Studia Celto-Slavica 9:Journal of the Learned Association Societas Celto-Slavica, volume 9

Mark Ó Fionnáin

This volume provides accounts of well-established themes of general Celtic inquiry from new theoretical perspectives, in addition to addressing new areas of research that have remained largely unexplored. The collection includes contributions by both established and young scholars on diverse aspects of culture, literature and linguistics, reflecting the multidisciplinary character of current trends in Celtology. The linguistic section of the book includes chapters dealing with Welsh phonology and possible areas of influence of the Brittonic language on English, as well as with the issues of translating culture-specific aspects of medieval Welsh texts and the problems of standardising Irish orthography and font. The second part of the volume is devoted to literature and considers neglected, and heretofore unexplored, aspects of Welsh-language poetry, fiction and children’s literature, the work of John Cowper Powys, and Scottish film in the theoretical context of post-humanism. Approaching these issues from different angles and using different methodologies, the collection highlights the connections between long-established academic areas of interest and popular culture, broadening the horizon of Celtic scholarship. Chapter One Prolegomena to a Study of Welsh Vocalism Sabine Asmus and Cormac Anderson Chapter Two Diphthongs in the North of Wales PawełTomasz Czerniak Chapter Three Translating or Mistranslating Celtic Law in the Polish versions of the “Four Branches of the Mabinogi” Katarzyna Jaworska-Biskup Chapter Four Revision of the Most Known Celtic Features of English Ireneusz Kida Chapter Five An Irish Solution to an Irish Problem: The (Neverending) Issue of Standardising Irish Mark Ó Fionnáin Chapter Six Hiraeth as Allegorical Form: Fflur Dafydd’s Atyniad Aleksander Bednarski Chapter Seven Is this Machine Alive? Machine-like, Biotic, Autopoietic Systems in Contemporary Cinema Maciej Czerniakowski Chapter Eight A Postcolonial Traveller? A Take on the Poetry of Iwan Llwyd Siôn Pennar Chapter Nine Barti Ddu: A Welsh Colonial Hero in a Post-colonial Text? Awen Schiavone Chapter Ten “History is not to be trifled within this way”? Re-contextualising John Cowper Powys’ Owen Glendower (1940) Angelika Reichmann

CELTIC FROM THE WEST 3. The Celtic languages and groups called Keltoi (i.e. ‘Celts’) emerge into our written records at the pre-Roman Iron Age. The impetus for this book is to explore from the perspectives of three disciplines—archaeology, genetics, and linguistics—the background in later European prehistory to these developments. There is a traditional scenario, according to which, Celtic speech and the associated group identity came in to being during the Early Iron Age in the north Alpine zone and then rapidly spread across central and western Europe. This idea of ‘Celtogenesis’ remains deeply entrenched in scholarly and popular thought. But it has become increasingly difficult to reconcile with recent discoveries pointing towards origins in the deeper past. It should no longer be taken for granted that Atlantic Europe during the 2nd and 3rd millennia BC were pre-Celtic or even pre-Indo-European. The explorations in Celtic from the West 3 are drawn together in this spirit, continuing two earlier volumes in the influential series.

Studia Celto-Slavica 9 Studia Celto-Slavica Journal of the Learned Association Societas Celto-Slavica Volume 9 - 2018 Centre de Recherche Bretonne et Celtique STUDIA CELTOSLAVICA Collection STUDIA CELTO-SLAVICA STUDIA CELTO-SLAVICA Journal of the Learned Association Societas Celto-Slavica volume 9 Centre de Recherche Bretonne et Celtique Université de Bretagne Occidentale – Brest 2018 Contents Statement from the Editorial Board Celto-Slavica in Heidelberg Jadranka Gvozdanović Which Linguistic Model for Brittany? Gary Manchec-German iii Semantic (Ir)regularities in Action Nouns in Irish Maria Bloch-Trojnar Opportunities Seized: from Tolstóigh to Pelévin Mark Ó Fionnáin Multiple Versions of Breuddwyd Pawl as a Source to Study the Work of Welsh Translators Elena Parina An Acoustic Study of Welsh and Slavonic Rhotics Sylwester Jaworski & Sabine Asmus i 1 31 63 79 101 Statement from the Editorial Board Studia Celto-Slavica is henceforth to become a new publication series at the Centre for Research in Breton and Celtic Studies (crbc ) of the University of Western Brittany (ubo ), Brest. This builds on long-term collaboration between research centres in Celtic Studies at Ulster University and UBO. Studia Celto-Slavica was originally launched in 2006 with the publication of the Societas Celto-Slavica inaugural colloquium proceedings at Coleraine.1 Altogether, eight volumes of the series have been published.2 The series invites contributions on topics such as Celto-Slavic isoglosses, Indo-European linguistic heritage and archaeological data, Celtic place-names in the Slavic countries, parallels in languages, literatures and cultures, as well as similarities between Celtic and Slavic narrative and folklore traditions. Developing from this original Celto-Slavica framework, the series now accepts articles on Celtic Studies topics in general. 1 Mac Mathúna, S., & Fomin, M., eds., Parallels between Celtic and Slavic. Proceedings of the First International Colloquium on Links and Parallels between Celtic and Slavic Traditions. Studia Celto-Slavica 1, Coleraine: TSO Publishers, 2006. 2 Mac Mathúna, S., Mikhailova, T., Fomin, M. & G. Bondarenko, eds., Proceedings of the Second International Colloquium of Societas Celto-Slavica. Moscow: Moscow State University Publishers, 2009; Brozović-Rončević, D., Fomin, M., & R. Matasović, eds., Celts and Slavs in Central and Southeastern Europe. Proceedings of the Third International Colloquium of Societas Celto-Slavica held at IUC, Dubrovnik, 18–19 September 2008. Zagreb: Institute for Croatian Language and Linguistics, 2010; Stalmaszczyk, P., & M. Fomin, eds., Dimensions and Categories of Celticity: Studies in Language. Proceedings of the Fourth International Colloquium of Societas Celto-Slavica. Part 1. Łódź: Łódź University Press, 2010; Fomin, M., Jarniewicz, J., & P. Stalmaszczyk, eds., Dimensions and Categories of Celticity: Studies in Literature and Culture. Proceedings of the Fourth International Colloquium of Societas Celto-Slavica. Part 2, Łódź: Łódź University Press, 2010; Fomin, M., Blažek, V., & P. Stalmaszczyk, eds., Transforming Traditions: Studies in Archaeology, Comparative Linguistics and Narrative. Proceedings of the Fifth International Colloquium of Societas Celto-Slavica held at Příbram, 26–29 July 2010. Łódź: Łódź University Press, 2012; Johnston, D., Parina, E. & Fomin, M., eds., ‘Yn llawen iawn, yn llawn iaith’: Proceedings of the Sixth International Colloquium of Societas Celto-Slavica. Aberystwyth: University of Wales Centre for Advanced Welsh and Celtic Studies, 2015; Jones, A. Ll., & Fomin, M., eds., ‘Y geissaw chwedleu’: Proceedings of the Seventh International Colloquium Societas Celto-Slavica, Bangor University: School of Welsh. i Celto-Slavica in Heidelberg Jadranka Gvozdanović The 8th International Colloquium of the Learned Association Societas CeltoSlavica was held between 1–3 September 2016 at the University of Heidelberg in Germany. Heidelberg was an old Celtic stronghold. Inhabited since around 5,000 bc , the earliest clearly recognisable Urnfield (around 1,200 bc ) and early La Tène (since 600 bc ) cultures have been attested on the so-called Holy Hill (Heiligenberg) on the opposite side of the Neckar river, visible on the picture below. Figure 1. Heidelberg (photo credit: Liza Walther) The local Celts had a highly developed and steady culture on the Holy Hill, a fact that is demonstrated by an Iron-age well (Bittersbrunnen), ritual sites and remnants of a wall. In the fifth century BC, the Holy Hill was the political centre of the lower Neckar valley. The Celts left the area around 100 bc, when Germanic tribes (especially the Sueben) took over their inhabitation areas. iii INTRODUCTION The colloquium started with a workshop on archaeogenetic and linguistic evidence about the earliest Celtic migrations. There was a lively discussion about the alternative Near Eastern vs. Steppe hypotheses, for which David Bradley’s research group presented new evidence that genetically and chronologically, clearly discernible Celtic ancestors came from the Steppe around four millennia ago. There was also an earlier migration from Near Eastern areas (about seven millennia ago), but Celtic lineage could not be ascertained for this migration. Paul Heggarty of the Jena Institute for the History of Mankind discussed a larger picture of Indo-European migrations, including Indo-Arian, for which linguistic evidence can be plotted even earlier in prehistory. John Koch discussed evidence for early Celtic presence in the Atlantic regions, and Eugenio Lujan presented the newest evidence on Celtiberian based on a project collecting and analysing a relatively complete corpus of the available texts. Tatyana Mikhailova discussed cultural construal of the early Celtic origins, while Václav Blaźek discussed critically the existing methodology for identifying pre-Indo-European substrate elements based on exclusion from the common Proto-Indo-European phonological pattern. The next two days of the colloquium were dedicated to cultural, linguistic and literary aspects of Celtic traditions. The specific topics included the origins of Celtic languages and language contact, Celtic archaeology and toponymy of Central and South-eastern Europe, historical poetics of Celtic narrative traditions, translations to and from Celtic languages, folklore as a tool for the internal identity construction as well as the medium for the absorption of foreign influences, and teaching of Celtic Studies. The 8th colloquium of the Societas Celto-Slavica Learned Association has shown again how vital Celtic studies are in the Slavic countries, represented at the event by scholars from Russia, Poland and Czech Republic, fully comparable to the tradition developed in the Celtic lands, including the event delegates from Ireland, both North and South, Wales as well as Brittany. Professor Jadranka Gvozdanović Chair, The 8 International Colloquium of Societas Celto-Slavica at the University of Heidelberg th iv Which Linguistic Model for Brittany? Gary Manchec-German1 1. Introduction unesco has classified Breton, the Celtic language spoken in the West of Brittany in northwestern France, among the ‘seriously endangered’ languages of the world.2 As such, Breton is one of the thousands of minority languages and dialects which may disappear by the end of the 21st century (cf. infra). In the first part of this article, I identify some of the major socio-economic, historical and sociolinguistic causes for the rapid decline in the use of Breton over the past 150 years. I then present some of the technical and practical problems hindering communication between speakers of the naturally transmitted, basilectal varieties of Breton3 and the normalized variety currently taught in the schools. Finally, one of the major objectives of this paper is to outline a project for creating pedagogical tools targeting native and passive speakers who make up, by far, the largest pool of speakers and potential speakers of the language today. The goal of this article is thus as social as it is linguistic, namely to assist those who want to better understand the function of the varieties language they still speak and, in the case of passive speakers, to provide educational 1 The author lives in the parish of Saint Yvi, southern Cornouaille (Finistère) and has been exposed to the Breton of this region since childhood. The subject of his 1984 thesis was the Phonology and morphology of southern Cornouaillais Breton within the context of Pierre Le Roux’s Atlas Linguistique du la Basse-Bretagne, unpublished, University of Western Brittany. 2 http://www.unesco.org/languages-atlas/index.php 3 In this article, the terms “vernacular Breton”, “dialect speakers”, “traditional speakers” are used interchangeably with the general meaning of “basilectal speakers”. The term “badume” adds an extra dimension to the concept of “basilect”. First coined by Le Berre and Le Dû (1996), it refers to the purely oral, naturally transmitted, highly stigmatized and fragmented regional and social varieties of Breton used, often with covert prestige and with a strong affective sentiments, in socially stigmatized speech communities. These days, Breton speakers tend to restrict their use the language to people they know intimately. Fañch Broudic (1999) describes it as “la langue de la convivialité”. Jean Le Dû (pc.) compares it to “tutoiement”, i.e. one only speaks Breton to people with whom one has strong bonds (family, neighbours, etc.). This has also been my own experience. 1 WHICH LINGUISTIC MODEL FOR BRITTANY? resources which would enable them to (re)learn the varieties they heard during their childhood. Currently, virtually none exist to assist them. At another level, the project might also prove useful for those seeking to build bridges and consolidate ties between the surviving linguistic networks and clusters of traditional Breton speakers as well as potentially interested learners of the language who wish to communicate with the dialect speakers in their areas. If only in a small way, I also hope that some of the points presented in this paper may be of use to those working on other minority languages and dialect communities elsewhere in the world. 2. Numbers of speakers As mentioned in the introduction, the future of Breton is bleak and the numbers of speakers have been in constant decline since the end of the 19th century. In 1900, Breton was the first language of 90% of the inhabitants of Western Brittany, 50–60% of whom were monolingual (Broudic 1999). By 1950, there were still an estimated 1,100,000 speakers of whom 700,000 used the language as their primary medium of communication. 100,000 of these were monolinguals (Gourvil 1952). By the 1980s, nearly all of the latter had passed away (Favereau 1991).4 The good news, however, is that it has been estimated that there are still around 120,000–150,000 native speakers of Breton today, more than in any other Celtic-speaking nation except Wales.5 Broudic (2007) put this figure at 240,000 in 1997 (297,000 including Breton speakers living in other regions of France) while, two years later, an insee study estimated this number to be slightly higher, 257,000 (two-thirds of whom were over 50 years of age at the time, ibid.) On the down side, the decline in the number of speakers is accelerating at a very rapid pace. By 2007 this number had dropped to 172,000, 103,000 of whom were over 60 years of age (Broudic 2009). In this same study, Broudic 4 Marie-France Kerjos, a secretary at the town hall of Saint Yvi, Finistère, informed me that the last monolingual speakers in the parish died in the mid-1980s. 5 It must be kept in mind that the French government has always refused to take into account the number of minority language speakers in France in the national census. For this reason, no official statistics exist regarding the number of minority language speakers in France, only estimates. Furthermore, given that the language has been so stigmatized, many older people I have known, who have an excellent working knowledge of the language, claim they have “never spoken” it or “have forgotten it”. This fact could considerably affect the statistics and must be taken into account. 2 GARY MANCHEC-GERMAN predicted that these figures would drop to 122,000 by 2017 and to around 100,000 by 2020 (ibid.).6 The number of speakers who actually use Breton on a daily basis is another matter with numbers estimated at 70,000 in 1999 and only 35,000 in 2007 (ibid.), a figure that strikes me as excessively low, but which certainly must not be dismissed. As a rule of thumb, most fluent speakers of the traditional varieties today are a) over 70 years old, b) live in the poorer areas of rural Western Brittany, c) are less formally educated than the average citizen and d) tend to be confined to lower-paying jobs. This cocktail is well-known to sociolinguists around the world and it is important to highlight that negative attitudes towards Breton often have far more to do with condescending attitudes on the part of French-speaking urbanites towards older, economically deprived members of the surrounding rural communities (i.e. les sans dents ‘the toothless ones’ as President Holland was quoted to have said in recent years) than with any inherent shortcomings in the varieties they speak. The position adopted in this article is the same as for most linguists: the vernacular Breton varieties are clearly coherent linguistic systems in and of themselves (and standard Breton is merely another variety that has been added to the mix). 3. The social and historical background for the language shift Although the speed of the language shift has mystified observers, the reasons which motivated it are relatively straightforward and, as we have just seen, are intimately linked to the social history of the Breton people and their language. I shall now attempt to outline some of the major stages of the passage towards French. The key event which triggered the decline steady in the use of Breton probably results from the gallicization of the Breton aristocracy during the 11th and 12th centuries. Unlike Wales, Brittany never developed an elevated, secular, Breton-medium literature (written or oral) comparable to the poetry produced by the Beirdd y Tywysogion (Poets of the Welsh Princes, 12th–13th centuries). Vernacular Breton remained the language of an impoverished peasantry concentrated in the western areas of the peninsula7 while Latin and French, on the other hand, retained a high status. In this regard, the 6 For more on this subject, see Broudic 1997, 1999, 2001, 2003 [2004], 2007, 2008, 2009, 2010. 7 Having said this, the petty nobility of Western Brittany continued to speak Breton, mainly for practical interactions with the peasantry, but also for religious purposes, until well into the 20th century. 3 WHICH LINGUISTIC MODEL FOR BRITTANY? history of Breton is similar in many ways to that of Early Middle English.8 As Jean Le Dû and Yves Le Berre (1996) have demonstrated, vernacular Breton has always evolved within the context of an extensive web of orally transmitted basilects or badumes.9 Although two major regional standard forms of Breton were developed by the Catholic clergy, based on the Léonais and Vannetais varieties, the Church only ever considered these ecclesiastical standards as tools for teaching and promoting Christianity to monolinguals (from the Middle Ages down to the mid-20th century), not for any other functional or educational purposes (Le Pipec fc.). After François I signed the Villers-Cotterêt Ordonnance in 1539, French became the official language of the kingdom of France, and Latin was gradually occulted from the secular sphere. Le Pipec (ibid.) clearly demonstrates that Breton was rarely ever used for any official purposes in written documents or public inscriptions, even in the heart of Breton-speaking Brittany. French has long been used for such purposes, proving that it has served as the societal norm in Brittany for centuries. Breton has thus never benefited from any public or official recognition whatsoever. Unlike Protestant Wales, where three-quarters of the population could read the Welsh Bible by the end of the 18th century (Clement 1971), aside from a small ecclesiastical and administrative elite, Bretons remained illiterate in both Breton and French until the end of the 19th century. In 1869, French military authorities estimated that the percentage of conscripts from Finistère who were illiterate in French to be about 70% overall (including the cities). These statistics were far higher in rural towns where the majority of the population resided: Scaer 92%, Chateauneuf 85%, Fouesnant 80%, Rosporden 87%, Arzano 93%, Bannalec 80% . . .10 In the countryside, these statistics reach nearly 100%. On account of similarly high illiteracy rates in other regions of France, the Loi Jules Ferry was passed in 1881 the outcome of which was the establishment of free, French-medium public schools throughout France. One of the 8 Indeed, a parallel can be drawn with England following the Norman Conquest, a period during which Early Middle English was viewed as the language of a conquered nation, with French and Latin the language of the elites. This situation persisted in England well into the 14th century and, in this respect, at least for a few centuries, both English and Breton speakers shared a similar sociolinguistic fate. 9 See footnote 3 for a definition of “badume”. 10 I thank Fañch Postic for having provided the source for these statistics: Situation des conscrits sous le rapport de l’instruction, Tableau de 1869, Finistère, pp 56-58, Archives départemental du Finistère. 4 GARY MANCHEC-GERMAN government’s main goals in founding a public school system was to teach standard French as a means of providing basic education to all children with a view to elevating the intellectual level of the population as a whole. Because minority languages and French dialects (patois) were seen to be a hindrance to learning by educators, it was felt that they had to be erased from the linguistic landscape. The mindset at the time was that French was the sole language of culture and refinement. Eleven years after the FrancoPrussian War, in a climate of fervent nationalist resurgence, it goes without saying that teaching the Breton language, history and culture were utterly irrelevant in the context of the national French curriculum and, in this regard, little has changed. Consequently, most Bretons remained (and have remained) ignorant of their own history and culture and never received the slightest formal instruction in their language. In much the same way that parents round the world today encourage their children to master English, suffice it to say that most Bretons felt it was necessary, for the future well-being of their children, to learn French . . . and to learn it well. For this reason, the mass of the population enthusiastically embraced the opportunity to enroll their children in a free public school. This was not at all because they did not appreciate or enjoy their own language or culture, but simply because they saw learning French as an economic and social necessity and as the sole path out of the dire poverty in which most people lived (see Déguignet 1999).11 Ironically, among the harshest critics of those individuals who did not learn French properly were socially ambitious Bretons themselves. Laggards were viewed with utter contempt (ibid.) and were mocked for speaking poor French (called galleg saout, lit. ‘the French of cows’). As in many societies (Chambers & Trudgill 1980, Guillou 1998, Broudic 2007), women tended to be in the forefront of this shift to French (which they often viewed in terms of their own emancipation) the consequence being that, during the course of the 20th century, younger monolingual farmers found it increasingly difficult to find wives.12 11 An 80-year-old neighbour in Saint Yvi confided that, for him, the Breton language and culture was an anchor around one’s neck that dragged everyone downward. He stressed that, as a socialist, he owed everything to the French Republic which provided him with a free education, and allowed him to master the French language (which he could not speak until he went to grade school) and, through French, an understanding of the world which he would never have had otherwise. For him Breton was an “intellectual ghetto”! In my experience, this is a viewpoint that is almost unanimously shared by people of his age, whether on the political right or left. 12 Cf. the popular song Ar pôtr yaouank koz (literally, “the young-old lad”, i.e. the bachelor) 5 WHICH LINGUISTIC MODEL FOR BRITTANY? In public and private schools throughout France and the colonies, public humiliation through the use of the symbole 13 was a highly efficient incentive for learning French and, in the case of Brittany, the majority of Breton speakers passed from total illiteracy in their native language to literacy in French in a remarkably short time (Broudic 2005). It should be added, that the use of the symbole should not downplay the efforts and dedication with which the pupils learnt French or the diligence with which the school masters taught it.14 As we have already noted, in 1900, French was a foreign language for 90% of the population of Western Brittany with 50 to 60% of these being monolinguals (Broudic 1999).15 Today, the vast majority of the grandchildren and great-grandchildren of these very same people often know only a few basic words (if any) in Breton.16 Because of the pressure traditionally placed on children to succeed academically, it is a well-known fact that Brittany is today one of the best educated regions of France.17 The two World Wars, which resulted in the mobilization of virtually every able-bodied young man for service in the French army (many of them monolinguals when they were conscripted), clearly acted as a further about a poor farmer who cannot find a wife because the women are only interested in marrying French-speaking men (Loeiz Roparz pc., 1980). 13 Children were encouraged to inform on one another other in the schoolyard and denounce whoever was speaking Breton. The child who was caught was given the symbole (often a wooden shoe, piece of slate, etc.) which was worn round his/her neck. The child who ended up with the symbole at the end of the day remained after school for punishment. The bonnet d’âne (a hat with ass’s ears) was the symbole used in Elliant. 14 An example is that of my main informant, Léontine Manchec, who spoke no French before going to school. When she left school at the age of 13, she was able to read, write and speak French at a respectable level. 15 My principle informants in Saint Yvi, old enough to remember, told me during the late 1970s that the first time they heard French (spoken by native speakers) was in 1914 when Belgian refugees were housed in their village. 16 In my classes at the University of Brest, I often ask the students to give me the meanings of basic words like bread (bara), meat (kig), people/family (tud ), etc. Only a handful of students ever know the answers, even those from the rural areas. 17 Jean-Louis Duchet, former Dean of the School of Arts and Humanities, University of Poitiers (pc.) informed me in 2015 that teachers assigned to the Académie de Rennes are considered privileged. Indeed, Finistère, has the highest percentage of PhDs in all of France. The national competitive exams also show that the Bretons have among the best academic results in the country. In 2016, for instance, the Télégramme daily newspaper reported that the success rates for the Baccalauréat exam were 99% for 30 Breton high schools with a further 15 having a 100% success rate. Bretons also excel in the CAPES and Agrégation national competitive exams in all disciplines. 6 GARY MANCHEC-GERMAN catalyst in the language shift in that, for many, this was the first time that they had been plunged into a uniquely French-speaking environment. Their military training may also have reinforced their sense of French nationalism given that a high percentage of Bretons were assigned to combat units18 with many serving up to five years during the war years (both World War I and World War II ). It is difficult to estimate the extent to which nationalistic fervour (combined with fiercely hostile reactions against pro-Nazi Breton nationalist groups such as the Breiz Atao) may have stifled support for the language and culture following World War II. Some historians (cf. Coativy 2017) estimate that supporters of the Breton language and culture of all political stripes were often branded as potential traitors for at least 20 to 30 years after the war.19 This corresponds to my own observations. Seen from another perspective, perhaps an even more decisive factor which sealed the fate of the Breton language as a societal language was the modernization of the economy and society which started after World War I. This process of the economic expansion accelerated rapidly after the Second World War leading to the industrialisation of agriculture and fishing. The direct effect was the demise of traditional family farming and fishing which, in turn, tore apart the social fabric which maintained Breton as a community language. The outcome was the largescale out-migration of the poorest, unemployed, rural dwellers to the large urban centres of France. As we have already said, these were often the best speakers of traditional Breton. An often-overlooked factor which further hastened the decline of the language was the passing of the last monolingual speakers during the 1960s and 1980s. While they were alive, families and friends were obliged to speak Breton in the household, regardless of their views on the use of Breton.20 However, after these monolinguals passed away, and on account of the growing social space occupied by the French language in everyday life (radio, television, newspapers, schools, government administration, business matters . . . ), bilingual Breton speakers increasingly drifted towards the use of French. In this sense, in most families, the language shift occurred almost seamlessly and, in some respects, unconsciously.21 18 For example, 60% of the French fusiliers-marins commandos who took part in the D-Day landings in Normandy on June 6, 1944, were Bretons. 19 In local Cornouaillais Breton one still hears people say Breiz atao, mad da lao (lazhañ) ‘Breiz atao (members) good for the slaughter’. 20 I observed this first hand as a child when visiting my great-grandparents in Saint Yvi, who had great difficulty in expressing themselves in French. 21 Léontine Manchec (born 1902) angrily rejected my claim that her nieces, born during the 7 WHICH LINGUISTIC MODEL FOR BRITTANY? As a result, the contexts in which Breton can be used naturally have been steadily eroded. In most cases today, native Breton speakers no longer bother to speak Breton to one another and spontaneously address their peers in French, particularly in public places where non-Breton-speakers are omnipresent. This means that occasions to speak Breton during a normal day have been radically reduced over the past fifty years.22 The consequence is that many older native speakers are actually forgetting their native language and feel more comfortable speaking French.23 To conclude this section, Breton no longer survives nowadays as a community language but rather is used within an array of disparate networks and clusters scattered throughout the country, including the large cities of Upper Brittany. For this reason, the time for revitalizing Breton as a community language has long past. Though it is true that French is the only language permitted for official purposes, this should not be taken to imply that the authorities forbid Breton speakers from using the language in public places. Rather, the self-imposed social constraints are so powerful that most Breton speakers, except in more militant circles, generally choose not to do so. For this reason, it is very conceivable that a visitor passing through the hinterland of Breton-speaking Brittany might never hear a word of the language. In this respect, one might describe Breton today as a hidden or secret language or, perhaps better, a language of intimacy. 4. Passive speakers For all of the sociolinguistic, economic and historical reasons indicated above, by the 1950s, the majority of parents had mastered spoken French to varying degrees (often with heavy Breton substrate influence) and consciously refused to speak Breton to their own children ‘for their own good’. As a result, the baby-boomers were largely conditioned by their parents to forsake the family language, a move that went hand in hand with French governmental initiatives in favour of cultural and linguistic assimilation. ‘Progress’ was the leitmotif and the language, and all things Breton, were portrayed to be a drag on economic, social and intellectual development of the region. Just as Breton was associated in people’s minds with an impoverished 1950s, could not speak Breton. ‘That’s impossible! Of course, they can! They simply prefer to speak French!’ came the reply (circa 1980). Indeed, though they could understand Breton, they could not speak it. 22 Today, it is virtually impossible to start a Breton language conversation in a public place with a stranger. 23 One informant, born in 1900 and who immigrated to Paris in 1920, told me years ago that he realized this process had been achieved when he began dreaming in French. 8 GARY MANCHEC-GERMAN past, the French language and culture were presented and perceived to be the keys to a radiant future. This view was largely vindicated in economic terms by the period known as the Trente Glorieuses, that is, the thirty years of rapid economic growth spanning the years from 1950 to 1980. An important fact to keep in mind is that, although the shift to French was very rapid, it was not entirely complete in the sense that, even though young people were actively discouraged from speaking Breton, large numbers of children born between 1945 and 1970, now between 45 and 70, understood the local varieties in their areas perfectly, even if they could not speak or chose not to speak the language. Despite dramatically enhanced attitudes toward Breton and Breton culture over the past 50 years, the social and economic reality described above largely explains the stigmatization which is still attached to speaking Breton today, especially in the minds of the oldest speakers (i.e. the inheritors of naturally-transmitted Breton). Having said this, and we have implied above, speaking Breton has become more fashionable in some circles today, particularly (but not solely) among younger, formally-educated members of the middle class, but to some extent among older Breton-speaking women (Guillou 1998, Broudig 2007) who now see learning Breton as ‘fashionable’. This tendency is perhaps part of a world-wide phenomenon which is well described in Bud Khleif’s 1978 article entitled “Ethnic Awakening in the First World”. Indeed, since the Second World War, Western nations have witnessed the rise of huge numbers of university-educated people and the foundation of what he calls the ‘knowledge-class’. This knowledge class is largely composed of baby boomers and their children whose origins are generally to be found in the working class. For this reason, the members of this class have tended to be more sensitive, particularly since the 1960s, to the plight of minority cultures around the world. In the case of Brittany, an increasing number of people in this category would like to relearn the local forms of Breton they heard around them during their childhood. This may be partly due to a rise in cultural, linguistic and ethnic self-awareness which Khleif considers to be a reaction against globalization, modernization and what is often perceived to be the rise of a cold, impersonal new world order: [Ethnicity] can be regarded as a search for roots, for identity . . . for coping with issues of alienation in a mass society. The resurgence of ethnicity cannot only be understood as a tool for social mobility but also as a widespread 9 WHICH LINGUISTIC MODEL FOR BRITTANY? quest for community, a search for authenticity in the face of the overwhelming forces of modern life that are thought to be conducive to depersonalization, bureaucratization, and unresponsiveness (Khleif 1978: 103-4). Furthermore, he describes this rise of ethnicity as being intimately associated with decolonisation and the decline of the former European colonial powers in the wake of the World War II. The former colonial powers and, now, the United States, which is seen by many critics of colonialism to be their successor, are often portrayed as the major culprits of cultural and linguistic repression. Since the 1960s, politically liberal activists have attempted to portray the Bretons (and other European minorities) as victims of the same colonial forces that once dominated peoples around the world.24 The desire to return to one’s roots and to encourage the use of minority languages thus appear to be part of a dual process: on the one hand, to counter the past effects of colonialism and, on the other, to resist the cultural alienation provoked by the socio-economically dominant nation states around the world (including former colonial powers) which are currently leading the drive towards globalism. 