial syllable (as with the beggan type). However, this overlooks the evidence regarding the -amhail suffix discussed above. Besides these adjectives, other items with the geanamhail pattern are rare, and two of them (bunnys < bunadhas and ellyn < aileamhain) appear to support Green’s claim. However, as discussed below, these may be special cases (§5.1.4.1). On the whole, then, it appears there is no strong evidence for a difference in treatment between the geanamhail and the dearmhad types. 5.1.4.1 Exceptions The two exceptions to the general rule that ‘contracted words’ show stress shift regardless of initial syllable weight (apart from -amhail adjectives) are bunnys ‘almost; most’ (G. bunadhas) and ellyn ‘behaviour’ (G. aileamhain).318 Both of these cases can be explained by lexicalization of post-lexical light stress and/or stress retraction in collocations. As an adverb ‘almost’, bunnys would often have light sentential stress, and as a noun ‘majority’ it would often be followed by a genitive or prepositional phrase bearing greater stress: 318 For this etymology see Thomson (1953: 205), Ó Maolalaigh (2006: 72). Broderick (HLSM II: 145) derives it less plausibly from ealadhain ‘art, craft, skill’ which would be expected to give */al-/ or */ol-/ (§2.1.3) rather than /elʲ-/. The only attested transcribed form is u̜ lʹən (JTK), which has the expected slender /lʲ/ but apparent confusion in the initial vowel. The semantic development from aileamhain ‘nurturing, upbringing’ to ‘good / bad manners, behaviour’ also seems more straightforward. 380 Aspects of the historical phonology of Manx ˌbunnys ny Creeˈsteenyn (SW: 178) yn ˌbunnys dy Chreeˈsteenyn (SW: 183) ‘most Christians’ The noun ellyn ‘behaviour, manners’ is rarely found outside the collocations ellyn mie ‘good behaviour’ and drogh-ellyn ‘bad behaviour’. Certainly in the first of these, and possibly the second, heavier stress on the adjective would be expected, i.e. ˌellyn ˈmie. Phillips has the latter phrase as ellyn mei (Moore and Rhŷs 1895 I: 452), showing the prosody of the later form, as well as elúyn on its own translating ‘nurture’ (Moore and Rhŷs 1895 I: 414), which appears to show the expected development with forward stress on a long vowel /eˈlʲuːnʲ/. It is possible that there had been a semantic split between these two forms, and that they were no longer recognized as the same lexeme. 5.1.5 First syllable conditioning: further details 5.1.5.1 Heavy sonorant clusters Ó Sé (1991: 162) describes the following category of initial syllables as counting as heavy and causing stress shift: Words in which a short vowel is followed by a cluster of sonorant consonant + voiced consonant (e.g. ordóg ‘thumb’) have tended to be treated in Manx like words with a long vowel in the first syllable (e.g. fág(bh)áil ‘leaving’). Vowels preceding such syllables were regarded as half long (síneadh meadhónach) by mediaeval Irish grammarians (Greene 1952). Clusters which are not voiced throughout do not have this lengthening effect on the first syllable (e.g. altóir). (Ó Sé 1991: 162) However, it appears that some combinations of sonorant + (historically) voiceless consonant can also count as heavy and cause stress shift. Ó Sé includes some of these in his ‘VRC’ (‘Short vowel + cluster of sonorant consonant and voiced consonant’) category, presumably on the strength of the Classical Manx spelling, namely chyndaa ‘turn’ (G. tiontódh),319 undaag(agh) ‘nettles’ (G. neanntóg). 319 Medial voicing may be early here; Philips has <nd>, and cf. ScG. tionndadh. 381 Aspects of the historical phonology of Manx cheinjean ‘bonfire’, G. teinteán chyndaa ‘turn’, G. tiontódh gamshoge (K.) ‘buffoon’, G. gaimseog injeig ‘paddock, enclosure’, G. inseog minjeig ‘bundle of heather; she-goat, young hind’, G. minnseóg molteyr ‘deceiver’, G. mealltóir undaagagh, ondaagagh ‘nettles’, G. neanntóg pundaig (Cr.) ‘hard stem of grass’, cf. Ir. puntán Stress shift is also found irregularly after certain other clusters: cartage ‘gadder’, G. cart ‘tan, scrape clean, clear away’ + óg? gorteog (K.) ‘stingy woman’, G. goirteóg (perhaps after Shaw 1780) fockleyr ‘dictionary’, G. foclóir kercheen ‘underling’, G. ceirt ‘rag’, ceirtín (eDIL), cf. ceirteachán (ÓD) kishteig, kishteen (K.) ‘casket’, cisteóg, -ín raghtaneys (Ecclesiasticus 10. 21), ‘roughness’, G. r(e)acht + án + as shughlaig (Cr.), shughlage, shulchaag (K.) ‘sorrel’, cf. G. sealgán etc., ScG. sealbhag, samhrag But the same clusters may also be followed by vowel shortening: braghtan ‘bread with butter etc.’, G. breachtán partan ‘crab’, G. partán carthan ‘tick’, G. (s)c(e)artán 5.1.5.2 Exceptions Initial stress and post-tonic vowel shortening in heavy-heavy items: aashag ‘seat to rest on, a seat made of matted straw’, G. áis + óg milljag (Cr.) ‘a sweet drink, ale before the hop is added, mead’, G. milseóg muiltchin (Cr.) ‘two year old mutton’, muiltin (K.) ‘eunuch’, G. muiltín neaynin (K.) ‘daisy’, G. nóinín runtag (Cr.) ‘round lump of a thing’, Manx runt, Eng. ‘round’ + óg scoidan (Cr.) ‘sheet of sail’, G. scód, ScG. sgòdan skeaban ‘(small) brush’, G. scuabán stramlag (Cr.) ‘crankled or awkward thing’, G. sraimleóg strumpag ‘strumpet’, cf. ScG. strùmpaid teaystag ‘dumpling’, G. taos + óg tholtan ‘ruin’, G. *tolltán 382 Aspects of the historical phonology of Manx Some of the above exceptions may be late formations from semi-productive use of the unstressed diminutive suffixes (as also with some of the occurrences of stressed terminations below, cf. Blankenhorn 1981: 245), or may represent stress retraction, or uniformity with the transparent stem (teayst ‘dough’, skeab ‘brush’). Strumpag is evidently adapted from English ‘strumpet’. Exceptions – stress shift in light-heavy items: cliegeen ‘jewel’, G. cleitín falleays (Cr., K.), falleish (Bible) ‘gleam of light’, EIr. folés, ScG. faileas fynneig ‘whiting’, G. fionnóg (HLSM II: 181) fedjeen ‘feather of arrow’, G. eite + ín ke(e)illeig (K.) ‘pollock, whiting’, ScG. caileag? lheibeidjagh ‘unwieldy’, G. leibéiseach peajeog (Cr.), pitteog (K.), piddeog (Ecclesiasticus 31. 24) ‘miser, churl’, G. piteóg putage (K.) ‘pudding’, G. putóg pyshage (K.) ‘mew of cat’, G. pis + óg robaig, roibage (Cr.) ‘whisker’, G. ribeóg, ScG. ribeag, roibean For lack of shortening in -een in cliegeen, fedjeen, kercheen, we may perhaps compare maintenance of length in this suffix in certain Ulster Irish varieties (Ó Dochartaigh 1984: 48–9). The word fynneig ‘whiting’ (HLSM: 181), which does not appear in the dictionaries, is doubtful. It is from Ewan Christian of Peel, a ‘semi-speaker’ (Broderick 1999: 5) ‘who first learned Manx from two old ladies in the same street when he was about five years old, and later from farmers and fishermen in and around Peel’ (ibid.: 75). Christian was apparently well-acquainted with ‘the Manx names of various birds and fish’ (ibid.: 75), so his information may be genuine, but the form recorded fɪˈnʹeːg with medial slender /nʲ/ suggests f(h)ynneig (Cr.), finneig (K.) ‘pod, capsule, small skiff’, i.e. G. fíneóg, an item with an expected original heavy initial syllable. The nouns bwaane (G. bothán) and bwaag (G. bothóg), which might have been expected to show vowel shortening (i.e. /bohaːn/ > */bohan/ > */boː.an/), as in crooag ‘maggot’, G. cruimheóg (/ˈkruṽʲoːɡ/ > /ˈkruṽʲaɡ/ > /ˈkrũː.aɡ/), appear rather to have 383 Aspects of the historical phonology of Manx early loss of intervocalic /h/ (</θ/), and retention of length on the suffix, with reduction of /o/ to non-syllabic /w/ (/ˈbohaːn/ > /ˈbo.aːn/ > /bwɛːn/). 5.1.5.3 Stress-shift in items with fricative vocalization in initial syllable: There is forward stress in a number of items with original short vowel + fricative in the initial syllable; it is not certain whether this is because of the long vowel or diphthong resulting from the vocalization of this fricative, or because the original medial clusters (/ṽr/, /vl/ etc.) were heavy as in the sonorant-initial clusters discussed above: abane ‘ankle’, G. adhbhrann, *adhbarn, *adhbán (§3.4.7.4) arrane ‘song’, G. amhrán farrain (Cr.) ‘wild parsnip’, G. feabhrán, odhrán gollage ‘pitchfork; earwig’, G. gabhlóg jiulean ‘small farmer’, G. deidhbhleán lourane ‘leper’, G. lobhrán (but also louyran ‘small castling’, Cr.) lhemeen, lhemyn (Cr.) ‘moth’, G. leaghman etc. liehbage (Cr.), liabage (K.) ‘flounder, fluke’, G. leadhbóg onnane ‘thistle’, G. fo(bh)thannán etc. 5.1.5.4 Verbal nouns in -ail, -eil, -al (G. -áil) In native items and older borrowings first syllable conditioning can be detected with the verbal noun forming suffix -áil,320 as shown in the following examples: light initial syllables, -al: brebbal ‘kick’, G. breabáil chebbal ‘offer’, Eng. ‘chap, cheap’ toiggal ‘understand’, G. tuigbheáil laccal ‘lack, want’ soghal ‘sob, groan’, Eng. ‘sough’ troggal ‘build, raise’, G. tógáil, ScG. t(r)ogail 320 For the history of this and related morphemes, see Ó Cuív (1980). See also §4.4.7. 384 Aspects of the historical phonology of Manx heavy initial syllables, -ail, -eil: baarail ‘spend’, Eng. ‘wear’ faagail ‘leave’, G. fágbháil farrail ‘fare’ pointeil ‘appoint’ sauail ‘save’, G. sábháil waiteil ‘wait’ Later -al becomes the ‘the grand Manksifier-general of English verbs; as, trying, TRYAL; fixing, FIXAL, &c., &c.’ (Cregeen: ix, original emphasis), and is used productively in numerous loans irrespective of initial syllable weight: dreamal, campal, layal, spiceal, walkal, plantal, weighal (all in Bible) There is evidence of fluctuation between stressed and unstressed reflexes of this ending in manuscript and non-standard sources, as in blakal : bla-caile ‘stare’ (Thomson 1995: 131), K. blakail (usu. blakey, cf. G. spléachadh, ScG. spleuchdadh). 5.1.5.5 Items with original heavy third syllables In general heavy third syllables do not attract stress, but show vowel shortening, irrespective of the weight of the preceding syllables: bwinnican ‘egg yolk’, G. buidheac(h)án, with influence from buinne, ScG. buidhean? feayragan ‘fan, parasol’, ScG. fuaragan foillycan (Cr.), folican (K.) ‘butterfly’, G. féileacán fynnican ‘egg-white’, G. *fionnacán, cf. Ir. gealacán; ScG. fionnagan ‘crowberry’ Jurynan, orynan (Ph.) ‘Jordan’, G. Eórthanán, Orthan(n)án (cf. Thomson 1995: 127) laaraghyn, laueraghyn (K.), but loagh(t)rane (Cr.), ‘handle of flail’, ScG. làmhrachan, Ir. lámhchrann lhaihaghan (K.) ‘lecture’, G. *léigheachán lheunican (Cr.), lionican (K.) ‘sty (on eye)’, ScG. leamhnagan, Ir. sleamhnán (ny) lomarcan ‘alone’, cf. G. lomrachán Manninan, name of legendary figure, G. Manannán monnaghan (K.), ‘a fat greasy fellow, a bloated monk’ (K.), manachán 385 Aspects of the historical phonology of Manx nieeaghyn, niaghyn ‘washing’, G. nigheachán ommidan ‘fool’, G. amadán oohagan (K.) ‘custard’, G. ughagán panshaghan (K.) ‘paunch’, ScG. painnseachan shommarcan (K.) ‘primrose’, cf sumark (Cr.), Ir. samhaircín, EIr. sobaircín ‘primrose’, seamróg ‘shamrock’, ScG. samhaircean, seòbhrach ‘primrose’ tuarystal, tooarystal ‘shape, appearance’, G. tuarascbháil ynrycan, ynrican ‘only’, G. aonracán Also all items with the agentive suffix -eyder /әderʲ/, G. -adóir, such as fuinneyder ‘baker’ (G. fuineadóir), kiaulleyder ‘musician’ (G. ceól + adóir), coyrleyder ‘advisor, counsellor’ (G. comhairle + adóir), ooashleyder ‘worshipper’ (G. uaisle + adóir). Some original trisyllables have bisyllabic forms by syncope: cughlhin (Cr.) ‘cone’, ?cochall + ín corlan, curlan ‘earthnut, pignut’, G. cúlarán etc., ScG. cutharlan creoghan (K.) ‘gadfly; harsh creditor’, G. cruadhachán earkan ‘lapwing’, G. adharcán foldyr, foldyr ‘mower’, ScG. fàladair loagan ‘stagger’, G. lámhacán, ScG. làmhagan mwatlag ‘welk’, ScG. maighdealag, Ir. maighdeog oghsan ‘rebuke, reproof’, G. achmhasán, EIr. athchomsán oikan (Bible, Cr., K.); oinkan, inkan (Cr.) ‘infant’, G. *naoidheacán, *naoidhneacán, ScG. naoidheachan roagan (Cr.), raucan (K.), ‘scallop’, G. ?*rothacán udlan ‘swivel’, G. udalán Sometimes there is final stress in synchronically bisyllabic items, presumably as a result of early syncope: phadeyr ‘prophet’, G. fáidheadóir scrudeyr ‘writer, scribe’, G. scríobhadóir Parlane ‘Bartholomew’, G. Parthalón Synchronically heavy third syllables are found only in loanwords and derivations involving stressed suffixes: emperúyr (Ph.) ‘emperor’ offishear ‘officer’ 386 Aspects of the historical phonology of Manx onnoroil ‘honourable’, G. onóir + amhail spyrrydoil ‘spiritual’, G. spioradamhail 5.1.6 Quality of post-tonic shortened vowels There is regularly /a/ in the reduced suffixes -an, -ag, -al judging by the orthography and by frequent occurrence of [a] in the phonetic data, although there is sometimes also reduction to [ә]. In some final syllables with shortened long vowels, the Phillips orthography appears to show /a/ which may have been reduced to /ә/ by the Classical Manx period: cassid ‘accuse’, Ph. kassad, kasaid (4), ghassaid, G. casaoid, ScG. casaid /kasadʲ/ Final -adóir is regularly reduced to -eyder in the Classical Manx orthography, which appears to suggest /e/ rather than /ә/ or /a/. 5.1.7 Irregular stress in reeriaght ‘kingdom’ Unexpected stress patterns may be lexically conditioned in certain instances. Notably, reeriaght ‘kingdom’ is found as /riˈriː.axt/ in Late Manx (Rhŷs: 166; HLSM II: 364), perhaps from a blend of ríoghacht and ríoghraidheacht. The influence of the rhythm of reciting the Lord’s Prayer may also be relevant (cf. dty ennym ‘thy name’ /dә ˈenәm/, dty aigney ‘thy will’ /dә ˈaɡʲnʲә/, dty reeriaght ‘thy kingdom’ /dә riˈriː.axt/). 