5. Practical barriers to language maintenance and revitalization Yet another obstacle preventing the maintenance of Breton as a societal language is linked to the overwhelming technological changes and advances that have been sweeping over the world, particularly during the past 30 years. This is especially the case regarding the development of technology, in particular the computer and Internet. Multitudes of concepts linked to the modern age are totally foreign to Breton, a language that was, until very recent times, primarily adapted to rural and maritime cultures. This fact alone has resulted in a dearth of native neologisms in all domains linked to modern life: science, IT technologies, economics, business, mechanics, geopolitics and so on. In this sense, it would be inaccurate to say that Breton has been ‘replaced’ by French (as one often hears) since Breton never developed native vocabulary in any of these fields. Rather, it would be more precise to assert that the language has been progressively squeezed out of existence by French, and now English, as new technologies and lifestyles continue to emerge and enter into common usage. 24 Cf. Coadic 2013, Hechter 1975. 10 GARY MANCHEC-GERMAN The fact is that, in the decades following World War I, the creation of neologisms by native speakers progressively ceased as Breton began to yield ground as a community language. A few examples which I collected decades ago from speakers born between 1890 and 1910 are as follows: marc’h-du ‘locomotive’ (lit. ‘black-horse’), marc’h-houarn ‘bicycle’ (lit. ‘iron-horse’, later replaced by French velo), karr-tan ‘automobile’ (lit. ‘fire-cart’, often replaced by French oto). Today, in practical terms, even fluent speakers unconsciously switch to French to discuss more technical subjects such as automobile repair, banking, etc. Current attempts by University educated specialists to fill the lexical void have probably arrived too late and, perhaps worse, recent neologisms are frequently calqued on abstract French-influenced reasoning (with one word encapsulating one concept) rather than metaphorical compounds which are common in Breton. The result is that the new vocabulary often strikes native Breton speakers as peculiar and foreign. Despite brave attempts to introduce the new vocabulary, only a few stalwart defenders of the language have actually succeeded in incorporating recent neologisms into their speech. Paradoxically, using such vocabulary in a conversation with a native speaker is often enough to cut the conversation short. In addition to these lexical difficulties, the highly fragmented dialects (cf. Figure 2 below) create real barriers to communication between native speakers in various parts of Brittany. Having said this, the differences between the dialects are often grossly exaggerated by the speakers themselves who often believe that they cannot carry on a conversation with people from outside of their respective areas.25 The reality is far more complex and the supposed obstacles regarding inter-comprehension are, in my view, overly emphasised and much more closely linked to social and psychological considerations than purely linguistic factors. 26 It is largely on account of the stigmatization associated with the dialects 25 Furthermore, such impressions of unsurmountable linguistic barriers between speakers of different regions, even if exaggerated, are every bit as significant as real linguistic obstacles and, as one can imagine, such opinions have seriously contributed to inhibiting inter-dialectal use between native Breton speakers. These real or imagined frontiers have long encouraged and served as a pretext for Bretons to shift to French as a lingua franca. 26 My informants in Saint Yvi claimed that they felt relatively comfortable speaking Breton as far Pont Aven (30 km to the east), as far as Carhaix (50 km to the northwest), Leuhan/ Poullaouen (30 km to the north) and Ergué Gaberic (10km to the west), the area west of Quimper being considered off limits linguistically (according to them). The people of Bro Vigouden were literally considered to be a different ethnic group described locally as being the descendants of Mongols! The same was true of the Vannetais who were said to be 11 WHICH LINGUISTIC MODEL FOR BRITTANY? and the low social status of rural speakers that Gwalarn militants, led by Roparz Hemon and others, concluded during the 1920s that the only way to stop the decline of Breton was to create a modern independent Breton state with a national linguistic norm structured along the lines of those of other modern European nations. While the political objective for an independent Breton state failed miserably, along with the collapse of the Third Reich (with whom these nationalists were closely allied ), the notion of the single, literary, standard Breton model, spelt in the unified peurunvan orthography they supported, has prevailed and is now the linguistic model taught in nearly all the private and public schools across Brittany.27 Extreme dialectal differences have thus been presented as unsurmountable obstacles by defenders of the new Breton norm and this standardized linguistic model is seen as the only hope for the survival of the language. Paradoxically, opponents of the Breton language use the same argument to promote the use of standard French. One of the main points defended by some militants has long been that, prior to the Treaty of Union with France in 1532, Breton was formerly a unified language and that the current dialects are the result of degenerative French influence. For this reason, the rich dialectal variety observed in the spoken vernaculars is considered as a menace to the existence of the language. On the other hand, the new norm represents a resuscitated and rejuvenated form of Breton that will allow it to regain its former vitality and status. While French influence on Breton, especially lexical influence, is undeniable, the counter-argument put forward by those more sympathetic to dialectal Breton is that such views are not only prescriptive but linguistically and historically fallacious. It is known, for instance, that the current dialects existed long before the unification with France (Guyonvarc’h 1984) and may even have their roots in Armorican Gaulish (Falc’hun 1963, 1981, Fleuriot 1980, German 1984, 1991, Evans 1990). Indeed, some go as far as to say, with a hint of sarcasm, that the new Breton norm is largely inspired by the standard French model itself, a linguistically unified vision that is utterly foreign to the sociolinguistic situation that has characterized Brittany since the Middle Ages. In very broad terms, the most radical advocates of “incomprehensible” (German 1984). Yet, in the late 1960s, I accompanied older male relatives (masons by trade) to the Trégor region (Plouaret, Tonquedec and La Roche Derrien) where they spontaneously communicated with the Trégorrois in Breton with relative ease. 27 Much of the tension surrounding the use of peurunvan (unified ) spelling as opposed to the skolveurieg orthographe universitaire (devised by François Falc’hun in the 1950s), deals with this highly emotional, ideological and political conflict, a conflict which still rages to this day. 12 GARY MANCHEC-GERMAN this view defend the following position regarding Breton and its future: a ) ‘New’ or ‘neo’ Breton is not really ‘Breton’ at all, but rather a sort of ‘Esperanto’ that is far too distant from the naturally transmitted forms of the language; b ) The dialects are thus the only legitimate forms of Breton; c ) Since the dialect speakers will pass away within the next ten to fifteen years, these dialects themselves are doomed to disappear; d ) Trying to maintain them is, at best, a ‘rear-guard’ action that is condemned to failure; e ) Supporters of the dialects (as well as of the new norm) would be better off accepting their certain demise and simply mourn their passing (Le Berre 1989).28 For obvious reasons, the two positions on Breton are irreconcilable. 6. Regional cultural and linguistic identities It is interesting to note that the fragmentation of the dialects mentioned above corresponds very closely to cultural and geographical divisions within Brittany itself. In turn, this explains why many older Bretons are far more attached to their regional Breton identity than to any sense of a national or political Breton identity per se. Unlike the nationalist sentiments often expressed in Wales, Ireland or Scotland, feelings of Bretonness29 have not generally translated into the desire for Breton nationhood or political independence from France (except for a handful of Breton activists). Indeed, one point that has often been overlooked by some specialists is that, since the French Revolution (especially since the late 19th century), most Bretons feel equally Breton and French. 28 Fañch Morvannou, the founder of Etrerannyezhel orthography and Professor of Breton at the University of Western Brittany, told me a few years ago that “Breton will certainly die, but at least it will die ‘loved’.” 29 The idea of ‘ethnic’ or ‘racial’ identity is a concept rejected by the French state. This explains why questions regarding ethnic origins are absent from the national census. The only defining criterion for being ‘French’ is citizenship. Nevertheless, there are very real, if latent, feelings of what could describe as ethnic belonging in Brittany, as well as in other regions of France. For instance, physically, the typical Breton is often described as being relatively short, stocky, having high cheekbones, light coloured-eyes, dark or reddish hair. In terms of character, they are variously described as tough, stubborn to a fault, hot-tempered, hard workers, hard drinkers, brawlers, competent and courageous seamen and soldiers, poets and dreamers. Breton women are portrayed as strong-minded and fiercely independent. Whatever the merits of these stereotypes, they exist and are known throughout France itself. 13 WHICH LINGUISTIC MODEL FOR BRITTANY? Poll after poll demonstrates this reality which can be explained by the fact that, for over two centuries, the French State has inculcated republican ideals of citizenship into the population via the school system and society at large. In this respect, feeling both Breton and French is quite comparable to feeling simultaneously, say, Texan and American in the United States. In general terms, the Breton sense of identity is thus far different from that of the modern Welsh, Irish and Scottish where cultural distinctiveness is often defined in terms of historical animosity towards England and the English. For most Bretons, the British and Irish situations are quite foreign and expressions of hostility towards France are rare, even if more and more Bretons do indeed proclaim their cultural uniqueness these days.30 The traditional sense of identity is thus more closely bound to the local region in which they grew up. These areas often steeped in rich cultural lore and preserve original linguistic characteristics as well. Examples of such areas are the following pays or broiou: Glazig, Melenig, Bigouden, Pagan, Bidar, Pourlet, etc. This is far more representative than any notion of national political identity.31 This micro-vision of Breton identity is thus far more revealing of the nearly tribal-like way in which Breton culture and language have been experienced by the older generations than is generally recognized. Directly and indirectly, the visceral attachment of speakers to their local variety of Breton can be very simply explained by their love of region. The map of the bro in Figure 1 corresponds remarkably well with the map in Figure 2 featuring the dialects and micro-dialects of Brittany.32 One of Denis Costaouec’s informants in La Forêt Fouesnant summarizes the situation nicely and echoes what I have often heard in nearby Saint Yvi: “the only form of Breton that we are attached to is the one we speak at home, between us, and which we know cannot be learnt except by speaking the language” (Costaouec 2002).33 30 Militants, however, might argue that this lack of political will is directly linked to the ignorance of most Bretons about the conquest of Brittany by the French crown during the late 15th century and Breton history more generally which they see as a long period of political indoctrination, acculturation and colonization. 31 As the older generations pass away, even this micro-vision of Breton identity is now disappearing. 32 Figures 1 and 2 were adapted on the basis of the original maps available at geobreizh.bzh. Despite the editorial team best efforts, no contact was established. Thus, the maps have been completely rebuilt, without however distorting their original look and feel. 33 My translation. 14 GARY MANCHEC-GERMAN 7. Normalized Breton vs traditional Breton For all of the reasons presented above, it is estimated that only 4–6% of Bretons below the age of eighteen know any Breton at all (a rise of 4 points since 2000, however) at all, and most of these have learnt it at school, not in the household. According to some guesstimates, there are approximately 10,000 to 15,000 fluent speakers of the Breton standard ranging in age from 10 to 50 years old. Most of these are university-educated, middle-class urban dwellers for whom French is the native language. Because of their youth, their formal education, their literacy in Breton and social status, not to mention their willingness to use the language in public places, the impetus is clearly on the side of those who are adopting the new Breton norm. On the negative side, many of these young people, but certainly not all, speak Breton in much the same way French learners often speak English, that is, with phonological and grammatical interference from French. In addition, as mentioned above, they tend to use numerous neologisms that are not understood by native speakers. However, these are difficulties that educators could correct over time. But the task will be difficult. French prosodic influence on Breton is particularly striking and can be partly explained by the fact that younger learners have been unconsciously conditioned to feel that the Breton accent, with the strong tonic stress on the penultimate syllable,34 sounds ‘backwards’, ‘ugly’, ‘rough’ and even ‘Germanic’.35 As one young man put it to me, ‘who wants to speak like your grandparents?’ The conclusion is astounding: even though young Breton speakers can often accurately imitate the accents of older family members, French sociolinguistic norms and negative value judgments about the stress system on the penultimate, which is inherent to the language, are so powerful and psychologically ingrained that new speakers often cannot bring themselves to adopt it when speaking Breton. Indeed, even hearing a pronounced Breton accent in French can still provoke reactions of amusement and even derision. 34 Generally, stress is placed on the final syllable in Vannetais Breton. 35 In my experience, most of these kinds of remarks are made by younger women. Men tend to think of Breton as a ‘manly’ language, etc. 15 Manche n Haut Léon Trégor Léo Tr ég or ё é Pa e ys d Bas Léon l è ays Plougastel Do P Kemevodez Fañch Crozon Rousik Mené e l Cornouail é Cap Sizun Penn é Sardin Glazik Pays ё bigouden V a n n eta i c s St-M ieu ai Br nn St- Pays de Pa ys alo Re s è é Í Océan Atlantique é Pays Nantais Figure 1. The traditional cultural regions of Brittany Brezhoneg hiziv n Bro Leo Bro Dreger Bro Gernev Mor Breizh Bro Sant-Brieg Bro Zol Bro Roazhon Bro Sant-Maloù Bro Wened Meurvor atlantel Figure 2. Breton-language dialect map of Brittany in relation to the bishoprics 16 France Pays Pagan é GARY MANCHEC-GERMAN 8. Barriers to communication between generations I want to stress that the point here is not to criticize the young learners; quite the contrary. The efforts made by young people to learn Breton border on the heroic given that the older generations have adamantly refused to help them. The effect, however, is that the traditional dialectal models are generally inaccessible to them. In addition to the cultural, educational and age differences between older traditional speakers and new speakers, the linguistic hurdles between the two groups are considerable and can discourage communication between them, particularly in regions where the linguistic distance between the written and spoken languages is particularly marked. This is the case for Cornouaille and Trégor, for instance, where the 58% of all native speakers live — 39% and 19% respectively (Broudic 2009). Indeed, Breton-language teachers administering the Breton language option of the national French baccalauréat exam tell me that only about 10% to 15% of students passing the orals would feel at ease carrying on a conversation with older dialect speakers in most areas. 9. Which linguistic model for Brittany? This brings us to the next point: which linguistic approach is best suited to teaching the Breton language: a ‘bottom-up’ approach, in which renewed support for the sociolinguistically stigmatized dialects of the traditional speakers (90% of all speakers) would be advocated, or the present ‘top-down’ approach, that is to say, the continued advocacy of the new normalized model? Clearly, endorsing the current standard language offers the distinct advantage of linguistic uniformity and thus enhanced mutual comprehension between learners. Furthermore, it is far easier pedagogically-speaking to teach a linguistic norm sharing the same grammatical rules and possessing a common lexicon. But the downside is that the breton des livres, as it is often called, is frowned upon and viewed as unnatural by traditional speakers.36 In defence of the new standard, some argue that all languages change and that the emergence of the new Breton normalised variety is no different from the rise of the French and English standard models. The counter-argument is that the Gwalarn inspired model was not the work of native speakers at all, but rather a majority of learners (albeit excellent learners) who lacked an intuitive feel for the language, hence the accusations of artificiality. 36 More harshly, some call it breton chimique or ‘chemical Breton’. 17 WHICH LINGUISTIC MODEL FOR BRITTANY? Regardless of one’s opinions, it seems fair to say that the rise of the new standard model has led to the appearance of a new form of diglossia in Brittany, that is tied not only to linguistic differences but also educational, age and class distinctions which contrast with the older sociolinguistic division between a native-speaking clergy (on which the new standard is based ) and a largely illiterate but native-speaking peasantry (cf. German 2009). The sociolinguistic situation in Brittany today could thus be viewed as a multilevel system with standard French as the national norm at the summit, standard Breton as the regional norm and the vernacular Breton and French varieties representing the paritary forms of both languages. To summarize, when people talk about ‘defending’, ‘saving’37 or even ‘revitalizing’ the Breton language, the fundamental question should be this: what are they really talking about — the naturally transmitted varieties, the new standard model or both? The point is crucial (not only in the case of Breton, but all threatened languages and dialects) because what underpins the entire discussion about the use of Breton is how we define what ‘Breton’ actually is — as opposed to what it ‘should be’. Opinions about this vary radically and discussions on the subject are often so emotionally charged that constructive exchanges are sometimes impossible. 10. Shifting sands Undoubtedly, one of the major achievements of Breton-language activists has been the creation of full-emersion Diwan language schools since the late 1970s. This success sparked the creation of competing bilingual programmes in the both Catholic and public schools. Without them there would be virtually no Breton spoken by anyone under 30 years old today. As a result, neither would there be university-level degree courses in Breton or the capes and certainly not the new Breton Agrégation exam. Moreover, as mentioned above, the more positive attitudes towards the Breton language and culture observed today can be understood as part of a broader movement which arose in Europe and North America in the 1960s, the aim of which was to defend minority rights (cf. Khleif 1978). For some, this has been expressed as an interest in Breton music, dancing, regional history and the like. For others, far less numerous, the new-found interest in Breton culture has led to an increase in the desire to learn the ‘language’. However, in the minds of most, this can only mean learning 37 Speaking about ‘saving’ a language is a false analogy which attributes living qualities to an inanimate entity or concept. Languages do not ‘die’ or ‘survive’ and are not ‘saved’ as such. It is the speakers of the concerned languages who choose (often on account of social or economic pressures) to speak them or not. 18 GARY MANCHEC-GERMAN standard Breton, just as one would study any other modern ‘language’ such as French, German or English. Yet, as we have seen, the Breton situation is complex and the choices are complicated by the reality on the ground. Despite all of these hurdles, significant numbers of people, both young and old, want to learn or relearn ‘Breton’. This is an incontrovertible fact. The task for young learners has been relatively simple: to study the standard model taught at school. However, for thousands of passive speakers or native speakers seeking to write or read ‘Breton’ (in this case, meaning their home dialect) the task is daunting owing to a dearth of study materials. Aside from the well-known confusion provoked by initial consonant mutations,38 simple attempts to look up basic dialect words in a standard Breton dictionary can often result in failure. The examples listed below are purposely restricted to comparisons between southern Cornouaillais words and their equivalents in the major normalized dictionaries, especially those favouring peurunvan spelling.39 The difficulties listed here, however, are similar for passive and native speakers throughout Western Brittany. 10.1 The spelling is often too distant from their pronunciation to be recognized An example of this difficulty can be seen in map 96 of Jean Le Dû’s Nouvel Atlas de la Basse-Bretagne (2001). In the major standard dictionaries, one encounters hiziv in peurunvan orthography, hiziw in etrerannyezhel (inter-dialectal) orthography and hiziou in skolveurieg (orthographe universitaire) forms which are in fact closer to Welsh heddiw than to the actual vernacular forms heard in the living language.40 Not one of the 198 points of this atlas provides an example of a pronunciation even remotely approaching it. Rather, going from the North to South, one encounters variants such as heye, hirie, hirio, hiniou, hiou, chiou and so on. Other examples of this kind abound: standard abalamour da betra ‘why, for which reason’ corresponds to southern Cornouaillais blam bra, standard a-us ‘above’ for southern Cornouaillais heuc’h, standard teuziñ ‘to melt’ for southern Cornouaillais toeñi, standard lazhañ ‘to kill’ for southern Cornouaillais lao. 38 When encountering the mutated word for “dog” after the possessive adjectives, for example, Ma c’hi ‘my dog’, da gi ‘thy dog’, he c’hi ‘her dog’, e gi ‘his dog’, ho ki “your dog”, the learner does not necessarily know whether to look the word up under C’H, K or G. The word is found in the dictionary under Ki. 39 In particular, Hemon & Huon 2005. 40 For more on the topic of competing orthographies in Brittany see Wmffre 2008. 19 WHICH LINGUISTIC MODEL FOR BRITTANY? The forms of verbs such as kavout ‘to get/have’ and bezañ ‘be’ are kaout and bout in southern Cornouaillais respectively. Truncated local forms render them unrecognizable with regard to their standard equivalents as can be seen in the following table showing the forms of the past habitual of kaout: Standard Breton (Peurunvan) Southern Cornouaillais Translation 1. am bez (me) mef ‘I habitually have’ 2. az pez (ti) tef ‘you (sg.) habitually have’ 3. en (he) devez (haoñ / hi) nef ‘s/he habitually has/gets’ 1. hor bez (neñ) nef ‘we habitually have’ 2. ho pez (hwi) pef ‘you (pl.) habitually have’ 3. o devez (hènn/hè) nef ‘they habitually have’ 10.2 The dialect word they know is simply not found in the dictionary and/or the semantic distinctions regarding a given concept are not defined in enough detail One simple example is brumenn and morenn (cf. Hemon & Huon 2005 s.v.) which translate the generic French word brume ‘fog’. In southern Cornouaille, brumachenn (not found in most dictionaries) is sometimes used as the generic. Nevertheless, fulaienn is the common word for fog in a few parishes of southern Cornouaille including Saint Yvi and is not known elsewhere. For this question, my current Saint Yvi informants (pc. 2018; all between 65–75 years of age for this question) give examples such as fulaienn-mor ‘sea-mist’ and fulaienn-heol-tomm (lit. ‘hot sun fog’), the early morning haze that announces a hot summer day. Distinctions such as these generally tend to be absent from most dictionaries. Indeed, fulaienn is not to be found in any of the dictionaries I have consulted (online or otherwise) except in Alan Heusaff’s dictionary of Saint Yvi Breton (Heusaff 1995). Morenn is a kind of ‘vapour’ or ‘sea-mist’ (which is unknown to them or perhaps forgotten).41 Distinctions such as these generally tend to be absent from most dictionaries. 41 Alan Heusaff (1995: 117) describes words for fog in terms of percentages of humidity: fulaienn > 91%; lusenn > 70%–90%; morenn < 75%; he does not seem certain about mogidell ? Personally, I see this as wisps of fog (resembling puffs of smoke, but this may be a personal interpretation influenced by moged ‘smoke’). Lusenn, morenn and mogidell are unknown to my current Saint Yvi informants. 20 GARY MANCHEC-GERMAN 10.3 The Breton word proposed for a given concept or object is simply not used or even understood in their particular dialect The differences in the use of the following interrogative pronouns provide some simple examples of how basic lexical differences can be a hurdle to communication between not only the regional varieties and the standard language but also between dialects themselves: standard pegoulz or peur ‘when’ vs. southern Cornouaille pevare;42 standard penaos ‘how’ vs southern Cornouaille piseurt mod, pe mod; standard perak ‘why’ vs. southern Cornouaille praskôz (< petra a zo kaoz). The standard forms would be understood in the Léon and Trégor regions, however. 10.4 The word exists both in the standard and in the dialect but is used with a different meaning One common example is the use of standard hag all to translate ‘etcetera’. For most native speakers it means ‘and all’ as in the following example: Sar’ ho peg betram me lammo barz boutou hag all ! ‘Shut your mouth or I’ll jump in it (wooden) shoes and all!’ Another example is Standard Breton diforc’h ‘different/différence’ but also ‘abortion’ (cf. Hemon & Huon 2005) which is often used to avoid the French borrowing diferañs (in universal use in the vernacular). Although diforc’h is indeed used with the meaning of ‘difference’ in parts of the Vannetais dialect area, for most speakers, diforc’h only means ‘an aborted cow or pig fetus’. Finally, gweladenn is often defined in standard Breton to mean a ‘visit’ (touristic or otherwise) or an ‘interview’ whereas in southern Cornouaille it refers to the ‘inspection’ of a future bride’s farm by the groom’s parents (usually his father). Again, there is a vast number of examples such as these. 10.5 Neologisms unknown in traditional language often used by standard speakers In such case, native speakers revert to French words. Standard Breton urziataer vs. Fr. ordinateur ‘computer’, pronounced as in French Standard Breton skinwel for Fr. television, pronounced [ˌtɛlẹviˈzẹin] ː Standard Breton skingomz vs. Fr. post or radio, pronounced as in French Standard Breton pellgomz vs. Fr. telefon, pronounced as in French43 42 In addition, benn pevare (with future reference) vs. pevare (with past reference). 43 All the standard examples are taken from Hemon & Huon 2005. 21 WHICH LINGUISTIC MODEL FOR BRITTANY? These few examples are just the tip of the iceberg and, because most native speakers are generally illiterate in Breton, using any normalized dictionary can prove to be an exasperating experience and users tend to give up after a few futile attempts.44 11. Bridging the gaps Now that I have presented this rapid sketch and shown some of the practical obstacles that complicate communication between learners and native speakers, I would like to propose a project to publish at least six to seven (and ideally as many as nine) pedagogically coordinated dialect dictionaries along with accompanying dialect grammars. Each one of these dictionaries and grammars would correspond to one of the major geolinguistic areas of Western Brittany corresponding to Figures 1 and 2. This approach would require a dedicated team effort and discussions are currently under way with interested parties and institutions. As we have seen, a rich array of pedagogical materials is available to teach standard Breton. The main argument in favour of the approach proposed here is that it would provide pedagogical resources to assist those seeking to master the varieties spoken in their home region. a) The first category concerns the roughly 100,000 to 150,000 (my own estimate) native dialect speakers mentioned earlier, 90% or more of whom are functionally illiterate in Breton. Ideally the existence of dialect dictionaries would disprove the widely held prescriptive view that their varieties are worthless patois (sometimes described as trefouedach ‘gibberish’ by purists) in relation to a supposed superior standard model. Rather, they would be shown to be coherent linguistic systems in their own right, having their own phonological, morphological, syntactic and lexical functions and rules. Realistically, however, for the sociolinguistic reasons outlined in this paper, only a handful of people in this category would be inclined to study their own varieties. b) Secondly, a very important slice of the population, namely passive speakers, has been largely ignored until now. They are the target population in this article. Interestingly, in 2007, 22% of those questioned in Broudic’s study, claimed to understand Breton while only 13% claimed 44 One notable exception is Francis Favereau’s excellent dictionary, Geriadur ar Brezhoneg a-vremañ, 1992 [2000]. However, it is mainly intended for people who can already read Breton (i.e. his examples are not translated into French). 22 GARY MANCHEC-GERMAN to actually use the language (Broudic 2009). These figures have certainly dropped since 2007 but we could estimate their number today at somewhere between 50,000 to 100,000 people. Most of these people are over 40 years of age and were often discouraged from speaking Breton by their parents, relatives and the society at large. Nevertheless, because they heard the language around them on a daily basis spoken by older relatives and neighbours while growing up, many often understand their dialects perfectly, but cannot speak the language themselves. As already explained, the reasons for this are simultaneously social, linguistic and often psychological. Elements of this population could potentially provide a numerically significant pool of new speakers. It must be said that many people in this category have a far more positive self-image in terms of Breton identity than their parents and grandparents did. Furthermore, they often feel a sense of loss and frustration at not speaking the family language. Consequently, there is a significant social demand on their part to reconnect with their own dialects to which they are often still viscerally attached. As mentioned earlier, this is part and parcel of the affective ties with the local micro-cultures seen in Figure 1 above. Significantly, many of these people are not necessarily interested in learning the standard language, a variety which has very different social functions and applications. If provided with the proper resources, encouragement and backup, many of the people in this category could be fluent within a relatively short time, some within a matter of months. On the downside, the networks in which they can use the language are rapidly shrinking and, in some areas, no longer exist! c) Finally, many young learners who have studied the standard language may want to use the resources to interact with what is left of the core population of traditional speakers in their areas.45 At this point, I should add that even under the best of circumstances it would be unrealistic to imagine that Breton will ever become a community language again. At best, what we are talking about here is providing a means 45 A few years ago, André Le Gac, conseiller général, and Annaig Daouphars, chargée de mission (Finistère) launched a project entitled “Quéteurs de mémoires” (Memory seekers) the goal of which was to bring children who are learning Breton into regular contact with older native speakers. It has been widely praised but its success has been limited. 23 WHICH LINGUISTIC MODEL FOR BRITTANY? to fulfil a real need on the part of several thousand potential new speakers who might, indirectly or directly, reinforce pre-existing linguistic networks and clusters i.e. citizens seeking to relearn their regional language and culture as a way of (re-)discovering their own past. The good news is that the raw materials for the dictionaries and grammars exist for all the regions of Brittany.46 Since the 1980s, numerous doctoral theses focussing on the grammar of regional Breton varieties have been written and many of these have been published or are available online. High-quality dialectological fieldwork is also currently being conducted independently by teams of young Breton scholars (without the support of any university funding). In order to be successful, however, both the dictionaries and the grammars would need to be designed following the same structure and format so as to allow both standard and dialect speakers to compare the vocabulary and grammatical features which distinguish the dialects a) from each other and b) from the standard language itself. 