5.2 Apocope Loss of final /ә/ has been noted especially in Manx and Scottish Gaelic (O’Rahilly: 138–9; Watson 1985: 128), although it also more sporadically occurs in Irish (e.g. Ó Curnáin 2007: 117–19). In Manx final /ә/ is usually retained in bisyllables, as in e.g. arrey ‘watch’, G. aire caashey ‘cheese’, G. cáise 387 Aspects of the historical phonology of Manx lurgey ‘leg’, G. lurga kiuney ‘calm’, G. ciúine sniaghtey ‘snow’, G. sneachta thanney ‘thin’, G. tana It tends to be lost in bisyllables after sonorant clusters /RN/, /RL/, /RLʲ/. However, it is retained baarney ‘gap’ (G. bearna), farney ‘black-alder’ (G. fearn(a)). arn /aːrn/ ‘sloe’, G. airne Baarle /beːrl/ ‘English language’, G. Béarla chiarn /tʃaːrn/ ‘lord’, G. tighearna coyrle /kõːrlʲ/ ‘advice’, G. comhairle oarn /oːrn/ ‘barley’, G. eórna As observed by O’Rahilly (138), apocope in Manx is more widespread in items with more than two original syllables (this category may include coyrle, chiarn above): aghin ‘petition’, G. athchuinge Boaldyn ‘May’, G. Bealltaine chaghter ‘messenger’, G. teachtaire eshlyn, eshlys ‘shroud’, G. eisléine feanish ‘witness, evidence’, G. fiadhnaise feysht ‘question, examine’, G. faoiside immyr ‘bed or butt of land’, G. iomaire kemmyrk ‘refuge’, G. coimirce Lhunys ‘Lammas’, G. Lughnasa maggle ‘testicle’, G. magairle magher ‘field’, G. machaire muinneel ‘sleeve’, G. muinchille roayrt ‘springtide’, G. rabharta, robharta, ScG. reothairt sharvaant ‘servant’, G. searbhónta shilleeid (Cr.), shelleed (K.), ‘snail, slug’, G. seilchide skibbylt ‘nimble, light of foot’, G. sciobalta sproghil ‘dewlap’, G. sprochaille staghyl ‘clumsy person’, G. stachaille troar, troayr ‘crops’, G. treabhaire 388 Aspects of the historical phonology of Manx The following are variable:321 -aghey, -agh, vn. termination, G. -aghadh, ScG. -achadh enney, enn ‘recognition, knowledge’, G. aithne bochilley, bochil ‘shepherd’, G. buachaille boandyr, boandyrey ‘nurse’, G. banaltra dunver, dunverey ‘murderer’, cf. G. dúnmharbhthóir firriney, firrin ‘truth’, G. fírinne skaaley, skaal ‘flat dish, saucer’, G. scála Items with an original termination -t(h)a, -t(h)e (past participles, old genitives of verbal nouns, etc.) often have apocope, as in: losht ‘burn, burnt’, G. loisc, loiscthe nasht ‘betrothed’, G. naiscthe skeilt ‘cloven’, G. scoilte Jeheiney-Cheays ‘Good Friday’, G. Aoine an Chéasta dooinney-poost, ben-phoost ‘bridegroom, bride’, G. pósta, gen. pósadh sheelt ‘sober’, G. síobhalta, Eng. ‘civil’ skibbylt ‘nimble, active’, G. sciobalta But /-ә/ is sometimes retained: cailjey ‘lost’, G. caillte (of sheep etc., otherwise usu. caillit) custey ‘cursed’, Eng. curse + G. -ta sailjey ‘salted’, G. saillte Note that loss / reduction of the participle ending leads to new forms with regular -it, sometimes reduplicated -jit (cf. Thomson 1970: 149): currit ‘put’, G. cuir, curtha + Manx -it riojit ‘frozen’, G. reóite + Manx -it A few loanwords which in other Gaelic dialects often have an excrescent final schwa (cf. Ó Curnáin 2007: 123–4) lack this in Manx, and some of these are assimilated into native paradigms: 321 As far as can be discerned, these are to be interpreted as variant underlying or citation forms; there appears to be no evidence of the ‘caducous vowel’ found in Scottish Gaelic (Watson 1985; also Borgstrøm 1940: 50). 389 Aspects of the historical phonology of Manx giat ‘gate’, gen. giattey, G. geata (invariable in singular) oast, in thie-oast ‘inn’, fer-oast ‘hostelier’, G. ósta pann, panney ‘pan’, gen. panney (Bible), G. panna pot ‘pot’, pl. pooiyt, G. pota spot ‘spot’, pl. spuitt, G. spota sole ‘threshold’, G. sóla But others have added /ә/: attey ‘crown’, G. hata, Eng. hat, Norse hattr barrey ‘bar’, G. barra cloagey ‘cloak’, G. clóca boandey ‘bond’, G. banda bolley ‘boll’, G. bolla cooiney ‘coin’ foalsey ‘false’, G. fallsa paggey ‘pack’, G. paca sthartey ‘job, spell of work’ (EDD s.v. start 11) thunney ‘ton’, G. tunna tubbey ‘tub’, G. tuba Emphatic suffixes / clitics -sa, -se always have apocope or metathesis: mish ‘I, me’, G. mise uss ‘you’, G. thusa ish ‘she, her’, G. ise shiuish ‘you’, G. sibhse aym’s ‘at me’, G. agamsa my ennym’s ‘my name’, G. m’ainm-se dty egooishys ‘without you’ (CS), G. i d’ fhéagmhais-se The following irregular verb forms have apocope: cheayll, geayll ‘heard’, G. chuala vaik, naik ‘saw’, G. faca ren ‘did, made’, G. rinne Other original bisyllables with apocope: bine ‘drop’, G. boinne drease, dreast (Cr.) ‘after a while’, ScG. an-dràsta failt ‘welcome’, G. fáilte 390 Aspects of the historical phonology of Manx faare ‘close, near’, G. farradh feoilt ‘generous’, G. faoilte(ach) eayst ‘moon’, G. éasca insh ‘tell’, G. innse reaisht ‘span’, G. réise Final -adh (in verbal nouns etc.) is usually retained, except where it coalesces into a long vowel or diphthong resulting from vocalized fricatives. Note that genitive -aidh (Manx -ee /i/) may nevertheless appear: craa ‘shaking’, G. crathadh, gen. craaee loau ‘rotting’, G. lobhadh, gen. loauee screeu ‘writing’, G. scríobhadh, gen. screeuee sneeu ‘spinning’, G. sníomhadh, gen. sneeuee This termination may spread by analogy to other verbal nouns without original -adh: snaue ‘swimming, crawling’, G. snámh, gen. snauee There is variable loss of -adh in freaylley, freayll ‘keep’ (G. friotháladh), and also in the termination -agh(ey) (see above). 5.3 Syncope The following concerns phonologically or lexically conditioned syncope in post-tonic syllables (cf. Watson 1985: 125–6). Syncope within morphological paradigms (as found generally in Gaelic dialects), is not considered,322 nor is syncope in trisyllables with original final heavy syllables (§5.1.5.5), or syncope in pre-tonic syllables (§5.1.3). Syncope is regular in final unstressed -rra(i)n(n), -rtha(i)n(n), -rra(i)ng (cf. O’Rahilly 1942a: 120): ayrn ‘part’, G. earrann faarn ‘rain leaking through roof’, G. fearthain er-mayrn ‘left, remaining’, G. marthain 322 E.g. cossyn ‘win, gain’ (G. cosain), verbal noun cosney (G. cosnadh), or millish ‘sweet’ (G. milis), comparative ny s’miljey (G. níos milse). 391 Aspects of the historical phonology of Manx keirn ‘rowan’, G. caorthann tayrn ‘pull, draw’, G. tarraing yiarn ‘iron’, G. iarann323 Syncope is also found in the following. Note that some items have both contracted and uncontracted by-forms. Some of the contracted forms may be back-formations from syncopated plurals (cf. Thomson 1999: 401–2). faarkey ‘bathe’, G. fothragadh, influenced by faarkey ‘sea’, G. fairrge? fess(t) ‘spindle’, G. fearsaid feysht ‘question, examine’, G. faoiside geaysh ‘hair, fur’, G. gaoiside insh ‘tell’, G. innis jeelt ‘saddle’, G. díollaid, diallaid mooads, mooadys ‘amount’, cf. G. méad + as shleayst, slheeayst, slheeassid ‘thigh’, G. sliasaid sleayst ‘shovel’, G. sluasaid taggloo ‘talk’, G. agallamh yindys ‘wonder’, G. iongantas, Ir. iontas Compare also Yernagh ‘Irish, Irishman’ (G. Éireannach), but Nherin ‘Ireland’ (an / in Éirinn) and Sostnagh ‘English, Englishman’ (G. Sa(c)sanach). 5.4 Epenthesis It has been noted that epenthesis or svarabhakti is more restricted in Manx than in most other Gaelic dialects (Marstrander: 70–1; O’Rahilly: 203; Jackson: 60; Thomson 1960; Ó Baoill 1980: 94, 101–2; HLSM III: 143–5). Notably it is absent in the following sonorant-stop clusters: 323 Originally O.Ir. íarn, with insertion of an epenthetic vowel in Middle Irish; for discussion see O’Rahilly (1942a). It seems more likely that the Manx form represents secondary loss of the vowel in the second syllable, as in the other items listed here, and with reduction of /iә̯/ to /ja/ as in chiarn ‘lord’ (G. tighearna) (perhaps with the motivation of avoiding an over-heavy syllable), than maintenance of original iarn (see §4.6.2). 392 Aspects of the historical phonology of Manx /lb/ scolb /skolb/ ‘chip, break shell; stir’, G. scolb /lɡ/, /lʲɡʲ/ bolg /bolɡ/ pl. builg /bulʲɡʲ/ ‘belly’, G. bolg324 jolg, jialg /dʒolɡ/ pl. jilg /dʒilʲɡʲ/ ‘thorn’, G. dealg tilg /tilʲɡʲ/ ‘throw’, G. tilg /rb/ borb /borb/ ‘harsh, rough, severe’, G. borb /rɡ/, /rʲɡʲ/ jiarg /dʒaːrɡ/ ‘red’, G. dearg farg /faːrɡ/ ‘anger’, G. fearg s’merg /smerʲɡʲ/ ‘woe’, G. is mairg325 Early epenthesis appears in monosyllables in the following clusters consisting of two sonorants, or a sonorant followed by a stop or a voiced fricative, and is already attested in Phillips: /lm/, /lʲm/ Collym /kolәm/ name, G. Colm (HLSM III: 144) hellym /helʲәm/ ‘sounded’, G. seinm (with dissimilation) 324 Ó Baoill (1980: 101–2) notes the anomalous apparent presence of an epenthetic vowel in Manx data from Wagner (LASID IV: 174, 188) in the cluster /lɡ/ where it is otherwise unattested: ‘[ən wulag uʃtʹə] an bolg uisce? [vɔlagən ʃeːdʹə] bolgán séide?’ (Ó Baoill’s interpretations). The clear vowel in these is suspicious as it looks like the diminutive /aɡ/ (G. -óg) rather than epenthesis where /ә/ would be expected. The first of these looks like yn vullag ushtey ‘the water keg’ (G. mullóg) (the feminine gender of mullag would explain the lenition, and the informant may have misunderstood Wagner’s prompt to translate ‘water-bag’, a term relating to the calving of cattle). The second may be confused with trisyllabic bellyssyn ‘bellows’ (Ifans and Thomson 1979–80: 150; HLSM II: 28), and or phonetically similar bollag ‘skull’ (G. ballóg), mullag ‘keg’, or mollag ‘buoy’ (G. meallóg); in any case this is a deviation from expected builg-sheidee (Jeremiah 6:29). Broderick also notes bolg my vaggleyn ‘my scrotum’ (‘the belly / bag of my testicles’, G. magairle), boləg mə ˈvɑgələn TC (HLSM II: 39), suggesting that ‘[t]he epenthetic vowel in bolg may be influenced from the central syllable in vaggleyn, so as to assist in the flow of the phrase’ (HLSM III: 144). Ó Baoill’s (1980: 102) suggested explanation is rather weak: ‘I would favour the view that the forms with epenthesis are the oldest and that the retention of the epenthesis in these forms is due mainly to their semantic relationship with the original stem being obscured or forgotten. What I am suggesting is that the form bolg on its own and in the phrase an bolg uisce may not be related to each other in the native speaker’s mind.’ It should also be borne in mind that Wagner appears to have had a tendency to misanalyse Manx and transcribe ghost features from Irish (§1.5.8). 325 But note Ph. sh’marig etc. The here may represent epenthesis, or perhaps simply palatality? 393 Aspects of the historical phonology of Manx /rm/, /rʲm/ gorrym /ɡorәm/ ‘blue’, G. gorm orrym /orәm/ ‘on me’, G. orm gerrym /ɡerʲәm/ ‘cock-crow’, G. gairm sterrym /sterʲәm/ ‘storm’, G. stoirm /nʲm/ ennym /enәm/ ‘name’, G. ainm (§4.4.3) /nʲb/ kennip /kenʲәp/ ‘hemp’, ScG. cainb, Ir. cnáib /rʲv/ Ph. teryuf /terʲәv/ ‘bulls’, CM terriu /terʲu/, G. tairbh Ph. meirif /merʲәv/ ‘dead’ (pl.), CM merriu /merʲu/, G. mairbh There was presumably epenthesis prior to vocalization of fricatives in e.g. jalloo ‘picture’ (G. dealbh) */dʲalәv/ > CM /dʒalu/, of which Early Manx teryuf etc. is the last remnant (Thomson 1960: 122). When further syllables are added, the epenthetic vowel may be absent:326 e.g. enmyn ‘names’ (ScG. ainmean), enmaghey ‘to name’ (ainmeaghadh), gormid ‘blueness’ (G. goirme + id), stermagh ‘stormy’ (G. stoirmeach). Where there is forward stress there is no epenthesis in a cluster preceding the stressed syllable (Thomson 1960: 121).327 Note the maintenance of /v/ in this position: colmane /kolˈmɛːn/ ‘dove’, G. colmán sharmane /ʃarˈmɛːn/ ‘sermon’, G. searmóin marvaanagh /marˈvɛːnax/ ‘mortal’, G. marbhánach shirveish /ʃirʲˈveːʃ/ ‘service’, G. seirbhís 326 As also with historical vowels, e.g. currym ‘duty’ (G. cúram), curmaghey ‘to charge’; corrym ‘equal’ (G. comhthrom), cormal ‘to compare’. 327 The existence of epenthesis in monosyllables such as gorrym but not in polysyllables with forward stress such as colmane is treated by Ó Baoill (1980: 101–2) as a puzzle in need of a solution. He suggests that epenthesis only occurred in stressed syllables, and so must postdate stress shift. However, he does not note the polysyllabic morphologically complex forms without stress shift which also show absence of epenthesis, such as enmaghey, stermagh etc. Since this category of epenthesis is restricted to original monosyllables, an obvious motivation for the development is to break up the cluster in the complex coda; this motivation would not exist in polysyllables, regardless of their stress pattern, if syllable boundaries fall within the cluster, i.e. col.ˈmane, ˈster.magh. Alternatively, there would be greater motivation for syncopation of the epenthetic vowel in longer words. 394 Aspects of the historical phonology of Manx It is unclear whether epenthesis in the cluster /rx/ was an early development; it is spelt <rgh> in Phillips, but as Thomson notes, this is also sometimes the case where a historical vowel is expected, as in karghey ‘repair, correct’, CM karraghey (G. coireaghadh, ScG. càireachadh). According to Thomson (1960: 121), it first appears in the surname Faragher (G. Mac Fearchair) in 1649 (cf. Moore 1903: 23; Quilliam 1989: 76). /rx/ dorraghey /dorәxә/ ‘dark’, G. dorcha orraghey /orәxә/ ‘bow-shot’, G. urchar (§4.2.2) Similar clusters /Nx/, /nx/ are only found in one item in Ph., apparently without (Thomson 1960: 124), and in a place name: /Nx/, /nx/ kranghyr (Ph.) /kraN(ә)xәr/ ‘lot’, G. crannchor Connaghyn /ˌskʲilʲә ˈkonәxan/ ‘Kirk Conchan’ (PNIM II: 510) IV: 361; HLSM There is epenthesis in /lx/ in one item, although this may have been reanalysed as -achán (cf. the Scottish form): /lx/ ollaghan /olәxan/ ‘treadle of spade’, G. ealchaing, ScG. ealachainn Later, epenthesis appears in original polysyllables in other clusters, in which a stop or a fricative is followed by a sonorant. Some of these are variable in Phillips (Thomson 1960: 124)328 and later, as shown by orthographic and metrical evidence (Thomson 1960: 125; Lewin and Wheeler 2019: 4). This type corresponds to the ‘secondary epenthesis’ characteristic of Munster Irish (e.g. Noyer 1990). 328 Thomson (1960: 124) also notes variation in derived or inflected forms with expected syncopation such as doccaragh ‘toilsome’ (G. docrach), focklyn ‘words’ (G. focla), feeacklyn ‘teeth’ (G. fiacla). In these items, however, it is uncertain whether we have original lack of syncope, epenthesis, or forms reconstructed from the stem. 395 Aspects of the historical phonology of Manx /br/ dobberan /dobәran/ ‘mourning’, G. dobrón /dr/ maarderagh /meːrdәrax/ ‘fornicator, whoremonger’, G. meirdreach mad(y)ran /mad(ә)ran/ ‘morning’, ScG. maidnean? /kr/ accyrys (Bible),329 accrys, /ak(ә)rәs/ ‘hunger’, G. ocras, but EIr. occoras /sr/ fysseree /fisәri/ ‘knowledge’, G. fiosraighe gassree, gadyree, gadyrey /ɡasәri/ ‘heat in bitches’, ScG. gasraidh glasseraght /ɡlasәraxt/ ‘vegetation’, G. glasrach losserey, pl. lossreeyn /losәrә/ ‘herb’, G. lusra /ʃrʲ/ casherick /kaʃәrʲәkʲ/ ‘holy’, G. coisrigthe shesheragh /ʃeʃerʲax/ ‘plough-team’, G. seisreach /xr/ (er-)shagh(y)ryn /ʃax(ә)ran/ ‘astray’, G. seachrán ogh(y)rish /ox(ә)rәʃ/ ‘bosom’, G. fochras Epenthesis is also attested in other medial clusters (Marstrander: 66; Thomson 1960: 120; HLSM III: 144–5), though usually not written (except in the items with -yragh): /nr/ maynrey ‘happy’, maynrys ‘happiness’ /meːnrә(s)/, G. méanar, mendrɑ̜ SK, mɛː̃ ndrə, meːndərəs JW (HLSM II: 293) /mnʲ/ > /mrʲ/ lheimyragh(t) /lʲeːmәrʲax/ ‘leap, jump’, cor-lheimyragh ‘skip’, G. léimneach breimaragh (Cr.), bremmeragh (K.) /bremәrʲax/ ‘fart’, G. broimneach330 329 So consistently in the Bible, but adjective accryssagh ‘hungry’. These may be influenced by the category of verbal / abstract nouns with G. -aireach(t), and words of similar shape such as fynneraght ‘coolness, breeze’ (G. fionnuaracht). 330 396 Aspects of the historical phonology of Manx /mn/ (>/ml/) > /mr/ (apparently a northern development) chymney /tʃimnә/ ‘will’, G. tiomnadh, tʹʃĩmnə, tʹʃɪmːnɑ̜ NM, tʹʃïmərə TC, tʹʃɪmərə HB (HLSM II: 79) famlagh (Bible, Cr.), famyragh (Cr.) /famlax, famәrax/ ‘seaweed’, G. feamnach, famlax, fɑ ̣mlɑ ̣x NM, famləx, fömərəx WA, famərax J:JTK (HLSM II: 158–8) /rɡ/ margey /marɡә/ ‘market’, G. margadh, maːgə and similar; vörəgə TC (HLSM II: 290) An epenthetic vowel may also be inserted in initial /mn > mr/, /mlʲ/, each occurring radically in one item each (actually fossilized eclipsis in the latter case), from Phillips (>[mәn-]) (Thomson 1960: 120) through to Late Manx (>[mәr-]) (HLSM III: 145): /mr/ mraane /mrɛːn/ ‘women’, G. mná, ScG. mnàthan, Ph. mynáyn (5), mynayn (2), mynanyn [sic], mᵊˈreːn(ʹ) S, mrɛːn S (HLSM III: 145), also gen. sg. Ph. myny, myníí (G. mná, but form probably = dat. mnaoi) /mlʲ/ my-leeaney331 (Bible), myleeaney (Cr., K.) /m(әˈ)lʲiә̯nә/ ‘this year’, G. i mbliadhna məˈlʹinə N/S (HLSM III: 145) 5.5 Vowel shortening and lengthening Both shortening and lengthening of vowels are attested in Manx in certain lexical items. The former is often shown in spelling, but the latter is not generally represented and may be a late development. Late spoken Manx is also characterized by not a few alterations in the quantity of stressed vowels […] In a number of words an originally short vowel has been lengthened, e.g. lhiābbee, sniāghtey, fākin, brēh,332 brīshey, bōght, pōbbyl [sic], in contrast to E[arly] Mod[ern] Ir[ish] leabaidh, sneachta, faicsin, breith, briseadh, bocht, pobal, respectively. On the other hand, originally long vowels 331 The initial cluster here may have been reanalysed as preposition my ‘about’ (ScG. mu). This item does not belong here, but rather shows regular lengthening of a synchronically final vowel from historical /Vh/ sequences (cf. §2.2.4). 