12. A dialect dictionary for South-central Cornouaille The advantage of dialect dictionaries is that they would allow users to avoid certain pitfalls mentioned above because the entries would only include the dialect words from the speaker’s region and these would be written in an adapted orthography recognizable to the inhabitants of a given area. In terms of presentation, a typical dictionary entry could be organized as follows: a) The entry, transcribed in adapted orthography so that the native speaker or passive speaker of a given area can easily recognize the word s/he is reading; b) For those who can read the ipa , the dialect word is followed by a phonetic transcription along with variant local pronunciations (this would be particularly useful for those focussing on tonic stress, degrees of vowel aperture, etc.); 46 Several dialect dictionaries already exist such as Jean Le Dû’s (2012), Le Trégorrois à Plougrescant. In October 2017, a thematically organized dictionary was published by M. Bouzec, J. Goapper and Y. Souffez, in which I actively participated for over six years, entitled Le Breton des rives de l’Aven et du Bélon, Blaz ha blazig c’hoarzh. Another example is Christian Fagon and Yann Riou’s (2015) Bredoneg ar Gear, on Teuzor. These three examples are all immensely valuable contributions to the local vernaculars and the cultures they reflect. 24 GARY MANCHEC-GERMAN c) Whenever possible or pertinent, the collector and/or informant’s initials and his/her parish are also indicated; d) The word is then written in peurunvan orthography to show the connection with the standard language, thus permitting a two-way exchange between traditional speakers and young learners; e) The definition(s) and, finally, the word used in context. The presentation is detailed and is perhaps even cumbersome, but it does offer the advantage of being relatively complete.47 In order to provide the reader with some concrete examples, here are extracts from the first draft of my forthcoming dictionary on the Breton of Saint Yvi. The examples (with my English translations) are thematically-organized and come from a section on the human anatomy. the anatomy The human skeleton skeuletenn [skǝ'lɛtən] BL skeledenn — skeleton framm ’ c’horf [ˌfxãm ə ˈxɔəf] BL framm ar c’horf — idem, lit. frame of the body eskornou korf ’n den [ɛsˈkɔənu ˌkɔəf nˈdɛ̃ːn] — (MFK-SY) the human skeleton, lit. the bones of the body eskornou ’n den [ɛsˈkɔənu nˈdɛ ː̃ n] (MK, MFK-SY)BL eskern an den — idem, lit. the bones of a human being/person Bones erchen ['ɛəʃən] (GG-SY, AH-SY, MK-SY), ersen ['ɛəsən] (MK-SY), pl. erskenou [ɛəsˈkɛnu], eskornou [ɛsˈkɔənu] (GG-SY, MK-SY) BL eskern — bone, bones (erchen is historically a plural and has been reanalysed as singular by some). Askorn is used for the singular in many other dialect areas, including in standard Breton. mél-ersen [ˌmẹːl 'ɛəsən] (MK-SY) BL mel-askorn — bone marrow (lit. the honey of the bone(s) ) bouédenn ['bwẹːdən] (AH-SY) BL bouedenn — idem (also the pulp / flesh a fruit) 47 These are extracts from the fieldwork that I have conducted over the past thirty-five years in south-central Cornouaille (Finistère), an area stretching eastward between Quimper and Rosporden. It focusses on a core area centring on and around the parishes of Saint Yvi and Elliant. To date, I have collected upwards of 10,000 lexical items and popular rhymes, riddles, sayings as well as idiomatic expressions. 25 WHICH LINGUISTIC MODEL FOR BRITTANY? the anatomy Cartilage and joints kranken [ˈkxãŋkǝn] (AH-SY) BL kranken — cartilage (also a crab); ’c’hranken ba penn ’n erchen braz ‘cartilage found at the end of a large bone’ BL ar c’hranken e penn un askorn bras Bé zo kranken ba kig sal. (MF-SY) BL Bez’ ez eus kranken e kig sal. There is cartilage in the bacon (Fr. lard). koubl [kubl] (AH-SY) BL koubl — the joint between two bones; koubl i vrec’h, e r’har [ˌkubl i vɣeːχ/ kubl i ˈɣæːə] BL koubl e vrec’h, e c’har — (his) arm or leg joint mailh [maiλ] BL mailh, pl. mailhou [ˈmaiλu] (MK-SY) BL mailhoù – finger joint or knuckle; also, large mallet, hammer (fig. a fist) Hém’ dapo ’n tol mailh ganim vo ket pell! (MK-SY) BL Hemañ a dapo un taol mailh ganin-me bremaik! Lit. He will catch a blow of a mallet ( fist/knuckles) with me soon I am going to punch him before long! Flesh and muscles béo [bẹọ] (AH-SY) BL bev — the quick (sensitive area of flesh, esp. under the nails) ’ Skilfenn zo ouét ba’m béo. (JLS-Brc) BL Ar skilfenn a zo aet er bev. The splinter went right into the quick. kig [kik] (MK-SY) BL kig — meat, flesh, muscles kig frost [ˌkik ˈfxɔs] (MK-SY) BL kig fraost — flabby, soft muscles kig stert [ˌkik ˈstɛət] (MK-SY) BL kig start — hard, powerful muscles (lit. hard meat) Haoñ neus kig stert ba i vrec’h ! (MK-SY) BL Eñ en-deus kig start en i vrec’h! His arms are solid muscle! Lit. He has hard meat in his arms. gẅ een [gɥɛ ː̃ n] (AH-SY, MK-SY) BL gwevn — well-muscled, sinewy, tough and wiry 13. Dialect grammar Naturally, corresponding dialect grammars would also need to be created. These would be similarly presented in local adapted orthography with a presentation of core grammatical features common to all dialects. The key characteristics specific to each area would thus be highlighted to facilitate communication with native speakers. If successful, these tools could eventually be developed into teaching manuals targeting a variety of different regional audiences and would supplement the corresponding pedagogical materials already available. 26 GARY MANCHEC-GERMAN 14. Conclusion I want to stress that the objective here is not to propose the replacement of the normalized language as it now stands. On the contrary, it is hoped that this approach would enrich the knowledge of the spoken language of all regions by permitting Breton speakers of all backgrounds and opinions to examine the connections between the standard language and the living dialects which, after all, are the result of the natural transmission of the language over several millennia. Given that language is first and foremost an oral phenomenon, this is not a mere detail that can be swept under the carpet, but rather a highly significant factor underlying the legitimacy of any living language. If this is not respected, the ‘language’ being promoted risks being perceived as little more than an artificial creation or invention, devoid of any historical and social depth. Indeed, most natural languages have varieties and, among these varieties, registers ranging from the formal (langue disparitaire, Le Berre & Le Dû 1996) to informal (langue paritaire, ibid.) that are governed by socially-determined criteria. In this regard, Breton should be no different. Solely promoting a high register of standard language is tantamount to putting a new roof on a dilapidated old house with crumbling walls and foundations. Without shoring up the supporting structure, the roof, regardless of the quality of the workmanship, serves little purpose. At the heart of this project is the desire to respond to a very real societal need on the part of many individuals who, for reasons ranging from nostalgia for the past and love of region to simple intellectual curiosity, would like to learn, relearn, or simply familiarize themselves with the native dialects spoken by their older family members and neighbours. As we have seen, when taken collectively, there are currently somewhere from 150,000 to 250,000 traditional speakers and passive speakers combined who currently do not have the linguistic tools which would permit them to explore their own varieties. Even if only 2% to 5% of this target population actually made the effort to study/relearn the local varieties using the resources proposed here, this could assist somewhere between 3,000 and 12,000 people. I do not want to end this paper on a pessimistic note but, despite the possible benefits of such an approach, the obstacles are formidable, not the least of which is the time it would take to prepare and publish the dictionaries and grammars. Sadly, the majority of the last generation of traditional native speakers (that is, those born in a monolingual Breton society) will be gone in the next 10 to 15 years. 27 WHICH LINGUISTIC MODEL FOR BRITTANY? Furthermore, the approach itself, as well as the premises on which it is based, would certainly be rejected by many, primarily on the grounds of linguistic propriety. Finally, considering the profound demographic changes in Brittany linked to the current effects of globalism (i.e. considerable in-migration of non-Bretons and the out-migration of younger native Bretons), this proposal may have little resonance among significant segments of the population. However, some of the lessons gleaned here may be of help, if not for Brittany, for other linguistic communities in the world facing similar difficulties. This is my sincere hope. University of Western Brittany References Bouzec, M., Goapper, J., Souffez, Y. 2017, Le Breton des rives de l’Aven et du Bélon, Blaz ha blazig c’hoarzh, Lannion: An Alarc’h. Broudic, F., 1997, L’interdiction du breton en 1902. La IIIe République contre les langues régionales, Spézet: Coop Breizh. Broudic, F., 1999, Qui parle le breton aujourd’hui? Qui le parlera demain? Brest: Brud Nevez. Broudic, F., 2001, ‘Le breton, langue en danger ?’, Parlons du breton. Association Buhez, Rennes: Ed. Ouest-France, 7–26. Broudic, F., 2003 [2004], ‘L’opinion des Bretons concernant leur langue’, Bulletin de la Société archéologique du Finistère, CXXXII, 435–45. Broudic, F., 2005, ‘Une polémique entre bretonnants en 1908 sur l’usage du «symbole»’, Langues de l’histoire, langues de la vie. Mélanges offerts à Fañch Roudaut, Association Les amis de Fañch Roudaut, 319–30. Broudic, F., 2007, Le breton: une langue en questions, Brest: Emgleo Breiz. Broudic, F., 2009, Parler breton au XXIe siècle. Le nouveau sondage de TMO Régions, Brest: Emgleo Breiz. Broudic, F., 2008, ‘Parlera-t-on encore breton en 2030 ?’, Bretagne[s] 10, 30–32. Broudic, F., 2010, ‘Les nouveaux sondages sur la pratique du breton et celle de l’occitan en Aquitaine’, La Bretagne Linguistique 15, 185–93. Chambers, A. K., Trudgill, P., 1980, Dialectology, Cambridge: Cambridge University Press. Clement, M., 1971, ‘The Campaign against Illiteracy’, in: Roderick, A. J., ed., Wales throughout Ages — from 1485 to the beginning of the 20th Century, Vol. 2, Landybie: Christopher Davies, 86–93. 28 GARY MANCHEC-GERMAN Coadic, R., ed., 2013. De la domination à la reconnaissance: Antilles, Afrique et Bretagne, Rennes: Presses universitaires de Rennes. Coativy, Y., 2017, ‘Brittany from the 13th to the 21st century: an Overview of Breton History’ (oral communication), Summer School in Breton Linguistic and Cultural Heritage Studies, held in Quimper, University of Western Brittany. Costaouec, D., 2002, Quel avenir pour le Breton populaire? Enquête à la Forêt Fouesnant, Brest: Emgleo Breiz. Déguignet, J.-M., 1999, Mémoires d’un Paysan Bas-Breton, Ar Releg-Kerhuon: An Here. Evans, E., 1990, ‘Insular Celtic and the Emergence of the Welsh Language’, in: Bammesberger A., Wollmann, A., eds., Britain 400–600: Language and History, Heidelberg: Carl Winter. Fagon, Ch., & Riou, Y., 2015, Bredoneg ar Gear, on Teuzor, Lannion: An Alarc’h. Falc’hun, F., 1951, Le système consonantique du breton: avec une étude comparative expérimentale, Rennes: Imprimeries Réunies. Falc’hun, F., 1963, Histoire de la langue bretonne d’après la géographie linguistique, Seconde édition, Paris: Presses Universitaires de France. Falc’hun, F., 1981, Perspectives nouvelles sur l’histoire de la langue bretonne, Paris: Union Générale d’Edition. Falc’hun, F., 1982. Les noms de lieux celtiques: première série — vallées et plaines, Paris: Slatkine. Favereau, F., 1991, ‘Quatre générations de bretonnants’, La Bretagne Linguistique 7, 31–52. Favereau, F., 1992 [2000], Geriadur ar brezhoneg a-vremañ, Morlaix: Skol Vreizh. Fleuriot, L., 1980, Les Origines de la Bretagne, Paris: Payot. German, G., 1984, Etude phonologique et morphologique du parler de Saint Yvi en Cornouaille et sa place dialectale selon les données de l’Atlas Linguistique de la Basse-Bretagne, unpublished PhD dissertation, Université de Bretagne Occidentale de Brest, Brest. German, G., 1991, ‘Une méthode dialectomérique (assisté par ordinateur) pour l’analyse des atlas linguistiques’, La Bretagne Linguistique 7, 177–214. German, G., 2003, ‘The French of Western Brittany in Light of the Celtic Englishes’, in: Tristram, H. L. C., ed., Celtic-Englishes III, Heidelberg: Carl Winter, 390–412. German, G., 2009, ‘The Story of Breton: Past, Present and ?’, invited lecture at the School of Scottish and Celtic Studies, University of Edinburgh. Gourvil, F., 1952, Langue et littérature bretonnes, Paris: Presses universitaires de France, Que sais-je. Guillou, A., 1998, ‘Les femmes et le breton’, La Bretagne Linguistique 11, Brest: CRBC, 227–31. 29 WHICH LINGUISTIC MODEL FOR BRITTANY? Guyonvarc’h, C., 1984, Aux origines du breton: Le glossaire vannetais du Chevalier Arnold Von Harff, voyageur allemand du XVe siècle, Rennes: Ogam-Celticum. Hechter, M., 1975, Internal Colonialism: The Celtic fringe in British National Development: 1536–1966, London: Routledge and Kegan Paul. Hemon, R., & Huon, R., 2005, Dictionnaire breton-français, français-breton, Al Liamm. Heusaff, A., 1995, Geriaoueg Sant-Ivi (Kernev-Izel ), Hor Yezh. Khleif, B., 1978, ‘Ethnic Awakening in the first World: the Case of Wales’, in: Williams, G., ed., Social and Cultural Change in Contemporary Wales, London: Routledge and Kegan Paul, 102–19. Le Berre, Y., 1989, ‘Langue et littérature bretonne depuis la revolution: une acculturation paradoxale’, Beiträge zur Romanischen Philologie, 28, 271–8. Le Berre, Y., & Le Dû, J., eds. 1996. Badume-Standard-Norme: Le double jeu du langage, Groupe de Recherche sur l’Economie Linguistique de la Bretagne, Brest: Centre de Recherche Bretonne et Celtique. Le Coadic, R., ed., 2013. De la domination à la reconnaissance: Antilles, Afrique et Bretagne, Rennes: Presses universitaires de Rennes. Le Dû, J., 2001. Nouvel Atlas Linguistique de la Basse-Bretagne. Volumes 1 & 2, Brest: Centre de Recherche Bretonne et Celtique. Le Dû, J., 2012, Le trégorrois à Plougrescant, Dictionnaire breton français/ Dictionnaire français breton, Brest: Emgleo Breiz. Le Pipec, E., forthcoming, ‘Le breton et l’école: 1499–1794, analyse d’un échec, ou pourquoi le breton n’est-il pas devenu une langue de scolarisation?’, Colloque Ecolang, Quimper. Le Roux, P., 1963, L’Atlas Linguistique de la Basse-Bretagne, Rennes: Plihon et Hommay. Tanguy, B., 1977a, Aux origines du nationalisme breton: Le renouveau des études bretonnes au XIXe siècle, Paris: Union Générale d’Edition. Tanguy, B., 1977b, Aux origines du nationalisme breton: Essai sur l’histoire de la langue bretonne par Hersart de la Villemarqué, Paris: Union Générale d’Edition. UNESCO, 2017, UNESCO Atlas of the World’s Languages in Danger, accessed 15 September 2018: http://www.unesco.org/languages-atlas/index.php. Wmffre, I., 2008, Breton Orthographies and Dialects: The Twentieth-century Orthography War in Brittany, Contemporary Studies in Descriptive Linguistics, Peter Lang. 30 Semantic (Ir)regularities in Action Nouns in Irish Maria Bloch-Trojnar 1. Introduction Cross-linguistically, verbal nominalizations display a close semantic and syntactic affinity to their corresponding predicates (Comrie 1976; Comrie and Thompson 1985; Koptjevskaja-Tamm 1993), as obtaining between (1a) and (1b): (1) a . The professor examined the students. b . the examination of the students by the professor Nouns derived from verbs are most readily associated with the interpretation of “an action or process” (Comrie 1976: 178), which regularly characterizes nominals related to dynamic verbs. When stative verbs act as the base, the derived nominal will denote a state (Melloni 2011: 2), and we will observe that the same thematic relations obtain between the verb and the noun with respect to the thematic roles borne by their syntactic satellites (Rozwadowska 1997). (2) a . John loves Lucy. b . John’s love of Lucy Another characteristic feature of action nominalizations is that they exhibit the process/result dichotomy:1 (3) a . The examination was on the table. b . She was his first love. 1 Anderson (1984) introduces the distinction between concrete and abstract nouns, MalickaKlepars-ka (1988) regular nominalizations and lexicalizations, and Walińska de Hackbeil (1984) clausal nominals with a full Θ-grid and Θ-nominals. Grimshaw (1990) distinguishes between Complex Event Nominals, with an associated argument structure, and Simple Event or Result Nominals, which lack it. The basis for this classification is, among other things, their ability to take obligatory arguments, to license event-related PPs and to pluralize. This distinction is upheld in other frameworks as Argument-Structure Nominals (AS-Nominals) and Referential Nominals (R-Nominals) in Borer (2003), whereas Alexiadou (2001) refers to them as “process nominals” and “result nominals” respectively. 31 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH The nominals in (3) denote concrete entities (a result in (3a), a person in (3b)), lack aspectual characteristics and accompanying satellite phrases, and cannot be directly associated with the thematic roles of Agent and Theme or Experiencer and Source borne by the corresponding verbs. They are also count nouns. The linguistic debate concerning nominalizations focuses on nouns with transpositional or situation-denoting semantics, whereas the result or referential nominals are largely an uncharted area. The aim of this paper is to examine the range of semantic interpretations displayed by verbal nouns in Irish. In the process of lexicalization, the meanings of nominals drift away from the core Action/Event/State reading and come to denote “something material connected with the verbal idea (agent, instrument, belongings, place or the like)” (Marchand 1969: 303). According to Kastovsky (1986: 596), additional readings of derived nominals develop in accordance with the following hierarchy: “Action/Fact → Result → Locative → Instrument → Agent”. Malicka-Kleparska (1988) views lexicalization as the absorption or incorporation into the meaning of a given nominalization of the thematic role which is most object-like, i.e. Experienced, Causer, Instrument, Location, Source, Goal and, if none of these is available, that of Theme. Recently, we have witnessed the rise of more elaborate models of semantic interpretation, which tackle the vagueness of the theta role system, such as Pustejovsky’s (1995) Generative Lexicon or Lieber’s (2004) Decompositional Lexical Semantics. The machinery of both these models has been deployed by Melloni (2011) to pinpoint the constraints on the formation of entity-denoting deverbal nominals in Italian. She also hypothesizes that similar, though not entirely matching, patterns of polysemy can be observed in different languages. Her hypothesis will be put to the test with a view to establishing the range of non-eventive semantic interpretations displayed by verbal nouns in Irish. In section 2 I sketch the theoretical background for an empirical and comparative analysis to be developed in section 3. The scope of research ranges from verbal nouns formed by means of overt nominalizers to verbal nouns formed by means of conversion. The primary source of data is the most comprehensive Irish-English dictionary, i.e. Ó Dónaill (1977), containing about 2300 verbal entries. For more examples we also consult the online version of de Bhaldraithe (1959) and sporadically the New English-Irish Dictionary, recently launched by Foras na Gaeilge.2 2 All three dictionaries are available at http://www.focloir.ie/en/ [last accessed 09.02.2017]. 32 MARIA BLOCH-TROJNAR 2. Classification of referential nominals (R-nominals) Non-eventive nominals lacking an associated argument structure typically denote the product or result of the event denoted by the base verb (e.g. construction in the sense of ‘edifice’). However, they can exhibit a varied mix of readings which do not match up with this description. For example, administration refers to a group of people and obstruction is more likely to be viewed as the causer/means, while drink clearly has an object interpretation. Therefore, in this paper, following Borer (2003), we will use the term R-nominals in the sense of referential rather than result nominals. We will primarily focus on cases of logical/inherent polysemy in deverbal nouns, along the lines proposed by Pustejovsky (1995). However, polysemy can also result from paradigmatic sense extension or metonymic transfer. In addition to this, some interpretations of E-nominals, though not strictly actional (such as factual, temporal, manner), stem from the vagueness or underdetermination of eventive nominals and arise in the syntactic-pragmatic context. Yet another class of readings are lexicalized/ idiomatized senses, i.e. products of unpredictable semantic drift. 2.1 Context-dependent readings (manner, temporal and factual, etc.) Some readings of action nouns are not instances of actual polysemy; i.e. they do not arise in the lexicon but on a higher level of syntactic-semantic composition. They result from the vagueness of the nominals and depend on a specific predicative context (or container into which they are inserted, as first observed by Vendler (1967) and later confirmed by Langacker (1991) and Pustejovsky (1995)). Action nominalizations can profile various facets of the reified event, such as the manner in which the action is carried out (4a), its duration (4b), its propriety (4c) and its factuality (4d), after Langacker (1991: 32): (4) a . Harvey’s taunting of the bear was merciless. b . Harvey’s taunting of the bear lasted three hours. c . Harvey’s taunting of the bear was ill-advised. d . Harvey’s taunting of the bear came as a big surprise. The examples above indicate that derived nominals can be accompanied by event participants on these interpretations, their presence being indicative of the regular status of the nominalization. Hence, they should be regarded as sense extensions of action nominals. 33 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH Although extremely frequent in cross-linguistic terms, the factual interpretation of action nouns seems to be unattested in Irish. Doyle (2002: 101) observes that when verbal nouns (vn s) are used as ordinary nouns, they are hardly ever accompanied by a complement, and he provides an example which, in his opinion, borders on ungrammaticality: (5) */? Chuir briseadh na gcloch ionadh orm. put-past break-vn the stones-gen. surprise on-me The breaking of the stones surprised me. Notably, the superordinate predicate selected by Doyle (2002), i.e. surprise, triggers a factual interpretation. Searches for translations of ‘the fact that’ interpretations in Irish invariably return dependent clause renderings, e.g. (6) a . His appointment as captain surprised us. Bhí iontas orainn cionn is go ndearnadh captaen de. Bhí iontas orainn gur ceapadh é mar chaptaen. b . I was surprised by her admission that she was ill. Chuir sé iontas orm nuair a dúirt sí go raibh sí tinn. c . I was surprised at his late arrival. Bhí iontas orm go raibh sé déanach ag teacht. Another context-dependent interpretation which seems to be virtually unattested with Irish verbal nouns is temporal. In many languages an action noun can denote the time span that it covers, especially when preceded by temporal extent expressions such as during, e.g. during the construction of the house. Again, attempts at extracting nominals with a temporal denotation from the dictionaries predominantly yield clausal equivalents (7c). Cases where the noun retains an actional interpretation are very rare (7a). I have found one such example in de Bhaldraithe (1959), and there are a few more such present-day cases as evidenced by the data from the New English-Irish Dictionary. Complex prepositions referring to duration, such as le linn, i rith and i gcaitheamh, can precede lexicalized verbal nouns which are like ordinary nouns (7b). 34 MARIA BLOCH-TROJNAR (7) a . ‘during the process of dismantling it’ le linn a bhainte anuas (de Bhaildraithe 1959) ‘during the equipment of the building’ le linn fheistiú an fhoirgnimh ‘during take-off and landing’ le linn an éirí de thalamh agus na tuirlingthe ‘during the mixing of ingredients’ le linn mheascadh na gcomhábhar b . ‘in the course of the meeting’ i gcaitheamh an chruinnithe (de Bhaildraithe 1959) ‘the rock was faulted during the earthquake’ éascadh an charraig le linn an chreatha talún ‘silence must be maintained during the examination’ ní mór fanacht ciúin i rith an scrúdaithe c . ‘during the fitting of the locks’ nuair a bhí na glais á bhfeistiú ‘I’ll talk to you during the break’ labhróidh mé leat nuair a bheidh an briseadh ann, labhróidh mé leat le linn an tsosa ‘during the clearance of debris’ nuair a bhí an bruscar á ghlanadh ar shiúl ‘during his convalescence from the illness’ agus é ag teacht chuige féin ón tinneas, le linn dó bheith ag teacht chuige féin ón tinneas ‘during his stay at the farm’ le linn dó bheith ag fanacht ar an bhfeirm On the other hand, action nouns in Irish freely show the manner interpretation, as can be seen in examples in (8) below, taken from Ó Dónaill (1977): (8) a. Tá siúl díreach aige. He walks straight. ‘manner of walking, gait’ b. Níl gearradh ná déanamh ar na cultacha aige. His suits are neither well cut nor well made. ‘style of cutting’ c. Is olc an fhéachaint atá air. He looks bad. ‘the way one looks, appearance’ d. Is olc an úsáid a thug siad don leanbh. They badly mistreated the child. ‘manner of handling, treatment’ e. Is deas an scríobh atá aige. He writes a nice hand. ‘(hand-)writing’ An interesting property of action nouns in Irish is the ability reading. Whereas in other languages this interpretation can be observed in nominals related to stative verbs only, in Irish it can be also observed with nouns derived from verbs relating to dynamic situations. Such verbal nouns denote the potential or the ability to perform an action. The examples in (9) come from Ó Dónaill (1977): 35 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH (9) a . Tá amharc maith agam. ‘I have a good sight.’ b . Tá éisteacht mhaith aige. ‘He has a good sense of hearing.’ c . Tá déanamh gnó ann. ‘He is well able to transact business, to manage.’ d . Níl tarraingt na gcos ann. ‘He can hardly drag his feet along.’ e . Níl léamh ná scríobh aige. ‘He can neither read nor write.’ 2.2 Sense extensions Melloni (2011) discusses Agentive-Collective and Locative sense extensions. The nominal amministrazione ‘administration’, just like its Polish equivalent, can denote a group of people or an institution involved in the performance of administrative activities and can refer to the place where such activities are carried out. Action nouns can show only the additional locative interpretation, as in entrata ‘entrance’. The collective reading can also be extended to nouns that refer to different systems, plants or instruments, such as riscaldamento ‘heating’, illuminazione ‘lighting’. Such sense extensions are possible if the morphological system lacks other means of expressing the collective/ locative senses in deverbal word formation (see “paradigmatic extensions under pragmatic pressure” in Booij & Lieber 2004). I have not come across cases where action nouns show collective sense extensions which would refer to animate entities (cf. téamh lárnach ‘central heating’, soilsiú sráide ‘street lighting’), unlike locative senses, which are exemplified in (10) below: (10) Verb Nominalization with a locative sense réitigh ‘level, smooth’ réiteach ‘clearing, level space’ réiteach i gcoill ‘forest clearing’ réiteach agus sliabh ‘plain and mountains’ doirt ‘pour, spill, shed’ doirteadh ‘watershed, slope’ snámh ‘swim, float’ snámh ‘swimming-place’; ‘deep pool’; ‘deep water, sea’ snámh abhann ‘swimming-place (fish-pool in river)’ amuigh ar an snámh ‘out in deep water’ Ná tit sa snámh. ‘Don’t fall into the deep.’ léim ‘jump, fly up’ léim ‘chasm, promontory’ 36 MARIA BLOCH-TROJNAR cart ‘clear away’ cartadh ‘dump’ luigh ‘lie’ luí ‘lie, slope, slant’ luí an chnoic, ‘the lie of the hill’ luí an chósta ‘the lie of the coast’ luí na tíre ‘the lie of the land’ 2.3 Lexicalized senses Cases of regular polysemy have to be teased apart from lexicalized senses. According to Beard (1987) “semantic drift” affects stored items in both principled and random ways. Semantic irregularity may be the offshoot of evolution from primary transparent meanings (e.g. construction, painting), or an idiomatic meaning may be added to the output of a regular process (e.g. revolution, transmission ‘gearbox’).3 Verbal nouns listed in (11) have disengaged from the word formation rule which generated them and so show an idiosyncratic non-actional interpretation: (11) Verb–vn Nominalization (plural) imigh ‘go, go on’ — imeacht imeachtaí an lae ‘events of the day’ buail ‘hit’ — bualadh bualaí cloiche ‘bruises from stones on the feet’ bris ‘break’ — briseadh bristeacha ‘breakers (waves)’ sáigh ‘thrust, stab’ — sá sá ‘stake’ sáite ciseáin ‘stakes of basket’ oscail ‘open’ — oscailt oscailt ‘first strip (in ploughing)’ 3 In the automotive sense transmission cannot have evolved slowly from the primary, transparent meaning ‘process of transmitting’. Carstairs-McCarthy (1992: 185) explains that “there is nothing in either the derivation itself or our experience of the world which might tell us that it relates to the transmission of power from the engine to the wheels in cars, rather than (say) the transmission of a message from the speaker to the hearer in a telephone conversation (whereby transmission might mean ‘handset’), or the transmission of an inherited characteristic from one generation to another (whereby transmission might mean ‘gene’).” Thus, the application of this term to a particular referent is the result of a conscious choice of a speaker and its subsequent institutionalization. 37 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH éist ‘listen to, hear’ — éisteacht Tá éisteacht anseo inniu. ‘Confessions are being heard here today.’ teagmhaigh ‘connect’ — teagmháil teagmhálacha leictreacha ‘electric contacts’ rith ‘run’ — rith rití fuachta ‘shivering spells’ siúil ‘walk’ — siúl siúl ‘travel, journey’ ar mo shiúlta ‘in my travels’ ardaigh ‘raise, ascend’ — ardú arduithe ‘horseshoes’ arduithe seaca ‘frost-nails’ glan ‘clean’ — glanadh glanadh ‘(clearance of ) afterbirth’ trácht ‘go, proceed, travel’ — trácht trácht ‘traffic’ croch ‘hang’ — crochadh crochadh ‘pothook’ oir ‘suit, fit, wish, need’ — oiriúint oiriúintí oifige ‘office accessories’ Apart from non-actional semantics, the count status of the noun may be an indicator of its lexicalized status; i.e. not all count deverbal nouns show unpredictable, idiomatized senses (see section 3 below). 2.4 Core readings In this analysis we are primarily interested in cases of logical/complementary polysemy, as put forward in Pustejovsky & Anick (1988) and Pustejovsky (1995, 2005). Logical or complementary polysemy can characterize both simplex and morphologically complex items. It refers to situations where a given lexeme has systematically related senses, which can be focused in distinct environments without being mutually exclusive (12a–b) and which can cluster together, producing the so-called dot-object, which allows a less specific, generic interpretation, as illustrated in (12c): (12) a. Mary enjoyed the book. (‘information object’) b. The book has a red cover. (‘physical object’) c. Mary likes the book. (‘physical object.information object’) 38 MARIA BLOCH-TROJNAR Pustejovsky (1995) argues that the semantics of action nouns in English is a reflection of the sub-events composing the event structure of the verb. An accomplishment verb, which is a left-headed transition, will lexicalize two sub-events, i.e. process and resulting state. The additional sense generated in the lexicon (i.e. dot-object reading) is an event reading, as represented in (13a) below (Pustejovsky 1995: 170–71): (13) a. The house’s construction was finished in two months. (process.result (event)) b. The construction was interrupted during the rains. (process) c. The construction is standing on the next street. (result) The result interpretation of nominals derived from verbs of creation may refer to the resulting state or the object created, i.e. the individual artefact resulting from the process (Pustejovsky 1995: 172).4 The nominal purchase derived from an achievement verb, which is a right-headed transition focusing the result state and not the process leading up to it, will show the event-object polysemy. Product or result nominals form the core of denotations of R-nominals. Pustejovsky’s proposal that the semantics of R-nominals reflects the aspectual properties of base verbs is pursued further in Melloni’s (2011) analysis, where the core semantics of R-nominals encompasses the product, means and entity in state readings, where entity in state can be viewed as a macro-category, since it captures the state element of meaning shared by all of them. The macro-category of entity in state is available to states and transitions, in Pustejovsky’s classification (Melloni 2011: 183). The path and measure interpretations are also closely linked to the aspectual properties of base verbs. 4 A more detailed discussion of the polysemy characterizing process-result nominals such as construction, translation or development is offered in Melloni (2011) and Melloni & Ježek (fc.). The (result-)state interpretation is available for verbs which express the process and the state, e.g. isolation. It is not available for creation verbs and redescription verbs, e.g. construction, translation. Such nouns cannot refer to the state of being constructed or translated. Instead, they denote concrete or abstract objects. Therefore, their inherent polysemy should be more appropriately classified as event.(result-)object rather than process.(result-)state. In this analysis, object and state readings will be regarded as the hyperonymic categories of result. 39 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH The product interpretation is prototypically associated with result nouns and refers to the entity or object coming into existence as a result of the process denoted by the base verb (construction, painting).5 R-nominals related to achievement verbs rivenimento ‘finding/discovery’, vincita ‘win’ and acquisizione ‘acquisition’, denote entities associated with the punctual/instantaneous changes of states. R-nominals can refer to objects or entities which play different roles in the situation denoted by the base predicate. The means interpretation, discussed by Bierwisch (1991), denotes the material/object used to perform the action: (14) Die Isolation des Kabels war defekt. The insulation of the cable was defective. Melloni (2011: 112) points out that the nominal can also denote the material/object resulting from the action (otturazione ‘a tooth filling’, argenatura ‘silver plating’, decorazione ‘decoration’). The nominals preferenza ‘preference’ and cognoscenza ‘knowledge’, related to unambiguously stative verbs preferire ‘prefer’ and conoscere ‘know’, denote the internal objects of base verbs. The R-nominal will refer to (animate/ inanimate) objects/entities associated with the states. With psychological verbs (which in aspectual terms can be causative/inchoative or stative), the R-nominal might refer to the entity/object corresponding to the source or stimulus of the psychological situation — to the exclusion of the Experiencer (e.g. sorpresa ‘surprise’). path nominals are derived from verbs with a path component in their semantics (Jackendoff 1990) which lexically involve a scalar structure (Hay, Kennedy & Levin 1999). Degree achievements (Dowty 1976) such as prolungare ‘extend’ and verbs of directed motion such as dicendere ‘descend’ give rise to nominals, which in addition to the product meaning, can refer to a spatial or directional path (prolungamento ‘extension’, discesa ‘descent’). The underlying scalar structure can also be observed in measure interpretations characterizing nominals derived from verbs with stative interpretations (apertura ‘span’). R-nominals do not correspond to a fixed argument position within the verb’s Argument Structure. It appears that all syntactic arguments can be 5 Product nominals can be subject to further semantic shifts, which can be also observed in the lexicon of simplex nouns. Just like book can be understood as a physical/concrete object and an ‘information object’, so can nominals related to verbs of writing and image creation activities. For example, translation can have an abstract information-object sense which is primary and a concrete one which is derived by metonymic transfer. 