332 397 Aspects of the historical phonology of Manx or diphthongs have been occasionally shortened, as in freogh, Gailic [sic], geayl, in contrast to E. Mod. Ir. fraoch, Gaoidhealg, gual, respectively; cf. further Manx fidder with Ir. fíodóir (< figheadóir). (O’Rahilly: 118–9)333 Jackson notes that these lengthenings and shortenings are ‘characteristic’ of Manx, and suggests that the length contrast in Manx is not very robust in any case. In principle there are long and short vowels, but the long vowels are sometimes little more than half-long. This is especially true in the case of Common Gaelic short stressed vowels which have been secondarily lengthened in Manx in certain circumstances. These lengthenings are a characteristic feature of Manx as distinct from Irish and Scottish Gaelic. […] On the other hand, equally characteristic, […] is […] the frequent shortening of original éa, ao, aoi, ua, uai. (Jackson: 9–10)334 Broderick’s observations are similar (HLSM III: 122): The long vowels are about three-quarters the length of their counterparts in Irish, especially original short stressed vowels which have been secondarily lengthened. This feature of secondary lengthening is a characteristic of L[ate] S[poken] M[anx] […] Equally characteristic, though not so prevalent, is the proclivity of LSM to shorten original stressed long vowels […]. All such long vowels can be affected by secondary shortening. (HLSM III: 122) The apparent reduction in the length of Manx long vowels noted by Broderick and Jackson could conceivably have contributed to fluctuation in quantity. However, compare Scouller (2017: 235–7) for the difficulties of making generalizations about vowel length. Scouller notes that Colonsay phonemically long vowels are marked as ‘half-long’ in SGDS, but he suggests that this may have been due to the single informant’s ‘clipped’ speech style and ‘in natural speech, vowel length can be extremely variable, and that the listener’s perception of a vowel as ‘long’ or ‘short’ is 333 The examples appear to be taken from Rhŷs. In this passage Jackson (9) also mentions the ‘not very common […] shortening of original stressed long vowels before final ᵈn and ᵇm’, which is discussed under the phenomenon of preocclusion (§4.5.2). 334 398 Aspects of the historical phonology of Manx more important than precise measurement of its duration’. Nevertheless, it is tempting to see the variability of vowel length in Manx as signs of an incipient reanalysis of vowel length distinctions as contrasts of quality on the lines of the English tense-lax contrast; compare also the difficulty of analysing length in diphthongs (§3.9.1). 5.5.1 Vowel shortening Vowel shortening may be found in a number of items. In most cases, the short vowel is shown in the Classical Manx orthography, as in firriney ‘truth’ (G. fírinne), in contrast to feer ‘true, very’ (G. fíor). A number of spellings in Phillips may suggest short realizations also (cf. Rhŷs: 166), although the picture is not entirely clear as vowel length is not marked consistently in this orthography. In addition to these cases, Rhŷs notes a number of examples of shortening of other items where the long vowel is marked in the orthography, including keead ‘hundred’ (G. céad) (Rhŷs: 7), eayl ‘lime’ (G. aol), meayl ‘bald’ (G. maol) (Rhŷs: 21), vooar (G. mhór) (Rhŷs: 67), dy bragh ‘ever’ (G. go bráth) (Rhŷs: 67). This shortening may be variable, as in the following case: Such a word as freoagh ‘heather’ (Goi. fraoch) should be pronounced frȳgh [frәːx] according to analogy, and I have occasionally heard it so, but much oftener it is fry̆ gh [frәx] with a short vowel. (Rhŷs: 18) Broderick (HLSM III: 122–40) gives numerous further examples of sporadic shortening in the Manx of the terminal speakers. These are mentioned by Stockman (1986: 12– 3), who compares them to similar developments in Ulster dialects. Table 107 contains those items where the evidence suggests that shortening was wellestablished in Classical and Late Manx: 399 Aspects of the historical phonology of Manx Table 107. Long vowel shortening Phillips cheddin, keddin; myrgeddin cheet335 ed (3), æd (2), ied (3), iedd CM pronunciation /çedәn/, /mәrˈɡʲedәn/ etymology English HLSM céadna; mar a gcéadna same; likewise i NM, e̜ TC, JW, ï HK /tʃi(ː)t/ tigheacht come /çed/ céad first e̜ HK, JW, ø TC, ɪ JK, NM, i, ɛ JTK, e HB e̜ (HLSM I: 50) /kurәm/ cúram duty, charge currym íít (7), iit (4), ít (2), ît, it (6), itt ie (6), iê, ei (5), êî ur, or, urr (5) eaddagh yd (12) /edax/ éadach clothes earroo er (8), ér (7), err, ær (3), ǽr (5) ydd, æd (2), ed (5), yd (5), yth, éth, eyd, ýd edr (5) /erʲu/ áireamh number ï JK, eː JK, e JTK, NM, e̜ TT e̜ WQ, ɪ TC /edәn/ éadan face ø TC, ɪ NM /edrәm/ éadrom light /enʲax/ éigin(each) some, certain e̜ HK, e JTK, ɛ W:N əˈnʹax NM, ɛnax HB /fiderʲ/ figheadóir weaver ɪ JK, TT, e̜ TT íírr, irr (18), yr, ir (2), ier (2), iyrr (2), iirr, yir, eir od (39), odd (13) /firʲәnʲә/ fírinne(ach) truth; true ẹ TC, ï HK, ɪ NM /fodi/ féad, ScG. faod can, may, maybe phalỳder /foldәrʲ/ ScG. fàladair mower; scythe o̜ JW, HB, ɪ, ɛ NM, ɔ TC, JK, NM, ə NM, o̤ HB o JK (adj. Gaelgagh) gellgah, golgkagh (?gel[Thomson]) /ɡelʲkʲ/, /ɡelʲɡʲ/ Gaoidhealg etc. Manx language chied eddin eddrym ennagh egnagh (3), ægnagh, egyn (3) fidder firriney; firrinagh foddee folder, foldyr, yiarn foldyragh Gaelg (Cr.), Gaelck (K.), Gailck 335 chit frequent in Bible MSS and elsewhere. 400 i JTK, HB, JK, W:N, WQ, NM, ɪ TC, NM, W:S, J:JK, J:NM, e̜ JW, e W:S, ᵘe Fa Aspects of the historical phonology of Manx eatt (4), æatt, eytt, yæth, eadth (4), etd (4), edth (3), eadh, eath (2), êath (3), êâth (2), edtg aink (16), áink (8) eid (2), éid, ǽid, eidg (2), æids, æ̀ dj, ǽd, éad, edg (2), edj (2), ed, id /ɡʲedәn/, /fedәn/ ScG. faotainn get, find ø TC, e NM, JK, JTK, HB, ɪ NM /heŋʲkʲ/ tháinig came /hedʲ/, /hem/ théid, théighim, théigh muid will go irree irr (27), ir̆ r̆ , írr, ir, írr /irʲi/ éirghe rise jean, jannoo (jean) ean (28), éan, eán, (jannoo) an (9), ān (2), ean (4), ian /dʒen/, /dʒenu/ déan, déanamh do, make jeig éeg (4), eeg (2), ieg (yeig) (5), éyg /dʒeɡ/ déag teen karraghey ar (3), arr (2), ayr, ar̆ r̆ /karaxә/ mend, repair kirree irr (5), ir /kiri/ cóirigh, ScG. càraich, càirich caoirigh, caoraigh mám, ScG. mam mám, ScG. mam i TC, TL, NM, e NM, e̜ TT (hem) e̜ TT, HK, ɪ TC, JW, i JW, (hem main, mayd) ɪ TC, ï, ɛ NM, (jed) e̜ TC, (jem) e̜ HK, (jem mayd) e̜ HK ɪ TC, i NM, JK, iː JK, ĭː NM (jannoo) e̜ TC, JW e JK, NM, W:N, o̤ JTK, HB, W:N, W:S, ɪ TK, DC, TL, EL, NM, a NM, JW, EC, ẹ JW, ɛ SK, Wa, J:EK, J:TL, (jeant) e̜ TC, (jean impv.) ɪ, e̜ , (jean fut.) e̜ , ɪ e̜ TC, Fa, Co, Wa, e JK, TK, HK, i WQ, e HB e̜ NM, JK, e TK, ɪ TK, ẹ JW, TK, HK, ï HK i, iː NM, ï HK, ɪ TC geddyn, feddyn haink, daink hed, jed, hem, jem, hemmayd, jemmayd mam /mam/ mam /mam/ sheep (pl.). handful (of both hands) blain, blister, boil 401 Aspects of the historical phonology of Manx milley (1 Chronicles 21. 12)336 mirril il (5), iyl. iil /milʲә/ míle thousand, million iː TK, JK, NM ‘mile’ ir (3), yr (2), ír /mirәlʲ/ míorbhail miracle e̜ JW, HK /mwari/ East Ulster Ir. móiridhe (Dinneen), móraí (Ó Dónaill) grandmother wɔ̜ːri, wɑːri NM, mŭā̱ri [mwãːri], wo̱ rə [wõrә] (Rhŷs notebook 6: 79, 149) a TC mwarree (Bible, Cr.), moirree (K.) prash ass (4) /praʃ/ *prais, prás brass queig; queiggoo queig (3), queigg, (wheiggoo) queiggu, quygu, quygú, quyggu, quyggú /kweɡʲ/, /xweɡʲu/ cúig, cóig; cúigmheadh five; fifth /reŋʲkʲ/ ráinig réad, raod, rud reached, arrived thing raink red ed (10), edd (4), yd (5), ydd, ǽd /red/, /rud/ rhum (Cr.),337 room (Bible), rooym (K.) reamlagh (Bible), rimlagh (Cr., NBHR) saillym umm, úym, úy̆ m /ru(ː)m/ 336 al (11), all (10), ail (15), âil, âîl, aill (3), áil, áill room kweg NM, TK, HK, kwɛg HK, fᵘe̜ g Wa; kwe.gu NM ɪ JK, HB, J:TL, JTK, NM, SK, ï JK, HB, W:S, SK, o̤ JK, HB, NM, SK, W:Peel, ø TC, u HB, NM, SK, ö J:EK, ə NM u, ũː, uː TC /rɨә̯mlax/, ?/rimlax/ ruaimneach fishing-line ɪ NM, ï HK, TK, ü W:S /salʲәm/ is áil le wish, want ɛː TC, aː JW, J:JK,338 HK, i JW, ɑ ̣ HK, ɪ TM meeil(l)ey ‘mile’ (Bible, Cregeen). rhum-aarlee ‘kitchen’ (Cregeen). 338 ‘In myr ’s ailliu [sic] hene, “as you yourself like,” = mö saːlʹʲu ˈhiːn [John] KN[een ], the Ir. áil, ScG. àill, seems to have had its vowel shortened and subsequently lengthened secondarily, as otherwise ɛːlʹ would presumably be expected; a short vowel seems indicated by the Manx spelling’ (Jackson: 25). 337 402 Aspects of the historical phonology of Manx scribider (Cr.), cf. screebeyder ‘scratcher’ (Cr.) sheill, sheiltyn (Bible, Cr.), sheltyn (Cr.), shein (Cr.), shell, shellagh, sheillagh, shillym (K.) skynn íyn /skribәderʲ/ scríobadóir grater /ʃe(ː)lʲtʃәn/ saoil, cf. Ir. dial. síl think vn. ʃĩn JW, ʃïᵈn NM, ʃiːn JK /skʲin/ scian, dat. sciain, ScG. sgian, dat. sgithinn339 spliúchán, ScG. spliùchan, spliuchan stán, stáin, ScG. staoin, stàin ScG. stàilinn, Norse stál knife iː JK, NM, i NM, TL, W:N, JK splughan (Cr.) ?/splʲuxan/ stainney /stanʲә/ steillin, steillyn àl /stelʲәnʲ/ pouch tin ɑ̃ ̣ː TC steel e̜ TC stheg (Cr.) /steɡ/ tooran ‘pronounced thurran’ (Cr.), tooran (K.) toshiaghjioarey /turan/ ?túrán, cf. torrán turret, stack of corn /ˌtoʃaxˈdʒoːrә/ taoiseach, tóiseach deóradh aon coroner one oː JTK, ɛː, Eː JK, uː TK, o̜ ː, u̜ ː J:JTK340 bhéas, bhias, bhíos, ScG. will be iː TC, ‘written vees but un vees yn (6), ýnn (3), ynn (3), yn̄ (2), únn, un̄ , ún, un, hyn (2) víís (6), vîîs, viss (4) /un/ /vi(ː)s/ steak 339 Cf. gen. and pl. forms with short vowel, scine, scena (eDIL). The HLSM examples apparently represent secondary re-lengthening, although it is short according to Rhŷs (18) even when stressed: ‘the case of […] un ‘one’ […], is the same [as that of freoagh ‘heather’, G. fraoch], except that the brevity of its vowel is sufficiently accounted for by the fact of its being a proclitic, though it may now sometimes have the stress but without restoration of its long vowel’. 340 403 Aspects of the historical phonology of Manx bhitheas, bhios ymmyd ym, ým /imәd/ adhmad use ynrycan, ynrican ynr (5) /inrәkan/ aonracán only pronounced vy̆ s [vәs]’ (Rhŷs: 166) ɪ JW, HK, ẹ HK, e HK ĩ JW The above Manx examples often correspond to cases of vowel shortening in other dialects, such as in frequently occurring verb forms (haink, cheet, hed, jean, jannoo, geddyn, irree, raink, sheill, saillym) (Quiggin 1906: 14; Mac Gill-Fhinnein 1966: 52– 6; Ó Curnáin 2007: 79). The latter may alternate as ‘unaccented’ (neamhaiceannta) forms alongside accented variants, used for example in answers to yes-no questions (Mac Gill-Fhinnein 1966: 52–6). In some Ulster dialects, however, the reduced forms appear to have been generalized: Stockman (1986), commenting on such shortenings in Ulster Irish, notes that Gaelic verbs often lack strong sentential stress, and suggests that reduced forms have spread by analogy to fully stressed positions.341 It is likely that many of the Manx shortenings represent similar lexicalization of post-lexical stress. Red ‘thing’ (G. réad, raod), is generally short rud in the modern dialects (and the written standards), presumably owing to lack of stress in collocations such as rud ar bith ‘anything’ (Manx red erbee). Rhŷs (18, 127) gives a similar explanation for shortening in proclitic un ‘one’ (G. aon) and dy bragh ‘ever’ (G. go bráth), from phrases such as dy bragh beayn (G. buan), dy bragh farraghtyn (G. mair), ‘everlasting’; lack of stress may also explain shortening in jeig ‘teen’ (G. déag), toshiagh-jioarey ‘coroner’ (G. taoiseach deóra), rhum-aarlee ‘kitchen’. The shortening, and vowel quality, in queig ‘five’ (G. cóig, cúig) may be attributed to the influence of jeig, and to postlexical destressing. There is a small group of nouns with initial éa- or ao- in Gaelic where shortening seems to be well established (eaddagh, eddin, eddrym), also feddyn (lenited form eddyn), and earroo may also have been influenced by these and by earish ‘time, E.g. ‘/ә danikʹ ʃә ˈraʃ gә foːLʹ? ha ˈdanikʹ/ An dtáinig sé ar ais go fóill? Cha dtáinig’ [‘Has he come back yet? No’] (Stockman 1986: 13), where the verb has a short vowel in both positions, despite being fully stressed in the answer. 341 404 Aspects of the historical phonology of Manx weather’ (G. iris, EIr. aires). Ymmyd ‘use’ (G. adhmad) may also belong here, although it may also have been influenced by semantically similar ymmyrch ‘need’ (ScG. imir), ymmyrchagh ‘necessary, useful’. There is apparently a semantic split between meeilley ‘mile’, with vowel length retained, and milley ‘million, thousand’ (both G. míle), the latter possibly influenced by English million (the usual word for ‘thousand’ is the borrowing thousane). Note, however, the curse my veelley mhillee ort, interpreted by Cregeen as ‘my dirty mile on thee, or my bad wish on thee’, but more readily explicable as ‘my thousand(fold) destruction on you’ (mo mhíle millidh ort). Compare also mirril (G. míorbhail), which may be similarly influenced by English ‘miracle’. In other cases there may be no obvious motivation for the shortening, although the bisyllabic cases may be attributed to a tendency observed in Ulster Irish for initial heavy syllables to be shortened in polysyllabic words (Stockman 1986). This phenomenon is discussed by Green (1997: 75–9) under the label ‘trochaic shortening’ (see Hayes 1995: 145–9; Prince 1990: 359–70). This phenomenon is explained by the observation that light-light trochees are cross-linguistically better formed than heavylight ones owing to a preference in trochaic languages for even duration (Hayes 1995: 79–85) (§5.1.1.6).342 The contractions to monosyllabic forms in haink, daink ‘came’ (G. tháinig), raink ‘reached, arrived’ (G. ráinig) and Gaelg ‘Manx’ (G. Gaoidhealg) may be regarded as a further stage of reduction; in the case of the verbs, see also cheayll, geayll ‘heard’ (G. chuala) and vaik ‘saw’ (G. faca). 