40 MARIA BLOCH-TROJNAR bound in derivation: internal and external (corresponding to the subject and object positions) and adjuncts (expressing manner/temporal/measure modification) of the base verb. Since in some cases the R-nominal cannot be associated with a syntactic argument at all, as in e.g. traduzione ‘translation’, the minimal requirement is that the R-nominal should correspond to a participant in the Lexical Conceptual Structure (lcs ) of the base verb. Since not all transition predicates give rise to logically polysemous nominals, there must be additional lexical constraints. In the course of her analysis, Melloni (2011) establishes that the product interpretation depends on the presence of an effected participant in the lcs of the base verb; i.e. the verbs must denote a creation event. This event may be broadly understood as representation or modification/alteration of the object’s referent. Effected arguments correspond to incremental arguments; i.e. their emergence is gradual and is mapped directly or indirectly onto the time course of the corresponding accomplishment event. Path interpretation in nominals depends on the presence of a non-effected incremental Path argument in the base verb. The formation of R-nominals in non-dynamic verbs depends on the presence of a non-sentient argument of the base; i.e. the R-nominal can refer to the non-Experiencer entity in a state (Melloni 2011: 115–16). Since the product-oriented interpretation is available to nominals which are characterized by incrementality and effectedness, the following classes of verbs are ruled out as potential bases: state verbs; activity verbs, which lack a state component (administrare ‘administrate’); directed motion verbs, describing situations that do not affect or modify the object that corresponds to a holistic Theme and implies a spatial path (salire ‘climb’); accomplishment and achievement verbs with an affected Theme, which do not lead to the creation of new entities or modifications conceptualizable as creations (asciugare ‘dry’); accomplishment and achievement verbs which involve the removal of the object (verbs of killing, verbs of removal and disappearance).6 We will now attempt to establish which aspects of verbal semantics are conducive to the emergence of R-nominals in Irish and how they affect the denotation of the nominal. We will discuss verbs belonging to various aspectual types and semantic classes to see whether the semantic generalizations formulated for Italian by Melloni also hold for Irish. 6 This would explain why there is an imbalance between E- and R-interpretations in that the former is the default denotation and the latter is not always available; e.g. the deverbal nominals such as abolizione ‘abolition’ or sfruttamento ‘exploitation’ have no R-interpretation. 41 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH 3. Base verb semantics and the interpretation of R-nominals In this section we will pay special attention to various semantic classes of verbs which fall into the category of transitions, i.e. accomplishments and achievements in the Vendler-Dowty taxonomy, since they are supposed to give rise to product and object nominals. 3.1 Verbs of creation According to Melloni (2011: 185) only creation verbs which express “the (potential) ability of putting into existence a new entity, i.e. the resulting product/artifact” can give rise to R-nominals with the product reading. Creation verbs subsume three subclasses of verbs, which predominantly are accomplishments with an incremental Theme or Path argument, along the lines proposed in Tenny (1994) and Dowty (1991). These are “Creation (/Result-Object) verbs” such as construire ‘build’, whose object is effected, not affected, changed or modified but created as a result of the action denoted by the base verb. “Creation by Representation” verbs such as tradurre ‘translate’ give rise to an entity-denoting nominal which is a representation of the Source argument, typically mapped onto the direct object position. “Creation by Modification” verbs such as correggere ‘correct’ motivate nominals which denote a new entity besides the affected (Patient/Theme) object. 3.1.1 Creation verbs Prototypical creation predicates denote events in which an Agent causes an entity to come into existence. Unlike Agent-Patient verbs, which also have an Agent in their theta grid, they do not take a Patient or affected object (wash a shirt, drive a car) but an effected object or object of result. Effected objects prototypically realize incremental Themes or Paths, i.e. those lcs participants which measure out or delimit the event described by the verb. The R-nominals corresponding to these verbs typically denote effected entities of creation events. The Irish data corroborate this. The verb tóg ‘build’ will give rise to a noun with the regular actional interpretation, as in tógáil tithe, droichead ‘construction of houses, of bridges’ and an entity-denoting count noun tógáil ‘structure’, as in (15) below: (15) Is breá na tógálacha iad. They are fine structures. 42 MARIA BLOCH-TROJNAR Some more examples are provided in Table 1 below:7 Table 1. R-nominals with a product interpretation related to creation verbs Verb Nominalization (plural) tóg ‘lift, raise, build’ tógáil ‘structure, building’ táirg ‘produce, manufacture’ táirgeadh, táirge ‘product’ léirigh ‘produce’ léiriú ‘production’ grean ‘engrave’ greanadh ‘engraving’, greanadh líneach ‘line-engraving’ However, they are outnumbered by nominalizations related to creation verbs, which do not give rise to entity-denoting nominals. These include, among others, déanamh ‘do, make.vn ’, cruthú ‘create.vn ’, foirgniú ‘construct.vn ’, ceapadh, cumadh ‘invent.vn ’, ullmhú ‘prepare.vn ’, seiftiú ‘provide.vn ’ and inscríobh ‘inscribe.vn ’.8 This may be so because the product interpretation is rendered by a derivative based on the action noun or the verbal root itself by means of some other suffix. (16) v–vn Nominalization with a specialized suffix cruthaigh–cruthú ‘create’ cruthúchán ‘creation, thing created’ foirgnigh–foirgniú ‘build, construct’ foirgneamh ‘building, structure’ inscríobh–inscríobh ‘inscribe’ inscríbhinn ‘inscription’ Tá inscríbhinn Laidine air. ‘It bears a Latin inscription.’ 7 In this section we examine cases of regular polysemy, which means that for each nominalization the regular interpretation of Action/Event/State is accessible. Only the entity-denoting interpretation is provided in the gloss. 8 In order to save space, nouns which are confined to the action or event reading are glossed here as ‘verb.VN’. 43 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH 3.1.2 Creation verbs by representation Representation verbs denote a situation where a new object/entity is created ‘besides’ or ‘in relation with’ the existing one. The internal argument of such verbs denotes an already existing entity. The result does not correspond to the entity denoted by the verbal object. Their internal argument can be characterized thematically as a Source and it is neither affected nor effected by the event. They are representation-source themes (Dowty 1991). Verbal nouns related to representation verbs which give rise to R-nominals are listed in Table 2 below. Table 2. R-nominals with a product interpretation related to verbs of creation by representation Verb Nominalization (plural) faisnéis ‘relate, recount, narrate’ faisnéis ‘information, intelligence, report’ faisnéis na haimsire ‘weather report’ faisnéisí trádála ‘trade returns’ aithris ‘imitate, mimic, mock’ aithris ‘imitation’ Níl inti ach aithris. ‘It is only a makeshift.’ Fainic aithrisí. ‘Beware of imitations.’ léirigh ‘clarify, explain, illustrate’ léiriú ‘clarification, illustration, representation’ léiriú grafach ‘graphical representation’ luaigh ‘mention, cite, quote’ lua ‘mention, citation, reference’ lua foinse ‘reference to source’ deimhnigh ‘certify, affirm, assure, check’ deimhniú ‘certification, certificate’ deimhniú cáilíochta ‘certificate of qualification’ Nominals derived from representation verbs and confined to the actional interpretation include, for example, aistriú ‘translate.vn’, cóipeáil ‘copy.vn’, brionnú, falsú ‘forge, falsify.vn’, clóscríobh ‘type.vn’, canadh ‘sing.vn’, péinteáil ‘paint.vn’, tarraingt ‘pull.vn’, líniú ‘draw.vn’ and aithris ‘narrate, recite.vn’. As in the case above, we may be dealing here with the phenomenon of blocking. The development of an entity denotation in action nouns is thwarted by the existence of a lexical item based on the same verbal stem with a concrete denotation or of a simplex noun which motivates the verb, e.g. 44 MARIA BLOCH-TROJNAR (17) v–vn Nominalization with a specialized suffix /Simplex noun motivating the verb aistrigh — aistriú ‘translate’ aistriúchán ‘translation’ Is aistriúchán ó Spáinnis go Béarla é. ‘It’s a Spanish-English translation.’ cóipeáil — cóipeáil ‘copy’ cóip ‘copy’ brionnaigh — brionnú ‘falsify, forge’ brionnúchán ‘something that has been forged’ clóscríobh — clóscríobh ‘type’ clóscríbhinn ‘type-script’ péinteáil — péinteáil ‘paint’ péintéireacht ‘painting’ péintéireacht ola ‘oil painting’ línigh — líniú ‘draw’ líníocht ‘(line-)drawing’ líníocht phionsail ‘pencil-drawing’ 3.1.3 Creation verbs by modification Modification verbs denote a situation where a tangible modification (involving the breaking, addition or subtraction of material) in the existing object takes place. The interpretation of the R-nominals corresponding to verbs which involve the addition of material is the means interpretation (material used to carry out the action, which can also be understood as the concrete result or product of the action). Verbal nouns related to modification verbs which give rise to R-nominals are listed in Tables 3a–c below. Table 3a. R-nominals with a product interpretation related to verbs of creation by modification Verb Nominalization (plural) ceartaigh ‘correct, rectify, amend’ ceartú ‘correction, amendment, adjustment’ ceartúchán ‘correction’ ceartúcháin a dhéanamh ar cheacht ‘to make corrections in a lesson’ ceartúcháin a dhéanamh ar scríbhinn ‘to make corrections in a written document’ athraigh ‘change, alter’ athrú ‘change, alteration’ athruithe ó bhun ‘sweeping changes’ 45 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH plátáil ‘plate’ plátáil ‘plating, sheeting, armour’ plátáil airgid ‘silver coating’ ceangail ‘bind, tie’ ceangal ‘tie, bond, obligation, fetter’ ceangail chaorach ‘sheep fetters’ nasc ‘tie, bind, link’ nascadh ‘binding, bond, obligation’ leasaigh ‘amend, reform, improve; season, flavour; fertilize’ leasú ‘amendment, improvement, reform’; ‘seasoning, flavouring’; ‘manure, fertilizer’ leasú reachta ‘amendment of statute’ leasú sailéid ‘salad dressing’ leasú nítrigineach ‘nitrogenous fertilizer’ teagasc ‘teach, instruct’ teagasc ‘body of teaching, doctrine, precept’; ‘magic formula, incantation’ teagasc na hEaglaise ‘the teaching of the Church’ briochta agus teagasca ‘spells and incantations’ múin ‘teach, instruct’ múineadh ‘moral, lesson’ múineadh scéil ‘the moral of a story’ clúdaigh ‘cover, wrap’ clúdach ‘covering, cover, wrap, lid’ clúdach boird ‘table-cover’ clúdach leabhair ‘book-cover’ clúdach piliúir ‘pillowcase’ clúdach buidéil ‘casing of bottle’ clúdach litreach ‘envelope’ cumhdaigh ‘cover, protect, upholster, enshrine’ cumhdach ‘cover, protection’; ‘(of relic) Shrine’ Cuir cumhdach ort féin. ‘Put on some protective clothing.’ líon ‘fill’ líonadh ‘filling’ líonadh i bhfiacail ‘filling in tooth’ tarrtháil ‘rescue, save, deliver, salvage’ tarrtháil ‘salvage’ tarrthálacha ‘things salvaged’ Also, events which tangibly/physically modify the object are expected to give rise to product R-nominals, providing that they do not give rise to the creation or modification that could be interpreted as a new entity. This class subsumes degree achievements such as méadaigh ‘enlarge’ and giorraigh ‘shorten’ as opposed to triomaigh ‘dry’ and dírigh ‘straighten’. 46 MARIA BLOCH-TROJNAR Table 3b. R-nominals with a product interpretation related to verbs of creation by modification (degree achievements) Verb Nominalization (plural) laghdaigh ‘lessen, decrease, diminish’ laghdú ‘decrease, diminution, reduction’ gor ‘heat, warm’ goradh ‘heating, warming, heat, warmth’ goradh dearg ‘to raise iron to a red heat’ goradh geal ‘to raise iron to a white heat’ goradh a thabhairt d’ iarann ‘to raise iron to heat’ méadaigh ‘increase, multiply, enlarge, grow bigger’ méadú ‘increase, multiplication, enlargement’ méadú ar shiopa ‘extension to shop’ Is mór an méadú a tháinig air. ‘He has grown a lot.’ giorraigh ‘shorten’ giorrú ‘abbreviation’ Nominals confined to the actional interpretation are in the majority and they include: sciamhú, breáthaigh ‘beautify, embellish.vn’, maisiú ‘adorn, decorate. vn ’, múnlú ‘mould, shape.vn ’, feabhsú ‘improve.vn ’, saibhriú ‘enrich.vn ’, bacadh ‘balk, hinder.vn ’, leathnú ‘widen, extend.vn ’, idirdhealú ‘differentiate.vn ’, claochlú ‘deteriorate.vn ’, ísliú ‘lower.vn ’ and ardú ‘raise, ascend. vn ’. Lexical items which block the emergence of the R-interpretation are given in (18): (18) v–vn Nominalization with a specialized suffix/ Simplex noun motivating the verb maisigh — maisiú ‘adorn, decorate, beautify’ maisiúchán ‘adornment, decoration’ maisiúcháin na Nollag ‘Christmas decorations’ feabhsaigh — feabhsú ‘improve’ feabhas ‘excellence, improvement’ Tá feabhas ar an aimsir. ‘The weather has turned fine.’ feabhsúchán ‘(act of ) improving, improvement’ deimhniú feabhsúcháin ‘improvement certificate’ bac — bacadh ‘balk, hinder’ bac ‘balk, hindrance, barrier’ 47 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH Another subcategory within modification verbs consists of verbs like break, which is a pure change of state verb, and cut, which additionally contains the manner element. Importantly, these verbs create a change of state in the verbal object, a change which is associated with the creation of a new object/entity, a by-product of the event denoted by the verb. Other prominent classes with these characteristics include verbs of bending and tearing/ separating. Table 3c. R-nominals with a product interpretation related to verbs of creation by modification (change of state verbs of breaking, separating, bending) Verb Nominalization (plural) bris ‘break’ briseadh ‘break’; ‘change’; ‘battle, defeat’ briseadh aife ‘(ruffle caused by) start of ebb-tide’ briseadh airgid ‘small change’ briseadh na Bóinne ‘the defeat at the battle of the Boyne’ deighil ‘divide, separate’ deighilt ‘division, partition, segmentation’ deighilt (a dhéanamh) idir dhaoine ‘to cause a split, a cleavage, between people’ rois ‘rip, rend’ roiseadh ‘rip, rent, tear’; ‘ragged cutting’ roiseadh i stoca ‘ravel in stocking’ roiseadh in éadach ‘rent in cloth’ roiseadh i gcraiceann ‘tear in skin’ oscail ‘open’ oscailt ‘opening’ oscailt idir charraigeacha ‘an opening between rocks’ oscailt súl ‘eye-opener’ fill ‘bend, fold’ filleadh ‘bend, fold’ filleadh in éadach ‘fold in cloth’ feac ‘bend’ feacadh ‘bend, bent posture’ feacadh i rud ‘a bend, a twist, in something’ fiar ‘bend, twist, distort’ fiaradh ‘warping, distortion’ 48 MARIA BLOCH-TROJNAR gearr ‘cut, shorten, levy’ gearradh ‘cutting, cut’; gearradh ar mhéar ‘cut on finger’ gearradh i mbléin ‘cut, incision, in groin’ gearradh rothaí i mbóithrín ‘wheel-ruts in lane’ na gearrthacha a íoc ‘to pay the rates’ Bhí na gearrthacha trom orainn. ‘The rates were a burden on us.’ scor ‘cut, slash’ scoradh ‘cut, slash, gash’ scoradh claímh ‘sword-cut’ This time the verbs which belong to the same classes and lack the entity-denoting interpretation in the nominal are in the minority, and they include réabadh ‘tear, rend, shatter.vn ’, dealú, pl. dealuithe ‘separate.vn ’ and ciorrú ‘cut, hack, maim.vn ’. It has to be underlined that verbs of varied aspectual classes, such as accomplishments and achievements but also activity verbs which do not take incremental themes but affected objects, contrary to the prediction of Melloni (2011), can give rise to R-nominals. The product interpretation can arise any time our system of conceptual and pragmatic knowledge detects a causally related by-product or effect associated with the event denoted by the verb. This is so with verbs of damage (e.g. loit ‘injure’), verbs of contact by impact (e.g. leadair ‘strike’) and verbs of exerting force (e.g. brúigh ‘press’) whose theta grid contains an Agent and Patient/Affected Object. The related R-nominals are fairly numerous and tend to denote the mark left on the Patient. Table 4. R-nominals with a product interpretation related to verbs with Affected Objects (verbs of damage, verbs of contact by impact and verbs of exerting force) Verb Nominalization (plural) loit ‘injure’ lot ‘hurt, wound, damage, violation, breach’ lot aithne ‘breach of commandment’ lot rialach ‘breach of rule’ goin ‘wound, stab’ goin ‘wound, stab’ goin bháis ‘death-wound’ 49 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH gortaigh ‘hurt, injure’ gortú ‘hurt, injury’ Bhí gortú ar mo chos. ‘I had an injured leg.’ Cá bhfuil an gortú ort? ‘Where are you hurt?’ leadair ‘smite, beat, hack, hew’ leadradh ‘beating, wound’ leadradh lainne ‘sword-stroke, sword-wound’ basc ‘bash’ bascadh ‘bashing, severe injury’ buail ‘hit, beat’ bualadh ‘beating, striking, fight’ speir ‘hamstring’ speireadh ‘crippling injury, fierce scrimmage’ gránaigh ‘scratch’ gránú ‘scratch, scrape, graze’ Níl ann ach gránú. ‘It is only a scratch.’ scríob ‘scrape, scratch’ scríobadh ‘scratch, scrapings; skin abrasion’ dóigh ‘burn, scorch’ dó ‘burn, burning sensation’ dó gréine ‘(discoloration from) sunburn’ dó laidhre ‘inflammation between the toes’ dó laidhre ‘(of crops) frostblight’ brúigh ‘press, push, shove, crush’ brú ‘pressure’ brú aeir ‘air pressure, atmospheric pressure’, brú fola ‘blood pressure’ fáisc ‘press, squeeze’ fáscadh ‘press, pressure, exertion, effort’ D’ imigh sé sna fáscaí. ‘He set off at great speed.’ fáscaí gáirí ‘fits of laughter’ den chéad fháscadh ‘at the first attempt’ srac ‘pull, jerk’ sracadh ‘piece torn off; piece, strip’; ‘impulsive act’; ‘pluck’ sracadh adhmaid ‘strip of wood’ sracadh talún ‘strip of land’ sracadh fir ‘manly impulse, manly act’ Tá sracaí aisteacha ann. ‘He does extraordinary things at times.’ sracadh caorach ‘heart, lights and liver of sheep’ As expected, pure contact verbs such as caress or touch, which cannot be associated with a visible trace or mark on the patient, do not give rise to nominals with R-interpretations. This is the case with cuimilt ‘rub, stroke.vn’, slíocadh ‘stroke.vn ’, cíoradh ‘comb.vn ’ and muirniú ‘fondle, caress.vn ’. In addition to this, if the verb lexicalizes a series of iterated atomic events (flail, trounce, 50 MARIA BLOCH-TROJNAR pummel ), an R-interpretation is not attested, even though there may be marks left on the Patient. This is true of giolcadh, sciúradh ‘cane.vn’, greadadh, greasáil, leasú, liúradh, rúscadh, stánáil ‘beat, trounce.vn ’, cíorláil ‘tousle.vn ’, radach ‘pelt.vn ’ and lascadh ‘lash, whip, flog.vn ’. No product interpretation is available even for forms which can pluralize, such as flípeáil ‘flip, strike; beat severely.vn ’ and súisteáil ‘flail, thresh.vn ’. There are no R-nominals with verbs whose theta grid contains an Agent and Patient (Recipient or Beneficiary). Consequently, nominals which denote verbs of social interaction are confined to the actional/eventive interpretation, e.g. cuidiú, cúnamh, fóirithint, fortacht ‘help.vn ’, cuimilt, sciúradh, scrabhadh ‘scold.vn ’, feannadh ‘criticize’ and raiceáil ‘maltreat.vn ’. Consumption/destruction verbs such as squander, waste, destroy, drink, consume and eat are semantic opposites of creation verbs. The internal arguments are incrementally affected by the gradual change induced by the event. Melloni (2011: 228) explains that “the state resulting from the completion of the event implies the complete removal of the object; consequently, when the action is accomplished there is nothing left for the R-nominal to denote”. Since there is no effected object, there can be no R-nominal with a product reading. This generalization holds for the majority of verbal nouns falling into this category: diomailt ‘waste.vn ’, meilt ‘chew, eat.vn ’, cnaí ‘gnaw, corrode.vn’, seargadh ‘waste, wither.vn’, pléascadh ‘explode.vn’, scriosadh, díothú ‘destroy.vn ’, milleadh ‘spoil.vn ’, ídiú ‘use up.vn ’, ithe ‘eat.vn ’, leá ‘melt.vn ’, galú ‘vaporize.vn ’, marú ‘kill.vn ’, cealú ‘cancel.vn ’ (cf. cealúchán ‘cancellation’) and dreo ‘decay.vn ’. R-nominals related to verbs of this category are few and far between. They denote an entity resulting from the destruction event. Jackendoff (1990) observes that destruction verbs could be interpreted as creation verbs. The only exception seems to be the verb ól ‘drink’, which takes an incremental affected object, and it is this object that is bound in the derivation of lexical meaning: Table 5. R-nominals with a product interpretation related to verbs of destruction and consumption Verb Nominalization (plural) raiceáil ‘wreck’ raiceáil ‘wrecking, wreck’ raiceáil loinge ‘the wreck of a ship’ goid ‘take away, remove, steal’ goid ‘thing stolen’ goid a ithe ‘to eat stolen food’ 51 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH cart ‘clear away’ cartadh ‘cleared-out material’ ól ‘drink’ ól ‘drink, drinking-liquid’ Is milis an t-ól é. ‘It is a sweet drink.’ 3.2 Non-prototypical creation verbs In this section we will examine the potential for forming R-nominals in verbs which can be classified as non-prototypical creation verbs. Non-prototypical creation verbs include cooking verbs, verbs of assembling, speech act verbs, mental action verbs and verbs of emission. Cooking verbs can acquire the function of verbs of creation and their corresponding nominals may show the object interpretation, as in róst ‘roast’ — róstadh ‘roasting, roast’ and grioll ‘grill, broil’— griolladh ‘broil, broiled meat’. However, no R-interpretations are available to fiuchadh, bruith ‘boil.vn ’, cócaráil ‘cook.vn ’ and bácáil ‘bake.vn ’, so the evidence from this category is inconclusive. Semantically, verbs of assembling and combining describe situations of assembling/grouping/associating of animate and inanimate entities and imply the gradual coming into existence of a resulting entity. Therefore, they are potential candidates for the product R-interpretation, as corroborated by the data in Table 6 below. Table 6. R-nominals with a product interpretation related to verbs of assembling and combining Verb Nominalization (plural) díolaim ‘glean, gather, collect, compile’ díolaim ‘collection, compilation’ díolaim dána ‘anthology of verse’ cruinnigh ‘gather, assemble’ cruinniú ‘gathering, assemblage’; ‘(needlework) pucker, gather’ cruinniú daoine ‘a gathering of people’ cruinnithe a chur in éadach ‘to pucker cloth’ bailigh ‘collect, gather’ bailiú ‘collection, accumulation’, bailiú daoine ‘gathering, assembly’ cnuasaigh ‘gather food (from woodland, sea-shore), pick (potatoes, etc.), collect, store’ cnuasach ‘garnered food’; ‘collection’ cnuasach coille ‘gleanings of woodland’ cnuasach trá ‘gleanings of seashore’ cnuasach mara agus tíre ‘food gathered from sea and land’ 52 MARIA BLOCH-TROJNAR tacair ‘glean, gather’ tacar ‘set’ tacar iata closed set’, tacar iomlán ‘complete set’ tacair scartha ‘disjoint sets’ saothraigh ‘earn’ saothrú ‘earnings, wages’ Tá saothrú maith aige. ‘He has good earnings.’ fiach ‘hunt, chase’ fiach ‘game’ fiach a fháil ‘to find game’ seilg ‘hunt, chase’ seilg ‘game, prey, quarry’ an tseilg a thabhairt isteach ‘to bring home the quarry’, seilg an lae ‘the day’s catch (of fish, etc.)’ There is no R-interpretation available for carn ‘heap, pile.vn’, eagrú ‘arrange. vn’ (cf. eagraíocht ‘organization’) and meascadh ‘mix.vn’ (cf. meascán ‘mixture’). Speech act verbs, verbs of communication and of the transfer of ideas, in overwhelming majority give rise to nominals with referential readings.9 These verbs take the Agent and the internal argument, which is a proposition expressed by an embedded clause, pp or a dp . Such a proposition can be interpreted as the speech artefact which is created through the accomplishment of the speech event. Being information objects, such nominals can give rise to metonymic displacements where abstract content is viewed as a concrete manifestation (a statement can be oral or written). Table 7. R-nominals with a product interpretation relatd to speech act verbs, verbs of communication and of the transfer of ideas Verb Nominalization (plural) abair ‘say, utter, speak’ rá ‘saying, utterance’ rá Bhearcháin ‘Bearchan’s saying, prophecy’ ráite béil ‘statements, remarks’ luaigh ‘mention, cite’ lua ‘mention, citation, reference’ lua foinse ‘reference to source’ 9 The following were found to be confined to the actional interpretation: fógairt ‘call out, proclaim.vn ’, aontú ‘assent, agree.vn ’, beannú ‘bless.vn ’ (cf. beannacht) and cúiseamh ‘accuse.vn ’. 53 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH maígh ‘state, declare’ maíomh ‘statement, assertion’ admhaigh ‘acknowledge’ admháil ‘acknowledgement, admission’; ‘receipt’ admháil creidimh ‘profession of faith’ foirm admhála ‘receipt-form’ deimhnigh ‘certify; affirm, assure’ deimhniú ‘certification, certificate, affirmation’, deimhniú cáilíochta ‘certificate of qualification’ deimhniú inniúlachta ‘certificate of competence’ dearbhaigh ‘declare, affirm, assure’ dearbhú ‘declaration; affirmation; confirmation’ mol ‘praise, commend’ moladh ‘praise’; ‘eulogy, panegyric’ moladh mairbh ‘panegyric for deceased person’, moltaí ‘lauds’ fianaigh ‘attest, testify’ fianú ‘attestation’ gearán ‘complain’ gearán ‘complaint, grievance’ tagair ‘allude’ tagairt ‘reference, allusion’ trácht ‘mention’ trácht ‘mention (of )’; ‘discourse’; ‘comment’ fiafraigh ‘ask’ fiafraí ‘inquiry, question’, Cén a fiafraithe sin ort? ‘Why all these questions?’ iomardaigh ‘reproach’ iomardú ‘reproach, accusation, challenge’ tairg ‘offer’ tairiscint ‘offer, bid’ tairngir ‘prophesy’ tairngreacht ‘prediction, prophecy’ tuar ‘prophesy’ tuar ‘sign, omen’ tuar cogaidh ‘portent of war’ tuar tubaiste ‘foreboding of tragedy’ conspóid ‘argue’ conspóid ‘argument, dispute, controversy’ impigh ‘entreat’ impí ‘entreaty, intercession’ Déanaim impí ‘I supplicate.’ díotáil ‘indict’ díotáil ‘indictment’ 54 MARIA BLOCH-TROJNAR éiligh ‘claim, demand’ éileamh ‘claim, demand’ éilimh ainmheasartha ‘unreasonable demands’ socraigh ‘agree, arrange’ socrú ‘settlement, arrangement, agreement’ socruithe sochraide ‘funeral arrangements’ bagair ‘utter a threat’ bagairt ‘threat’ bagairtí a spalpadh ‘to thunder out threats’ éar ‘refuse, deny, repel’ éaradh ‘refusal, denial’ eitigh ‘refuse’ eiteach ‘refusal’ eiteach dearg ‘a flat refusal’ ordaigh ‘order’ ordú ‘order, command, injunction’ orduithe máirseála ‘marching orders’ meil ‘talk, discuss, incessantly’ meilt ‘incessant talk’ Activity verbs which denote dynamic situations of the emission of light, sound and substances can give rise to an R-nominal, which refers to the emitted entity and so can be viewed as an effected object. Again, it has to be stressed that verbs belonging to this class do not satisfy the requirements of effectedness and incrementality put forward by Melloni (2011). Table 8. R-nominals with a product interpretation related to verbs of emission Verb Nominalization (plural) astaigh ‘emit’ astú ‘emission’ tál ‘(of milk) field, shed, pour, secrete’ tál ‘lactation’; ‘yield, flow’ tál breá bainne ‘good yield of milk’ tál deor ‘flow of tears’ gleadhair ‘beat noisily, pelt, pummel’ gleadhradh ‘noisy beating, clatter’; ‘blaze, flare, glare’ gleadhradh cos ar an urlár ‘clatter of feet on the floor’ gleadhradh cainte ‘din of talk’ gleadhradh tine ‘blazing fire’ gleadhradh solais ‘glare of light’ 55 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH liúigh ‘yell, shout’ liú ‘yell, shout’ liú a ligean ‘to give a yell’ liú catha ‘battle-cry’ blosc ‘crack, explode’ bloscadh ‘crack, explosion’ éagnaigh ‘complain’ éagnach ‘moan, groan’ éagnach a dhéanamh (le tinneas) ‘to moan (with sickness)’ Lig sé éagnach as ina chodladh. ‘He gave a groan in his sleep.’ crith ‘tremble’ crith ‘tremble, shiver, quiver’ creathanna fuachta ‘cold shivers’ tabhair ‘give’ tabhairt ‘grant, assignment, delivery’ cúitigh ‘requite, repay, compensate’ cúiteamh ‘requital, compensation’ cúiteamh ar chaillteanas ‘compensation for loss’ cúiteamh ar shaothar ‘reward for labour’ However, we do not observe a product interpretation in díoscadh ‘creak, grate.vn ’, brúchtadh ‘belch, burst, erupt.vn ’, ligean ‘release.vn ’ and sceitheadh ‘pour forth, discharge, erupt.vn ’. Directed motion verbs such as climb and go down indicate a change in location along a directed, spatial Path. Dowty (1991) argues that they have holistic (non-incremental) Themes. However, directed motion verbs are likely to show the non-eventive meaning — the directed Path covered by the moving entity (which is a Theme). Stative verbs with a degree interpretation will give rise to the measure reading. I have found only a handful of examples which illustrate this regularity: Table 9. R-nominals with a product interpretation related to verbs with degree interpretation Verb Nominalization (plural) tit ‘fall’ titim ‘fall’ titim aille ‘slope of cliff’ tarraing ‘pull, haul’ tarraingt ‘drag, haul, haulage, traction’ tarraingt cloch ‘haulage of stones’ Is fada an tarraingt é. ‘It is a long haul.’ 56 MARIA BLOCH-TROJNAR tuirling ‘descend, land’ tuirlingt ‘descent, landing’ slog ‘swallow, gulp down’ slogadh ‘swallow’ Tá slogadh fada aige. ‘He takes a long draught (has a large capacity for drink).’ lig ‘let, release, extend’ ligean ‘scope, stretch, extension’ ligean sciathán ‘wing-stretch measure’ croch ‘hang, raise up, lift’ crochadh ‘pitch, amount of slope’ crochadh áirse ‘the pitch of an arch’ crochadh dín ‘the pitch of an roof’ 3.3 Activities and states We now turn to the overview of verb classes which in Melloni’s (2011) analysis are not expected to give rise to R-nominals on account of their inherent aspectual and semantic characteristics. Activity verbs should not produce results, neither effected objects nor modifications. Since they lack an incremental Path component of meaning they cannot display a path reading either. This prediction is corroborated by the Irish data. The following verbal nominalizations related to activity verbs do not have entity denotations: rómhar, tochailt ‘dig.vn ’, foghlaim ‘learn.vn , taighde ‘research.vn ’, scuabadh ‘sweep.vn ’, iascach ‘fish.vn ’, rialú ‘rule, govern.vn ’ (cf. rialúchán ‘regulation, control’, rialtas ‘government’), stiúradh ‘steer.vn ’, riar ‘administer, manage.vn ’ (cf. riarachán ‘administration’) and fanacht ‘wait.vn ’. Mental action verbs relate to the opinion or judgement somebody may have in reaction to something. They are usually classified as states with a proposition or a cognate object as an internal argument. If attested at all,10 a corresponding R-nominal expresses the abstract content of a proposition, as in (19) below: 10 The nominalizations confined to the regular interpretation far outnumber those with resultative semantics: ceapadh, síleadh ‘think, consider.vn ’, meas ‘estimate, consider.vn ’, breathnú ‘observe, examine, judge.vn ’, samhlú ‘imagine.vn ’ and réasúnú, meabhrú ‘reason.vn ’. 57 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH (19) Verb Nominalization (plural) smaoinigh ‘think, reflect, consider’ smaoineamh ‘thought, reflection, idea’ smaointe casta ar chúrsaí an tsaoil ‘complicated thoughts on the affairs of life’ barúil ‘think, be of the opinion (that)’ barúil ‘opinion’ measúnaigh ‘assess’ measúnú ‘assessment’ State verbs, which describe non-dynamic situations, hardly ever give rise to R-nominals. R-nominals refer to the (animate and inanimate) entities associated with the states, e.g. teastaigh ‘be wanted, needed’ — teastálacha ‘(pl.) necessaries’, as in teastálacha beaga a bhí uaim ‘little things I needed’. With verbs which take a Subject Experiencer, the corresponding R-nominal can refer to the propositional content if the verb takes a subordinate clause or to the Source argument, e.g. meall ‘beguile, deceive’ — mealladh ‘deception, disappointment’, tnúth ‘long for, desire’ — tnúth ‘expectation, desire’. 4. Conclusion It is impossible to state the restrictions on the development of concrete senses in action nouns in absolute terms; i.e. we can only speak of tendencies. As rightly observed by Melloni (2001), it is not possible to predict the polysemy of nominals only on the basis of their event-structure properties. Reference has to be made to deeper levels of their semantics (be it qualia structure in the model of Pustejovsky (1995) or Body in the model of Lieber (2004)). We started our discussion by separating cases of inherent polysemy from context-dependent meanings, sense extensions and lexicalizations. We have observed that action nouns in Irish can give rise to manner and ability interpretations depending on the type of superordinate predicate. However, unlike verbal nouns in many other languages, they do not show the factual interpretation and the temporal reading is infrequent. In terms of sense extensions, action nouns can undergo metonymic transfer to denote locations and inanimate instruments but not animate collections. In this paper we have managed to identify several regularities pertaining to the emergence of result interpretations, which in Melloni’s terms can be viewed as product, object, means or entity in state. Melloni’s claim that R-nominals bear the feature [-dynamic] and cannot denote Agents or Instruments, i.e. lcs participants agentively involved in dynamic situations, 58 MARIA BLOCH-TROJNAR finds support in the Irish data. R-nominals assume the function of some other lcs participant in the situation. In many cases it is the effected object/Theme which comes into existence as a result of an incremental activity. Therefore, broadly understood, verbs of creation give rise to product readings. These include verbs of creation, representation and modification. R-nominals related to verbs of creation are not as numerous as expected on account of semantic blocking. Non-prototypical creation verbs relating to assembling deserve to be mentioned in this respect, unlike for example degree achievements. In the course of our analysis it has transpired that verbs, which need not satisfy the requirements of effectedness and incrementality are equally, or even more, important. Verbs of damage, contact by impact and exerting force, which in aspectual terms need not be classified as transitions and which take affected rather than effected objects, can give rise to product interpretations in R-nominals. Another prominent class which almost always gives rise to R-nominals consists of verbs relating to speech acts, communication and transfer of information. They denote the contents of the proposition which acts as the complement of the verb. In addition, activity verbs which relate to emission, understood in a broad way, can form R-nominals since the activity brings about effects/by-products in the form of emitted substances, sounds, movements, etc. Directed motion verbs and verbs of consumption which have an incremental argument have an R interpretation corresponding to the path. Manner of motion verbs which lack an incremental path are excluded as sources for R-nominals. Nevertheless, such nouns are attested. Due to the fact that the inventory of stative verbs in Irish is impoverished compared to other languages, the evidence relating to verbs expressing mental or psychological states, spatial configuration and propositional attitude is inconclusive and must be left for future research. The John Paul II Catholic University of Lublin 59 SEMANTIC (IR)REGULARITIES IN ACTION NOUNS IN IRISH References Alexiadou, A., 2001, Functional Structure in Nominals. Nominalization and Ergativity, Amsterdam. Anderson, M., 1984, ‘Prenominal Genitive NPs’, The Linguistic Review 3, 1–24. Beard, R., 1987, ‘Lexical Stock Expansion’, in: Gussmann, E., ed., Rules and the Lexicon. Studies in Word Formation, Lublin, 23–42. Bierwisch, M., 1991, ‘Event Nominalizations. Proposals and Problems’, Acta Linguistica Hungarica 40 (1–2), 19–84. Booij, G. & Lieber, R., 2004, ‘On the Paradigmatic Nature of Affixal Semantics in English and Dutch’, Linguistics 42, 327–57. Borer, H., 2003, ‘Exo-skeletal vs. Endo-skeletal Explanations. Syntactic Projections and the Lexicon’, in: Moore, J. & Polinsky, M., eds., The Nature of Explanation in Linguistic Theory, Stanford, CA, 31–67. Carstairs-McCarthy, A., 1992, Current Morphology, London and New York. Comrie, B., 1976, ‘The Syntax of Action Nominals. A Cross-language Study’, Lingua 40, 177–202. Comrie, B., & Thompson, S. E., 1985, ‘Lexical Nominalizations’, in: Shopen, T., ed., Language Typology and Syntactic Description, Cambridge, 349–96. de Bhaldraithe, T., 1959, English-Irish Dictionary, Dublin. Dowty, D., 1976, Word Meaning and Montague Grammar.The Semantics of Verbs and Times in Generative Semantics and in Montague’s PTQ, Dordrecht. Dowty, D., 1991, ‘Thematic Proto-roles and Argument Selection’, Language 67, 547–619. Doyle, A., 2002, Covert and Overt Pronominals in Irish, Lublin. Grimshaw, J., 1990, Argument Structure, Cambridge. Hay, J., Kennedy, C., & Levin, B., 1999, ‘Scalar Structure Underlies Telicity in Degree Achievement’, in: Matthews, T., & Strolovitch, D., eds., The Proceedings of SALT IX, Ithaca, 127–44. Jackendoff, R. S., 1990, Semantic Structures, Cambridge. Kastovsky, D., 1986, ‘The Problem of Productivity in Word Formation’, Linguistics 24, 585–600. Koptjevskaja-Tamm, M., 1993, Nominalizations, London. Langacker, R., 1991, Foundations of Cognitive Grammar, Vol. II. Descriptive Application, Stanford. Lieber, R., 2004, Morphology and Lexical Semantics, Chicago. Malicka-Kleparska, A., 1988, Rules and Lexicalisations. Selected English Nominals, Lublin. Marchand, H., 1969, The Categories and Types of Present-Day English WordFormation. A Synchronic-Diachronic Approach, München. 