342 For another potential factor, see Ó Maolalaigh (2014b: 19), who links vowel shortening in Stockman’s (1986) Ulster items to the presence of a light sonorant following the vowel in many of them. 405 Aspects of the historical phonology of Manx 5.5.2 Vowel lengthening According to Jackson (9), unhistorical vowel lengthening is found in /a/ and /o/: These lengthenings […] occur only before present-day single consonants, and seem to affect only original stressed a, ai, o, and ea in its a pronunciation (Jackson: 9) This is noted also by Marstrander (61, 68). Rhŷs likewise reports this primarily in /a/: Open a, long. This is approximately the ā of the English word ‘father,’ and it is not uncommon in Manx, especially when an a which, etymologically speaking, is short, has been lengthened in an accented syllable, as for example, in the word […] fakin ‘to see’ (Ir. faicsin, feicsin, ScG. faicsinn, faicinn), […] lhiabbee ‘bed’ (Ir. leabadh, ScG. leaba), […] cliaghtey to ‘be wont, a habit or custom (Ir. cleachtadh, cleachd, ScG. cleachd), […] clashtyn ‘to hear’ (Ir. cloisdin, ScG. claistinn), […] shassoo ‘to stand’ (Goi. seasamh). […] I have sometimes heard this vowel [Open a, long] in monosyllables ending with s, such as glass ‘green, blue, grey’ (Goi. glas), pronounced just like its Welsh equivalent glâs and so in […] jiass (Goi. deas). But more usually the a in these words and the like is decidedly short or of an intermediate length. (Rhŷs: 3) According to Rhŷs (45–6), the stressed vowel in moddey ‘dog’ (G. madadh) is long in the singular, but short in the plural moddee: Short [nasal] y [ә]̃ […] occurs in […] moddee ‘dogs’ the tone vowel of which is always short and this differs both in quality and quantity from that of the singular […] moddey. (Rhŷs: 45) According to Broderick (HLSM III: 122), however, all short vowels can be lengthened: This feature of secondary lengthening is a characteristic of L[ate] S[poken] M[anx] and can affect all (orig[inally] stressed) short vowels. (HLSM III: 122) Thomson (1999: 391) notes early evidence for secondary lengthening in the form of diacritics in the 1707 transcriptions of Manx speech for Edward Lhuyd (§1.5.1) which 406 Aspects of the historical phonology of Manx suggest lengthening, again in /a/ and /o/. It is possible that these lengthenings began in /a/, as there is no risk of merger with /aː/ from historical á, which has become /ɛː/ (§2.2.2). On the other hand, Jackson’s and Broderick’s descriptions suggest that secondarily lengthened vowels may have remained shorter than historical long vowels, so full mergers may have been avoided. In a few items, unhistorical vowel lengthening appears to be long-established and is shown in the orthography (Table 108): Table 108. Vowel lengthening shown in orthography Phillips fairaig (Cr., K.), fairage (K.)343 sheeiney, sheeint(yn) (Cr., Bible) sooree spagey, spaagey strep, strepey, strebin agg CM pronunciation /faˈrʲɛːɡ/ etymology English HLSM faireóg /ʃiːnʲә/ sine gland, waxkernel, ‘a lump in the groin or armpit’ (Cr.) teat, nipple i TC /suːrʲi/ suirghe court, woo /spɛːɡә/ spaga, Eng. bag (O’Rahilly 1927: 27) ScG. streap, Ir. dreap bag, scrip ?/streːpә/, ?/streːbәnʲ/ struggle, wrestle, wallow uː JW, NM, HK, SK, u W:N, uː J:EK i: TC 343 The stress marked on the first syllable by Cregeen, but see §5.1. If there is in fact stress shift as implied by the spellings, this suggests the strong likelihood of an earlier long vowel in the first syllable (§5.1); there may have been confusion with fáir, fáireóg ‘nest’ (although this is not attested in Manx). 407 Aspects of the historical phonology of Manx 5.6 Vowel nasalization Phonological vowel nasalization is widely attested across Gaelic dialects (Ó Maolalaigh 2003a; Ó Curnáin 2007: 291–361), although many varieties, particularly in Ireland, and parts of south-west Scotland appear to have lost this feature, or to have been in the process of losing it, relatively recently, i.e. in the last two centuries (Quiggin 1906: 64; Pedersen 1909: 386; Grant 1987: 58; Ternes 2006: 103; Ó Curnáin 2007: 325–332). Besides the prototypical case of vowel adjacent to original nasal consonants (including mh */ṽ/), vowel nasality (and denasalization) in Gaelic arises by a number of processes (Ó Curnáin 2007: 319–24), which produce a system replete with complexity, exceptions, unpredictability and considerable dialectal and idiolectal variation. Alongside categorical phonemic nasalization, speakers (including those lacking phonemic nasalization or having a reduced system) may exhibit a number of other types of vowel nasalization, including co-articulatory (partial) nasalization (Ternes 2006: 104–5; Morrison 2018: 5), phonological perseverative and anticipatory spread of nasality (Ó Curnáin 2007ː 293–5, 301–11), and paralinguistic nasalization and nasal speech setting (ibid.: 311–316, 1860–4). These complexities can make analysis very difficult (ibid.: 310–332; Oftedal 1956: 41), especially when combined with ongoing denasalization (Ó Curnáin 2007: 310–1, 324). Jackson’s (63–4) and Broderick’s (HLSM III: 147) descriptions of terminal Manx report only sporadic vestigial remains of vowel nasalization. However, the evidence of Rhŷs suggests that vowel nasality was much more widespread in the speech of his informants. He devotes a whole chapter to ‘nasal vowels’ (Rhŷs: 31–48), and presents a much more complete system, recognizable as similar to those described e.g. by Ternes (2006) and Ó Curnáin (2007). Rhŷs provides evidence of a wide range of processes found in other Gaelic dialects, including perseverative spread of nasality (Rhŷs: 35); vowel nasality after initial clusters cn-, gn-, mn-, sn- and tn- where n is synchronically /r/ (Rhŷs: 33–4); nasality in items where the nasal consonant is elided (Rhŷs: 35–6; also Jackson 63; HLSM 408 III: 147); and a number of apparent cases of Aspects of the historical phonology of Manx nasality arising through rhinoglottophilia344 (Lewin 2019a: 85–9), a phenomenon previously noted by Ó Maolalaigh (2003a: 116–7) in Manx injil ‘low’ (G. íseal). The near absence of vowel nasality in the terminal speakers recorded in the twentieth century might be considered a semi-speaker feature of incomplete acquisition, comparable to the absence of feminine gender concord or the lack of control of initial mutation (§1.6.9.1); however it may also in part represent the end point of a trend towards denasalization across the Gaelic world, perhaps connected with language contact (Ó Curnáin 2007: 359). For reasons of space this topic cannot be discussed in further detail here, and the reader is directed to my discussion of vowel nasalization in Lewin (2019a: 82–9). 344 The term rhinoglottophilia refers to the relationship between glottal or laryngeal and nasal articulations, which have acoustic and perceptual similarities (Matisoff 1975; Ohala 1983). In the Gaelic context, see Ó Maolalaigh (2003a) and Ó Curnáin (2007: 323). 409 Aspects of the historical phonology of Manx 410 Aspects of the historical phonology of Manx Chapter 6 Conclusion 6.1 Topics covered From the outset, the aim of this doctoral research project has been to cover as wide a range of sub-topics within the historical phonology of Manx as time and space would allow to be treated in adequate depth. It was recognized that practically all aspects of the topic (and indeed all areas of Manx linguistics) required in-depth re-evaluation and fresh analysis in order to bring them to a state where they can be solidly engaged with by scholars of the Gaelic languages on a basis comparable to descriptions of other dialects and periods (§§1.1, 1.5). In view of the breadth of areas in critical need of attention, and the acknowledged hindrance that a lack of adequate descriptions of Manx presents to Gaelic linguistics (e.g. Ó Maolalaigh 1997: 5, 11), it was felt that an approach focusing on a number of (only loosely related) topics in reasonable depth, rather than on a single, narrower topic in exhaustive depth, was the right one. In principle, however, a number of the chapters or sections in the present work could form the basis of full theses in their own right. It was difficult to prioritize topics, and precedence was given to those considered to be of the widest interest within the pan-Gaelic context (including the development of ao(i) and ua(i), and the sonorant consonants); those which seemed to be the most complex and intractable (or to have suffered the most misanalysis in the previous literature), and therefore in most urgent need of reappraisal and resolution; and simply those which had long pricked the author’s curiosity. Initially, topics which seemed to have been covered somewhat more extensively than most in the existing literature, notably the complex developments in the stress system (§5.1), were intended to be covered only briefly. However, as the project progressed, closer examination showed that none of the existing analyses of the topic of stress were fully adequate, and some contained significant misapprehensions, such that it was decided that suprasegmental and prosodic topics merited a full chapter of their own. 411 Aspects of the historical phonology of Manx On the other hand, the intricate topic of the “new” diphthongs (and triphthongs) formed within Gaelic dialects during the modern period from the vocalization of medial and final fricatives (abh > /au̯ /, etc.), and with which Manx in particular ‘teems’ (Rhŷs: 2), although certainly of equal interest to the other subjects covered in the thesis,346 has been reluctantly omitted for reasons of space, although one of the most complex parts of this topic is discussed briefly in the chapter on G. ao(i), ua(i) (§3.9.1). Apart from Chapter 4 on the sonorants, developments in the consonant system have not been covered in depth within this thesis. This is despite the fact that the topic of medial voicing and fricativization (‘secondary lenition’), in particular, is one of the most intriguing areas of Manx phonology, and one in which change can clearly be seen in progress during the attested period of the language, and which was still very much in a state of fluctuation among the terminal speakers (Thomson 1984: 314–5; HLSM III: 4–13; Williams 1994b: 712; Green 2006). It seems likely that secondary lenition would correlate with sociolinguistic factors, as well as dialect and idiolect, and that some of the quantitative approaches introduced in this thesis could fruitfully be brought to bear on this topic. Similarly, the degree to which the broad-slender patalization contrast had broken down or entered a situation of ‘wild allophonic variation’ (Broderick 1999: 81–6; see also Williams 1994b: 712, 737; Lewin 2017a: 156, 187–8) in the language of the terminal speakers has not yet been quantified (although see §4.4.3). 6.2 Progress made Although the main ‘outlines’ of the distinctive developments of Manx phonology have been known to scholarship since Rhŷs’s pioneering treatise published in 1894, the existing descriptions have been difficult to interpret, and inaccessible and misleading to the scholar who is not a Manx specialist, but wishes to make reference to the language in broader contexts. The main achievement of this thesis, then, is to clarify and describe in a systematic fashion a good deal of what has previously been only 346 See e.g. Ó Maolalaigh (2006) for aspects of this topic in a pan-Gaelic context. 412 Aspects of the historical phonology of Manx known in broadest outline or assumed about Manx phonology. For example, it has long been known that G. ua(i) /ua/ is often fronted in Manx, but the environments and lexical items in which this takes place have not before been clearly described. Furthermore, while the superficial similarity here between Manx and Ulster Irish developments has been noted, it has not been previously pointed out that the conditioning factors appear to be quite different in the two cases (§3.8). That ao(i) and ua(i) have some tendency to merge had been noted by previous scholars, but the full range of evidence for these developments from across the attested period of the language has not before been collated and analysed quantitatively (Chapter 3). While some uncertainties and ambiguities in the evidence remain (which, given imperfections in the data, may never be resolved completely), a solid basis for reference and further research has nevertheless been provided, and a significant step forward has been taken in removing some of the “noise” in the ‘mass of raw phonetic Manx data’ (Ó Maolalaigh 1997: 5) which comprises existing datasets. By comparing a number of different sources quantitatively (the Phillips orthography; the eighteenth-century orthography; maximally complete lexical sets for individual phones or developments drawing on the dictionaries; the independent descriptions of Rhŷs, Marstrander, Jackson and Broderick; the corpus of transcribed data in Broderick’s HLSM; and instrumental analysis of the recordings of the terminal speakers), it is possible to identify patterns which were not immediately obvious in previous, impressionistic analyses. Even if results based on a single dataset can only be tentative, firmer conclusions can be drawn when a number of independent datasets point towards similar patterns, as notably is the case with the topic of ao(i) and ua(i). With the exception of Rhŷs, whose work traces the development of Manx sounds from Phillips through to the speech of his own informants, most research on Manx phonology has drawn almost exclusively on data recorded or transcribed from the last remaining speakers in the twentieth century, compared directly with Classical Irish cognate forms, with relatively little consideration of the evidence of the historical orthographies. These seem to have generally been presumed to be idiosyncratic and erratic (§1.6.2), too irregular and ambiguous to provide much usable evidence. At any rate, scholars — especially those commenting on Manx in passing rather than 413 Aspects of the historical phonology of Manx specializing in it — have known too little about the conventions of Manx spelling to make reliable use of orthographic data in their analyses, and when they have tried, there has been a tendency to ascribe implausible realizations to orthographic representations owing to deficient understanding of robustly attested patterns and conventions (see §§1.5.7, 5.1.1.4, 5.1.1.5 for examples). Redressing the balance, and to a certain extent “rehabilitating” the Manx orthographies, has therefore been a central concern of this thesis. Throughout, it has been shown that there are often striking regularities in the orthographies which can be shown to correlate strongly with particular realizations (e.g. the contrasts between /uː/ ~ /uә̯/, /iː/ ~ /iә̯/, §2.2.6, and between /eː/ and /ɛː/ §2.2.3), although it is certainly true that there are also many frustrating ambiguities (for example, the lack of clarity with regard to sonorant contrasts, Chapter 4). The somewhat negative attitudes of many scholars towards Manx, and in particular assumptions regarding its ‘anglicized’ or even ‘creoloid’ nature (Lewin 2017a), have resulted in a tendency to assume that certain features of the language as recorded in the twentieth century were also characteristic of earlier periods, when it is likely that they are in fact symptomatic of incomplete acquisition and/or rustiness in a situation of language shift and obsolescence, and belong to the last generation or two of speakers only (ibid.: 180–93). Making this distinction clearer, and bringing an understanding of the processes of language shift based on contemporary empirical research on language contact and bilingualism to bear on the study of Manx, has been a major concern of my research (Lewin 2014b; 2017a; 2019a) and informs many aspects of the present thesis. Nevertheless, it is not always easy to distinguish between internal developments, contact features, and features related to obsolescence (e.g. the reductions in the sonorant consonant inventory discussed in Chapter 4). Claims that Manx is in some way exceptional, such that contemporary phonological frameworks applied to other languages are unsuitable for the analysis of Manx data, have been treated with scepticism (§1.5.9), although it is readily admitted that analysis of historical linguistic data is not always simple, and that conclusions reached are, by necessity, sometimes tentative. 414 Aspects of the historical phonology of Manx This brings us to another resource which has been under-used and to some extent dismissed (e.g. Jackson: 4): the descriptions and transcriptions made by John Rhŷs in the 1880s and 90s, including his recently documented notebooks (§1.5.2). Although in some respects less scientific and more rudimentary than his successors in the twentieth century, Rhŷs’s descriptions are of paramount importance because they pertain to an earlier generation of informants, at least some of whom were Manx-dominant speakers who acquired Manx in communities where knowledge and use of English was not yet universal. They attest to features such as more consistent and productive use of initial mutations, the /r ~ rʲ/ contrast (§4.2.1), more extensive preservation of coda /r/, /rʲ/ (§4.2.3), more conservative reflexes of ao(i) and ua(i) (§3.5.1), extensive vowel nasalization almost completely absent in the terminal speakers (§5.6; Lewin 2019a: 82–9), as well as other linguistic features such as noun gender (Lewin 2019a: 79–82). Although considerations of space and scope have not allowed too much discussion of wider theoretical implications, insights from the cross-linguistic theoretical and typological literature have been brought to bear where appropriate, and have proved particularly enlightening in the analysis of stress shift (§5.1), vowel shortening in polysyllables (§5.5.1), and the development of earlier geminate sonorants (including the emergence of preocclusion, §4.5). The latter discussion in particular constitutes a significant contribution in placing Manx within the context of prosodic developments within the north-west European linguistic area (cf. Wagner 1964; Salmons 1992; Iosad 2016b). In terms of situating Manx in the wider Gaelic dialectological context, it is hoped that the analyses in this thesis will provide a more solid basis for comparison than has hitherto been available. It has been shown that, as would be expected, Manx developments often show similar patterns, trajectories and conditioning factors to analogous developments in the other Gaelic dialects — for example, some of the short vowel developments have similar conditioning factors to those evidenced in Ó Maolalaigh (1997) — although, as noted above in relation to fronting of ua(i), such parallels should not be taken for granted. The sonorants also show similar developments to those found elsewhere (e.g. early merger of /l̪ / and /ḻ/ to the dental lateral (§4.3.2), which is widely attested in Ireland and Scotland), as well as specific 415 Aspects of the historical phonology of Manx parallels with Scottish Gaelic (merger of broad and slender lenis /n ~ nʲ/ in certain items, §4.4.3), but there are also specific Manx developments (e.g. depalatalization of lenis */lʲ/ in certain limited circumstances, §4.3.2.2). An important connection to neighbouring dialects is revealed in the analysis of the Manx development of G. ao(i) which is shown to give a front-central mid unrounded (or only weakly rounded) vowel /әː/, which also represents agh etc. (Chapter 3). This bears a strong resemblance to the situation in south-western Scottish dialects (§3.2.1; Lewin 2018), in contrast both to the rest of the northern Gaelic dialect area (Ulster and the rest of Scotland), where ao > /ɯː/, contrastive with agh > /ɤː/, and southern Irish dialects, with ao > /iː/ or /eː/ and agh > /әi̯ /. This connection has not been picked up by previous analyses. Other similarities, however, are less likely to be related to any historical affinity or contact. The developments of lengthening, rounding and diphthongization of short vowels before original geminate sonorants in monosyllables might at first glance invite association with similar developments in Munster and the northern Hebrides, but the evidence of Phillips’ orthography shows that in the early seventeenth century these developments were only incipient (§4.6), and, on the whole, Manx realizations of this period would have resembled those found in the conservative varieties of Ulster and southern Argyll today. If the Manx developments were largely a development of the mid seventeenth century, there is some evidence that the analogous developments in Munster and Scotland had taken place at least a century or more earlier (O’Rahilly: 51–2). It is unlikely for sociohistorical reasons (§1.3) that this development could have spread into Manx from these areas at such a late date; rather, such parallel developments should be seen as arising from a limited set of options for realizing common inherited features (§1.4). Similar considerations are relevant in the case of stress shift and post-tonic long vowel shortening (§5.1). 416 Aspects of the historical phonology of Manx 6.3 By-products The development of this thesis naturally involved considerable amounts of background research and data-gathering, and investigation of tangential topics, not all of which has found its way into the permitted space, or is relevant to the central focus of the project. Some of this material has already appeared in print in an article on the cross-Gaelic dialectological and typological development of ao(i) (Lewin 2018), and an evaluation of Rhŷs’s work as a fieldworker (Lewin 2019a), including notably a discussion of aspects of the topic of vowel nasalization in Manx which are only briefly discussed in the thesis (§5.6). It is hoped that further analyses of related topics, including especially a discussion of the topic of fricative vocalization mentioned above, can be published soon. Throughout this thesis extensive tables of lexical material are given, with Gaelic or Early Irish cognates provided, or at least tentatively suggested, as far as possible. The most important previously-available sources for etymologies of Manx words are Thomson’s (1953; 1954–59) glossary of Phillips’ prayer book, and Broderick’s dictionaries of ‘Late Spoken Manx’ (HLSM II) and Rhŷs’s notes (Broderick 2019). These are restricted, however, to items which happen to appear in the material on which they are based. Other etymologies are given in varia notes on Irish and Scottish Gaelic lexical items by a number of scholars, notably O’Rahilly (see Lewin 2017a: 147) (and also in his Irish Dialects Past and Present), and in notes to editions of Manx texts. The former are restricted to items which happen to be of interest from the perspective of the other Gaelic languages, however, and the latter again to forms which appear in particular texts. Since datasets in the present study incorporate numerous lexical items which are attested only in texts for which glossaries are unavailable (e.g. they may be found in the Bible, but not in Phillips or HLSM), or else are only attested in Cregeen’s and Kelly’s dictionaries, the above sources may be of little help, and etymologies for these items had to be identified. During the course of this work, a sizeable collection of additional lexical and etymological data has been assembled, which is likely to prove useful in future research. There are several hundred items whose etymology remains obscure which require further investigation. In these cases, 417 Aspects of the historical phonology of Manx the insights into the development of Manx phonology and orthography gained during the present research have the potential to provide valuable clues. 6.4 Prospects for future research As mentioned above, some of the innovative methodologies employed in this project could certainly be applied productively to other aspects of Manx phonology. The instrumental analysis of recordings of the terminal speakers, which involves timeconsuming annotation of spectrograms, has only been applied in a couple of areas (§§2.2.3, 3.7, 4.5) but has obvious potential for wider application, including to other sets of recordings beyond those of the Irish Folklore Commission. The quantitative analyses of written material, as well as written phonetic transcriptions, have similarly been applied parsimoniously in the present work, and doubtlessly have the potential to reveal further insights. A full quantitative analysis of all vowel representations in Phillips is a desideratum, for example. The written material on this which much of this thesis is based is largely restricted to readily available printed material which has been digitized, including the Bible and the two main dictionaries. The only manuscript source of which extensive use is made is Phillips’ Prayer Book, for which we have an edition and a full glossary. The analysis of the eighteenth-century orthography is largely restricted to the “standard” as represented in the later editions of the Bible, and the dictionaries. Divergent forms found in less standardized printed sources (such as Coyrle Sodjeh, the earliest printed book in Manx published in 1707), and in manuscript sources such as the translators’ drafts of the Bible (Lewin 2019b), have only been referred to when they happen to have come to my attention. Clearly, a more systematic engagement with these sources would be fruitful. The majority of the printed Manx texts have now been digitized in one format or another, an effort to which the present author has contributed (e.g. Lewin and Wheeler 2017; 2019a). In due course it is hoped these texts can be brought together in a similar online format to the available corpora of Irish (Corpas Stairiúil na Gaeilge and Nua- 418 Aspects of the historical phonology of Manx Chorpas na Gaeilge) and Scottish Gaelic (Corpas na Gàidhlig). The manuscript material presents a more significant challenge, and includes, as mentioned, a sizeable portion of the original drafts of the Bible translation, around 700 sermons, and ~40,000+ lines of carvals (religious ballads). Relatively little of this material has been edited,348 or digitized, although the sermons and the carvals have at least been recently catalogued; see Lewin (2015b) for an edition of the earliest known manuscript sermon. Much of this material is in quite divergent orthography, and has obvious potential to provide much additional information on phonological change, dialect, and the development and refinement of the orthography within the social networks of eighteenth- and nineteenth-century Man. An obvious priority would be to trace the development of the orthography in the manuscript sources from the early to mid eighteenth century, prior to its the standardization in the Bible in the 1770s, and to investigate possible changes in the orthographic practices of key figures, such as Philip Moore, for whom we have surviving early sermons from the 1720s, and who went on to train many of the other clergy and Bible translators and to oversee and edit the Bible translation for publication (Butler 1799: 186–205). Another important source of information is personal and place names, especially the extensive data contained in Broderick’s Placenames of the Isle of Man (PNIM). These have been referred to at certain points in the thesis, but await more systematic analysis (§1.5.10). It is hoped in due course to complete a fuller description of Manx historical phonology, including those areas (primarily the consonantal system) which have had to be omitted from the present thesis. Similarly, other areas of Manx linguistics (such as morphology and syntax) await fresh treatment, and these are also areas in which I have taken an interest and made some progress (Lewin 2014b; 2016a; 2016b), and to which I hope to return. Questions relating to the medieval origins and development of Manx and its relationship with other Gaelic dialects (§§1.3, 1.4), and questions of language contact throughout its history (Lewin 2017a), also deserve fuller treatment. 348 With the important exception of significant parts of the folksong manuscript corpus, which have been edited by Broderick, Thomson and others, and the original prose writings of Edward Faragher (Broderick 1981a; 1982b; Lewin 2014). 419 Aspects of the historical phonology of Manx 420 Aspects of the historical phonology of Manx Bibliography Primary sources AG – Aght giare dy heet gys tushtey as toiggal jeh’n chredjue Creestee. John Clague (1814), translation of An Introduction to the knowledge of the Christian religion, Henry Crossman (1742). < https://books.google.co.uk/books?id=lMoHAAAAQAAJ&pg=> Bible – Unless otherwise specified references are to the 1819 (repr. 1979) edition of the Manx Bible Bible MSS – original translators’ drafts, with editorial emendations. Manx National Heritage Library MS 5690C. CS – Coyrle Sodjeh. Lewin and Wheeler (2017) (eds) HLSM – transcribed texts and vocabulary in A Handbook of Late Spoken Manx (Broderick 1894–86) Hymns – Lioar dy Hymnyn. [Manx hymn books] 1795, 1799, 1830, 1846. Lewin and Wheeler (2019) (eds) PC – Pargys Caillit. Thomson (1995) (ed.) PC 1796 – ‘Thomas Christian, Pargys Caillit (1796)’, ed. by Max W. Wheeler (2017) <https://www.academia.edu/31310779/Thomas_Christian._Pargys_Caillit._1 796_edition> [accessed 11.11.2018] PC 1872 – ‘Thomas Christian, Pargys Caillit, as published in Manx Miscellanies No. 1, Douglas: The Manx Society, 1872’, ed. by Max W. Wheeler (2017). <https://www.academia.edu/31310780/Thomas_Christian._Pargys_Caillit._1 872_edition> [accessed 11.11.2018] Scanned from PDF of original publication: <books.google.co.uk/books?id=Kws5AQAAMAAJ> [accessed 11.11.2018] Phillips, Ph. – Bishop Phillips translation of the Book of Common Prayer (c. 1610). Moore and Rhŷs (1895) (eds); Thomson (1953) (glossary). PSD – Plain and Short Directions and Prayers [appendix to Coyrle Sodjeh]. Lewin and Wheeler (2017) (eds) 421 Aspects of the historical phonology of Manx SW – Sharmaneyn liorish Thomase Wilson [sermons by Thomas Wilson]. 1783 (Bath: R. Cruttwell) <https://books.google.co.uk/books?id=ROQGAAAAQAAJ&dq> NBHR – ‘Manx Stories and Reminiscences of Ned Beg Hom Ruy’ Broderick (1981b) (ed.) Rhŷs, John. 1886–93. Manx notebooks. National Library of Wales GB 0210 JORHYS A2/5; A3/9, 11, 12; A4/10. Trad. Ballad – ‘Manx Traditionary Ballad’. Thomson (1960–62) (ed.) Secondary sources Ahlqvist, Anders. 1994. ‘Litriú na Gaeilge’, in Stair na Gaeilge, ed. by Kim McCone and others (Maynooth: National University of Ireland), pp. 23–59 Archangeli, Diana and others. 