60 MARIA BLOCH-TROJNAR Melloni, Ch., 2011, Event and Result Nominals. A Morphosemantic Approach, Bern. Melloni, Ch., & Ježek, E., fc., ‘Complex Types in the (Morphologically) Complex Lexicon’, in: Bouillon, P., ed., Proceedings of the GL2009, 5th International Conference on Generative Approaches to the Lexicon, Sept.15–17, Pisa, Italy (available at http://attach.matita.net/elisabettajezek/file/Jezek-Melloni-GL2009.pdf ) [last accessed 09.02.2017]. New English-Irish Dictionary (available at http://www.focloir.ie/en/). Ó Dónaill, N., 1977, Foclóir Gaeilge-Béarla [Irish-English dictionary], Dublin. Pustejovsky, J., & Anick, P., 1988, ‘On the Semantic Interpretation of Nominals’, in: Vargha, D., ed., Proceedings of the 12th Conference on Computational Linguistics. Vol. 2, Budapest, 518–23. Pustejovsky, J., 1995. The Generative Lexicon, Cambridge, ma . Pustejovsky, J., 2005. ‘A Survey of Dot Objects’, Brandeis University (available at http: //www.cs.brandeis.edu/~jamesp/dots.pdf/) [last accessed 09.02.2017]. Rozwadowska, B., 1997, Towards a Unified Theory of Nominalizations. External and Internal Eventualities, Wrocław. Tenny, C., 1994, Aspectual Roles and the Syntax-Semantics Interface, Dordrecht. Walińska de Hackbeil, H., 1984, ‘On Two Types of Derived Nominals’, in: Testen, D., Mishra,V., & Drogo, J., eds., Papers from the Parasession on Lexical Semantics, Chicago, 308–32. Vendler, Z., 1967, Linguistics in Philosophy, New York. 61 Opportunities Seized: from Tolstóigh to Pelévin Mark Ó Fionnáin 1. Introduction From the earliest days of the Celtic Revival, various Russian authors began appearing in Gaelic guise. As the Irish language was finding its feet again after centuries of neglect and there was a movement underway to quickly produce new, written literature in a language which — apart from poetry and folk songs — had never been much cultivated in recent times, one of the easiest and quickest ways to do so was to translate: as the original text already existed, all that had to be done was to put it into Irish, and lo and behold, a short story, poem or novel instantaneously came into being in Irish. And thus, several of the major Russian authors of that era made it into print in many of the myriad publications at that time.1 The earliest translation would appear to be that of Lev Tolstoy from 1909, entitled An Maistín agus an Geampa Aráin ‘The Imp and the Lump of Bread’, and over the following years, Tolstoy would, numerically, seem to have been the most popular Russian author to be translated, with Chekhov not too far behind. Other major Russians who were presented to the Irish-language readership at that time included Pushkin, Turgenev and Dostoyevsky. This statement, however, does come with some necessary qualifications. The first of these is the fact that there was quite a lot of repetition of the stories translated, with a given story frequently undergoing several different versions over the years. The translations of Tolstoy, for example, include at least three versions each of two stories (Как Чертенок Краюшку Выкупал2 and Упустишь огонь, не потушишь3), and Pushkin’s Пиковая Дама ‘The Queen of Spades’ also underwent three different versions. 1 See Ó Fionnáin (2015) for a more detailed listing of the Russian texts that appeared in Irish during this period, and for more general information on issues of translating into Irish, including from Russian, and the questions raised, see O’Leary (1994, 2004, 2011). 2 ‘How a Little Devil Redeemed a Crust of Bread’, but usually entitled ‘The Imp and the Crust’ or ‘Promoting a Devil’ in English. The first version produced was the translation from 1909 mentioned in the main text. 3 ‘If You Miss the Fire You will not Quench it’, but usually entitled ‘Quench the Spark’ or ‘A Spark Neglected Burns the House’ in English. 63 OPPORTUNITIES SEIZED: FROM TOLSTÓIGH TO PELÉVIN There is also a further caveat involved. Most of these translations were merely described as being ‘a translation of a story by . . . ’, with no unambiguous statement regarding the original language, whether it was from the original Russian text or from an extant English translation. Examples from various publications include the following: “Sgeul on Ruisis: aistriú é seo ar Sgeul Rúisise do cheap Anton Tchehov” ‘A story from Russian: this is a translation of a Russian story composed by Anton Chekhov’, “‘Fiachra Éilgeach’ do chuir Gaedhilg air” ‘Translated into Irish by ‘Fiachra Éilgeach’’ or “Tolstoí na Rúise do scríobh, an “Seabhac” do chuir i nGaedhilg” ‘Tolstoy of Russia wrote it, ‘An Seabhac’ translated it into Irish’. It is thus unclear how many of these translations were done from the actual Russian and how many from existing English versions. There have been, however, several Irish speakers who did know Russian and who did thus translate from the original text, and the aim of this paper is to take a brief look at some of these translations in order to see how these translators dealt with the problem of foreign names and nouns and to see what purpose, if any, they saw their translations as serving. 2. Dia, Diabhail agus Daoine One of the first books to include translations of Russian authors was Father Gearóid Ó Nualláin’s (1874–1942) Dia, Diabhail agus Daoine ‘God, Devils and People’, from 1922. Ó Nualláin was the author of the four-part Studies in Modern Irish, a series that looked at the grammar of Modern Irish and analysed it in painstaking detail. He continued in this vein in Dia, Diabhail agus Daoine, a collection containing five of his own compositions and two translations from Russian. Each story in this volume is accompanied by exhaustive grammatical notes and observations, as he clearly saw his texts as having a didactic goal as opposed to an entertainment purpose, in keeping with his previous publications. The two translations in the book from the Russian are Pushkin’s ‘Snowstorm’ (entitled in Irish Síon agus Sneachta ‘Bad Weather and Snow’) and Tolstoy’s ‘What Men Live By’ (An Fiosrú, literally ‘The Visitation’).4 Ó Nualláin seizes the opportunity offered to him to provide his readership with copious endnotes containing a wealth of detailed insider knowledge on life and the customs in Russia at that time, including Russian culture (“Russian use of the word brat ‘brother’ sometimes means little more than 4 It should be noted here that on the website ainm.ie, a compendium of more than 1700 biographies of people who had links with the Irish language, Dia, Diabhail agus Daoine is described as being “aistrithe ó shaothar Rúisise Leo Tolstoy” ‘translated from Leo Tolstoy’s Russian work’, even though there is only one story by Tolstoy. 64 MARK Ó FIONNÁIN the English ‘friend’ and can just mean ‘kinsman’”, Ó Nualláin 1922: 93), food and drink (“kvas is a fermented drink made from rye and malt, or from ryeflour”, ibid., 94), accommodation (“The stove, in the Russian peasant’s house, is a large brick erection, the space on top forming a kind of loft, where several people would have room to lie down. It is, of course, a cosy spot”, ibid., 94), units of measurement (“the verst is .66 of a mile and the equivalent of 500 sachine (i.e. 500x7 feet)”, ibid., 108), items of clothing (a kaftàn is a “long, wide, old-fashioned overcoat worn by men”, ibid., 108), linguistics (“The final consonant in . . . Russian spelling . . . is sharpened in speech into the corresponding ‘voiceless’ sound”, ibid., 94), and the literal translation of some Russian idioms (“M. mumbled something under her breath. — The Russian has ‘under her nose’”, ibid., 93; “‘they are as wax in the candle to me.’ So the Russian. ‘They are the light of my life’ is the normal English way of expressing it”, ibid., 96). Regarding nouns and proper names themselves, the movement of stress in Russian is highly problematic, with nine possible patterns for nouns alone.5 Any learner of the language thus has to not only learn the six cases in the singular and plural for any given noun, but learn the stress pattern for that one noun and then remember it in use. It is very much a challenge. Most translators (and thus, by extension, their translations) tend to ignore marking the stress in Russian nouns and names, instead letting the reader pronounce Russian names such as Tolstoy or Ivan any way they wish. It is, however, generally understood that in transcriptions from Russian “the marking of stress position [is] highly desirable” (Comrie and Corbett 1993: 55). Ó Nualláin, exceptionally for his era, clearly felt so as well and, continuing in his vein as a didactician, he helpfully places the stress on place names and nouns “for the convenience of the reader” (Ó Nualláin 1922: 107). It is not clear how convenient it actually is for the reader to be able to pronounce (fictitious) Russian placenames such as Ненарадово/ Nienaràdovo correctly, but an incorrectly placed stress in a Russian word would clearly worry such a man of pedantry and grammar. Unfortunately, Ó Nualláin’s multiple pages of notes and comments on ‘correct’ Irish do somewhat distract from the joy of reading Pushkin or Tolstoy in Irish,6 a feeling that is echoed in Muiris Ó Droighneáin’s later comments on other compositions of Ó Nualláin’s that there is ‘a trace of the coldness of the man of grammar and logic on the fingers of the author and translator’ (“mar a bheadh iarracht 5 Ryan and Norman (1996: vii). 6 The translation of Pushkin takes up 20 pages and is accompanied by 6 pages of detailed notes; Tolstoy takes up 34 pages and also has 6 pages of notes. 65 OPPORTUNITIES SEIZED: FROM TOLSTÓIGH TO PELÉVIN d’fhuairneamh fhir an ghraiméir agus na laoighice ar mhéireanna an ughdair agus an aistrightheora,” Ó Droighneáin 1936: 66). 3. Dánta Próis Another author who knew Russian and who translated from the original texts was Liam Ó Rinn (1886–1943). Earlier on in his career he translated two works by Tolstoy, namely ‘God Sees the Truth but Waits’ and ‘A Prisoner of the Caucasus’,7 although these were probably done from the English, and in the 1930s he produced several translations of stories by Chekhov.8 In 1933, however, his original translations of Ivan Turgenev’s Стихотворения в Прозе, entitled Dánta Próis ‘Prose Poems’, was published. In his book Ó Rinn took an interest in correctly and accurately transliterating Russian for his reader. As such, Ó Rinn first presents his reader with a transliteration system, which he accredits to ‘two Irish-speakers’ (“beirt Ghaedhilgeoirí,” Ó Rinn 1933: 7). He (ibid.) explains this by saying: Is gnáth le gach éinne focail Rúisise do scríobh do réir foghraíocht a theangan féin i slí gur deacair uaireanta a cur thar n-ais sna litreacha Rúisise. Tá beirt Ghaedhilgeoirí, áfach, tar éis córus cruinn do cheapadh chun a scríbhte i litreacha Rómhánacha agus is dá réir sin atá gach focal Rúisise sa leabhar so. Everybody usually writes Russian words according to the phonetics of their own language in a way that makes it difficult at times to put them back into Russian letters. Two Irish-speakers, however, have come up with a precise system to write them in Roman letters and it is according to that system that every Russian word in this book is written. These two Irish-speakers, however, were very outward-looking, and did not limit themselves to the traditional Irish alphabet and phonemes. Aside from Irish (to represent some of the vowel sounds), they borrowed from English (ch and sh for the sounds /tʃ/ “mar atá sa bhfocal church” and /ʃ/ “mar atá sa bhfocal sheep”) and French (using j for the traditional English zh, i.e. /ʒ/, “j mar j sa bhfocal jour nó s sa bhfocal pleasure”). Ó Rinn, like Ó Nualláin, also points out the fact that Russian is devoiced at the end of words (“Deineann 7 Бог правду видит, да не скоро скажет appeared as Chíonn Dia an Fhírinne, acht Feitheann and Кавказский пленник as Ina Phrisúnach ar Shliabh Caucais. Both of these were printed in the newspaper Sinn Féin in July and August 1914. Ina Phrísúnach . . . was left unfinished. Chíonn Dia an Fhírinne . . . was republished in So Súd (Ó Rinn 1953) after Ó Rinn’s death. 8 Оратор ‘The Orator’ appeared as An Cainnteoir and Пари ‘The Bet’ as An Geall in the periodical Humanitas in the early 1930s. An Cainteoir [sic] was also republished in So Súd. 66 MARK Ó FIONNÁIN p, t, k, f, s agus sh de b, d, g, v, z agus j fé seach i ndeire focail do ghnáth”), but then goes into even more depth by mentioning the hard and soft signs na consainí eile leathan ach amháin roimh ghuth chaol nó roimh an gcomhartha caol (‘). Cuirtear an comhartha leathan (’) i ndiaidh consaine chun í dhéanamh leathan d’ainneoin caol-ghotha ina diaidh the other consonants are broad except before a slender vowel or the slender sign (‘). The broad sign (’) is put after a consonant to make it broad despite a slender vowel coming after it. (Ó Rinn 1933: 7) and broad and slender vowels (broad vowels are ‘broad at their front and end’ “leathan ina dtosach agus ina ndeire”, whilst slender vowels are ‘narrow at their front and broad at their end’ “caol ina dtosach agus leathan ina ndeire”). Furthermore, even though it is not specifically mentioned in his transliteration system, Ó Rinn also adopts the method of showing the stressed syllables in Russian words and names, allowing the reader to correctly pronounce Tatíschev, Prokopóvich, Sumarokóv, Vorontsóva-Dáshkova, Zapíski Okhótnika, Mërtvyia Dúshi and Revizór, amongst others.9 Ó Rinn prefaces his translations with a 16-page brief overview of the history of Russian literature, and takes the opportunity afforded to make a few political and cultural points. He starts off by defending ‘recent’ events in Russia — and the Russians — from their (nameless) detractors, those who would call the Russians half-barbaric (“leath-bharbartha”), and who are dismissive of Russian’s great noble literature, claiming that it had merely sprung up overnight (“fás aon-oíche an litríocht mhór uasal atá acu”). He also, however, sees in their literature a possible source of inspiration for the nascent Irish-language literature, especially in the case of translations, a theme close to Ó Rinn’s heart:10 An mhuintir go bhfuil eagla orthu go ndéanfaidh mórán aistriúcháin díobháil do thréithe dúthchasacha litríochta na Gaedhilge cuimhnídis dá mhéid leabhar do haistríodh go Rúisis nár bhain sé pioc o dhúthchasaíocht litríocht na teangan san: bíodh a 9 However, his transliteration system as presented is not complete, since he does not give any indication to his reader what sound ‹ë› /jo/, as in Mërtvyia Dúshi, is meant to represent, for example. 10 See Ó Fionnáin (2014) for a detailed look at Ó Rinn’s interest and approach to translation and Daltúin (2013) for a general overview of Ó Rinn and his works, including his translations. 67 OPPORTUNITIES SEIZED: FROM TOLSTÓIGH TO PELÉVIN fhianaise sin ar litríocht Rúisise na naoú aoise déag — atá chó bunúsach agus chó dúthchasach ar a laighead le haon litríocht dar cumadh in iarthar na hEorpa san aois chéanna. Brisfidh an dúthchas trí shúilibh an chait Ghaelaigh, leis, ach foidhne bheith againn. Those people who are afraid that too much translation will damage the native aspects of Irish literature, let them remember that however many books were translated into Russian, it did not take anything away from the nativeness of that language’s literature: as proof of that there is the Russian literature from the 19th century, which is at least as basic and as native as any literature that was composed in western Europe in the same century. Everything will happen naturally in Irish as well, if only we are patient. (Ó Rinn 1933: 16) He also seizes the opportunity to have a go at those who feel that Irish should remain pure and unsullied from foreign influences, especially in relation to the coinage of new words and neologisms:11 ach an dream a mholann dúinn focail iasachta do sheachaint . . . ba chóir dóibh mhachtnamh . . . d’ainneoin a usachta do sna Rúíseánaigh focail nua do cheapadh as préamhacha a dteangan féin, gur beag scéal le Chékhov, cuir i gcás, atá glan ó fhocail iasachta . . . agus iad litrithe do réir foghraíochta. but the people who advise us to avoid foreign words . . . should remember that . . . however easy it is for Russians to invent new words from the roots of their own language, there are few stories by Chekhov, for example, which are unsullied by foreign words . . . and which are spelt phonetically. (Ó Rinn 1933: 15–16) On a less political note, he also states that he feels that Turgenev is not as good as Tolstoy or Dostoyevsky and does not move his reader as Chekhov does (ibid., 24–5). He also notes in a footnote that Ukraine is a separate ‘nation’ and that Ukrainian is a sweeter language than Russian (“Náisiún fé leith atá sa Rúis Bhig agus teanga acu is binne ná an Rúisis”, ibid., 20). It is worth noting that the State publishers An Gúm only accepted this book for publication due to the scarcity of original translations from Russian 11 See Mag Eacháin (2014) for a detailed account of the issues at that time regarding the creation of new Irish terminology. 68 MARK Ó FIONNÁIN and not because the reading public might have much interest in it (Daltúin 2013: 210–1), although a review in The Cork Examiner (29 November 1933) did call it Cnósacht de ghiotaí beaga gleoite agus bíodh gur deacair an ghleoiteacht go léir do thabhairt isteach sa Ghaedhilg agus an dul céadna do bheith air, is iongantach mar éirigh leis an aistrightheoir é dhéanamh. Tuilleann sé an moladh mór dá bharr. A beautiful collection of small pieces and though it is difficult to bring all of the beauty across into Irish and have it have the same style, it is wonderful how the translator managed to do it. He deserves a lot of praise for this. (cit. from Daltúin 2013: 210) 4. Maighréad Nic Mhaicín Most of the extant translations from Russian into Irish, however, were done by Maighréad Nic Mhaicín (1899–1983).12 In the 1930s she published translations of Chekhov’s Вишнёвый Сад ‘The Cherry Orchard’ (1935) and a collection of Chekhov short stories entitled Gearr-Scéalta Cuid 1 ‘Short Stories Part 1’ (1939), although no second part ever appeared; in the 1950s she published translations of Turgenev’s Зaписки Охотника ‘A Huntsman’s Notebook’13 (1954) and, in conjunction with (the now late) Fr. Ó Nualláin, Scéalta ón Rúisis ‘Stories from the Russian’ (1955), a collection of four stories (two by Pushkin, one each by Tolstoy and Turgenev), with two translations contributed by her and two by Fr. Ó Nualláin. Unlike Ó Nualláin and Ó Rinn, Nic Mhaicín provided no endnotes, footnotes, introductions, commentaries or, indeed, personal opinions, leaving the reader with pure, unadulterated stories in Irish. However, her translations are not totally apolitical in that she deviated from what had been the norm up until then of writing Russian names and words in one of the many available English ways and actually made an attempt at Gaelicising the 12 Nic Mhaicín was a graduate of Celtic Studies and French in Belfast and visited Russia in 1932 when studying the language. While there in 1935 she married Irishman Patrick Breslin. However, after Nic Mhaicín returned to Ireland to give birth to their child in 1938 she was not allowed back into the USSR, and the Soviet authorities would not permit Breslin to leave. He eventually died of ill health in a Soviet camp in Kazan’ in 1942. In 1943 Nic Mhaicín became a lecturer in Russian at Trinity College, Dublin, retiring in 1969. For more on Nic Mhaicín, see Breatnach and Ní Mhurchú (1994: 61–2). 13 However, this collection contains translations of only 14 of the original 25 Russian stories. 69 OPPORTUNITIES SEIZED: FROM TOLSTÓIGH TO PELÉVIN names in use in her texts, thus trying to prove that Irish was just as capable of handling Russian names and nouns as any other Standard Average European language.14 Nic Mhaicín’s approach involved one of two ways. She either (very) occasionally used an Irish version of the name in question, for example Áine in place of the original Russian Аня /anʲa/ in her translation of Вишнёвый Сад, or else she Gaelicised the spelling. In other words, she limited herself to the 18 letters of the traditional Irish alphabet, and tried to adhere to the ‘caol le caol, leathan le leathan’ spelling rule as far as possible. This attempted Gaelicisation both worked and did not work. In the case of some of the easier, less problematic names, such as Тула ‘Tula’, there was no problem in inventing a Gaelic version like Túla (Nic Mhaicín 1954: 25, 209).15 With a name like Василий ‘Vasiliy’, she came up with Bheasailí (Nic Mhaicín 1939: 12), which, despite its awkward appearance, does give an approximate Irish pronunciation of the Russian.16 Less successful, though are many of her other attempts, for various reasons. In the case of Лизавeта ‘Lizavyeta’, for example, Nic Mhaicín solves the problem of the lack of a ‹z› and a ‹v› in Irish by using an ‹s› and the native Irish combination ‹bh›, and thus Gaelicising the name as Lisabheta (Nic Mhaicín and Ó Nualláin 1954: 44). Unfortunately, this leads to a further problem of how to pronounce said ‹s›: is it a slender /∫/ due to the slender ‹i› before it, a broad /s/ due to the broad ‹a› after it or, indeed, an English /z/, as in the name ‘Elisabeth’? In the case of the name Кoстя ‘Kostya’, rendered by Nic Mhaicín as Costia, how is her Irish ‹ia› to be pronounced? Is it /ja/ as in English or /i:a/ as in Irish? And in the case of a name, patronymic and surname coming together, the end result can get very much out of hand: Любoвь Андрeевна Ранeвская ‘Lyubov Andreyevna Ranevskaya’ from Вишнёвый Сад is Gaelicised by Nic Mhaicín as Liúbobh Andréemhna Rainémhscáidhea, although this unwieldy mouthful immediately raises several questions regarding orthography and pronunciation. Why is the broad/ slender orthographic rule broken? Why are both ‹bh› and ‹mh› used to represent /v/? Does the Irish síneadh fada accent show stress or a long vowel? 14 As one example of this variation, Чехов (Ch[y]ekhov) was variously written by Irish-language translators in the period in question as Tchekoff, Tchehov, Tsecheov, Tshechov, Chékhov, Tséchobh and Tsechobh. 15 The examples given in this section are a very small representative sample. A detailed analysis of all the names used in Nic Mhaicín’s translations (and by others) is beyond the scope of this work. 16 However, in a later story Vasiliy is written as Bheasilí, with a slender ‹s› giving /∫/ (Nic Mhaicín 1954: 120, 133, 181). 70 MARK Ó FIONNÁIN These issues were not exclusive to names: nouns and place names were also Gaelicised, but with the same orthographic and phonetic problems. Kvas was Gaelicised as cmheas (Nic Mhaicín 1939: 9), whilst the placename Крaсное Селo ‘Krasnoye Syelo’ appeared as Crasnoidhe Seló (Nic Mhaicín and Ó Nualláin 1955: 7). As Nic Mhaicín’s texts were also published in the Gaelic font, we are left with a situation where the potential Irish-language reader deliberately chooses to read these stories in Irish as opposed to the readily-available English-language version, in the Gaelic font, and in beautiful idiomatic Ulster Irish, but then has to decide whether to give the names in the text an Irish pronunciation or an English one or, indeed, a mixture of both. However, to give her her due, Nic Mhaicín, unlike Ó Rinn and Ó Nualláin, seems to have totally eschewed English letters for their Irish phonetic equivalents; for example, instead of the ‘English’ ‹v› she consistently uses ‹bh› or ‹mh› all the way throughout, and in this she was more successful and less inconsistent than other authors. Ris[t]eárd Ó Foghludha, for example, in his translation (probably from the English) of Chekhov’s The Bear (Ó Foghludha 1923) gives us Popobha as the surname of one of the main characters, with an Irish ‹bh› representing the traditional ‹v›, but we are then told that it is a play by Chekhov, with the traditional English spelling. The pedantic and hyper-correct Fr. Ó Nualláin (Nic Mhaicín and Ó Nualláin 1955: 96) presents us with Avdotya Semeonobhna in one of his translations, thereby giving a combination of a ‘foreign’ ‹v› and ‹y› and the native Irish ‹bh› and ‹eo› in the one name. A new book of translations of Russian authors into Irish — Scéalta ón Rúis (Mac Annraoi 2016) – has recently been published, although since they have been translated from English they will not be covered in this paper. However, the translator, Risteárd Mac Annraoi, has Gaelicised the names, presenting the reader with the likes of Gógal (Gogol’), Gointearov (Goncharov), Pilniac (Pilnyak) and Túirgéiniev (Turgenev), amongst others. As can be seen, he eschews the Irish ‹bh› for ‹v›, which is now acceptable in foreign words, and also avoids ‘foreign’ sounds by using, for example, ‹s› for ‹z› and a slender ‹t› for ‹ch›, e.g. Замятин (Zamyatin) is Gaelicised as Saimiaitin. Other authors given include Baibil (Babel), Dostaidheivscí (Dostoyevsky), Púiscin (Pushkin), Téachov (Chekhov) and Tolstái (Tolstoy). What is of more interest to us here is that Mac Annraoi also reproduces some of the earlier works by Ó Rinn and Nic Mhaicín, including Nic Mhaicín’s An Bhanríon Spád. However, he not only rewrites the title of her translation as An Bhanríon Spéireata (Nic Mhaicín and Ó Nualláin 1955: 39–67) and standardises a lot of the northern Irish forms she used, he also ‘improves’ 71 OPPORTUNITIES SEIZED: FROM TOLSTÓIGH TO PELÉVIN on her attempts at Gaelicisation, giving the reader Liosaivéata Iavanóvna, for example (Mac Annraoi 2016: 107–137). 5. Amón-Rá Regarding Modern Irish’s orthography, Campbell makes the bold and brave statement that “Irish has one of the least efficient writing systems in use” (1991: 632),17 and whilst this is patently untrue, Irish orthography would appear not to be ideally suited to transliteration, as can be seen from Nic Mhaicín’s brave and honourable failures above. In his foreword to Rashoomon, translations of several stories from Japanese into Irish, the translator Seán Ó Dúrois also reckons that Irish spelling is not suited to transliterating, and resorts to the common transliterated English version of names and places (Ó Dúrois 1995: 14–5). This is also the tack taken by Ó Fionnáin, stating in his introduction to Folcadán Airciméidéis, his translations of works by Kharms and Vvedénskii, that D’fhéadfainn na focail agus na hainmneacha Rúisise sa téacs a litriú de réir chóras fuaimnithe na Gaeilge, ach chonacthas dom go mbeadh leithéidi Púiscín, Tolstóí nó Vvieideanscaí ró-aduain dóibh siúd atá cleachtach cheana féin don léitheoir, go háirithe má tá sé cleachtach cheana féin ar leithéidí Pushkin, Tolstoi srl . . . I could have spelt the Russian words and names in the text according to the sound system of Irish, but it seemed to me that the likes of Púiscín, Tolstóí or Vvieideanscaí would be too strange for those which are already used by the reader, especially if they are used to the likes of Pushkin, Tolstoi, etc . . . (Ó Fionnáin 2004: vii–viii) This is also the approach taken in Ó Fionnáin’s translation of Victor Pelevin’s Омон Ра (2012).18 The titular character’s first name is taken from the acronym for the Russian riot police, omon (Отряд Mилиции (now Мобильний) Особого Назначения ‘Special Purpose Militia (now Mobile) Unit’). His second name is taken from the Egyptian sun god Ra. Thus, at first 17 This statement seems to have been expunged from later editions, however. In any case, the description of Irish itself in Campbell’s book has several errors in spelling, even in the later editions, such as mòr, na hein, De hAoine, aran, seachtan, and questionable grammatical forms, e.g. dosna clochaibh, tá thú ag obair, and should therefore in any case be approached with caution. 18 This is the most common Anglicised version of his name (Виктор Пелевин). Ó Fionnáin writes it in Irish as Víktor Pelévin, marking the stress positions in the name. 72 MARK Ó FIONNÁIN sight his name seems to consist of two separate units. However, this name also contains a third linguistic pun. Although it transliterates into English as ‹omon ra›, the correct pronunciation is /amon ra/, with an /a/ instead of the first, expected, /o/, and if pronounced in this way the reader will see (or, more probably, hear) the link to the great chief Egyptian god Amon-Ra.19 This linguistic pun is only available to those who pronounce the first ‹o› of the name as an /a/ in accordance with the rules of Russian stress and phonetics. Therefore, to be fully aware of the three meanings of the name, i.e. the riot police, the sun god Ra and the god Amon-Ra, the name, when given in a translation, should convey the very same information as far as possible. As the OMON is only known in Russia, any translation would need to explain the reference. Andrew Bromfield’s 1994 English version has the following on the back cover: “his name combines the Russian word for special police force and the Ancient Egyptian sun god”, whilst the Polish translation explains the reference in a footnote (“Otdieł [sic] milicyi osobogo naznaczenija — Specjalny Oddział Milicji ”, Rojewska-Olejarczuk 2007: 7). Neither of these two examples manages to convey the triple pun contained in the name. Ó Fionnáin, however, in an effort to impart to his Irish readers the triple meaning has added a sentence to the text, which reads: “Ainmníodh mé as an Arm Mílísteach d’Ócáidí Neamhghnácha” ‘I was named for the Militia Army for Special Occasions’ (Ó Fionnáin 2012: 1), thus linking the protagonist’s name with the special police force and, if the reader is au fait with the official Irish versions of Egyptian mythological names for deities, both Ra and Amon-Ra. Regarding Omon’s brother, Ovir, who was meant to become a diplomat and whose name is the acronym for the Russian Отдел Виз и Регистрации ‘Тhe Department of Visas and Registration’, different approaches are again employed: Bromfield ignores it totally, whilst Rojewska-Olejarczuk (2007: 8) once more explains this in a footnote (“Otdieł wiz i riegistracyi — Wydział Wizowo-Meldunkowy”). Ó Fionnáin, though, adds in extra information through a further rewriting of the text in order to, again, bring the Irish-language reader that bit of extra knowledge, rendering the brother’s name in Irish as ‘Arvac’, from the initial letters of An Roinn Víosaí agus Clárúcháin (Ó Fionnáin 2012: 1–2). Regarding the translation/ explanation of these unique Russian names, Vernitski has more to say on the topic, regarding Bromfield’s translation as a missed opportunity to give the English language reader further, deeper insight into the meaning of the names in the text (Vernitski 2000: 95), and stating with regard to ‘Omon’ and ‘Ovir’ that 19 Also variously spelt Amun-Ra or Amen-Ra in English. 73 OPPORTUNITIES SEIZED: FROM TOLSTÓIGH TO PELÉVIN The reader may wonder why this name [Omon] is not common, especially as on the same page the name of the protagonist’s brother, Ovir (in Russian, the abbreviation for Visa Section), is given without its being stated in the text that this is an unusual name. The English reader may well assume that neither name has any additional meaning. A further complaint made by Vernitski is the lack of explanation as to who Aleksándr Matrósov was. As Omon is returning to Moscow in order to become a cosmonaut, the lorry he is travelling in stops in the woods so that the driver can answer the call of nature. As he does so, Omon hears the sounds of machinegun fire in the distance and asks the driver what it is. He is told that it is from the Aleksándr Matrósov Infantry School, and the journey then continues. The black humour, however, is that just as Omon attends a flight school named after Alekséy Merése’ef, a pilot who lost his legs in the Second World War, and thus all students must have their legs amputated in his honour, those who attend the school named after Matrósov must learn to be like their very own Second World War hero, who stopped German bullets with his body so that his colleagues could storm a German position. This is the unstated ‘in-joke’ for Russian readers and should ideally be somehow brought across in translations. However, this reference to Matrósov is totally omitted in Bromfield’s translation, leaving the reader with a random incident in the woods, whilst the Polish text does reference Matrósov’s name but offers no further information as to who he was or what he did, leaving it to the Polish reader to find out what the reference — and thus the humour — is. The Irish reader, however, is again treated to a further explanation in the text, thus hopefully bringing the point home or, at the very least, giving the Irish-language reader the opportunity to make the connection themselves (Ó Fionnáin 2012: 33): [An C]oláiste Oiliúna in Ómós don Laoch Cogaidh Aleksándr Matrósov a thuill clú agus cáil as é féin a chaitheamh ar inneallghunna de chuid an namhad i rith an Dara Cogadh Mór. [The T]raining College in honour of the War Hero Aleksándr Matrósov who earned fame and renown for throwing himself on the enemy’s machinegun during the Second World War. Vernitski feels that this lack of explanation in the English translation leaves the whole episode of machine guns firing at a military school very random 74 MARK Ó FIONNÁIN and “more obscure and absurd” for the English-language reader (and, indeed, one could claim, for the Polish reader too). Overall, regarding Bromfield’s translation, Vernitski (2000: 98) feels that These examples suffice to show how a work which could have become a landmark translation into English . . . , a turning point in the Western appreciation of Russian culture, and which provide [sic] the ideal occasion to represent Pelevin to the English reading public became, instead, a missed opportunity. Hopefully, however, this was an opportunity seized for the Irish-language reader.20 6. Conclusion If the English translation of Омон Ра can be described as a series of ‘missed opportunities’, the Irish-language translations of works from Russian have generally better served their readership, even if such translations are few and far between. In the case of those few translators who actually did translate from the original Russian into Irish, stress was marked on names and nouns, explanations were given in the text or as endnotes, the reader was provided with historical developments of literature, political and cultural points, and, in the case of Nic Mhaicín, a brave attempt was made to show that Irish was capable of transliterating Russian, just as well as other, non-minority languages could. The Irish-language translators, far from missing opportunities, seem to have eagerly seized any chances provided and run with them. The John Paul II Catholic University of Lublin, 20 Indeed, Vernitski has a list of things about Bromfield’s translation of Омон Ра that he does not like and which he feels were missed opportunities, but a discussion of these and how they are dealt with in the Irish-language (and Polish) translation would be beyond the scope of this paper. As this paper has focused to a degree on the transliteration and handling of names in Irish translations from Russian, the discussion of Amón-Rá has been limited to examples of names. 