2011. ‘ATR in Scottish Gaelic tense sonorants’, in Formal Approaches to Celtic Linguistics, ed. by Andrew Carnie (Newcastle upon Tyne: Cambridge Scholars Publishing), pp. 283–306 Ball, Martin J. and Briony Williams. 2001. Welsh Phonetics (Lewiston, NY: Edwin Mellen) Barras, William. 2018. ‘Residual Rhoticity and Emergent r-sandhi in the North West and South West of England: Different Approaches to Hiatus-Resolution?’, in Sociolinguistics in England, ed by Natalie Braber and Sandra Jansen (London: Palgrave Macmillan), pp. 363–92 Barry, Michael V. 1984. ‘Manx English’, in Language in the British Isles, ed. by Peter Trudgill (Cambridge: Cambridge University Press), pp. 167–77 Bauer, Laurie. 1985. ‘Tracing phonetic change in the received pronunciation of British English’, Journal of Phonetics, 13: 61–81 Bauer, Michael. 2011. Blas na Gàidhlig The Practical Guide to Gaelic Pronunciation (Glasgow: Akerbeltz) Bergin, Osborn. 1946. ‘Irish Grammatical Tracts III and IV’, Ériu, 14: 165–257. Blankenhorn, Virginia S. 1981. ‘Pitch, quantity and stress in Munster Irish’, Éigse, 18: 225–250 422 Aspects of the historical phonology of Manx Boersma, Paul and David Weenink. 2015. Praat: doing phonetics by computer <http://www.praat.org> Borgstrøm, Carl Hj. 1937. ‘The Dialect of Barra in the Outer Hebrides’, Norsk Tidsskrift for Sprogvidenskap, 8: 71–242 ——— 1940. The Dialects of the Outer Hebrides (Oslo: H. Aschehoug & Co.) ——— 1941. The Dialects of Skye and Ross-shire (Oslo: H. Aschehoug & Co.) Breatnach, Liam. 1994. ‘An Mheán-Ghaeilge’, in Stair na Gaeilge, ed. by Kim McCone and others (Maynooth: National University of Ireland), pp. 221–333 ——— 2003. ‘On Words Ending in a Stressed Vowel in Early Irish’, Ériu, 53: 133– 42 Breatnach, Risteard A. 1993. ‘Cregeen’s Manx proverbs and familiar phrases’, Éigse, 27: 1–34 Breatnach, Risteard B. 1947. The Irish of Ring Co. Waterford: A Phonetic Study (Dublin: Dublin Institute for Advanced Studies) Broderick, George. 1981a. ‘Baase Illiam Dhone’, Celtica, 14: 105–23 ——— 1981b. ‘Manx Stories and Reminiscences of Ned Beg Hom Ruy’, Zeitschrift für celtische Philologie, 38: 113–78 ——— 1981c. ‘Manx Traditional Songs and Song Fragments I: Manx Museum MS. 263A (Clucas Coll.)’, Béaloideas, 48–9: 9–29. ——— 1982a. ‘Manx Stories and Reminiscences of Ned Beg Hom Ruy Translation’, Zeitschrift für celtische Philologie, 39: 117–94 ——— 1982b. ‘Manx Traditional Songs and Song Fragments II: Manx Museum MS. 450A’, Béaloideas, 50: 1–41 ——— 1983. ‘Boddagh Yn Cooat Laaghagh: A Manx Version of a Fenian Tale’, Béaloideas, 51: 1–10 ——— 1984a. ‘Ny Kirree Fo Niaghtey’, Celtica, 16: 157–68 ——— 1984b. ‘Berrey Dhone: A Manx Caillech Bérri?’, Zeitschrift für celtische Philologie, 40: 193–210 ——— 1984c. ‘Ec ny Fiddleryn’, Zeitschrift für celtische Philologie, 40: 211–27 ——— 1984–86. A Handbook of Late Spoken Manx. 3 vols. (Tübingen: Niemeyer) ——— 1990. ‘Fin as Oshin’, Celtica, 21: 51–60 423 Aspects of the historical phonology of Manx ——— 1993. ‘Manx’, in The Celtic Languages, ed. by Martin J. Ball and Nicole Müller, 1st edn (London: Routledge), pp. 228–86 ——— 1994–2005. Placenames of the Isle of Man. 7 vols (Tübingen: Niemeyer) ——— 1997. ‘Manx English an Overview’, in The Celtic Englishes, ed. by Hildegard L. C. Tristram (Heidelberg: Winter), pp. 123–34 ——— 1999. Language Death in the Isle of Man (Tübingen: Niemeyer) ——— 2005. ‘Manx Gaelic’, in Bausteine zum Studium der Keltologie, ed. by Helmut Birkhan (Wien: Praesens), pp. 239–41 ——— 2010. ‘Manx’, in The Celtic Languages, ed. by Martin J. Ball and Nicole Müller, 2nd edn (London: Routledge), pp. 305–56 ——— 2011. ‘The imperfect and secondary future in Late Spoken Manx’, Scottish Gaelic Studies, 28: 307–32 ——— 2015. ‘Manx Traditional Songs and Song-Fragments III: 1 MNH MD 900 MS 08307 and the Robert Gawne Collection?’ <https://sites.google.com/view/gailck-hasht/artyn?authuser=0> [accessed 21.06.2020] ——— 2016a. ‘Sir John Rhŷs and his visits to the Isle of Man (1886-1893) I: Catalogue & Lists of Informants’. <https://gaelg-hasht.000webhostapp.com/astoo/gb/John%20Rhys%20I.pdf> [accessed 23.10.2018] ——— 2016b. ‘Sir John Rhŷs and his visits to the Isle of Man (1886-1893) II: Diary of his visits to the Isle of Man (1886-1893)’. <https://gaelg-hasht.000webhostapp.com/astoo/gb/John%20Rhys%20II.pdf> [accessed 23.10.2018] ——— 2017. ‘The last native Manx Gaelic speakers. The final phase: ‘full’ or ‘terminal’ in speech?’, Studia Celtica Fennica, 14: 18–57 ——— 2018a. ‘Manx Gaelic: the end game’, Celtica, 30: 56–182 ——— 2018b. ‘Francis J. Carmody: The Manx Recordings’, Studia Celtica, 52: 157– 78 ——— 2018c. ‘Prof. Sir John Rhŷs in the Isle of Man (1886–1893): Linguistic material and texts’, in Proceedings of the second European symposium in Celtic Studies, held at Prifysgol Bangor University from July 31st to August 3rd 2017, ed. by Raimund Karl and Katharina Möller (Hagen: Curach Bhán), pp. 35–69 424 Aspects of the historical phonology of Manx ——— 2018d. ‘Carl Marstrander’s Field Notes from the Isle of Man 1929, 1930, 1933’, Celtica, 30: 183–205 ——— 2018e. ‘Manx Traditional Songs and Song-fragments in the End-phase of Manx Gaelic: From the Clague Music Collection (1890s)’, Studia Celtica Fennica, 15: 28–64 ——— 2018f. ‘Fin as Oshin: a reappraisal’, Zeitschrift für celtische Philologie, 65: 63–81 ——— 2019. Prof. Sir John Rhŷs in the Isle of Man (1886–1893): dictionary. Zeitschrift für celtische Philologie, 66: 15–73 Butcher, Andrew and Debbie Loakes. 2008. ‘Enhancing the left edge: the phonetics of prestopped sonorants in Australian languages’, Journal of the Acoustical Society of America, 124: 2527. Butler, Weeden. 1799. Memoirs of Mark Hildesley, D.D. late Lord Bishop of Sodor and Mann; and Master of Sherburn Hospital, under whose auspices the holy scriptures were translated into the Manks language (London: J. Nichols) Campbell, John Lorne (ed.). 1972. Saoghal an Treobhaiche (Inbhirnis: Club Leabhar) Campbell, Lyle and Martha C. Muntzel. 1989. ‘The structural consequences of language death’, in Investigating Obsolescence: Studies in Language Contraction and Death, ed. by Nancy C. Dorian (Cambridge: Cambridge University Press), pp. 181–196 Carmody, Francis J. 1947. Manx Gaelic Sentence Structure in the 1819 Bible and the 1625 Prayer Book (Berkeley: University of California Publications) ——— 1953. ‘Spoken Manx’, Zeitschrift für celtische Philologie, 24: 58–80 Carney, James. 1964. The poems of Blathmac, son of Cú Brettan, together with the Irish Gospel of Thomas and a poem on the Virgin Mary (Dublin: Irish Texts Society) Chaudhri, Talat. 2007. ‘Studies in the Consonantal System of Cornish’ (unpublished doctoral dissertation, University of Wales, Aberystwyth) Clancy, Thomas Owen (ed.). 1998. The Triumph Tree: Scotland’s Earliest Poetry AD 550–1350 (Edinburgh: Canongate Books) ——— 2010. ‘Gaelic in medieval Scotland: advent and expansion’, Proceedings of the British Academy, 167: 349–92 425 Aspects of the historical phonology of Manx Coates, Richard. 1987. ‘Historical Linguistics’, in New Horizons in Linguistics 2, ed by John Lyons, Richard Coates, Margaret Deuchar and Gerald Gazdar (London: Penguin), pp. 179–99 Cox, Richard A. V. 2011. ‘Eachdraidh is tùs a’ mhìr àireamhail ann an Gàidhlig (na h-Albann)’, Scottish Gaelic Studies, 28: 255–94 ——— 2013. ‘The history of the numerical particle in (Scottish) Gaelic’, in Lochlann: festskrift til Jan Erik Rekdal på 60-årsdagen: Aistí in ómós do Jan Erik Rekdal ar a 60ú lá breithe, ed. by Cathinka Hambro and Lars Ivar Widerøe (Oslo: Hermes Academic Publishing), pp. 247–81 Cregeen, Archibald. 1835. A Dictionary of the Manks Language (Douglas: Quiggin) Cumming, J. G. (ed.). 1859. An Account of the Isle of Man (Douglas: Manx Society) DASG: Corpas na Gàidhlig, Digital Archive of Scottish Gaelic (University of Glasgow) <http://dasg.ac.uk/corpus/> de Bhaldraithe, Tomás. 1945. The Irish of Cois Fhairrge, Co. Galway: A Phonetic Study (Dublin: Dublin Institute for Advanced Studies) de Búrca, Seán. 1970. The Irish of Tourmakeady, Co. Mayo (Dublin: Dublin Institute for Advanced Studies) Dillon, Myles. 1953. ‘Semantic distribution in Gaelic dialects’, Language, 29: 322–25 ——— 1962. ‘Phonetic analogy’, in Proceedings of the Fourth International Congress of Phonetic Sciences held at the University of Helsinki 4–9 September 1961, ed. by Antti Sovijärvi and Pentti Aalto (The Hague: Mouton), pp. 577–79 Dilworth, Anthony. 1996. ‘A comparison of a central western dialect with a peripheral one — western Mainland Inverness-shire and Perthshire’, Scottish Language, 14/15: 42–51 Dinneen, Patrick S. 1927. Foclóir Gaedhilge agus Béarla: An Irish–English Dictionary, reprinted 1996 (Dublin: Irish Texts Society) Dorian, Nancy C. 1977. ‘The Problem of the Semi-Speaker in Language Death’, International Journal of the Sociology of Language, 12: 23–32 ——— 1981. Language death: The Life Cycle of a Scottish Gaelic Dialect (Philadelphia: University of Pennsylvania Press) ——— (ed.) 1989. Investigating Obsolescence: Studies in Language Contraction and Death (Cambridge: Cambridge University Press) 426 Aspects of the historical phonology of Manx ——— 2010. Investigating Variation: The Effects of Social Organization and Social Setting (Oxford: Oxford University Press) Dressler, Wolfgang U. 1981. ‘Language shift and language death — a protean challenge for the linguist’, Folia Linguistica, 15: 5–28 Dwelly, Edward. 1911. A Gaelic dictionary, specially designed for beginners and for use in schools. Profusely illustrated (Camus-a-Chorra: the author) <http://www.dwelly.info/> EDD: The English Dialect Dictionary, Wright 1898–1905 eDIL: Toner, Gregory and others (eds). Electronic Dictionary of the Irish Language. <http://edil.qub.ac.uk/dictionary/search.php> Elsie, Robert. 1986. Dialect relationships in Goidelic (Hamburg: Helmut Buske) Etchingham, Colmán and others. 2019. Norse-Gaelic Contacts in a Viking World (Turnhout: Brepols) Faclair Beag: Bauer, Michael and William Robertson. Am Faclair Beag. <https://www.faclair.com> Fargher, Douglas C. 1979. Fargher’s English–Manx Dictionary (Douglas: Shearwater Press) Fellows-Jensen, Gillian. 1983. ‘Scandinavian Settlement in the Isle of Man and NorthWest England: The Place-Name Evidence’, in The Viking Age in the Isle of Man, ed. by Christine Fell and others (London: Viking Society for Northern Research), pp. 37–52 ——— 2015. ‘The Manx Place-name Evidence,’ in A New History of the Isle of Man. Volume III The Medieval Period 1000–1406, ed. by Seán Duffy and Harold Mytum (Liverpool: Liverpool University Press), pp. 257–80 Fougeron, Cécile and Caroline L. Smith. 1999. ‘French’, in Handbook of the International Phonetic Association (Cambridge: Cambridge University Press), pp. 78–81 Gelling, Margaret. 1971. ‘The place-names of the Isle of Man’, Journal of the Manx Museum, 7:82, 130–9, 168–75 ——— 1991. ‘The place-names of the Isle of Man’, in Language Contact in the British Isles. Proceedings of the Eighth International Symposium on Language Contact in Europe, Douglas, Isle of Man, 1988, ed. by P. Sture Ureland and George Broderick (Tübingen: Niemeyer), pp. 141–55 427 Aspects of the historical phonology of Manx George, Ken and George Broderick. 2009. ‘The revived languages – Cornish and Manx’, in The Celtic Languages, ed. by Martin J. Ball and Nicole Müller (London: Routledge) Gilchrist, A. G. and others. 1924–26. ‘Manx Collection’ (various individual titles). Journal of the Folk-Song Society, 28: 117–183, 29: 203–280, 30: 281–347. Gillies, William. 1994. ‘The Celtic languages: some current and some neglected questions’, in Speaking in our tongues: medieval dialectology and related disciplines, ed. by Margaret Laing and Ian K. Williamson (Woodbridge: D. S. Brewer), pp. 139–48 GOI: A Grammar of Old Irish. Thurneysen 2003 Grannd, Seumas. 2000. The Gaelic of Islay: A Comparative Study (Aberdeen: Department of Celtic) Grant, James. 1987. ‘The Gaelic of Islay: phonology, lexicon and linguistic context’ (University of Aberdeen, unpublished doctoral dissertation) Green, Antony D. 1997. ‘The Prosodic Structure of Irish, Scots Gaelic, and Manx’ (Cornell University, unpublished doctoral dissertation) ——— 2006. ‘The independence of phonology and morphology: The Celtic mutations’, Lingua, 116: 1946–1985 Greene, David. 1952. ‘Middle Quantity in Irish’, Ériu, 16: 212–18 ——— 1962. ‘Varia. I’, Ériu, 19: 111–13 ——— 1976. ‘The Diphthongs of Old Irish’, Ériu, 27: 26–45 ——— 1977. ‘Feras, banas, and some related abstracts’, Ériu, 28: 155–61 ——— 1983. ‘Cró, crú and similar words’, Celtica, 15: 1–9 Grinevald, Colette and Michel Bert. 2011. ‘Speakers and communities’, in The Cambridge Handbook of Endangered Languages ed. by Peter K. Austin and Julia Sallabank (Cambridge: Cambridge University Press), pp. 45–65 Hall, T. A. 2009. ‘Rule Inversion in a Regional Variety of Bavarian German’, Zeitschrift für Dialektologie und Linguistik, 76: 137–64 Hamer, Andrew. 2007. ‘English on the Isle of Man’, in Language in the British Isles, ed. by David Britain (Cambridge: Cambridge University Press), pp. 171–75 Harvey, Mark and others. 2015. ‘Contrastive and Non-contrastive Pre-stopping in Kaytetye’, Australian Journal of Linguistics, 35: 232–50 428 Aspects of the historical phonology of Manx Hayes, Bruce. 1995. Metrical Stress Theory: Principles and Case Studies (Chicago: University of Chicago Press) Henderson, John Noel. 1974. A phonetic study of the Irish of Tory island, Co. Donegal (Belfast: Institute of Irish Studies) HLSM: A Handbook of Late Spoken Manx, Broderick 1984–86. Holmer, Nils. 1938. Studies on Argyllshire Gaelic (Uppsala: Almqvist & Wiksells) ——— 1957. The Gaelic of Arran (Dublin: Dublin Institute for Advanced Studies) ——— 1962a. The Gaelic of Kintyre (Dublin: Dublin Institute for Advanced Studies) ——— 1962b. The Dialects of Co. Clare Part I (Dublin: Royal Irish Academy) Ifans, Dafydd and Robert Thomson. 1980. ‘Edward Lhuyd’s Geirieu Manaweg’, Studia Celtica, 14–5: 129–67 Iosad, Pavel. 2013. ‘Head-dependent asymmetries in Munster Irish prosody’, Nordlyd, 40: 66–107 ——— 2016a. ‘Tonal stability and tonogenesis in North Germanic’, In Beyond borealism: New perspectives on the North, ed. by Ian Giles, Laura Chapot, Chris Cooijmans, Ryan Foster, and Barbara Tesio (London: Norvik Press), pp. 80–97 ——— 2016b. ‘The northwest European phonological area: New approaches to an old problem’ (handout, paper delivered at University of Edinburgh Linguistic Circle, 29th September 2016) <https://www.research.ed.ac.uk/portal/files/52774442/handout.pdf> [accessed 19.05.2019] Jackson, Kenneth. 1953. Language and History in Early Britain (Edinburgh: Edinburgh University Press) ——— 1955. Contributions to the Study of Manx Phonology (Edinburgh: Nelson) ——— 1968. ‘The breaking of original long ē in Scottish Gaelic’, in Celtic studies: essays in memory of Angus Matheson 1912–1962, ed. by James Carney and David Greene (London: Routledge), pp. 65–71 Jardine, Adam and others. 2015. ‘Banyaduq Prestopped Nasals: Synchrony and Diachrony’, Oceanic Linguistics, 54: 548–78 Jones, Glyn E. 1982. ‘Central rounded and unrounded vowels in sixteenth-century Welsh’, Cardiff Working Papers in Welsh Linguistics, 2: 43–52. 429 Aspects of the historical phonology of Manx Jones, George. 2010. ‘The Gaelic of Jura: A Description’ (unpublished doctoral dissertation, Aberystwyth University) Kelly, John. 1804. A Practical Grammar of the Antient Gaelc; or, Language of the Isle of Mann, usually called Manks (London: John Nichols and son) ——— 1866. Fockleyr Manninagh as Baarlagh [‘Manx and English dictionary’] (Douglas: Manx Society) Kewley Draskau, Jennifer. 2008. Practical Manx (Liverpool: Liverpool University Press) Kneen, J. J. 1925–28. The Place-Names of the Isle of Man with their Origin and History, 6 vols (Douglas: Yn Cheshaght Ghailckagh) ——— 1927. Letter to editor, Revue Celtique, 44: 467 ——— 1931. A Grammar of the Manx Language (London: Oxford University Press) ——— 1937. The Personal Names of the Isle of Man (London: Oxford University Press) ——— 1938. English–Manx Pronouncing Dictionary (Douglas: Mona’s Herald) Kniezsa, Veronika. 1997. ‘The Origins of Scots Orthography’, in The Edinburgh History of the Scots Language, ed. by Charles Jones (Edinburgh: Edinburgh University Press), pp. 24–46 Ladefoged, Peter. 1967. Three Areas of Experimental Phonetics (London: Oxford University Press) Ladefoged, Peter and Ian Maddieson. 1996. The Sounds of the World’s Languages (Oxford: Blackwell) Ladefoged, Peter and others. 1998. ‘Phonetic Structures of Scottish Gaelic’, Journal of the International Phonetic Association, 28: 1–41 Lass, Roger. 2000. ‘Phonology and Morphology’, in The Cambridge History of the English Language, ed. by Roger Lass (Cambridge: Cambridge University Press), pp. 56–186 Lavin, Thomas J. 1957. ‘Notes on the Irish of East Mayo: 1’, Éigse, 8: 309–21 La Voie, Lisa. 1996. ‘Consonantal strength: results of a data base development project’, Working Papers of the Cornell Phonetics Laboratory 12: 207–32 LEIA: Lexique étymologique de l’irlandais ancien. Vendryes, Bachallery and Lambert 1959–1996 430 Aspects of the historical phonology of Manx Lenoach, Ciarán. 2012. ‘An Ghaeilge Iarthraidisiúnta agus a Dioscúrsa’, in An Chonair Chaoch: An mionteangachas sa Dátheangachas, ed. by Ciarán Lenoach, Conchúr Ó Giollagáin and Brian Ó Curnáin (Indreabhán, Leabhar Breac), pp. 19–109 Lewin, Christopher. 2011. ‘A sermon in Manx translated from the work of Thomas Wilson: Text and Commentary’, Scottish Gaelic Studies, 28: 175–210 ——— 2014a. Lioar-lhaih Ghaelgagh: Original Manx Gaelic Prose 1821–1907 (Douglas: Culture Vannin) ——— 2014b. ‘The persistence of grammatical gender in Manx Gaelic’ (unpublished undergraduate dissertation, University of Edinburgh) ——— 2015a. ‘Classical Manx, Revived Manx and English: competing standards’, in Minority languages in Europe and beyond – results and prospects, ed. by P. Sture Ureland and John Stewart (Berlin: Logos), pp. 23–31 ——— 2015b. ‘A Manx sermon from 1696’, Zeitschrift für celtische Philologie, 62: 45–96 ——— 2016a. ‘The syntax of the verbal noun in Manx Gaelic’, Journal of Celtic Linguistics, 17: 147–239 ——— 2016b. ‘Syntactic innovation in Manx and Sutherland Gaelic’, in Rannsachadh na Gàidhlig 8, ed. by Wilson McLeod, Anja Gunderloch and Rob Dunbar (Edinburgh: Dunedin), pp. 237–49 ——— 2016c. ‘The revivability of Manx Gaelic: a linguistic description and discussion of Revived Manx’ (unpublished masters dissertation, Aberystwyth University) ——— 2017a. ‘“Manx hardly deserved to live’: Perspectives on language contact and language shift’, Zeitschrift für celtische Philologie, 64: 141–205 ——— 2017b. ‘Scholarship and Language Revival: Language Ideologies in Corpus Development for Revived Manx’, Studia Celtica Posnaniensia, 2: 97–118 ——— 2018. ‘The vowel */әː/ ao in Gaelic dialects’, Papers in Historical Phonology, 3: 158–79 ——— 2019a. ‘John Rhys: a pioneer in Manx Studies’, Cambrian Medieval Celtic Studies, 77: 65–95 ——— 2019b. ‘The Manx Bible translation as a source for linguistic features and variation in vernacular eighteenth-century Manx’ (conference paper, 26th International Congress of Celtic Studies, Bangor University) 431 Aspects of the historical phonology of Manx Lewin, Christopher and Max W. Wheeler (eds). 2017. ‘The Principles and Duties of Christianity ... In English and Manks (Coyrle Sodjey), by Thomas Wilson, 1707’. <https://sussex.academia.edu/MaxWheeler> [accessed 18.01.2018] ——— (eds). 2019. ‘Lioar dy Hymnyn: A critical edition of the four Manx Gaelic hymn books printed 1795–1846’. <https://sussex.academia.edu/MaxWheeler> [accessed 31.01.2019] Lhuyd, Edward. 1707. Archæologia Britannica: Giving some account Additional to what has been hitherto Publish’d, of the Languages, Histories and Customs Of the Original Inhabitants of Great Britain, From Collections and Observations in Travels through Wales, Cornwal, Bas-Bretagne, Ireland and Scotland (Oxford: the author) Loakes, Debbie and others. 2008. ‘Phonetically prestopped laterals in Australian languages: a preliminary investigation of Warlpiri’, in Proceedings of the 9th Annual Conference of the International Speech Communication Association Brisbane, ed. by Janet Fletcher and others, pp. 90–3. <https://www.isca-speech.org/archive/interspeech_2008/i08_0090.html> [accessed 20.05.2019] MacBain, Alexander. 1911. An Etymological Dictionary of the Gaelic Language, 2nd edn (Stirling: Eneas Mackay) <http://www.ceantar.org/Dicts/MB2/> Mac Cárthaigh, Eoin. 2014. The Art of Bardic Poetry: A New Edition of Irish Grammatical Tracts I (Dublin: Dublin Institute for Advanced Studies) Mac Gill-Fhinnein, Gordon. 1966. Gàidhlig Uibhist a Deas (Téacsleabhar) (Baile Átha Cliath: Institiúid Ard-Léinn Bhaile Átha Cliath) Macquarrie, Charles W. 2015. ‘The Isle of Man in Medieval Gaelic Literature,’ in A New History of the Isle of Man. Volume III The Medieval Period 1000–1406, ed. by Seán Duffy and Harold Mytum (Liverpool: Liverpool University Press), pp. 281–304 Manx National Heritage. 2003. Skeealyn Vannin / Stories of Mann: The complete collection of Manx Language archive recordings made by the Irish Folklore Commission in 1948 (Douglas: Manx National Heritage) <https://www.youtube.com/watch?v=zOdn9m_EtAA&list=PLxvEbuf5XvSu bAhwdDxX9xBf14pfAxB5d> [accessed 26.04.2020] ——— 2017. ‘Manx language archives; wax cylinder recordings’. <https://www.youtube.com/watch?v=4cbdFPIL1Bk&list=PLxvEbuf5XvSuB CI4Wp4XFd2YUXt6ve7pT> [accessed 05.06.2019] 432 Aspects of the historical phonology of Manx Marstrander, Carl J. S. 1932. ‘Det Norske Landnåm på Man’, Norsk Tidsskrift for Sprogvidenskap, 6: 40–386 Matisoff, James A. 1975. ‘Rhinoglottophilia: the mysterious connection between nasality and glottality’, in Nasálfest: Papers from a symposium on nasals and nasalization ed. by James A.Matisoff, Charles A. Ferguson, Larry M. Hyman and John J. Ohala (Stanford, California: Dept. of Linguistics, Stanford University), pp. 265–87. McCone, Kim. 1994. ‘An tSean-Ghaeilge agus a réamhstair’, in Stair na Gaeilge, ed. by Kim McCone and others (Maigh Nuad: Roinn na Sean-Ghaeilge), pp. 61– 219 ——— 1996. Towards a Relative Chronology of Ancient and Medieval Celtic Sound Change (Maynooth: Department of Old Irish) McDonald, R. Andrew. 2007. Manx kingship in its Irish Sea setting, 1187–1229: King Rǫgnvaldr and the Crovan dynasty (Dublin: Four Courts Press) McKenna, Malachy. 2001. ‘Palatalization and Labials in the Irish of Torr, Co. Donegal’, in Béalra: Aistí ar Theangeolaíocht na Gaeilge, ed. by Brian Ó Catháin and Ruairí Ó hUiginn (Maigh Nuad: An Sagart), pp. 146–60 McLeod, Neil. 2000. ‘Kinship’, Ériu, 51: 1–22 McManus, Damian. 1994. ‘An Nua-Ghaeilge Chlasaiceach’, in Stair na Gaeilge, ed. by Kim McCone and others (Maynooth: National University of Ireland), pp. 335–4 Megaw, Basil. 1976. ‘Norseman and native in the Kingdom of the Isles: a reassessment of the Manx evidence’, Scottish Studies, 20: 1–44 Meisel, Jürgen M. 2009. ‘Second-language acquisition in early childhood’, Zeitschrift für Sprachwissenschaft, 28: 5–34 Menn, Lise. 1989. ‘Some people who don’t talk right: Universal and particular in child language, aphasia, and language obsolescence’, in Investigating Obsolescence: Studies in Language Contraction and Death, ed. by Nancy C. Dorian (Cambridge: Cambridge University Press), pp. 335–45 Mhac an Fhailigh, Éamonn. 1968. The Irish of Erris, Co. Mayo: A Phonemic Study (Dublin: Dublin Institute for Advanced Studies) Miller, Stephen. 2007. ‘“Here the Manx language lingers, and may linger some time longer”: Manx and English in Cregneash in 1901’, Zeitschrift für celtische Philologie, 55: 108– 21 433 Aspects of the historical phonology of Manx Montrul, Silvina A. 2008. Incomplete Acquisition in Bilingualism: Re-examining the Age Factor (Amsterdam: John Benjamins) ——— 2015. The Acquisition of Heritage Languages (Cambridge: Cambridge University Press) Moore, Arthur W. 1891. Carvallyn Gailckagh (Isle of Man: John Christian Fargher) ——— 1896. Manx Ballads and Music (Douglas: G. & R. Johnson) ——— 1903. Manx Names, 2nd edn (London: Elliot Stock) Moore, Arthur W. and John Rhŷs (eds). 1895. The Book of Common Prayer in Manx Gaelic. 2 vols (Oxford: Oxford University Press) Moore, A. W., Sophia Morrison and Edmund Goodwin. 1924. A Vocabulary of the Anglo-Manx Dialect (London: Oxford University Press) Morris-Jones, John. 1913. A Welsh Grammar Historical and Comparative (Oxford: Clarendon) Morrison, Donald Alasdair. 2018. ‘Vowel nasalisation in Scottish Gaelic: The search for paradigm uniformity effects in fine-grained phonetic detail’, in Supplemental Proceedings of the 2017 Annual Meeting on Phonology, ed. by Gillian Gallagher, Maria Gouskova and Sora Yin (Washington, DC: Linguistic Society of America) <https://journals.linguisticsociety.org/proceedings/index.php/amphonology/ar ticle/view/4236> [accessed 25.10.2018] Musil, Jakub. 2017. ‘Lateral Consonants in Scottish Gaelic: From dialectology towards a substance-free phonology’ (unpublished dissertation, University of Edinburgh) Ní Chasaide, Ailbhe. 1979. ‘Laterals of Gaoth Dobhair Irish and of Hiberno-English’, in Papers in Celtic phonology, ed. by Dónall P. Ó Baoill (Coleraine: New University of Ulster), pp. 54–78 Nic Fhlannchadha, Siobhán and Tina M. Hickey. 2007. ‘Acquiring an opaque gender system in Irish, an endangered indigenous language’, First Language, 37: 475– 99 Noyer, Ralph. 1990. ‘Secondary Epenthesis and Stress in Munster Irish’, Proceedings of the Harvard Celtic Colloquium, 10: 1–23 Ó Baoill, Colm. 1978. Contributions to a Comparative Study of Ulster Irish & Scottish Gaelic (Belfast: Queen’s University of Belfast) 434 Aspects of the historical phonology of Manx Ó Baoill, Dónall P. 1980. ‘Preaspiration, epenthesis and vowel lengthening— interrelated and of similar origin?’, Celtica, 13: 79–108 ——— 2012. ‘Tábhacht an athraithe ai go oi sa tSean-Ghaeilge mar léiriú ar fhorás canúna’, in Ilteangach, ilseiftiúil: Féilscríbhinn in ómós do Nicholas Williams, ed. by Liam Mac Amhlaigh and Brian Ó Curnáin (Dublin: Arlen House), pp. 410–29 Ó Briain, Micheál. 1923. ‘Hibernica’, Zeitschrift für celtische Philologie, 14: 309–34 Ó Buachalla, Breandán. 1988. ‘MacNeill’s Law and the Plural Marker –(e)an’, Proceedings of the Royal Irish Academy, 88C: 39–60. ——— 2002. ‘“Common Gaelic” revisited’, in Rannsachadh na Gàidhlig 2000, ed. by Colm Ó Baoill and Nancy R. McGuire (Obar Dheathain: An Clò Gaidhealach), pp. 1–12. Ó Cléirigh, C. R. 1958. Review of Jackson (1955), Éigse, 9: 139–142. Ó Cuív, Brian. 1944. The Irish of West Muskerry, Co. Cork: A Phonetic Study (Dublin: Dublin Institute for Advanced Studies) ——— 1957. ‘A Poem in Praise of Raghnall, King of Man’, Éigse, 8: 283–301 ——— 1980. ‘The verbal noun ending -áil and related forms’, Celtica, 13: 125–45 Ó Curnáin, Brian. 2007. The Irish of Iorras Aithneach County Galway. 4 vols (Dublin: Dublin Institute for Advanced Studies) ——— 2012. ‘An Ghaeilge Iarthraidisiúnta agus an Phragmaitic Chódmheasctha Thiar agus Theas’, in An Chonair Chaoch: An mionteangachas sa Dátheangachas, ed. by Ciarán Lenoach, Conchúr Ó Giollagáin and Brian Ó Curnáin (Indreabhán, Leabhar Breac), pp. 19–364 Ó Direáin, Séamas. 2015. A Survey of Spoken Irish in the Aran Islands, Co. Galway, 2nd edn (Dublin: Dublin Institute for Advanced Studies) <http://aranirish.nuigalway.ie/en/> [accessed 02.01.2018] Ó Dochartaigh, Cathair. 1978. ‘Unstressed long vowel shortening in Ulster: the evidence from Achill’, Éigse, 17: 331–58 ——— 1987. Dialects of Ulster Irish (Belfast: Institute of Irish Studies) ——— 1994–97. Scottish Gaelic Dialect Survey, 4 vols (Dublin: Dublin Institute for Advanced Studies) Ó Dónaill, Niall. 1977. Foclóir Gaeilge–Béarla (Baile Átha Cliath: An Gúm) <https://www.teanglann.ie/en/fgb/> 435 Aspects of the historical phonology of Manx Oftedal, Magne. 1956. The Gaelic of Leurbost Isle of Lewis (Oslo: H. Aschehoug & Co.) ——— 1963. ‘On “Palatalized” Labials in Scottish Gaelic’, Scottish Gaelic Studies, 10: 71–81 Ohala, John J. 1981. ‘The listener as a source of sound change’, in Papers from the Parasession on Language and Behavior, ed. by Carrie S. Masek, Roberta A. Hendrick and Mary Frances Miller (Chicago: Chicago Linguistics Society), pp. 178–203 ——— 1983. ‘The phonological end justifies any means’, in Proceedings of the XIIIth International Congress of Linguists, August 29–September 4, 1982, Tokyo, ed. by Shirō Hattori and Kazuko Inoue (Tokyo: Proceedings publication committee), pp. 232–43 ——— 1993. ‘The phonetics of sound change’, in Historical Linguistics: Problems and Perspectives, ed. by Charles Jones (London: Longman), pp. 237–78 Ó hIfearnáin, Tadhg. 2007. ‘Manx orthography and language ideology in the Gaelic continuum’, in Langues proches–Langues collatérales / Near Languages– Collateral Languages, ed. by Jean-Michel Eloy and Tadhg Ó hIfearnáin (Paris: Harmattan), pp. 159–70 Ó Máille, Tomás. 1910. The language of the Annals of Ulster. Manchester: Manchester University Press. Ó Maolalaigh, Roibeard. 1996. ‘The development of eclipsis in Gaelic’, Scottish Language, 14/15: 158–73 ——— 1997. ‘The Historical Short Vowel Phonology of Gaelic’ (unpublished doctoral dissertation, University of Edinburgh) <https://www.era.lib.ed.ac.uk/handle/1842/22535> [accessed 05.06.2019] ——— 2001. ‘Forás na nDeirí Díspeagtha -ean agus -ein i nGaeilge na hAlban’, in Béalra: Aistí ar Theangeolaíocht na Gaeilge, ed. by Brian Ó Catháin and Ruairí Ó hUiginn (Maigh Nuad: An Sagart), pp. 1–43 ——— 2003a. ‘Processes in Nasalization and Related Issues’, Ériu, 53: 109–32 ——— 2003b. ‘Varia II. A Possible Internal Source for Scottish Gaelic Plural -an’, Ériu, 53: 157–61 ——— 2003c. ‘Varia III. Vocalic Variation in air-, aur-’, Ériu, 53: 163–9 436 Aspects of the historical phonology of Manx ——— 2006. ‘Coibhneas idir consan (dh / gh) agus guta i stair na Gaeilge’, in Aistí ar an Nua-Ghaeilge, ed. by Aidan Doyle and Siobhán Ní Laoire (Baile Átha Cliath: Cois Life), pp. 41–78 ——— 2007. ‘Péisteoigín itheas éadach: The significance of leaghmhan ‘moth’’. in Fil súil nglais A Grey Eye Looks Back: a Festschrift in honour of Colm Ó Baoill, ed. by Sharon Arbuthnot and Kaarina Hollo (Ceann Drochaid, Perthshire: Clann Tuirc), pp. 213–40 ——— 2008a. ‘The Scotticisation of Gaelic: a reassessment of the language and orthography of the Gaelic Notes in the Book of Deer’, in Studies on the Book of Deer, ed. by Katherine Forsyth (Dublin: Four Courts Press), pp. 179–274 ——— 2008b. ‘‘Bochanan modhail foghlaimte’: Tiree Gaelic, lexicology and Glasgow’s Historical Dictionary of Scottish Gaelic’, Scottish Gaelic Studies, 24: 373–523 ——— 2013. ‘Gaelic gach uile/a h-uile and the genitive of time’, Éigse, 38: 41–93 ——— 2014a. ‘Am buadhfhacal meadhan-aoiseach meranach agus mearan, mearanach, dàsachdach, dàsan(n)ach na Gàidhlig’, Scottish Studies, 37: 183– 206 ——— 2014b. ‘Glottal and related features in the Gaelic languages’ (unpublished paper presented at 8th Celtic Linguistics Conference, University of Edinburgh) ——— 2019. ‘Fadó: a conservative survival in Irish?’, Éigse, 40: 207–25 ——— forthcoming a. ‘Cumhachd àireamhan: Cinneachadh nan taobhach caola: Rochainnteachd agus for-dheasachadh’, in Litreachas, Eachdraidh & Cànan: Rannsachadh na Gàidhlig 7, ed. by Sheila M. Kidd, Thomas Owen Clancy and Roibeard Ó Maolalaigh (Glasgow: Celtic & Gaelic, University of Glasgow), pp. 320–81 ——— forthcoming b. ‘Lexical and phonological variation in words for ‘ankle’ in Gaelic’ Ó Muircheartaigh, Peadar. 