75 OPPORTUNITIES SEIZED: FROM TOLSTÓIGH TO PELÉVIN References: Breatnach, D., & Ní Mhurchú, M, 1994, Beathaisnéis 4: 1882–1982. [Biography 4: 1882–1982], Dublin. Bromfield, A., trans., 1994, Omon Ra, London. Campbell, G., ed., 1991, Compendium of the World’s Languages, Volume 1, London. Comrie, B., & Corbett, G., eds., 1993, The Slavonic Languages, London. Daltúin, M., 2013, Liam Ó Rinn (1886–1943). Scríbhneoir Cathrach agus Criticeoir Iarchoilíneach [Liam Ó Rinn (1886–1943). Urban Writer and Postcolonial Critic], Unpublished PhD thesis, Dublin. Mac Annraoi, R., trans., 2016, Scéalta ón Rúis [Stories from Russia], Dublin. Mag Eacháin, C., 2014, Téarmaíocht Ghaeilge na hAthbheochana [Irish Terminology of the Revival], Dublin. Nic Mhaicín, M., & Ó Nualláin, G., trans., 1955, Scéalta ón Rúisis [Stories from the Russian], Dublin. Nic Mhaicín, M., trans., 1935, An Silín-Ghort le Anton tSechobh [The Cherry Orchard by Anton Chekhov], Dublin. Nic Mhaicín, M., trans., 1939, Anton tSechobh: Gearr-Scéalta Cuid I [Anton Chekhov: Short Stories Part 1], Dublin. Nic Mhaicín, M., trans., 1954, Scéalta Sealgaire [A Huntsman’s Stories], Dublin. Ó Droighneáin, M., 1936, Taighde i gComhair Stair Litridheachta na NuaGhaedhilge ó 1882 anuas [Research on the History of Modern Irish Literature from 1882 onwards], Dublin. Ó Dúrois, S., trans., 1995, Rashoomon, Dublin. Ó Fionnáin, M. trans., 2004, Folcadán Airciméidéis [Archimedes’ Bath], Dublin. Ó Fionnáin, M., trans., 2012, Amón-Rá, Dublin. Ó Fionnáin, M., 2014, Translating in Times of Turmoil: Liam Ó Rinn’s Irish Language Translations of Adam Mickiewicz, Lublin. Ó Fionnáin, M., 2015, ‘Na Ceithre Máistrí: Chekhov, Turgenev, Tolstoy and Pushkin and the translation of Russian into Irish’, in: Czerniak, T., Czerniakowski, M., and Jaskuła, K., eds., Representations and Interpretations in Celtic Studies, Lublin, 267–282. Ó Foghludha, R., trans., 1923, An Béar [The Bear], Dublin. Ó Nualláin, G., 1922, Dia, Diabhail agus Daoine [God, Devils and People], Dublin. Ó Rinn, L., 1914a, ‘Chíonn Dia an Fhírinne, acht Feitheann’ [God Sees the Truth, but Waits], Sinn Féin, July 11–August 1. Ó Rinn, L., 1914b, ‘Ina Phrisúnach ar Shliabh Caucais’ [A Prisoner on Mount Caucas], Sinn Féin, August 8–22. Ó Rinn, L., 1930, ‘An Geall’ [The Bet], Humanitas 1 (3), 13–17. 76 MARK Ó FIONNÁIN Ó Rinn, L., 1931, ‘An Cainnteoir’ [The Speaker], Humanitas 2 (1), 17–19. Ó Rinn, L., 1933, trans., Dánta Próis le Ivan Turgenev [Prose Poems by Ivan Turgenev], Dublin. Ó Rinn, L., 1953, So súd [This and That], Dublin. O’Leary, P., 1994, The Prose Literature of the Gaelic Revival, 1881–1921: Ideology and Innovation, Pennsylvania. O’Leary, P., 2004, Gaelic Prose in the Irish Free State, 1922–1939, Dublin. O’Leary, P., 2011, Writing beyond the Revival. Facing the Future in Gaelic Prose, 1940–1951, Dublin. Rojewska-Olejarczuk, E., trans., 2007, Omon Ra i inne opowieści [Omon Ra and other stories], Warsaw. Ryan, W. F., and Norman, P., 1996, The Penguin Russian Dictionary, London. Vernitski, A., 2000, ‘Viktor Pelevin and the Translation of Omon Ra: A Missed Opportunity’, New Comparison 29, 87–98. 77 Multiple Versions of Breuddwyd Pawl as a Source to Study the Work of Welsh Translators Elena Parina 1. Introduction Although the image of Middle Welsh prose is dominated both in academia and outside of it by the Four Branches of the Mabinogi and Culhwch ac Olwen, translations constitute the majority of text witnesses that have come down to us.1 The proportion of translated vs. native texts in medieval manuscripts cannot be ascribed to the vagaries of manuscript preservation but clearly manifests the importance of translations for Welsh readers. Whereas we find the Four Branches in their entirety in only two manuscripts, the translations of different versions of Historia Regum Britanniae by Geoffrey of Monmouth are found in 15 manuscripts alone in the period covered by the Rhyddiaith Gymraeg 1300–1450 project (Luft et al. 2013) and there are many more. Brut y Brenhinedd belongs to the texts that must have been so relevant for the Welsh readership that they were translated several times.2 The same is true for the text called Breuddwyd Pawl, or the Vision of Saint Paul, which will be discussed in this contribution. Having written on the translation found in Oxford, Jesus College ms. 119 (Book of the Anchorite of Llanddewi Brefi) elsewhere,3 I present in this article a preliminary approach for comparing it here with another Middle Welsh translation from a very similar Latin source, found in nlw, Llanstephan ms. 4. I argue that such multiple translations allow us important insights into the work of Middle Welsh translators and into the Middle Welsh language itself. Visio Pauli (vp ), known also as the Apocalypse of Paul, is a text written presumably in Greek in Egypt in the mid-third century.4 From this, 1 This research is part of the project ‘Übersetzungen als Sprachkontaktphänomene — Untersuchungen zu lexikalischen, grammatischen und stilistischen Interferenzen in mittelkymrischen religiösen Texten’, led by Prof. Erich Poppe at the Philipps-Universität Marburg, supported by the Fritz Thyssen Foundation. I am thankful to Prof. Erich Poppe for all his help in the preparation of this article. Advice from Dr. Sergey Ivanov was most important for the progress of this study. Needless to say I alone am responsible for the errors in this work. 2 On the relationship between several text witnesses, see Sims-Williams 2016a. 3 Parina 2016; Parina 2017. 4 For a useful introduction to the text in its different version, see Elliott 1999: 616 –19 and 79 MULTIPLE VERSIONS OF BREUDDWYD PAWL it was translated into Coptic, Syriac, Armenian, Ethiopic, Slavonic (where it is known as Слово о видении апостола Павла/Хождение апостола Павла по мукам) and Latin (in several redactions), from which it was in turn translated into European vernaculars. We are indebted to Caerwyn Williams for his analysis of the Middle Welsh Visio Pauli versions, as is also true of many other religious texts. In his 1962 article he proposed the existence of three independent translations into Middle Welsh, which he mapped onto the Latin tradition using the influential study of Silverstein (1935). One translation belongs to the Latin Redaction I and is found in Shrewsbury School ms . 11 (s. XIV/XV),5 nlw ms . Peniarth 32 (Y Llyfr Teg, s. XIV/XV), nlw ms Peniarth 50 (c. 1445), nlw ms . 5267 (1438)6 and nlw ms . Peniarth 267 (1640).7 I will not analyse this version in this contribution. My focus is on two translations of Redaction IV, the first found in nlw ms . Llanstephan 4 (s. XIV/XV; hereinafter Llst4)8 and the second (hereinafter the LlA version) in several manuscripts, the earliest of which are Oxford, Jesus College ms . 119 (Book of the Anchorite of Llanddewi Brefi, 1346, hereinafter LlA), nlw ms . Peniarth 3 (s. XIII/XIV), nlw ms . Peniarth 14iv (s. XIV1),9 nlw ms . Peniarth 15 (s. XIV/XV) and nlw ms . Llanstephan 27 (The Red Book of Talgarth) (s. XIV/XV).10 the Hell-on-Line web resource by Eileen Gardiner (http://www.hell-on-line.org/BibJC3.html#P [last accessed 1.05.2017]). See also Hilhorst 2007. 5 The dates for all Welsh manuscripts are taken from Huws (2000: 58– 64) unless otherwise stated. 6 Williams 1962: 112. 7 Ibid., 113. 8 It should be noted that Llst4 is written by the scribe designated by Daniel Huws as X91, who transcribed NLW MS. Peniarth 32 (Y Llyfr Teg) as well as parts of Oxford, Jesus College MS. 111 (The Red Book of Hergest), NLW MS. Peniarth 19 and NLW MS. Peniarth 190 (see TEI header for Llst4 in Luft et al. 2013; Huws 2000: 60). Llyfr Teg contains a translation of Redaction I of VP (see above), so that one could examine any possible contamination of the two redactions in Llst 4 and Llyfr Teg. This, however, remains a task beyond the scope of this contribution. 9 By Peniarth 14 I refer hereinafter to the fourth part of this composite manuscript (see Huws 2000: 59). 10 Williams believes that the texts in these manuscripts “are all derived from the same original, a Welsh translation of a Latin Redaction IV version of Visio Sancti Pauli, but there is an appreciable measure of variation in their readings, and it is difficult to establish a definite relationship between any of them other than the ultimate derivation from the same original” (Williams 1962: 117– 8). Poppe (fc.) has demonstrated that these texts show considerable linguistic variation, such as differences in word order, still following Williams’s hypothesis of a common ur-translation for the texts in these five manuscripts. 80 ELENA PARINA Finding Latin texts which could be regarded as similar to the supposed source for the Middle Welsh translations is difficult, given the extreme fluidity of the Visio Pauli tradition. When my findings presented elsewhere are summarised,11 it emerges that both versions go back to Latin texts of Silverstein’s Redaction IV, in the classification of the Jiroušková C-group texts of the so-called Hölle-Fassungen (Jiroušková 2006). The following opinion expressed in a recent study of the text with regard to one particular textual family could, in my opinion, be extrapolated to the whole range of our texts: “The changing internal affiliations of the B texts in relation to individual variants may be likened to the changing patterns of a kaleidoscope in response to each rotation” (Dwyer 2004: 93). Fortunately, these variants have been edited scene by scene from all the available manuscripts (48 for the version relevant for us) by Jiroušková (2006), so that I will quote several manuscript witnesses for each case or ‘rotation of the kaleidoscope’. The Latin sources for each of the two different translations of Redaction IV were definitely not identical, which can be shown with the following example: (1 ) nini a dywedwn dy uot ti yn vab y duw byw kan rodeist ti yni orffwys duw sul e hun. (l la : 132r) 12 We say that you are the Son of the living God because you gave us respite on Sunday. 13 Ni a’th vendig6n di vab du6 kanys ti a rodeist ynni gorffowys bop sul o boeneu ufferna6l. (l lst4: 38r) We bless you, Son of God, because you have given us respite of infernal pains on each Sunday. In Latin manuscripts we find, amongst others, variants such as the following: Benedicimus te, fili dei excelsi, qui donasti nobis requiem! (O5; Jiroušková 2006: 847) Et dicimus te filium dei vivi, qui dedisti nobis refrigerium diei huius, quod omne 11 See Parina 2017 for further details. 12 The texts are quoted after Luft et al. 2013. I have adapted punctuation and capitalisation. 13 All the translations are mine unless otherwise stated. I translate only the first Welsh quotation unless the differences between the Welsh versions or the Welsh versions and the Latin original are too numerous. 81 MULTIPLE VERSIONS OF BREUDDWYD PAWL tempus nostre vite in terra fuit! (L7; Jiroušková 2006: 846) Benedicimus te, fili dei, qui nobis donasti requiem die dominico tuo! (P7; Jiroušková 2006: 848) Et nos filium dei credimus, quia dedisti nobis refrigerium huius diei et noctis! (C6; Jiroušková 2006: 845) The LlA version must stem from a Latin text where benedicimus was changed to et dicimus, whereas in the Llst4 version we find a translation of the more common original variant. Other features in the same sentence link the Welsh versions to different Latin manuscript witnesses. Therefore, it is important to understand that we are not dealing with two translations of an identical text, as is often the case in translation studies dedicated to modern literature with a more stable printed source text, but with translations of different sources which must, however, have been similar enough to allow us some comparison.14 In this contribution, due to the extreme fluidity of the Latin Visio Pauli, I quote different Latin manuscript witnesses for different instances; for reasons of space, I cannot lay out the reasons for the choices here. The texts are quite short and additionally Llst4 lacks a folio,15 so that I have excluded the non-matching part of the LlA text from my study. The LlA text thus contains 1258 words; the Llst4 text contains 1213 words. For this study the texts have been aligned with the relevant Latin versions; both texts were also part-of-speech tagged with the help of Dr. Marieke Meelen.16 Despite these formal approaches, the size of the texts does not allow us to draw any wider conclusions on the language and style of the texts and my discussion will be by necessity anecdotal. 2. Comparing two versions The following example shows how a comparison of two translations contributes to our understanding of translators’ decisions and of the text style. 14 Nely van Seventer states that the two translations of Sibilla Tiburtina – the one in the Red Book of Hergest and the White Book of Rhydderch (van Seventer 2018) and the other in Peniarth 14 – are translated from a common source, but independently (van Seventer, 2017). 15 Williams 1962: 124. 16 The PoS tags also contain some morphological information, such as plural number for nouns and adjectives or tense, mood and person-number agreement for verbs. I am indebted to Raphael Sackmann for help in correcting the automatically assigned tags. 82 ELENA PARINA (2) Hyeu, medeu vn o’r diefyl, Llyma yr eneit a tremyga6d gorchymynnev duw a’e gyfureithev. Ac yna darllein chartyr a’e pechodeu a’e weithredoed drwc yn yscriuennedic yndi ac yn y varnv yg kyfuyrgoll. (l la : 131r) Yea, said one of the devils, here is the soul that despised God’s orders and His laws. And then he read/was read a charter with his sins and bad doings written in it and condemning him to perdition. Edrych6ch, heb 6y, pa wed y tremyga6d yr eneit orchymyn du6 tra vu yn y byt. A’r eneit truan a’e lythyr yn y la6 yn darllein. yn yr h6nn yd oed y bechodeu yn ysgriuenedic ac yn y varnu e hun. (l lst4: 36v) Look, said they, how the soul despised God’s orders as it was in the world. And the poor soul reading its letter in its hand, in which its sins were written and condemning itself. Latin text witnesses differ significantly in this passage. I choose one from Oxford, Merton College, MS. 13 (s. XIV ex. or XV in.), the text published in the 1894 edition of the Welsh LlA text. 17 Et dixerunt adinvicem: Vidimus, quomodo anima ista contempsit mandata dei in terra! Tunc legit anima ista cartam suam, in qua scripta erant peccata sua, et se ipsam iudicabat. (O5; Jiroušková 2006: 790) In London, British Library, MS. Royal 8.E.XVII (s. XIII ex.–XIV in.), one of the manuscripts that has a text usually closer to the LlA version, the word scripta is absent, but we probably have to suppose that it was there in both sources of the Welsh texts. Et iterum dixerunt: Ista anima contempsit preceptum domini et legit cartam suam, in qua erant peccata sua, et se ipsam iudicavit. (L7; Jiroušková 2006: 789) However intensive the variation within the Latin text witnesses is, it is safe 17 For other variants, see Jiroušková 2006: 789–92. 83 MULTIPLE VERSIONS OF BREUDDWYD PAWL enough to suppose that the Latin originals had ‘a letter’, carta, modified by a relative sentence introduced by in qua and containing also the participle scripta. On the lexical level, the word carta is translated differently in the two versions. In British Latin, c(h)arta means mainly a charter, “formal deed or instrument authenticated by witnesses or seal or both”, but meanings of “letter or informal note” are also attested (Latham 1997, s.v. charta). The LlA version has chartyr (LlA) (spelled as chartẏr in Peniarth 15, sartyr in Peniarth 3 and Peniarth 14, and syartyr in Llanstephan 27). gpc quotes this example under “siartr, siarter, &c.” and suggests that the lexeme is a loanword from M.E. chartre or directly from Old French. The meaning given is “(royal, &c.) charter; formal document; also fig.”, and indeed most of the early examples show its usage in a formal sense; cf. the following: (3) ef a gennhadawd y brenhin [ . . . ] y sartyr ydan y inseil ac inseil y legat. (Brut y Tywyssogyon, Peniarth 20, 282; Jones 1941: 218) The king sent [ . . . ] his charter under his seal and the seal of his legate. The word is generally rather rare; I have found only eight examples in the Rhyddiaith Gymraeg 1300–1425 corpus for <charter, chartyr, siartyr, sartyr>. Llst4 uses a more generic word for carta — llythyr — a polysemous word, meaning, as the English letter, both ʻan alphabetic character’ and ʻa written text’. In the Rhyddiaith Gymraeg 1300–1425 corpus, 385 examples of (l)lythyr are found. We can thus suggest that chartyr in the LlA version is chosen under the influence of the Latin text, whereas the Llst4 version translator was more target-oriented in his choice. The participle scripta is translated in both versions as ysgriuennedic. Adjectives in -edic are regarded as one of the strong markers of translation style.18 In a prescriptive discussion of the ‘natural Welsh language’ the form ysgrifennedig as translation from scriptum est is specifically mentioned and criticised.19 I will come back to the relationship between Latin perfect participles and –edic adjectives later (3.2.1). In the case of this instance, it is significant that both versions translate scripta as ysgriuennedic, and the data of the prose corpus (Luft et al. 2013) suggest that this lexeme was used quite frequently (55 examples are found for different orthographic variants). 18 See Luft 2016: 172. 19 Morgan 1970: 271; on -edic-adjectives, see also Nurmio 2015: 174–9, and Russell 1990: 76– 80, 103–8. 84 ELENA PARINA The relative clause containing scripta, on the other hand, is translated with two different constructions. In LlA we find chartyr a’e pechodeu a’e weithredoed drwc yn yscriuennedic yndi, lit. ʻa charter with his sins and his bad deeds written it itʼ, with no finite verb and with a conjugated preposition. In Llst4 we find a’e lythyr [ . . . ] yn yr h6nn yd oed y bechodeu yn ysgriuenedic, lit. ʻand his letter . . . in this his sins were writtenʼ with the relative construction described by Evans as that used “in translated works” (gmw 66) and with a preposition and the demonstrative yr hwnn.20 The exact distribution of these relative constructions will be analysed in depth in our project.21 What we can say about this example is that the LlA version appears in this instance to be more similar to the style of native texts, whereas Llst4 uses a construction that is much more frequent in translated texts. 3. Similarities I will first discuss two instances in which the two translations show similar behaviour. The first concerns lexical choice. In this preliminary study I can only address one word, but a systemic comparison of the vocabulary of the two versions would be rewarding. 3.1. Lexical choice (4) Ac yna pawl a welas gyr bronn pyrth vffernn deri tanllyt. (l la : 129r) And then Paul saw next to the doors of hell fiery trees. Sef y g6eles pa6l geyr lla6 porth uffern deri tanllyt. (l lst4: 35v) Vidit Paulus ante portas inferni arbores igneas (l 8; Jiroušková 2006: 664) In both versions Latin arbores is translated by Welsh deri. According to gpc , it is a plural of dâr which means “oak-tree; fig. foremost warrior, leader, mighty lord” (gpc online, s.v. dâr). However, the translations given in the dictionary do not suit our context. It would be strange to suppose such 20 On this type of relative clause, see also Sims-Williams 2016b: 151–2. 21 See footnote 1. 85 MULTIPLE VERSIONS OF BREUDDWYD PAWL a detailed knowledge of infernal flora in Welsh and to suggest that the translator deliberately changed a generic word for ʻtreeʼ to ʻoakʼ. It is much more likely that we find here, as in the case of the cognate derw (a collective noun),22 a generalisation of meaning,23 so that an adequate translation of the Welsh versions would be the hyperonym ʻtreesʼ. Therefore, the translation by Williams (1892: 635) —“And then Paul saw before the gates of hell a fiery oak-grove”— is to be rejected. An analysis of other usages of deri in Middle Welsh prose shows that the generalised meaning ʻtreesʼ is not unique to the Visio Pauli texts. An example from Brut y Brenhinedd and its translation by John Parry can serve as another illustration: (5) Ac yno ybu kyfranc kalet yryngthunt. allad llawer o bop tu. canys yno ybrethit y bryttannyeit ogysgot yderi. Ac yno y perys arthur llad y deri. (Brut y Brenhinedd, BL Cotton Cleopatra MS. B V part i, 78v) And then there was a fierce battle between them, and many were slain on all sides, for there the Britons were wounded by the shade of the oaks. And then Arthur had the oaks cut down. (Parry 1937: 158) Parry translates deri as oaks, but if we look at the Latin text, the general arbores are found there, which is more natural given the context of the episode: Conserto itaque proelio, stragem Britonibus faciunt, sese uiriliter defendentes. Vsi etenim arborum auxilio, tela Britonum uitabant. Quod Arthurus intuens iussit arbores circa illam partem nemoris incidi. (hrb IX.145) Once the battle was joined, they defended themselves valiantly and slaughtered the Britons. Moreover the trees permitted them to avoid the Britons’ weapons. Noting this, Arthur ordered the trees surrounding that part of the forest to be cut down. (hrb : 196) 22 GPC, s.v. derw —“oaks, oak-trees; (sometimes) terebinth tree (in bibl.); trees; plants having some resemblance to oak; transf. (in medieval poetry) oak coffin; fig. valiant man, stout warrior; of oak, oaken”. 23 See Geeraerts 1997: 68–79. 86 ELENA PARINA I take this example as proof of the second part of John Rhŷsʼ note in his preface to the Book of the Anchorite edition: The texts, being translations, cannot be regarded as the best models for Welsh prose, but they are important in the lexicographical sense, as helping to fix the exact meaning and connotation of words, the indefiniteness of which, when they occur in medieval Welsh poetry, leaves not a little room for doubt. (Morris-Jones and Rhŷs 1894: v) While the critical examination of the first statement on the syntactical qualities of the translated texts is the core of our project, the second statement on the importance of translated texts for semantic research remains true — so that despite the existing precision of lexicographical description of Middle Welsh in gpc , these texts can still yield more information. 3.2. Grammar Another aspect where one can find similarity is grammar. I will first address some more general issues and then turn to the question of the tense marking of the verb gwelet ʻseeʼ. Since we are dealing with two translations, one would expect to find features of the translational style, such as those noticed by Roberts:24 greater use of concord of a plural adjective and noun; concord of a plural verb with a plural subject following; the position of the adjective before the noun it modifies; the use of demonstrative pronouns as relatives. I have discussed an example of the last feature in section 1, and ex. 18 is another case of this. Concord of a plural subject and verb is discussed for one sentence in ex. 20. There are no more data in our sample for further comparison. The following can be said about the syntactic behaviour of adjectives: there are no plural adjectives in attributive positions, but this evidence is neutral because the adjectives found with plural nouns do not have plural forms (e.g. deri tanllyt ʻfiery treesʼ and y weithredoed da ʻhis good deedsʼ in both versions). As for the position of the adjective, in LlA adjectives precede the noun twice in 17 noun phrases with adjective in attributive function; in Llst4 this occurs in 3 out of 21 cases. These results are comparable, though not every individual noun phrase has an equivalent in the other version. Altogether there are five examples of adjectives preceding nouns in LlA and Llst4, and among these, there are three examples of phrases with ordinal numerals which always precede the noun (gmw 48): yn y pedweryd nef ʻin the fourth heavenʼ (LlA); 24 Roberts 1975: xxviii, quoted in Luft 2016: 172 and Nurmio 2016: 162, Russell 2017: 231. 87 MULTIPLE VERSIONS OF BREUDDWYD PAWL hyt y trydyd nef ʻup to the third heavenʼ; o’r nawuet awr ʻfrom the ninth hourʼ (Llst4). In both versions we also find examples of the adjective preceding the noun when the nominal phrase is used in a vocative function: wynnvydedic pawl ebostol ‘o blessed Paul apostle’ (LlA) and druanaf eneit ʻpoorest soulʼ (Llst4). A detailed analysis of the frequency of this feature in native texts and the texts of LlA is to be undertaken within our project; what can be demonstrated at the moment is the presence of some examples of adjectives preceding nouns (in the future ordinal numerals should be excluded from the statistics since they always precede the noun in m .w .), but the very limited number of these examples is also clear. The usage of the verb gwelet in the past is the second aspect in which we can see similarity in the grammar of our versions that is not just due to the fact that both are written in Middle Welsh but that is possibly due to the fact that they are both translations. Visio Pauli is, as the title suggests, a text focused on vision, so that Latin vidit is used in the text frequently enough to allow generalisations on how the Latin perfect of a perception verb is translated. In the corresponding parts of vp in LlA and Llst4 we find vidit six times. It is translated twice in both versions with the imperfect form gwelei: (6) Ac edrych a oruc pawl o bell y vrthaw, ac ef a welei eneit pechadur yn rwym gann seith gythreul wedy’r dwyn yr awr honno o’r corff, ac ynteu yn gweidi ac yn vdaw. (l la : 131r) And Paul looked afar off, and he could see a sinner’s soul bound by seven devils newly taken, at that hour, from the body, while he shrieked and howled (Williams 1892: 636) Ac edrych a oruc Pawl y’r nef ac y’r daear, ac ef a welei eneit pechadur y·r6ng seith kythreul yn vda6 ac yn dryc·yruerth g6edy y d6yn y dyd h6nn6 o’r korff. (l lst4: 36r) Post hoc asspexit longius, vidit animam peccatricem inter septem diabolos ululantem et exeuntem de corpore eo die. (L8; Jiroušková 2006: 784) Three times vidit is translated by the preterite form gwelas (in LlA)/gweles (in Llst4); see ex. 4 above or the following sentence. 88 ELENA PARINA (7) Ac yna y gwelas pawl y nef yn kyffroi. (l la : 131v) And there Paul saw the heavens stirred. (Williams 1892: 637) ac yna y g6eles pawl y nef yn kyffroi. (l lst4: 37r) Et tunc Paulus vidit celum subito moveri. (L8; Jiroušková 2006: 822) However, if we look at other text witnesses of the LlA translation, we can see variation there: Ac yna y gwelei Bawl y nef yn kyfroi (Peniarth 3: 29); ac yna y gwelei pawl y nef yn kyffroi (Llanstephan 27: 54r); Ac yna y gweles y nef yn kyffroi (Peniarth 14: 159). On one occasion on which the versions differ, LlA has a preterite form, while Llst4 has an imperfect. (8) Odyna y gwelas pawl gwyr a graged yn noethon. (l la : 130v) Then Paul saw men and women naked. (Williams 1892: 636) ac ef a welei yn y pyde6 g6yr a g6raged meibyon a merchet yn noethon. (l lst4: 36r) Et vidit Paulus in alio loco viros et mulieres nudos. (L7; Jiroušková 2006: 778) This variation seems to be indicative of a very fine semantic distinction between gwelei/gwelas forms. The significance of this imperfect/preterite variation becomes clear in comparison to the language of native prose. One usually encounters the following example of gwelei during the first days of learning Middle Welsh: (9) Ac ef a welei lannerch yn y coet, o uaes guastat. (Pwyll Pendeuic Dyuet; Thomson 1957: 1.13) 89 MULTIPLE VERSIONS OF BREUDDWYD PAWL And he could see a clearing in the forest, a level field. (Davies 2007: 3) D. Simon Evans lists a separate meaning “possibility” for the imperfect with verbs of perception (GMW 110), which we also see in the translation of Davies. However, without analysing the semantics of gwelei in a strict formal way, we can say that there are also numerous examples in which it is appropriately translated with the English past simple form saw.25 The preterite of gweled is also used in native prose, in subordinate clauses and more rarely in main clauses; see (10) Porth y gaer a welas yn agoret; ny bu argel arnei. (Manawydan uab Llyr; Hughes 2007: 6.180) She saw the gate of the fort open; there was no concealment on it.26 Therefore, we may suggest the existence of a fine semantic distinction between gwelei and gwelas in the native prose, with the preference for gwelei. The data of the Rhyddiaith Gymraeg 1300–1425 corpus (Luft et al. 2013) show the following distribution: total LlA LlA% Peniarth 4 Peniarth 4% (g)welas + (g) weles 1476 20 57,1 40 23,4 (g)welei 997 15 42,9 131 76,6 We intend to investigate this phenomenon in detail in the course of our project, but we can make the preliminary deduction from these data that gwelei is significantly more frequent in the part of the White Book of Rhydderch containing the Four Branches, Culhwch ac Olwen and the so-called three romances than it is in LlA. The language of LlA shows another distribution, preterite forms being more frequent. The hypothesis is that translators were influenced by Latin which did not have such a fine distinction between the meanings as gwelei vs. gwelas. The use of the two forms in both independent 25 Cf. Kyrchu i llys a oruc ynteu; ac yn y llys ef a welei hundyeu ac yneuadau (Pwyll Pendeuic Dyuet; Thomson 1957: l.75) “He made his way to the court. He saw sleeping quarters there and halls” (Davies 2007: 5). 26 “She found the gate of the fort open — it was ajar” (Davies 2007: 40). 90 ELENA PARINA translations suggests that both were used actively, but the variation between the translations as in ex. 8, as well as the variation between text witnesses of one version (in ex. 7), seems to indicate that usage was more fluid. 4. Differences In this section I will look at differences between the versions, again first in the domain of lexical choice and then in grammar. 4.1. Lexical choice The small scale of this study, as already mentioned, does not allow for any large-scale generalisations. In the following three examples, two different words are used in Welsh translations as equivalents of a Latin word — their occurrence in the same context shows that they are synonyms, and these Latin contexts can provide information on their exact meaning. Latin ovīle is translated by phald in LlA and keil in Llst4. (11) Et erant anime in illo loco una super aliam quasi oves in ovili. (L7; Jiroušková 2006: 778) A hynny pob vn ar warthaf y gilyd. megys deueit y mywn phalt. (l la : 131r) And those [souls] one above another as sheep in sheepfold. ar eneideu pob vn ar benn y gilyd megys deueit y my6n keil. (l lst4: 36r) Welsh ffald is a loanword from o .e . fald (as in Modern English sheepfold) and is used frequently in Middle Welsh prose inter alia in legal contexts which allow us to understand its meaning easily. In the following example the payments for burning a certain type of building are discussed: (12) Buarth, a thalgell, a chreu moch, a ffalt deueit, dec ar hugeint a tal pob vn. (Llyfr Blegywryd; Williams & Powell 1961: 96) A cattle yard, a lean-to, a pig-sty, and a sheep fold are each thirty pence in value. (Richards 1954: 93) 91 MULTIPLE VERSIONS OF BREUDDWYD PAWL The word keil is used not only in Llst4 but also in two text witnesses of the LlA version: megis deueit y mewn keil (Pen3, Pen14). I was only able to find one more attestation of this word in the prose corpus: (13) Kyrchwn heb ohir yr haner gwyr rackw a dilewn wynt val deveit y mewn keil wynt27 a rannwn wynt ar hyt y kyvoeth. (Brut y Brenhinedd; nlw Peniarth 21: 5r) Let us attack without delay those half-men and destroy them as sheep in a sheepfold and divide them throughout the realm. Compared to the standard edition of Historia Regum, the following translation seems to be very free: Armate uos, uiri, armate et per densatas turmas incedite. Nulla mora erit quin semimares istos uelut oues capiemus atque captos per regna nostra mancapabimus. (hrb I.20) To arms, men, to arms, close your ranks and advance. We shall soon capture these effeminates as if they were sheep, and make them slaves in our country. (hrb : 24) However, if we look at the variants of hrb , we will find in the so-called first variant version a construction closer to what we find in Welsh: Armate vos, o viri, armate celeriter et per turmas ordinatas ad pugnam incedite. Nulla mora erit quin semimares istos velut oves intra caulas capiemus et captos captivos per regna nostra mancipabimus. (Hammer 1951: 37, see also hrb : 14) It could be significant that here we find a Latin word caulas phonetically close to the Welsh one, which is also noted as one of the etymological hypotheses for the word in GPC (s.v. cail): “i’w gysylltu o bosibl â’r ail elf. yn bugail a bod cail < Clt. *koli̯a o’r gwr. *qel- ‘gyrru’, ond cf. Llad. caulæ ‘corlan’”—‘to be possibly related to the second element in bugail, cail < Celtic *koli̯a from the root *qel- ‘to drive’, but cf. Lat. caulæ ‘fold’’. Apart from these four instances in the Rhyddiaith Gymraeg 1300–1425 corpus, the word is attested in the following passage: 27 The second wynt is probably superfluous. 92 ELENA PARINA (14) Ena keuyt e giudaut en erbyn y gilid yg capadocia ac en oes h6nn6 e keytheir pampilia cany doeth trwy e drws yr keil. (Proffwydoliaeth Sibli Ddoeth; Isaac et al. 2013; Peniarth 14: 51) Then its citizens will rise against each other in Cappadocia, and at that time Pamphilia will be captured since they did not come through the gate of the fold.28 There are also some examples in poetry; see (15) Pe bawn, myn y Pab annwyl, / Yn y llwyn, anneall hwyl, / Cyd y bu’r gwr, cyflwr cail, / Ebwch gwae, wrth y baich gwiail, / Gwyn ac addwyn ei hwyneb, / –Gwae fi!–ni welwn i neb. (Dafydd ap Gwilym, 146: l. 29; Disgwyl yn Ofer) If I were in the woods, / in the dear Pope’s name, senseless condition, / as long as the man with the load of sticks, / state of imprisonment, exclamation of grief, / pure and tender is her face, / —woe is me!—I wouldn’t see anyone. (http://www.dafyddapgwilym.net/) But in this example cyflwr cail is rephrased with the Modern Welsh mewn cyflwr o gaethiwed ‘state of distress/imprisonment’ and cannot give us a clear indication of the lexeme’s meaning. Therefore, the vp examples show us the importance of translational texts as sources of information on the semantics of less frequent words; this information is provided both by comparison to the translations’ originals and, in the case of multiple translations, by equivalents from other versions. I will briefly address another point at which our translators differ in their lexical choice. Here is the list of seven plagues around a fiery furnace: (16) Ac yng kylch y ffwrnn yd oedynt seith pla. kynntaf oed eiry. Ar eil oed tan. Ar tryded oed ia. Pedwared oed waet. Pymhet oed seirff. Chwechet oed mellt. Seithuet oed derewant. (l la : 129r) 28 Nely van Seventer kindly supplied me with the part of her research concerning this passage. The corresponding Latin text is Tunc surget gens adversus gentem in Cappadociam et Pamphiliam captivabunt in ipsius tempore, eo, quod non introerit per ostium in ovile (Sackur 1898: 182–3) ʻThen people will rise against people in Cappadocia, and they will capture Pamphilia in that age, for this reason: because they did not enter through the door into the foldʼ (translation by N. van Seventer). In another translation found in the White Book of Rhydderch ovile is translated by dauatty, literally ʻsheep-houseʼ: Ac yna y kyuyt kenedyl yn y teyrnas a elwir Capadocia, atheyrnas Pampilia a geithiwant yn amser hwnnw am nat yntredant drwy drus y dauatty (NLW Peniarth 5, 13r). 93 MULTIPLE VERSIONS OF BREUDDWYD PAWL Around this furnace were seven plagues: the first was snow, the second was fire, the third was ice, the fourth was blood, the fifth was snakes, the sixth was lightning, the seventh was stink. Tr6y y seith fflam hynny y dyellir y seith poen yssyd yn uffern; nyt amgen eiry, ia, tan, g6res, nadred, tywyll6c[h], drewyant. (l lst4: 35v) Et VI plage sunt in circuitu fornacis: Prima est nix, secunda glacies, tercia sanguis, quarta ignis, quinta serpentes, sexta fetor. (C6; Jiroušková 2006: 671) In LlA the translator uses seirff (sg. sarff, a borrowing from spoken Latin *sarpans < Lat. serpens; gpc s.v. sarff) and in Llst4 nadred (a word of Celtic origin) for serpentes.29 A search through the word lists of the Rhyddiaith Gymraeg 1300–1425 corpus shows that both words for snakes are quite frequent words; different forms of neidr occur 133 times, and different forms of sarff are found 219 times. I do not think, however, that the difference in usage in the two versions is indicative of anything more than the closeness of the meaning of these words. Another argument for this is the use of a doublet in Llst4 where LlA has only nadred: (17) Odyna y gwelas pawl gwyr a graged yn noethon a phryfet. a nadred yn y bwyta. (l la 130v) Then Paul saw men and women naked, and worms and adders devouring them. (Williams 1892: 636) ac ef a welei yn y pyde6 g6yr a g6raged meibyon a merchet yn noethon, a phryfet a seirff a nadred yn eu knoi. (l lst4: 36r) Et vidit Paulus in alio loco viros et mulieres nudos et vermes et serpentes comedentes eos. (L7; Jiroušková 2006: 778) 29 Interestingly something similar happens with the translations of Latin plage in the same sentence: LlA has pla, while Llst4 uses poen ʻpainʼ. 94 ELENA PARINA 4.2. Differences: syntax I now turn to some instances of differences in syntax. 4.2.1. Translation of Latin perfect participles An example of a Latin perfect participle scriptum translated by ysgriuennedic in both versions has already been shown above (ex. 1). There is another example in this text in which the translations differ. (18) Ac yna dangos pydew a oruc idaw. a seith ynseil arnaw. (l la : 130v) And then he showed him a pit and seven seals upon it. [lost folio] yn pyde6 inseiledic o seith inseil. (l lst4: 36r) Tunc ostendit ei puteum signatum septem sigillis. (O5; Jiroušková 2006: 766) Although the Llst4 text is incomplete here, the nominal phrase is preserved. While the LlA version uses a prepositional construction and omits the translation of signatum ʻsealedʼ, the Llst4 version uses a construction very close to that of Latin, with inseiledic ʻsealedʼ following the noun it modifies and turning the Latin instrumental ablative of septem sigillis into a prepositional phrase governed by the preposition o, here in the sense “by, by means of” (gmw 204). 4.2.2. Relative clauses The number of different ways to construct a relative sentence in Middle Welsh accounts for the differences in the versions in this respect. In ex. 1, one such instance was shown; I will now discuss two other differences. (19) y rei hynny heb yr angel ny credassant ygrist y gwr a diodefawt anghev yr pobyl y byt. (l la : 130v) Y rei heb·y Mihangel. ny chredassant y vab du6 yr h6nn a diodefa6d yr y byt. (l lst4: 36r) Hii, qui non crediderunt in filium dei, qui passus pro salute mundi. (M4; Jiroušková 2006: 773) 95 MULTIPLE VERSIONS OF BREUDDWYD PAWL In the LlA version we find a construction using y gwr ʻthe manʼ as an antecedent (see gmw 69). This is a construction similar to that used in poetry to refer to God in Juvencus 930 or in Kyntaw geir a dywedaw (y gur am creuys e am nerth ʻthe one who created me is my strengthʼ).31 The Llst4 version uses the construction with the demonstrative pronoun yr hwnn, mentioned already in section 1. Here again, we can place Llst4 closer to the translated end of the ‘native vs. translated’ scale than the LlA version. The importance of weighting all the features against each other with regard to their significance and the need for a more systematic study emerges from the next example. Here is the first sentence of our texts: (20) Dyw sul dyd detholedic yw, yn yr hwnn y caffant yn y dyd hwnnw yr eneideu a uont yn y poenev, orffwys yn diboen trwy lewenyd. (l la : 129r) Sunday is a chosen day, whereon the souls that are in pains receive rest without pain through joy. (Williams 1892: 635) Pwy bynnac a vynno g6ybot p6y gyntaf a lauurya6d y beri gorffowys du6 sul y’r eneityeu a vei yn uffern. (l lst4: 35v) Whoever would like to know who was the first to bring about peace on Sunday to the souls that were in hell. The exact wording of the Latin source for both versions is not easily established.32 What is important here is that both versions end with a fragment which can be translated as ʻsouls that were in (pains/hell)ʼ. According to normative expectations, the form of the verb bod should not agree in number with the plural antecedent; however, as shown in Plein (2016: 197), even in the Mabinogion sub-corpus, 23% of examples of bod in relative clauses following a plural nominal antecedent show agreement, and shorter non-Mabinogion texts show such behaviour in 54% of cases. Here again, we find a feature which is related to the ‘native vs. translated’ scale, but the presence of concord is not a strong indication of the translational style, 30 See Falileyev 2016: 25. 31 See Falileyev 2012: 54, 230. 32 For the variants; see Jiroušková 2006: 654–659. 96 ELENA PARINA since almost a quarter of Mabinogion examples also show it. But it is nevertheless interesting that we find a variant in the LlA version that is closer to the translation style than the one used in Llst4. 5. Conclusions Studying various translations of one text — even given the extreme fluidity both of the source and of the target texts — is rewarding and delivers important results. First, as noticed already by Rhŷs in 1894, these texts are sources of valuable semantic information, since we can relate Middle Welsh lexemes to their equivalents found in other versions as well as to those in the Latin original. Secondly, as we have seen, several features have been identified in Welsh philology as marks of translation and therefore often as indicators of ‘unnatural’ translation style. By comparing our two Visio Pauli versions, we can see that some of these features are not found in our translated texts (adjective congruence with plural nouns — possibly due to the sample size), some appear occasionally (adjectives preceding nouns) but are equally rare in both versions, while some are present only in one version (relative clause with demonstrative). Our sample size does not allow us to make statistically valid statements on the frequency of these features. However, I hope to have demonstrated that the systematical study of such features can help us to draw a map of the language of Middle Welsh prose, seeing it not in a binary opposition of native vs. translated texts but as a continuum using these features as some kind of coordinates. While a quantitative comparison of the aforementioned syntactic features within the language of native texts has not been conducted for this contribution, it seems that the LlA version would be somewhat closer to the standard native texts on this continuum than the Llst4 version. Thus, a comparison of multiple translations helps to understand better the syntax and lexis of Middle Welsh as well as the personal choices of individual translators. Philipps-Universität Marburg/ Institute of Linguistics, RAS, Moscow 97 MULTIPLE VERSIONS OF BREUDDWYD PAWL Appendix I. List of Latin manuscripts mentioned in text C6 Cambridge, St John’s College, MS. D.20 (95) (s. XV) L7 London, British Library, MS. Royal 8.E.XVII (s. XIII ex.–XIV in.) L8 London, British Library, MS. Royal 8.F.VI (s. XV) M4 München, Bayerische Staatsbibliothek, clm 12728 (s. XV) O5 Oxford, Merton College, MS. 13 (s. XIV ex. or XV in.) P7 Paris, Bibliothèque Nationale, Ms. lat. 3529A (s. XIV in.) Abbreviations GMW Evans, D.S., 1964, A Grammar of Middle Welsh, Mediaeval and Modern Welsh Series. Supplementary volume. Dublin. GPC GPC Online. University of Wales Centre for Advanced Welsh & Celtic Studies, 2014. http://geiriadur.ac.uk/gpc/gpc.html [accessed 1.05.2017]. HRB Reeve, M. D., ed., 2007, Geoffrey of Monmouth. The History of the Kings of Britain, translated by Neil Wright, Arthurian Studies 69, Woodbridge. References Dafydd ap Gwilym — http://www.dafyddapgwilym.net [accessed 1.05.2017]. Davies, S., 2007, The Mabinogion, Oxford/New York. Dwyer, M.E., 2004, Contributions to the Textual History of the Medieval Latin Redactions of the Visio Pauli, Ph.D. thesis, University of Tasmania. Elliott, J.K., 1999, The Apocryphal New Testament: a Collection of Apocryphal Christian Literature in an English Translation, Oxford. Falileyev, A., 2016, Llawlyfr Hen Gymraeg, Aberystwyth. Falileyev, A., 2012, Из ранней валлийской поэзии [From early Welsh poetry], Sait-Petersburg. Geeraerts, D., 1997, Diachronic Prototype Semantics: a Contribution to Historical Lexicology, Oxford/New York. 98 ELENA PARINA Hammer, J., 1951, Geoffrey of Monmouth. Historia regum Britanniae. A Variant Version Edited from Manuscripts, Cambridge, MA. Hilhorst, A., 2007, ‘The Apocalypse of Paul: Previous History and Afterlife’, in: Bremmer, J.N., Czachesz, I., eds., The Visio Pauli and the Gnostic Apocalypse of Paul, Studies on early Christian apocrypha, Leuven, [Dudley, MA], 1–22. Hughes, I., 2007, Manawydan uab Llyr: Trydedd gainc y Mabinogi; golygiad newydd ynghyd â nodiadau testunol a geirfa lawn, Cardiff (2nd ed.). Huws, D., 2000, Medieval Welsh Manuscripts, Cardiff. Isaac, G.R., Kapphahn, K., Nurmio, S., Rodway, S., & Sims-Williams, P., 2013, Rhyddiaith Gymraeg o Lawysgrifau’r 13eg Ganrif/13th Century Welsh Prose Manuscripts, Aberystwyth, http://cadair.aber.ac.uk/dspace/handle/2160/11163 [accessed 1.05.2017]. Jiroušková, L., 2006, Die Visio Pauli. Wege und Wandlungen einer orientalischen Apokryphe im lateinischen Mittelalter unter Einschluss der alttschechischen und deutschsprachigen Textzeugen, Mittellateinische Studien und Texte 34, Leiden. Jones, Th., 1941, Brut y tywysogyon: Peniarth Ms. 20, Cardiff. Latham, R.E., & Howlett, D.R., 1997, Dictionary of Medieval Latin from British Sources: A–L, London. Luft, D., Thomas, P.W., & Smith, D.M., 2013, Rhyddiaith Gymraeg 1300– 1425. http://www.rhyddiaithganoloesol.caerdydd.ac.uk [accessed 1.05.2017]. Luft, D., 2016, ‘Tracking ôl cyfieithu: Medieval Welsh Translation in Criticism and Scholarship’, Translation Studies 9.2, 168–82. Morgan, T.J., 1970, ‘Cymraeg Naturiol Cywir’ [Correct natural Welsh], Ysgrifau Beirniadol 13, 254–75. Morris-Jones, J., & Rhŷs, J., 1894, The Elucidarium and Other Tracts: in Welsh from Llyvyr agkyr Llandewivrevi A.D. 1346 (Jesus College, MS. 119), Oxford. Nurmio, S., 2015, Studies in Grammatical Number in Old and Middle Welsh, unpublished PhD thesis, University of Cambridge. Parina, E., 2017, ‘A Welsh Version of Visio Pauli: its Latin Source and the Translator’s Contribution’, Apocrypha 28, 155–86. Parina, E., 2016, ‘Об одном средневаллийском переводе Visio Sancti Pauli’ [On one Middle Welsh translation of the Visio Sancti Pauli], Indo-European Linguistics and Classical Philology 20, 842–50. Parry, J.J., 1937, Brut y Brenhinedd. Cotton Cleopatra version, Cambridge, MA. Plein, K., 2016, Verbalkongruenz im Mittelkymrischen, unpublished PhD thesis, University of Marburg. Poppe, E., fc., Syntactic Variation in Middle Welsh: New Perspectives, in: Proceedings of the First European Symposium in Celtic Studies. Richards, M., 1954, The Laws of Hywell Dda: the Book of Blegywryd, Liverpool. Roberts, B.F., ed., 1975, Cyfranc Lludd a Llefelys, Dublin. 99 MULTIPLE VERSIONS OF BREUDDWYD PAWL Russell, P., 2017, ‘‘Go and Look in the Latin Books’: Latin and the Vernacular in Medieval Wales’, Proceedings of the British Academy 206, 213–46. Russell, P., 1990, Celtic Word-formation: the Velar Suffixes, Dublin. Sackur, E., 1898, Sibyllinische Texte und Forschungen. Pseudometodius, Adso und die Tiburtinische Sibille, Halle. Silverstein, Th., 1935, Visio Sancti Pauli. The History of the Apocalypse in Latin together with Nine Texts, London. Sims-Williams, P., 2016a, ‘The Welsh Versions of Geoffrey of Monmouth’s ‘History of the Kings of Britain’’, in: Harlos, A., & Harlos, N., eds., Adapting Texts and Styles in a Celtic Context. Interdisciplinary Perspectives on Processes of Literary Transfer. Studies in Honour of Erich Poppe, Münster, 53–74. Sims-Williams, P., 2016 b, ʻJohn Morris-Jones and his Welsh Grammarʼ, Transactions of the Honourable Society of Cymmrodorion 22, 134–53. Thomson, R.L., 1957, Pwyll Pendeuic Dyuet, Mediaeval and Modern Welsh Series 1, Dublin. van Seventer, N., 2018, ‘The Translation of the Sibylla Tiburtina into Middle Welsh’, in: Jones, A. Ll., & Fomin, M., eds., ‘Y Geissaw Chwedleu’. Proceedings of the 7 th International Colloquium of Societas Celto-Slavica, Studia CeltoSlavica 8, Bangor, 109–118. van Seventer, N., 2017, ‘The Middle Welsh Sibylla Tiburtina: One Text, Two Translations’, paper delivered at the XI Cardiff Conference on the Theory and Practice of Translation in the Middle Ages (Vienna). Williams, J.E. Caerwyn, 1962, ʻWelsh Versions of Visio Sancti Pauliʼ, Études Celtiques 10, 109–26. Williams, R., 1892, Selections from the Hengwrt Mss. Preserved in the Peniarth Library: Vol. II, London. Williams, St.J., Powell, J. E., 1961, Cyfreithiau Hywel Dda yn ôl Llyfr Blegywryd, Cardiff. 100 An Acoustic Study of Welsh and Slavonic Rhotics Sylwester Jaworski & Sabine Asmus 1. Introduction The word rhotics, or, informally, r-sounds, is an umbrella term that refers to a natural class of speech segments whose members share a number of phonological properties despite varying significantly with respect to place and manner of articulation. Common r-sounds found in the world’s languages include dental/alveolar and uvular trills [r ʀ], dental/alveolar taps [ɾ], voiced and voiceless uvular fricatives [ʁ χ], alveolar approximants [ɹ] and their retroflex variants [ɻ]. So diverse is the class of rhotics that Ladefoged & Maddieson (1996: 215) conclude that the only feature that r-sounds have in common is their representation of numerous variations of the base grapheme <r> in those languages that use the Latin alphabet. For instance, <ř> represents the fricated trill of Czech, <hr> stands for the fortis trill [r̥] in Irish orthography,1 while the digraph <rz> is a Polish spelling convention to represent the sound [ʒ] that has evolved from the palatalised trill /rʲ/. The most intriguing facts about rhotics is that markedly different sounds share a number of significant phonological properties. For instance, the manner of articulation of an r-sound does not seem to affect its phonotactic properties. Rhotics tend to have several distinct articulatory variants, yet they do not occur in complementary distribution. Sounds belonging to this natural class, irrespective of their articulation, pose a challenge to the speaker as they are usually among the last speech sounds mastered by children in the process of language acquisition (Vihmann 1996). The inherent difficulty of rhotics also makes them particularly susceptible to phonetic change, as evidenced by historical linguists (e.g. Carlton 1990; Rospond 1973). In addition, Labov’s (1966) classic study demonstrated that the frequency with which the rhotic sound occurs in the speech of New Yorkers can be related to the social group the speakers come from. In other words, the sound can be thought of as a social marker, as is potentially true for any sound. This paper examines samples of Welsh speech acoustically with a view to determining the extent of allophonic variation exhibited by the 1 For the use of this term, see Asmus & Grawunder (2017, fc.). 101 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS spirantised trill /r̥ʰ/.2 Being a sound of extraordinary complexity, /r̥ʰ/ is probably more susceptible to phonetic change than plain trills. It further follows that, in connected speech, this segment is likely to be replaced with various sounds that represent a lesser degree of articulatory complexity. Since the informants come from different parts of Wales, an attempt is made to establish whether the substitutions made by speakers of North Welsh differ from those produced by southerners. Finally, the Welsh spirantised /r̥ʰ/ is compared to the Czech fricated /r̝/ and the Russian palatalised /rʲ/ with respect to their susceptibility to phonetic reduction. The paper is structured as follows. Section 1 provides general information regarding the natural class of rhotics. Section 2 presents an outline of the phonology of rhotics from a cross-linguistic perspective. Section 3 is concerned with the acoustic properties of trills and explains why they are so prone to reduction. Section 4 introduces the data collection methods of the current study, which include a phonetic experiment involving 23 native speakers of Welsh and the analysis of recordings obtained from audiobooks for the Slavic languages. The remaining part of this section is devoted to the presentation of the findings. This is followed by Section 5 containing concluding remarks. 2. The natural class of rhotics As indicated above, due to the remarkable articulatory diversity of rhotics, finding a single acoustic feature that they have in common poses a challenge. A lowered value of the third formant (F3) appears to be the only acoustic property shared by several types of rhotics, but uvular r-sounds have a relatively high third formant and so do dental ones (Lindau 1985: 165). Having examined various types of rhotics, Lindau (1985: 166) states that “there is no physical property that constitutes the essence of all rhotics”. Although a satisfactory definition of rhotics has not been provided yet (Maddieson 1984; Lindau 1985; Ladefoged & Maddieson 1996; Wiese 2001, 2011), there is a body of statistical data that allows linguists to make generalisations as to which sound can be regarded as the prototypical rhotic. The list of general statements on rhotics shown in (1) is the result of an analysis of r-sounds found in a representative sample of 316 languages performed by Maddieson (1984: 82). 2 Considering that Welsh is an aspiration language and following others, e.g. Broderick (2016), Asmus & Grawunder (2017) use the IPA symbol [ṛ] for this spirantised rhotic in order to differentiate it from the aspirated allophone /r(h)/ of /r/, which is strong in the Bala Lake District. The transcription is debated and Thomas (2000) uses the sequence [hr] to represent the spirantised trill. Throughout this work, however, we use /r̥ʰ/ as suggested by Hannahs (2012: 43). 102 SYLWESTER JAWORSKI & SABINE ASMUS (1 ) a. An r-sound is likely to be voiced. 308/316 97.5% b. An r-sound is likely to be dental or alveolar. 273/316 86.4% c. An r-sound is likely to be interrupted. 244/282 86.5% d. A retroflex r-sound is likely to be a continuant. 20/38 52.6% e. An approximant r-sound is likely to be retroflex. 15/28 53.6% f. A fricative r-sound is likely to be retroflex. 5/10 50.0% The statements in (1) point to the conclusion that the prototypical rhotic is voiced, dental/alveolar, interrupted3 and continuant. The apical trill [r] is the only sound that meets these criteria (Lindau 1985). However, the articulation of [r] requires a great deal of precision (Solé 2002; Ladefoged & Maddieson 1996; Żygis 2005; Wiese 2011) Therefore, it is liable to phonetic change. 2.1. An outline of the phonology of rhotics It is paradoxical that sounds as diverse as rhotics exhibit striking similarities in the domain of phonology. Wiese (2003, 2011) provides a list of generalisations suggesting that rhotics do constitute a natural class. These include phonotactic properties, syllabicity, synchronic and diachronic alternations and phonological consequences of such alternations (see also Hall 1997; Walsh Dickey 1997). As far as the first generalisation is concerned, rhotics tend to occupy privileged, vowel-adjacent positions within the syllable. It follows that they typically occur in syllable-initial, intervocalic and syllable-final position. In those languages that allow for consonant clusters, rhotics usually follow the CrVrC sound pattern, where the onset and coda C slots can be occupied by more than one consonant. Another phonological feature characteristic of rhotics is that, in many languages, they have a syllabic variant. For instance, non-vowel-adjacent rhotics of Czech, Croatian and American English have syllabic variants, as in Cz krtek ‘mole’, Cr krk ‘neck’, AE runner, pronounced [ˈrʌnɹ ]. On the other hand, in some other languages, e.g. Polish, Russian and Ukrainian, rhotics found in such sound combinations are classified as obstruentised 3 In Maddieson’s (1984) terminology, both trills and taps/flaps are referred to as “interrupted” as they consist of, at least, one complete closure. Obviously, the other manners of articulation are labelled “uninterrupted”. 103 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS rhotics (Gussmann 2007; Jaworski 2014). R-sounds of this type are found, for instance, in P krtań ‘larynx’ [krtaɲ], R рта [rta] ‘mouth’ (gen. sing.) and U ртуті [rtutʲi] ‘mercury’. As aforementioned, synchronically and diachronically, rhotics exhibit a very strong tendency towards phonetic change. Lindau (1985) argues that the usual pattern of change involves a gradual decrease in the degree of constriction, whose final outcome is the deletion of a rhotic segment as illustrated graphically in Figure 1. Although this graph covers many of the attested changes, it does not make any reference to the /r/ > /ʀ/ change that has taken place in a number of European languages, e.g. French, Danish and Swedish (Demolin 2001). ɾ˳ ɾ r˳ r ɽ ʐ ʋ ɻ ɹ ʀ˳ ʀ χ ʁ ʁ˕ ɚ ɑ˞ Figure 1. Associations between various types of rhotics (after Lindau 1985). Relatively common phonological changes, most of which are represented in the graph, include the following: (i) neutralisation of the /r/-/ɾ/ distinction observed in the speech of Spanish-speaking Filipinos (Lipski 1987a: 32) or in the Spanish of Equatorial Guinea (Lipski 1987b: 11); (ii) neutralisation of the /r/-/rʲ/ contrast that affected the sound systems of several Slavonic languages (Żygis 2005; Kavitskaya et al. 2009); (iii) uvularisation, or the /r/ > /ʀ/ change, that occurred in a number of European languages, e.g. French, Swedish, Danish and German (Demolin 2001); (iv) lateralisation, i.e. replacing rhotics with laterals, which occurs regularly in many varieties of Spanish, e.g. in harto ‘full’ pronounced [alto] or mar ‘sea’ pronounced [mal] (Hualde 2005: 188); (v) metathesis, i.e. reordering of segments, which affected 104 SYLWESTER JAWORSKI & SABINE ASMUS the sound systems of all Slavonic languages, e.g. the Proto-Slavonic word *korva ‘cow’ was rendered kráva in Slovene and Czech, krowa in Polish and Lower Sorbian, krava in Slovak and kruva in Upper Sorbian (Carlton 1990). The alternations observed both synchronically and diachronically do not seem to affect the phonotactic properties of rhotics (Wiese 2003: 12). In other words, in the theoretical CrVrC syllable, the rhotic slots can be occupied by rhotic segments ranging from trills to approximants. The statement remains true even if an apical rhotic is replaced with a velar one. This fact can be easily confirmed by listening to Poles who have a speech impediment and cannot pronounce an apical trill. In their speech, /r/ is usually realised phonetically either as the velar fricative [ʁ] or the post-alveolar approximant [ɹ]. 3. Trills A trill is produced when the active articulator is set in vibration by the airstream in the oral cavity. Trills are the only speech sounds for which the articulation of the movements of the active articulator are not controlled by any muscular action (Ladefoged & Maddieson 1996: 217; Solé 2002). Instead, the vibration results from the aerodynamic conditions created by an airstream passing through the aperture between the active and passive organ. When the organs separate, a certain volume of air flows through the aperture and, consequently, the pressure behind the place of articulation drops dramatically allowing the active articulator to spring back to its former position producing another closure.4 In many languages, e.g. Spanish, Italian, a typical trill consists of two or three closures, but the number of cycles depends on speaking style, on the position a trill occupies within the syllable and, also, on the place of articulation. According to Solé (2002: 669), the frequency of vibration of a voiced apical trill is in the range of 26–29 Hz, which is consistent with the data provided by other authors, e.g. Ladefoged & Maddieson (1996) and Recasens & Espinosa (2007). Although rhotics are said to be susceptible to change, trills appear to be rather common sounds among them. In the ucla Phonological Segment Inventory Database (upsid ), which includes information regarding the phonemic inventories of 451 languages, all of the world’s languages that have a rhotic in their consonant inventory seem to have a trill, and an overwhelming majority of the trills (99.1%) are produced in the alveolar/dental place of articulation (see also Maddieson 1984). 4 This phenomenon is referred to as the Bernoulli effect (Żygis 2005: 389). 105 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS 3.1. The plain apical trill /r/ As trilling crucially depends on the size and shape of the aperture as well as on the volume of airflow, minimal changes to one of the factors can result in a non-trilled realisation, i.e. either a fricative [ɹ̝] or an approximant [ɹ]. The results of an experiment described in Solé (2002: 672) strongly suggest that a pressure drop across the lingual constriction within the range of 2.5– 3.5 cm H2O is sufficient to impair trilling (Figure 2). This finding may account for the sound’s susceptibility to phonetic change as evidenced by synchronic and diachronic alterations of [r] to [ɹ] or [ɹ̝] that have been attested in many languages that have a trill in the inventory (Blecua 2001; Colantoni 2005; Wiese 2003; Verstraeten & van de Velde 2001; Solé 2002; Recasens & Espinosa 2007; Żygis 2005; Jaworski & Gillian 2011). Ps > Po > Pa 7.4 5.4 – 4.0 0 ΔP 2–3 4 Figure 2. Values of subglottal pressure (Ps), oropharyngeal pressure (Po) and atmospheric pressure (Pa), expressed in cm H2O, across the lingual constriction required for trilling and voicing. The ΔP parameter indicates the minimal pressure difference at the glottal and lingual constrictions (after Solé 2002: 675). As noted above, due to their complexity, trills may pose a challenge to speakers and, therefore, may exhibit a considerable amount of variation. In the Slavonic languages, the articulatory variants of /r/ can be assigned to five phonetic categories that correspond to the following manners of articulation: trill, tap, fricative, approximant and affricate (Jaworski 2018). The number of allophones could be further extended by including the voiceless variants of trills, fricatives and affricates that occur regularly in certain contexts, e.g. word-finally. A voiceless trill is sometimes heard in Polish words such as teatr ‘theatre’ pronounced [tɛatr̥]. However, in the same word, the final rhotic can also be realised as a fricative [tɛatɹ̝] or, less frequently, an affricate [tɛatɾᶴ] (see Jaworski 2018). As for trilled allophones, they could be further divided into three subcategories, namely fully voiced, fricativised and approximantised trills. The main criterion for distinguishing the three types of trills is the degree of constriction produced by the articulating part of the tongue, which can be referred to as (i) complete closure, (ii) close approximation and (iii) open approximation. 106 SYLWESTER JAWORSKI & SABINE ASMUS Fully articulated trills involve at least two constrictions of type (i). Fully voiced trills are composed of an alternating sequence of intermittent closures and vocalic elements, which makes them similar to a series of taps. A typical voiced trill, produced in the word раскаялся [rasˈkajəɫsə] ‘he repented’, is presented in Figure 3. It consists of three occlusions, the last of which has the form of an approximant. The three constrictions last 19, 21 and 22 ms respectively, while the vocalic intervals between them are slightly longer (23 and 24 ms). The vocalic intervals have a distinct formant structure that indicates that the vocalic elements can be classified as mid-high central vowels. The values of F1 measured at midpoint are almost identical (428 Hz and 431 Hz), and so are those of the second formant (1509 Hz and 1516 Hz). Figure 3. A trilled variant of Russian /r/ pronounced in the word раскаялся ‘he repented’. Fricativised trills differ from fully voiced trills in that a constriction of type (ii) is created between the alveolar ridge and the approximating apex. This rising gesture of the apex is termed close approximation by virtue of impeding the airflow to the extent that aperiodic noise is produced by the jet of air passing through the constriction. The auditory impression of trilling is thus made by an alternating sequence of periods of frication and vocalic elements. An initial fricativised trill, produced in the Ukrainian word рyки [r̝uˈci] ‘hand’ (gen. sg.) in one of the audiobook recordings, is depicted in Figure 4. The characteristic feature of such tokens is that they consist of two usually incomplete constrictions separated by a period of aperiodic friction. The two constriction phases of 21 and 19 ms are almost complete, yet the intervening segment is by no means a vocalic element as it does not have 107 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS any formant structure. The corresponding section of the waveform is composed of both periodic vibration and aperiodic friction; thus the element separating the constrictions should be classified as a voiced fricative element. Figure 4. A fricativised trill produced in the Ukrainian word рyки ‘hand’ (gen. sg.). It is worth stressing that fricativised allophones represent a moderate degree of phonetic reduction. Further reduction of the apical gesture gives rise to approximantised allophones. An example of an initial approximantised trill, produced in the Belorussian word pыбаловы ‘ fishermen’, is depicted in Figure 5. The sound consists of two incomplete constriction phases separated by a vocalic element. The intensity of the first constriction is noticeably lower than that of the following vocalic element. As for the other constriction, its phonetic properties hardly differ from those of the flanking vowels. However, the waveform shows that the amplitude of vibration drops when the approximant constrictions are produced, and it increases when the intervening vocalic element and the following vowel are articulated. It is these minute rising gestures of the apex that produce the auditory impression of trilling. 108 SYLWESTER JAWORSKI & SABINE ASMUS Figure 5. An approximantised trill produced in the Belorussian word pыбаловы ‘fishermen’. 3.2. Complex trills This section is concerned with complex trills, i.e. those in which trilling occurs simultaneously with an additional gesture. Although such features are usually referred to as secondary in the phonetic literature,5 the adjective additional is consistently used throughout the paper as the so-called secondary feature is sometimes preserved in various phonological contexts, whereas the primary one turns out to be redundant. 3.2.1. Spirantised trill /r̥ʰ/ The spirantised trill /r̥ʰ/, which is something of a rarity cross-linguistically, is found, for instance, in Welsh (Ball 2015).6 As suggested by the ipa symbol representing the sound, it can be thought of as an apical trill pronounced simultaneously with audible glottal friction. This sound /r̥ʰ/ is represented by the digraph <rh> in Welsh orthography. The phonological status of the 5 Alternatively, such sounds may be referred to as marked trills Żygis (2005) argues that, in comparison with unmarked counterparts, marked segments are (i) less frequent in the world’s languages (ii) more complex in terms of articulation, (iii) perceptually less salient, (iv) acquired later in the acquisition process, (v) phonetically unstable, and (vi) more narrowly distributed. These generalisations do not necessarily hold for all languages. For instance, in Welsh, whose grammar is based on a consonant mutation system that can be defined as “systematic morpheme-initial consonant alternations with a phonetically distinct consonant, not obviously caused by their phonetic and/or phonological context, but reflecting basic phonetic processes in the language, i.e. lenition, spirantisation and nasalisation” (Asmus and Grawunder 2017, fc.). Consequently, it is argued here that, in Welsh and Irish, a sound resulting from mutation cannot be regarded as marked. 6 The sound must be very infrequent indeed as the UCLA database does not list /r̥ʰ/. 109 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS segment arouses a certain degree of controversy. Jones (1984) maintains that both the plain apical trill /r/ and the spirantised apical /r̥ʰ/ belong to the phonemic inventory of Modern Welsh. Two distinctive rhotic phonemes are also distinguished in Ball’s (2015) analysis. A carefully pronounced token of a spirantised trill, pronounced in the Welsh word rhwymo [r̥ʰujˈmɔ] ‘bind’, is presented in Figure 6. On the spectrogram, the initial part of the sound in question is represented by a sequence of as many as seven closing gestures (the light vertical stripes) separated by periods of voicelessness. In this particular trill, the average duration of the incomplete closures is 8 ms, while the mean duration of the periods of friction is 16 ms. These temporal properties translate into the vibration rate of 43 Hz. This rate is much higher than that obtained for voiced trills, which is on the order of 25 Hz (Lindau 1985; Ladefoged & Maddieson 1996). Figure 6. A carefully pronounced token of the spirantised trill of Welsh produced in the word rhywmo ‘bind’. Trills of this type certainly constitute an articulatory difficulty as gestures produced with different organs must be executed with perfect timing. In addition to that, throughout the sound, the critical difference between subglottal pressure and oropharyngeal pressure must be maintained to allow the repetitive movements of the apex to be made (Solé 2002). This requirement, which calls for a great degree of articulatory precision, makes trills in general, and complex trill in particular, prone to phonetic change. 