2015. ‘Gaelic dialects present and past: a study of modern and medieval dialect relationships in the Gaelic languages’ (unpublished doctoral dissertation, University of Edinburgh) ——— 2016. ‘Fin as Ossian revisited: A Manx ballad in Belanagare and its significance’, Zeitschrift fur celtische Philologie, 63: 95–127 Ó Murchú, Máirtín. 1986. ‘R caol i dtús focal: blúire canúineolaíochta’, in Féilscríbhinn Thomáis de Bhaldraithe, ed. by Seosamh Watson (Baile Áth 437 Aspects of the historical phonology of Manx Cliath: Coiste Fhéilscríbhinn Thomáis de Bhaldraithe, An Coláiste Ollscoile), pp. 19–26 ——— 1989a. East Perthshire Gaelic (Dublin: Dublin Institute for Advanced Studies) ——— 1989b. ‘Some Irish phonological rules and their chronological order’, Ériu, 40: 143–6 Ó Murchú, Séamas. 1989. ‘Réimniú an bhriathair i nGaeilge Charna’ (unpublished doctoral dissertation, National University of Ireland, Dublin) O’Rahilly, T. F. 1926. ‘Etymological Notes’, Scottish Gaelic Studies, 1: 28–37 ——— 1927. ‘Etymological Notes. II’, Scottish Gaelic Studies, 2: 13–29 ——— 1931. ‘Etymological Notes III’, Scottish Gaelic Studies, 3: 52–72 ——— 1932. Irish Dialects Past and Present (Dublin: Browne and Nolan) ——— 1942a. ‘Iarann, lárag, etc.’, Ériu, 13: 119–27 ——— 1942b. ‘Notes, Mainly Etymological’, Ériu, 13: 144–219 Ó Searcaigh, Séamus. 1925. Foghraidheacht Ghaedhilge an Tuaiscirt (Béal Feirste: Brún agus Ó Nualláin) Ó Sé, Diarmuid. 1984. ‘Coimriú siollaí tosaigh sa Ghaeilge’, Éigse, 20: 171–86 ——— 1989. ‘Contributions to the study of word stress in Irish’, Ériu, 40: 147–78 ——— 1991. ‘Prosodic change in Manx and lexical diffusion’, in Language Contact in the British Isles. Proceedings of the Eighth International Symposium on Language Contact in Europe, Douglas, Isle of Man, 1988, ed. by P. Sture Ureland and George Broderick (Tübingen: Niemeyer), pp. 159–80 ——— 2000. Gaeilge Chorca Teangeolaíochta Éireann) Dhuibhne (Baile Átha Cliath: Institiúid Ó Súilleabháin, Pádraig (ed.). 1953. Rialachas San Froinsias (Baile Átha Cliath: Institiúid Árd-Léinn Bhaile Átha Cliath) Parry-Williams, T. H. 1923. The English Element in Welsh: a Study of English Loanwords in Welsh (London: Cymmrodorion) Patterson, Nerys W. 1990. ‘Patrilineal kinship in early Irish society: the evidence from the Irish law texts’, The Bulletin of the Board of Celtic Studies, 37: 133–65 Pedersen, Holger. 1909–13. Vergleichende Grammatik der keltischen Sprachen. 2 vols (Göttingen: Vandenhoeck und Ruprecht) 438 Aspects of the historical phonology of Manx Pickeral, John. 1988–90. ‘A Preliminary Phonology of Manx’, Orbis, 35: 81–97 PNIM: Placenames of the Isle of Man. Broderick 1994–2005 Polinsky, Maria. 2006. ‘Incomplete acquisition: American Russian’, Journal of Slavic Linguistics, 14: 191–262 Prince, Alan S. 1990. ‘Quantitative consequences of rhythmic organization’, Chicago Linguistic Society, 26: 355–98 Quiggin, E. C. 1906. A Dialect of Donegal (Cambridge: Cambridge University Press) Quilliam, Leslie. 1989. Surnames of the Manks (Peel: Cashtal Books) Rhŷs, John. 1894. The Outlines of the Phonology of Manx Gaelic (Douglas: Manx Society). Also published as an appendix to vol. 2 of Moore and Rhŷs (1895). Robertson, Charles M. 1902. ‘Sutherland Gaelic’, Transactions of the Gaelic Society of Inverness, 25: 84–125 Røsstad, Rune. 2011. ‘Kva fortel kjeldene om innovasjonen dl i norsk?’, Maal og Minne, 103, 96–119 Russell, Paul. 1995. An Introduction to the Celtic Languages (London: Longman) Salmons, Joseph C. 1992. Accentual change and language contact: comparative survey and case study of early northern Europe (London: Routledge) Sammallahti, Pekka. 1998. The Saami languages: an introduction (Kárášjohka: Davvi girji) Sandøy, Helge. 2005. ‘The typological development of the Nordic languages: I: Phonology’. In The Nordic languages: An international handbook of the history of the North Germanic languages, ed. by Oskar Bandle and others (Berlin: Walter de Gruyter), pp. 1852–71. Sasse, Hans-Jürgen. 1992a. ‘Theory of language death’, in Language Death: Factual and Theoretical Explorations with Special Reference to East Africa, ed. by Matthias Brenzinger (Berlin: Mouton de Gruyter), pp. 7–30. Sasse, Hans-Jürgen. 1992b. ‘Language decay and contact-induced change: Similarities and differences’, in Language Death: Factual and Theoretical Explorations with Special Reference to East Africa, ed. by Matthias Brenzinger (Berlin: Mouton de Gruyter), pp. 59–80. Schmidt, Annette. 1985. Young people’s Dyirbal. An example of language death from Australia (Cambridge: Cambridge University Press) 439 Aspects of the historical phonology of Manx Scouller, Alastair MacNeill. 2017. ‘The Gaelic Dialect of Colonsay’ (University of Edinburgh, unpublished doctoral dissertation) Sebba, Mark. 1998. ‘Orthography as practice and ideology: the case of Manx’ (Lancaster University Department of Linguistics and Modern English Language, Centre for Language in Social Life: Working Paper 102) <https://eprints.lancs.ac.uk/id/eprint/60448/1/clsl102.pdf> [accessed 27.06.2020] SGDS: Scottish Gaelic Dialect Survey. Ó Dochartaigh 1994–97. Shaw, John. 1971. ‘L’évolution de vieil-irlandais áe, óe, aí, oí dans les dialectes gaéliques’, Études celtiques, 12: 147–56 Shaw, William. 1780. A Galic and English Dictionary; containing all the words in the Scotch and Irish dialects of the Celtic, that could be collected from the voice, and old books and MSS. 2 vols (London: the author) Sommerfelt, Alf. 1922. The dialect of Torr, Co. Donegal (Christiania: Dybwad) ——— 1959. ‘The study of Celtic phonetics and phonemics’, Phonetica, 3: 51–63 Stockman, Gearóid. 1986. ‘Giorrú gutaí fada aiceannta i nGaeilge Chúige Uladh’, in Féilscríbhinn Thomáis de Bhaldraithe, ed. by Seosamh Watson (Baile Átha Cliath: Coiste Fhéilscríbhinn Thomáis de Bhaldraithe, An Coláiste Ollscoile), pp. 11–8 Stowell, Brian. 2005. ‘The case of Manx Gaelic’, in Rebuilding the Celtic Languages: Reversing Language Shift in the Celtic Countries, ed. by Diarmuid Ó Néill (Talybont: Y Lolfa), pp. 383–416 Stüber, Karin. 1997. ‘The Inflection of Masculine and Feminine N-Stems in Irish’, Ériu, 48: 229–37 Ternes, Elmar. 1973. The Phonemic Analysis of Scottish Gaelic. 1st edn (Hamburg: Helmut Buske) ——— 2006. The Phonemic Analysis of Scottish Gaelic. 3rd edn (Dublin: Dublin Institute for Advanced Studies) Thomson, Robert L. 1950. ‘The syntax of the verb in Manx Gaelic’, Études celtiques, 5: 260–92 ——— 1953. ‘Early Manx: A Contribution to the Historical Study of Manx Gaelic arranged as a supplementary volume to the Moore-Rhys edition of the Phillips Prayer Book (1610)’ (unpublished B.Litt. dissertation, University of Glasgow) 440 Aspects of the historical phonology of Manx ——— 1954–59. ‘A Glossary of Early Manx’, Zeitschrift für celtische Philologie, 24, 272–307; 25, 100–40, 264–308; 27: 79–160 ——— 1960. ‘Svarabhakti and some associated changes in Manx’, Celtica, 5: 116– 26 ——— 1960–62. ‘The Manx Traditionary Ballad’, Études Celtiques, 9: 521–48, 10: 60–87 ——— 1963. ‘Norse Loanwords in Manx’, Scottish Gaelic Studies, 10: 65–8 ——— 1968. ‘Edward Lhuyd in the Isle of Man?’, in Celtic studies: essays in memory of Angus Matheson 1912–1962, ed. by James Carney and David Greene (London: Routledge), pp. 170–82 ——— 1969. ‘The Study of Manx Gaelic. Sir John Rhys Memorial Lecture’, Proceedings of the British Academy, 55: 177–210 ——— 1970. Foirm na n-Urrnuidheadh. John Carswell’s Gaelic translation of the Book of Common Order (Edinburgh: Scottish Gaelic Texts Society) ——— 1976. ‘The stressed vowel phonemes of a Manx idiolect’, Celtica, 11, 255– 263 ——— 1981. Lessoonyn Sodjey ’sy Ghailck Vanninagh [‘Further lessons in Manx Gaelic’] (Douglas: Yn Cheshaght Ghailckagh) ——— 1983. ‘The Continuity of Manx’, in The Viking Age in the Isle of Man: Select Papers from the Ninth Viking Congress, Isle of Man, 4-14 July 1981, ed. by Christine Fel, Peter Foote, James Graham-Campbell and Robert L. Thomson (London: Viking Society for Northern Research), pp. 169–74 ——— 1984. ‘Manx’, in Language in the British Isles, ed. by Peter Trudgill (Cambridge: Cambridge University Press), pp. 306–17 ——— 1988. ‘Manx-Latin gilbogus again’, Celtica, 20: 141–44 ——— 1990. ‘The Revd Dr John Kelly as a Lexicographer’, Proceedings of the Isle of Man Natural History and Antiquarian Society, 9: 443–58 ——— 1991. ‘Foreign elements in the Manx vocabulary’, in Language Contact in the British Isles. Proceedings of the Eighth International Symposium on Language Contact in Europe, Douglas, Isle of Man, 1988, ed. by P. Sture Ureland and George Broderick (Tübingen: Niemeyer), pp. 127–40 ——— 1992. ‘The Manx language’, in Donald MacAulay (ed.), The Celtic Languages (Cambridge: Cambridge University Press), pp. 100–36 441 Aspects of the historical phonology of Manx ——— 1995. Pargys Caillit. An abridgement of John Milton’s Paradise Lost By Thomas Christian with the anonymous translation of Thomas Parnell’s The Hermit (Douglas: Centre for Manx Studies) ——— 1997. Paart dy homileeyn ny sharmaneyn oikoil Agglish Hostyn (Douglas: Yn Cheshaght Ghailckagh) ——— 1998. Yn Fer-raauee Creestee (1763) The Christian Monitor (1686): A Bilingual Edition with Notes and Introduction (Douglas: Yn Cheshaght Ghailckagh) ——— 1999. ‘Edward Lhuyd’s Geirieu Manaweg II’, Celtica, 23: 390–407 ——— 2000. ‘Manx’, in Glanville Price (ed.), Languages in Britain & Ireland (Oxford: Blackwell), pp. 58–89 ——— 2015†. ‘Language in Man: prehistory to literacy,’ in A New History of the Isle of Man. Volume III The Medieval Period 1000–1406, ed. by Seán Duffy and Harold Mytum (Liverpool: Liverpool University Press), pp. 241–56 Thomson, Robert L., and Adrian J. Pilgrim. 1988. An Outline of Manx Language and Literature (Douglas: Yn Cheshaght Ghailckagh) Thurneysen, Rudolf. 2003. A Grammar of Old Irish, 1st edn 1949 (Dublin: Dublin Institute for Advanced Studies). Trask, Larry. 1995. ‘Summary: Buccalization’. <https://linguistlist.org/issues/6/6960.html> [accessed 22.09.2019] Trebitsch†, Rudolf and Ulla Remmer. 2003. The collections of Rudolf Trebitsch Celtic recordings 1907–09. Audio CD and accompanying booklet (Vienna: Österreichische Akademie der Wissenschaften) Trudgill, Peter. 1974. ‘Linguistic change and diffusion: Description and explanation in sociolinguistic dialect geography’, Language in Society, 3: 215–46 Ua Súilleabháin, Seán. 1994. ‘Gaeilge na Mumhan’, in Stair na Gaeilge, ed. by Kim McCone and others (Maigh Nuad: Roinn na Sean-Ghaeilge), pp. 479–538 Unsworth, Sharon and others. 2014. ‘The Role of Age of Onset and Input in Early Child Bilingualism in Greek and Dutch’, Applied Psycholinguistics, 35: 765– 805 Vendryes†, J., E. Bachallery and P.-Y. Lambert. 1959–1996. Lexique étymologique de l’irlandais ancien. 7 vols (incomplete) (Dublin: Dublin Institute for Advanced Studies) 442 Aspects of the historical phonology of Manx Vikar, Anders. Contributions to the History of the Durham Dialect (Malmö: Röhr) Wagner, Heinrich. 1956. Review of Jackson (1955), The Modern Language Review, 51: 107–9 ——— 1958–69. Linguistic Atlas and Survey of Irish Dialects. 4 vols (Dublin: Dublin Institute for Advanced Studies) ——— 1959. Gaeilge Theilinn: Foghraidheacht Gramadach Téacsanna (Baile Átha Cliath: Institiúid Ard-léinn Bhaile Átha Cliath) ——— 1964. ‘Nordeuropäische Lautgeographie’, Zeitschrift für celtische Philologie, 29: 225–98 Watson, Seosamh. 1985. ‘Caducous and Fallen Vowels in Irish and Scottish Gaelic’, Ériu, 36: 125–36 Watson, William J. 1926. The History of the Celtic Place-Names of Scotland (Edinburgh: Blackwood) Wells, John C. 1982. Accents of English, vol. 2 The British Isles (Cambridge: Cambridge University Press) Wentworth, Roy G. 2002. ‘Na bolaichean aig na Geàrrlaich ’s an loch làn diubh: fòineimean taobhach ann an dualchainnt Ghàidhlig an Ros an Iar’, in Rannsachadh na Gàidhlig 1, ed. by Nancy R.McGuire and Colm Ó Baoill (Obar Dheathain: An Clò Gaidhealach), pp. 91–9 2005†. Rannsachadh air Fòn-eòlas Dualchainnt Ghàidhlig Gheàrrloch, Siorrachd Rois. Online publication (Dublin: Dublin Institute for Advanced Studies) <https://www.dias.ie/wp-content/uploads/2011/04/gearrloch.pdf> (accessed 21.12.2017) Whall, W. B. 1913. Ships, Sea Songs and Shanties (Glasgow: James Brown & Son) Wheeler, Max W. 2017. ‘Survey of grammatical gender in Classical Manx’ (unpublished paper) <https://www.academia.edu/32892848/Survey_of_grammatical_gender_in_C lassical_Manx> [accessed 14.05.2019] ——— 2018. Fockleyr Chregeen aa-orderit [Cregeen’s dictionary re-ordered]. 2nd edn <https://www.academia.edu/35892604/Fockleyr_Chregeen_aaorderit._New_edition._Feb._2018> [accessed 21.10.2018] 443 Aspects of the historical phonology of Manx ——— 2019. ‘Phillips’ Manx translation of the Psalms (MNH MS 00003) Diplomatic edition’ <https://www.academia.edu/41457167/Phillips_Manx_translation_of_the_Ps alms_MNH_MS_00003_._Corrected> [accessed 20.06.2020] ——— 2020. ‘Psalmyn Ghavid. A critical parallel text edition of the two Manx translations of the Psalms, alongside the English’ <https://www.academia.edu/41917797/Psalmyn_Ghavid._A_critical_parallel _text_edition_of_the_two_Manx_translations_of_the_Psalms_alongside_the_ English> [accessed 20.06.2020] Williams, Nicholas. 1994a. ‘Na Canúintí a Theacht chun Solais’, in Stair na Gaeilge, ed. by Kim McCone and others (Maynooth: National University of Ireland), pp. 447–78 ——— 1994b. ‘An Mhanainnis’, in Stair na Gaeilge, ed. by Kim McCone and others (Maynooth: National University of Ireland), pp. 703–44 Wood, Simon N. 2006. Generalized Additive Models (Boca Raton: Chapman and Hall/CRC) Wright, Joseph. 1898–1905. The English Dialect Dictionary. 6 vols (London: Henry Frowde) <http://eddonline-proj.uibk.ac.at/edd/index.jsp> 444 Aspects of the historical phonology of Manx 445 Aspects of the historical phonology of Manx 446 Aspects of the historical phonology of Manx 447

RELATED PAPERS

RELATED TOPICS

Log In

Aspects of the historical phonology of Manx

Aspects of the historical phonology of Manx

Related Papers

RELATED PAPERS

RELATED TOPICS