110 SYLWESTER JAWORSKI & SABINE ASMUS 3.2.2. The palatalised trill of Slavonic languages The palatalised trill of Slavonic languages poses a challenge for the speaker. Recasens & Pollarès (1999) explain that the articulatory difficulty involved in the pronunciation of [rʲ] stems from two antagonistic gestures that need to be made. When the plain trill is pronounced, the predorsum is lowered and retracted, whereas for [rʲ] it is raised and fronted (see also Ladefoged & Maddieson 1996; Solé 2002). These movements change the internal configuration of the speech organs reducing the likelihood of creating the aerodynamic conditions necessary for trilling. Consequently, the sound seems to be more prone to reduction than its plain counterpart. Figure 7. A trilled variant of the Ukrainian /rʲ/ produced in the word різко ‘sharply’. As regards the acoustic properties of [rʲ], they typically consist of two closures separated by a vocalic element. A representative example of a fully voiced trill, produced in the word piзко [ˈriskə] ‘sharply’, is presented in Figure 7. This particular token consists of two taps and a vocalic element.7 The duration of the first tap is 18 ms, whereas the other is 2 ms shorter. The separating vocalic element is 20 ms long. One full cycle of vibration lasts approximately 38 ms, which translates into a vibration rate of 26.3 Hz. The values of the formants of the vocalic element measured at mid point are 460 Hz and 2040 Hz for F1 and F2, respectively. Judging from these acoustic parameters, the vocoid should be labelled as a mid high front vowel. 7 As shown by the spectrogram, the vowel following the second constriction contains a substantial amount of friction that also results from the raised position of the dorsum. 111 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS However, this finding is hardly surprising given that the trill is pronounced with the dorsum held close to the palate. Predictably, trilled allophones of /rʲ/ constitute a minority variant in connected speech. Instead, speakers pronounce either fricatives or approximants. Interestingly, tapped variants of the phoneme are not as frequent as one might have expected (see Section 4 for details). 3.2.3. The fricated trill of Czech The articulation of the phoneme /r̝/ appears to be highly idiosynchratic as different researchers provide divergent descriptions of the segment. According to Dankovičová (1999: 71), the sound “starts as a trill and continues as a fricative”. The author claims that it usually involves a greater number of vibrations than in the case of /r/. By contrast, Šimáčková et al. (2012: 226) describe the fricated trill /r̝/ as “a period of friction interrupted at the beginning by a contact or contacts created by a retracted apico-alveolar gesture”. The friction component of /r̝/ appears to be more distinct than trilling as foreigners normally perceive the rhotic as the post-alveolar fricative [ʒ]. Howson et al. (2015: 125) explain that this auditory impression is due to the shape of the opening for the trill. More specifically, the acoustic effect is produced by the wide flat channel that is created when the apex is not making contact with the palate, which bears a close resemblance to the tongue configuration for [ʃ] and [ʒ]. Figure 8 shows a fricated trill produced by a native speaker of Czech in the word řekl [r̝ekl] ‘he said’. The spectrogram shows clearly that the informant managed to produce one complete closure (16 ms) that was followed by a period of friction in the mid and high frequencies that lasted 43 ms. This realisation is thus consistent with the account provided by Šimáčková et al. (2012). Judging from its acoustic properties, this token can hardly be labelled as a trill. By virtue of its acoustic properties, it should rather be called an affricated rhotic. Rhotics of the type presented in Figure 7 probably represent the most frequent allophone of /r̝/, but by no means should it be assumed that trilled variants no longer occur in the Czech language. In all likelihood, they are still produced in emphatic speech, declamatory styles and emotionally charged words. However, in word-initial position, the sound tends to be reduced to a voiced fricative that can be represented by the symbol [ɹʒ]. 112 SYLWESTER JAWORSKI & SABINE ASMUS Figure 8. Fricated trill in the Czech word řekl [r̝ekl] ‘he said’. 4. The study The objectives of the current study are fourfold: (i) to describe the acoustic properties of the spirantised rhotic of Welsh, (ii) to determine the range of allophonic variation of /r̥ʰ/ by specifying which sounds are substituted or /r̥ʰ/ in speech, (iii) to establish whether speakers coming from the South and the North of Wales produce the same allophones of /r̥ʰ/ and (iv) to ascertain whether the spirantised trill /r̥ʰ/ of Welsh, the palatalised /rʲ/ of Russian and Ukrainian, and the fricated /r̝/ of Czech exhibit a similar amount of variation.8 4.1. The participants and data collection methods In order to achieve the goals of the study, different methods of data collection were used. In the case of Welsh, 23 native speakers of the language participated in the experiment. There is a considerable difference between representatives of the northern variety of Welsh and those who speak the southern dialect of the language. The informants were asked to read a list composed of 16 target words, embedded in the carrier phrase Dw i heb ddweud X, ond Y!, which translates into English as ‘I didn’t say X, but Y!’.9 The list was read twice so that each item occurred in the X and Y slots.10 8 It should be borne in mind that, in the Wels language, the sound [r̥ʰ] is an allophone of /r/, but it is not investigated in this study. 9 Literally, ‘I am without say(ing) X, but Y’. 10 The second author prepared the wordlist, Sven Grawunder designed the experiment, while the acoustic analyses were conducted by the first author. 113 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS Altogether, the participants produced 736 tokens of /r̥ʰ/, 7 of which had to be excluded from analysis for technical reasons. The recordings were made in Szczecin, Poland, in 2014 and Leipzig, Germany, in 2014/2015.11 The Praat software (version 4.2.21) was used to make the recordings, digitise the data and produce the spectrograms and oscillograms. As for the Slavonic languages, audiobook recordings of the Old Testament were used to obtain material for analysis. The acoustic properties of one hundred tokens of /rʲ/ pronounced by two native speakers of Russian and Ukrainian and the same number of tokens of /r̝/ pronounced by two Czech speakers were examined for the purposes of the paper. Since audiobooks are sold commercially, an assumption is made that the analysed speech samples represents the standard variety of the languages in question. The two methods of data collection differed significantly, yet they also have certain things in common. First of all, in the four languages in question (Welsh, Russian, Ukrainian and Czech), samples of read speech were analysed. Also, only word or morpheme-initial rhotics were taken into consideration as the distribution of the Welsh spirantised rhotic is restricted to a greater extent than that of the Slavonic complex rhotics /rʲ/ and /r̝/ presented in sections 3.2.2 and 3.2.3. However, there is a considerable difference regarding the number of lexical items in which the rhotics occur. In the case of Welsh, there is a great disproportion between the numbers of informants speaking the northern and southern dialects, 19 and 4 respectively, yet the numbers of rhotics are sufficiently high to conduct a statistical analysis. 4.2. Results and discussion The acoustic analysis of the data revealed that the Welsh spirantised rhotic may undergo radical phonetic reduction. In the examined speech samples, three allophones of the phoneme /r̥ʰ/ can be distinguished. They differ with respect to the manner of articulation and were assigned to the following phonetic categories: trill, tap and fricative. The analysis also indicates that when /r̥ʰ/ is realised phonetically as a spirant (or fricative), the resultant sound may be either a sibilant or a non-sibilant fricative. The major acoustic feature of the former is aperiodic noise of high intensity with a distinct lower cut-off frequency. The cross-linguistic analysis of spirants reveals that this frequency is higher, the more anterior the place of articulation of a fricative is 11 For participation we would like to thank Aled, Andrew, Bryn, Catrin, Delyth, Dewi, Eiri, Fflur, Guto, Gwyer, Hywel, Ifan, Judith, Lowri, Lois, Marian, Nia, Peredur, Rhiannon, Rhys and Eleri. For technical assistance, we also express our gratitude to Swen Grawunder, University of Kiel, Germany and Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany. 114 SYLWESTER JAWORSKI & SABINE ASMUS (Ladefoged & Maddieson 1996: 163). In the case of non-sibilant fricatives, weak noise is present across the whole range of frequencies. This criterion is used in this study to classify allophones of /r̥ʰ/ that were pronounced as fricatives. The /r̥ʰ/ pronounced as a sibilant fricative, in the word rhaff ‘rope’, is depicted in Figure 9. The major reason for selecting this item is that it allows the reader to compare the acoustic features of sibilant and non-sibilant fricatives pronounced within one word. The spectrogram shows clearly that the rhotic segment does not include any closures. Instead, it consists of a period of high-intensity friction that lasts 158 ms with a distinct cut-off point in the spectrum at, approximately, 2500 Hz. Judging from its acoustic properties, the resultant fricative could be classified as a sibilant with a post-alveolar place of articulation. In other words, it has a quality similar to that of the [ʃ] sound of English. Another piece of evidence as to the sibilant status of the spirantised /r̥ʰ/ is provided by the oscillogram, which shows clearly that the amplitude of changes in air pressure produced by the sibilant is twice as high as that of the word-final [f]. Figure 9. Spirantised rhotic of Welsh realised as a sibilant fricative in the word rhaff ‘rope’. Non-sibilant realisations of the rhotic /r̥ʰ/ have acoustic properties similar to those of [f]. As depicted in Figure 10, which includes a spectrogram of the word rhaff ‘rope’ produced by a different speaker, the aperiodic noise of considerable duration (149 ms) extends across the whole frequency range, i.e. from 0 Hz to 5000 Hz. The intensity of friction increases steadily as the fricative approaches the onset of the vowel due to partial voicing of the 115 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS fricative. Neither of the fricatives has a noticeable cut-off point, which is a characteristic feature of non-sibilants. Figure 10. Spirantised rhotic of Welsh realised as a non-sibilant fricative in the word rhaff ‘rope’. The examined recordings also include a number of interrupted realisations of /r̥ʰ/ that cannot be called taps due to being made up of a closure phase followed by a relatively long period of friction. A representative allophone of this type, pronounced in the word rhan ‘part’, is presented in Figure 11. The 12-millisecond incomplete closure phase, represented by the light stripe in the spectrogram, is followed by a long period of friction (116 ms) occurring across the whole frequency range, which indicates that it is a non-sibilant fricative. Figure 11. An interrupted realisation of /rʰ/ produced in the word rhan ‘part’. 116 SYLWESTER JAWORSKI & SABINE ASMUS The token in Figure 11 constitutes a taxonomic problem as it does not match any descriptions of tapped rhotics, nor can it be regarded as representing a trilled variant. Given its acoustic features, i.e. a closure followed by friction, one might argue that it could be referred to as an affricated rhotic. However, it is worth stressing that this realisation of /r̥ʰ/ bears a striking resemblance to the allophone of the Czech fricated trill /r̝/ presented in Figure 8 above. Sounds having such characteristics are classified as trills in Czech despite having only one closure phase (see Šimáčková et al. 2012).12 Following this line of reasoning, trills made up of one closure represent a stage of reduction intermediate between trilled and fricativised variants. For lack of an appropriate label, throughout this work, they are tentatively referred to as one-tap trills and are represented by the symbol [ɾʰ] combining a tap and a non-sibilant fricative. With respect to the palatalised trill /rʲ/ of Slavic, trilled variants of the sound are hardly ever produced. Although in some languages, e.g. Belorussian, the sound merged with its plain counterpart in all contexts, the phoneme tends to be realised as a fricative element in those languages that have it in their sound inventories. The same process affects /r/, but to a lesser extent (Jaworski 2018). Kasatkin (2006: 40) explains that: При прoизношении [p], [p’] контакт кончика языка c нёбом может иногда oтсутствовaть, чаще встречается это у [p’]; в этом случае [p], [p’] мoгут реализоваться как щелевые coглacные c тесной и короткой щелью. Наиболее характерна такая артикуляция [p’] в заударной интервокальной позиции и в кoнцe cловa: бéрег, варúть, дверь, корь. In the articulation of [r] and [rʲ], sometimes there can be no contact between the apex and the palate, more often in the case of [rʲ]; in such cases, [r] and [rʲ] can be realised as fricatives with a narrow and short constriction. This type of articulation is most characteristic of [rʲ] in post-stress intervocalic position and at the end of the word, e.g. бéрег ‘shore’, варúть ‘to cook’, дверь ‘door’, корь ‘measles’.13 Although this experiment is concerned with word-initial rhotics, it should be remembered that, in connected speech, rhotics preceded by a word-final vowel are, de facto, intervocalic. This makes them more susceptible to fricativisation, especially so if they constitute the onset segment of an unstressed syllable. 12 As a matter of fact, Ladefoged & Maddieson (1996: 245) use the phrase one-tap trill, but the authors do not explain how they differ from taps. 13 Translation by the author (S.J.). 117 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS Figure 12. Fricativised allophone of the Russian phoneme /rʲ/ produced in the word peки ‘river’ (gen. sg.). Figure 12 presents a word-initial token of /rʲ/, produced by a native speaker of Russian in the word peки [ɹiˈki] ‘river’ (gen. sg.), which illustrates the fricativisation process described in Kasatkin (2006). Both the spectrogram and the corresponding section of the oscillogram provide evidence that the friction component is superimposed on changes in air pressure caused by the movements of the vocal cords. This acoustic consequence of combining the two sources of energy is the lack of a distinct formant structure. The /rʲ/ phoneme can also be pronounced as an approximant if the constriction between the articulators is too wide for the jet of air to produce friction, but allophones of this type are rather rare in prosodically strong word-initial position. An analysis of the acoustic properties of the three complex rhotics in question indicates that they differ with respect to allophonic variation observable in the examined recordings. As shown in (2), four articulatory variants of the Welsh spirantised rhotics were distinguished which include spirantised trills [r̥ʰ], one-tap spirantised trills [ɾ̥ʰ], sibilant fricatives [ɹʃ] and non-sibilant fricatives [ɹx].14 As for the Czech fricated trill, all the examined tokens fall into either the one-tap trill [ɾʱ] or the fricative category [ɹʒ].15 The reason for using slightly different symbols to represent the one-tap 14 For lack of phonetic symbols representing natural classes of speech sound, [ʃ] and [x] were selected arbitrarily to stand for sibilant and non-sibilant fricatives, respectively. By no means should it be understood that the trills are realised either as [ʃ] or [x]. 15 Dankovičová (1999) argues that the trill involves two or three closures, but such realisations were not encountered in the examined recordings. Therefore, this variant was bracketed to denote a possible, but unattested allophone. 118 SYLWESTER JAWORSKI & SABINE ASMUS trills of Welsh and Czech is that they differ with respect to the feature [voice ]. In our data, the Welsh sound is always [-voice ],16 whereas the Czech segment is always [+voice ], at least in this context. Finally, the palatalised rhotic of Russian and Ukrainian also has four major articulatory variants that are labeled as follows: palatalised trill [rʲ], palatalised tap [ɾʲ], palatalised ricativised rhotic [ɹ̝ʲ] and palatalised approximant [ɹʲ]. The approximantised variants encountered in the East Slavonic languages represent the most radical form of reduction as they are vowel-like segments in terms of acoustic features. (2 ) /rʰ/ ˳ /r/ ˕ /rʲ/ [rʰ] [ɾʰ] [ɹᶴ] [ɹx] ˳ ˳ ([r]) [ɾʱ] [ɹƷ] [rʲ] [ɾʲ] [ɹʲ] [ɹʲ] ˕ The allophones presented in (2) differ with respect to articulatory complexity, which definitely has an impact on their frequency of occurrence. Table 1 presents general statistical data regarding the distribution of the allophones in the four languages investigated in this study. The first conclusion that emerges from the analysis is that fricativised allophones, i.e. those that consist of a period of friction without any closure, constitute the majority variant in each language.17 By contrast, trilled variants that are supposed to be the articulatory target either do not occur in the recordings (in Czech), or their frequency of occurrence is unexpectedly low (in the case of Russian). In addition, voiced trills have more articulatory variants as fricativised allophones can be further reduced to approximants. 16 Asmus & Grawunder (2017) argue convincingly that, in the case of the Insular Celtic languages, the opposition voiced-voiceless should rather be replaced with lenis-fortis. 17 In this work, the terms ‘fricated trill’ and ‘fricativised trill’ are introduced to distinguish two sounds of Czech. The former is used in the phonetic literature to refer to the Czech rhotic phoneme /r̝/ that consists of trilling and friction produced simultaneously (see Dankovičová 1999). However, as shown in t is study, the /r̝/ phoneme may have an allophone pronounced as a period of friction without a single closure phase. In this work, such allophones are referred to as fricativised trills, which seems to be a convenient phonetic label for the resultant sound. The same label is used to refer to those allophones of the spirantised /r̥ʰ/ of Welsh and the palatalised /rʲ/ of Russian and Ukrainian that are realised phonetically as a period of friction. 119 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS Table 1. Distribution of allophones of complex rhotics Tokens Trill One-tap trill/Tap Fricative Approximant Welsh 729 92 31 606 - Czech 100 - 24 76 - Russian 100 4 18 63 15 Ukrainian 100 8 11 69 12 Language The spirantised trill of Welsh typically consists of two, less frequently three, incomplete closures separated by periods of friction. If two closures are produced, the duration of the constriction does not seem to be affected nor does the duration of the intervals of friction. The former is, on average, 8 ms long, whereas the mean duration of the latter is 15 ms. Unfortunately, a reliable statistical comparison of the temporal characteristics of trilled variants pronounced by different speakers cannot be performed due to an insufficient number of tokens. The generalisations presented in this section are based on the performance of Speaker 6, a person with a trained voice, who herself articulated 21 trilled allophones of /r̥ʰ/. As regards the one-tapped allophone of the Welsh spirantised trill, its spectral characteristics, i.e. a closure phase followed by a period of friction, resemble the partially reduced allophone of the Czech trill. However, the two sounds differ considerably with respect to the temporal characteristics of both the closure phase and the friction. In Welsh, the closure phases range from 9 ms to 13 ms, while in Czech the vary to a greater extent, i.e. from 11 ms to 18 ms. On the other hand, the Welsh allophones have much longer periods of friction, which range between 73 ms and 139 ms, while the period of friction is on the order of 50 ms. By contrast, Russian and Ukrainian tapped variants are never followed by a period of friction. They consist of a single tap released into the following vowel. The mean duration of the closure phase is 16.8 ms in Russian and 17.1 ms in Ukrainian. Despite a relatively low number of tokens, a statistical test comparing the durations was performed and it yielded an insignificant result (p = 0.3185). As for the fricativised allophones, there are three differences between Welsh and the three Slavonic languages. Fricativised variants are always fortis in Welsh, but never in the other languages, at least in word-initial position. 120 SYLWESTER JAWORSKI & SABINE ASMUS In addition, Celtic and Slavic trills differ significantly in duration, with the Welsh language having the longest and the East Slavonic languages the shortest fricativised allophones. In Welsh, such sounds last up to 156 ms, whereas in Czech their duration does not exceed 96 ms. By comparison, in Russian and Ukrainian, the period of friction can be as short as 25 ms and is hardly ever longer than 45 ms. In the case of the Welsh language, the informants were divided into northerners and southerners with a view to determining whether the allophones of /r̥ʰ/ follow similar distribution patterns in their speech. However, the representatives of the north outnumber those of the south by the ratio of 19 to 4, which might have affected the final outcome of the experiment. Another dependent variable that was taken into consideration was the position occupied by the word within the sentence. The underlying assumption was that, in syntactically exposed position, i.e. following a pause, the rhotic /r̥ʰ/ would be more resistant to phonetic change. Table 2. Articulatory variants of /rʰ̥ / in stressed position (following a pause) NORTH (19 speakers) SOUTH (4 speakers) Tokens % Tokens % Trill 72 23.7 5 7.8 One-tap trill 16 5.3 2 3.1 Fricativised rhotic 216 71.0 57 89.1 Table 2 presents the articulatory variants of /r̥ʰ/ produced by the informants in prosodically strong position. The two sets of data look similar in that fricativised allophones form a substantial majority, while one-tapped trills constitute the least frequent variant. In order to establish whether the allophones follow a similar distribution pattern, the data were compared statistically in a contingency table, which is a variant of the chi square test. The test did not confirm the initial impression as the differences between the two distribution patterns reached the level of significance (F = 9.1709; df = 2; p = 0.0102). Nevertheless, the relatively high value of p suggests that the result might have been different had the two groups been equal in number. 121 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS The /r̥ʰ/ sound was expected to be more susceptible to phonetic change when placed in unstressed, phrase-final position. As Table 3 shows, this assumption was confirmed, to a certain extent, as fewer trills and more fricativised allophones were produced by the northerners. As for the informants representing the south, there is hardly any change in comparison with the data presented in Table 2. The same statistical test performed on the data demonstrated that, in this context, the distribution patterns are almost identical (F = 0.1987; df = 2; p = 0.9054). However, a highly significant difference was achieved when the patterns of distribution produced by the northerners in the stressed position were compared with those obtained in the unstressed position (F = 48.9304; df = 2; p = 2.37E-11). In the case of speakers from the south, the distribution is very similar in both contexts (F = 1.2630; df = 2; p = 0.5318). Table 3. Articulatory variants of /rʰ̥ / in stressed position (following a pause) NORTH (19 speakers) SOUTH (4 speakers) Tokens % Tokens % Trill 13 4.4 2 3.2 One-tap trill 11 3.6 2 3.2 Fricativised rhotic 275 92 58 93.6 Having presented the results, it would be interesting to show how they relate to those of Thomas’s (2000) Welsh Dialect Survey. It must be stressed, however, that the two studies differ significantly with regard to research methodology, methods of data analysis and, most of all, the recorded subjects. Therefore, certain differences seem to be inevitable. The first noticeable discrepancy regards the number of allophones of /r̥ʰ/, which is greater in Thomas (2000). For instance, according to the phonetic transcription, the word rhaff ‘rope’ as the following variants: [kra:f], [sa:f], [gra:f], [hra:f], [ra:f] (Thomas 2000: 98). Of these, only [sa:f], i.e. a fricativised variant with a sibilant, is attested in our study, where it is transcribed as [ʃa:f] by virtue of the acoustic characteristics of the resultant fricative. None of the informants produced a deaspirated trill, as in [ra:f], nor did they pronounce a plosive-rhotic cluster as in [kra:f] or [gra:f]. Still, deaspiration appears to be the most frequent process applied Thomas’ (2000) informants. 122 SYLWESTER JAWORSKI & SABINE ASMUS As for the frequency of occurrence of [r̥ʰ] sounds, in Thomas (2000), approximately 50% of all tokens were classified as spirantised trills. By contrast, in the current investigation, only 21% of spirantised rhotics occurring in a prosodically strong position were trilled and as few as 4.2% of those that were found in an unstressed context at the end of a phrase were trilled. However, the results of both studies point out to the conclusion that the rhotic is less frequently trilled in the south of Wales than in the north. 5. Conclusion Confirming Ball (2015) and Jones (1984), Welsh clearly has two distinct rhotic phonemes, although the spirantised one is rather unstable. The acoustic analysis conducted for the purposes of the study confirmed that the spirantised rhotic of Welsh exhibits a considerable amount of variation. Its allophones distinguished in this study include spirantised trills, one-tap trills, sibilant fricatives and non-sibilant fricatives. The obtained results also suggest that /r̥ʰ/ may be undergoing a phonetic change. The claim is substantiated by the presented data, according to which fricativised allophones constitute the vast majority of the examined tokens not only in prosodically weak phrase-final position (92.2%), but also in a syntactically exposed position (74.2%). However, in order to identify the rhotic’s potential for phonetic changes, a perception study would have to be undertaken. Focusing on the spirantised rhotic, no attention was paid to possible variants of [r]. However, Asmus & Grawunder (2017) show that [r] itself comes in two phonemes, i.e. lenis and fortis, so that Welsh seems to feature three distinct trills: lenis and fortis /r/ and fortis /r̥ʰ/; these are probable remains of a development towards a fourfold sonorant system as found in Irish. This explains certain deviations of the Welsh spirantised trill which escapes various generalisations about trills and fricatives, e.g. their markedness. Nevertheless, as the Irish system is waning, a later perception study will show in which direction the spirantised trill may develop. So far, the trill seems to show more similarities in its potential change with the Slavonic languages, in which the palatalised and fricated trills tend to be fricativised, than with English. At least, no typical English /r/ phoneme could be identified as a replacement or /r̥ʰ/. The changes are shown graphically in Figure 13, which is a slightly modified version of the diagram presented in Figure 1. Our diagram includes the fricativisation as a possible stage of phonetic reduction affecting apical rhotics. The resultant sound, represented by the symbol [ɹ̝], has been shown to constitute an allophone of the phoneme /r/ in the Slavonic languages (see Jaworski 2018). On the other hand, the evidence presented in this paper indicates that the spirantised rhotic of Welsh 123 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS can also be rendered as a sibilant or non-sibilant fricative. The two allophones are represented in the graph by linking the Welsh rhotic /r̥ʰ/ to both [ɹ̝] and [ʐ]. The rationale behind establishing the links is that the articulation of both [ɹ̝] and [ʐ] involve the front part of the tongue and the latter sound may result from retroflection (see Hamman 2003). ɾ r˳ r r˳ h ɽ ˕ ʋ ɾ˳ ɹ ʐ ɹ ɻ ʀ˳ ʀ χ ʁ ʁ˕ ɚ ɑ˞ Figure 13. A modified version of Lindau’s (1985) associations between various types of rhotics that includes the aspirated rhotic of Welsh. With respect to pronunciation differences between speakers of the southern and northern dialects of Welsh, the results are rather inconclusive. Clearly, in order to clarify differences between northerners and southerners, further recordings need to be made, ideally with equal numbers of informants from both areas. Sylwester Jaworski University of Szczecin Sabine Asmus University of Szczecin and University of Leipzig 124 SYLWESTER JAWORSKI & SABINE ASMUS References Asmus, S. & Grawunder, S., 2017 (fc., working title), Vowel Length in Welsh Monosyllables, its Interrelation with Irish and Related Problems: An Acoustic Study and its Didactic Implications, Lewiston, 49–61. Ball, M. J., 2015, ‘How Many rhotic Phonemes does Modern Welsh Have’, in: Czerniak, T, Czerniakowski, M. & Jaskuła, K., eds., Representations and Interpretations in Celtic Studies, Lublin. Blecua, B., 2001, Las vibrantes del Espanol: manifestaciones acusticas y procesos foneticos, unpublished PhD dissertation, Universidad Autonoma de Barcelona. Broderick, G., 2016, ‘Hauptsprachliche Entwicklungen Im Britischen’ (pc.). Carlton, L., 1990, Introduction to the Phonological History of the Slavic Languages, Columbus, OH. Colantoni, L., 2005, ‘Increasing Periodicity to Reduce Similarity: An Acoustic Account of Deassibilation in Rhotics’, in: Diaz-Campos, M., ed., Selected Proceedings of the 2nd Conference on Laboratory Approaches to Spanish Phonetics and Phonology, Somerville, MA, 22–34. Dankovičová, J., 1999, ‘Czech’, Handbook of the International Phonetic Association: A Guide to the Use of the International Phonetic Alphabet, Cambridge, 70–74. Demolin, D., 2001, ‘Some Phonetic and Phonological Observations Concerning /ʀ/ in Belgian French’, in: Van de Velde, H & van Hout, R., eds., R-atics: Sociolinguistic, Phonetic and Phonological Characteristics of /r/, Etudes and Travaux 4, Brussels, 63–74. Gussmann, E., 2007, The Phonology of Polish, Oxford. Hall, A. T., 1997, The Phonology of Coronals, Amsterdam & Philadelphia. Hamman, S., 2003, The Phonology of Retroflexes, Utrecht. Hannahs, J. S., 2012, The Phonology of Welsh, Oxford. Howson, Ph., Kochetov, A. & van Lieshout, P., 2015, ‘Elimination of the Grooving Patterns of the Czech Trill-fricative’, Journal of Phonetics 49, 117–129. Hualde, J. I., 2005, The Sounds of Spanish, New York. Jaworski, S., 2014, ‘A Comparison of Croatian Syllabic [r] and Polish Obstruentised [r]’, in: Witkoś, J. & Jaworski, S., eds., New Insights into Slavic Linguistics, Frankfurt am Main, 179–192. Jaworski, S., 2018, Rhotic Sounds in the Slavic languages: An Acoustic Study, Hamburg. Jaworski, S. & Gillian, E., 2011, ‘On the Phonetic Instability of the Polish Rhotic /r/’, Poznań Studies in Contemporary Linguistics 47(2), 380–398. Jones, G. E. , 1984, ‘The Distinctive Vowels and Consonants of Welsh’, in: Ball, M. J. & Jones, G. E., eds., Welsh Phonology. Selected Readings, Cardiff. Kasatkin, L. L., 2006, Sovriemiennij ruskij izik: fonietika [Modern Russian language: Phonetics], Moscow. 125 AN ACOUSTIC STUDY OF WELSH AND SLAVONIC RHOTICS Kavitskaya, D., Iskarous, K., Noiray, A. & Proctor, M., 2009, ‘Trills and Palatalisation: Consequences or Sound Change’, in: Reich, J., Babyonyshev, M. & Kavitskaya, D., eds., Proceedings of the Formal Approaches to Slavic Linguistics 17, Ann Arbor, 97–110. Labov, W., 1966, The Social Stratification of English in New York City, Washington D.C. Lindau, M , 1985, ‘The Story of /r/’, in: Fromkin, V. A., ed ., Phonetic Linguistics: Essays in Honor of Peter Ladefoged, New York, 157-168. Ladefoged, P. & Maddieson, I., 1996, The Sounds of the World’s Languages, Chichester, West Sussex. Lipski, J. M., 1987a, ‘El español vestigial de Filipinas’, Anuario de Lingüística Hispanica (Valladolid) 3, 123–142. Lipski, J. M., 1987 , ‘Fonética fonología del español guineano’, Revista trimestral de cultura 2(1), 9–18. Maddieson, I., 1984, Patterns of Sounds, Cambridge. Recasens, D., & Espinosa, A., 2007, ‘Phonetic Typology and Positional Allophones or Alveolar Rhotics in Catalan’, Phonetica 63, 1–28. Recasens, D. & Pallarès, M. D., 1999. ‘A Study of /ɾ/ and /r/ in the Light of ‘dac ’ Coarticulation Model’, Journal of Phonetics 27, 143–169. Rospond, S., 1973, Gramatyka historyczna języka polskiego, Warszawa. Solé, M-J., 2002, ‘Aerodynamic Characteristics of Trills and Phonological Patterning’, Journal of Phonetics 30, 655–688. Šimáčková, Š., Podlipský, V. & Chládková, K., 2012 ‘Czech Spoken in Bohemia and Moravia’, Journal of the International Phonetic Association 42, 225–232. Thomas, A. R., 2000, The Welsh Dialect Survey, Cardiff. Walsh Dickey, L., 1997, The Phonology of Liquids, unpublished Ph.D. dissertation, University of Massachusetts, Amherst. Verstraeten, B. & van de Velde, H., 2001, ‘Socio-Geographical Variation of /r/ in Standard Dutch’, in: Van de Velde, H, and van Hout, R., eds., R-atics: Sociolinguistic, Phonetic and Phonological Characteristics of /r/, Etudes and Travaux 4, Brussels, 45–62. Vihman, M. M., 1996, Phonological Development. The Origins of Language in the Child, Oxford. Wiese, R., 2003, ‘The Unit and Variation of German /r/’, Zeitschrift fur Dialektologie und Linguistik 70, 25–43. Wiese, R ., 2011, ‘The Representation of Rhotics’, in: van Oostendorp, M., Ewen, C. J., Hume, E. & Rice, K., eds., The Blackwell Companion to Phonology, Oxford, 711–729. Żygis, M., 2005, ‘(Un)markedness of Trills: The Case of Slavic r-Palatalisation’, Zeitschrift fur Slawistik 50, 383–407. 126 T he volume contains articles from the 8th International Colloquium of Societas Celto-Slavica at the University of Heidelberg, including the colloquium summary by J. Gvozdanović (‘Celto-Slavica in Heidelberg’). G. German (‘Which linguistic model for Brittany?’) asks whether to promote the traditionally transmitted Breton language varieties spoken naturally, or the new standardized language. M. Bloch-Trojnar (‘Semantic (ir)regularities in action nouns in Irish’) demonstrates that the patterns of polysemy in verbal nouns in Modern Irish are constrained by the lexical semantics of the base verbs. M. Ó Fionnáin (‘Opportunities seized: From Tolstóigh to Pelévin’) takes a fresh look at several Irish-language translations from the original Russian. E. Parina focuses on ‘Multiple versions of Breuddwyd Pawl as a source to study the work of Welsh translators’, examining how a comparison between multiple translations from differing sources can provide us with new insights. S. Jaworski and S. Asmus (‘An acoustic study of Welsh rhotics’) analysed a variation of rhotic phonemes of 23 Welsh speakers, 21% of the tokens classified as aspirated trills. Sometimes replaced with various fortis fricatives, or with the glottal [h], Welsh aspirated rhotic resembles the Slavic palatalised /rʲ/ replaced with the fricative /ʒ/ in Polish. Sylwester JAWORSKI & Sabine ASMUS Elena PARINA Gary GERMAN Maria BLOCH-TROJNAR Mark Ó FIONNÁIN ISSN 2058-9050 ISBN 979-10-92331-43-1 9 79 1092 33 1 43 1 Prix : 15 € Centre de Recherche Bretonne et Celtique CS 93837 - 20, rue Duquesne 29238 Brest Cedex 3

RELATED PAPERS

RELATED TOPICS

Log In

Studia Celto-Slavica 9:Journal of the Learned Association Societas Celto-Slavica, volume 9

Studia Celto-Slavica 9:Journal of the Learned Association Societas Celto-Slavica, volume 9

Related Papers

RELATED PAPERS

RELATED TOPICS