Linguistische Berichte Herausgeber Günther Grewendorf (Frankfurt) Arnim von Stechow (Tübingen) Redaktion Günther Grewendorf Eric Fu~ Johann W. Goethe-Universitat Fachbereich Neuere Philologien Institut für Linguistik Grüneburgplatz 1 D-60629 Frankfurt am Main Tel. +49-69-798-32398 Fax +49-69-798-323 99 E-Mail: Ib@lingua. uni-frankfurt.de Beitrâqe aus Forschung und Anwendung 8eirat Peter Auer (Freiburg) Josef Bayer (Konstanz) Daniel Büring (Wien) Harald Clahsen (Colchester) Karin Donhauser (Berlin) Gisbert Fanselow (Potsdam) Caroline Féry (Frankfurt) Irene Heim (Cambridge) Ludger Hoffmann (Dortmund) Ekkehard Kôniq (Berlin) Jôrq Meibauer (Mainz) Gereon Müller (Leipzig) Susan Olsen (Berlin) Rosemary Tracy (Mannheim) Richard Wiese (Marburg) Ede Zimmermann (Frankfurt) Sprachtypologie Configurationality, Successive Cyclic Movement and Object Agreement in Kiribati and Fijian* Joachim Sabel Abstract Fijian and Kiribati Die Linguistischen Berichte sind in Bezug auf Gegenstande und Methoden der Linguistik auf maximale Offenheit hin ausgerichtet, halten im Hinblick auf die zugrunde gelegten wissenschaftlichen Standards aber an einem hohen theoretischen und empirischen Anspruch fest. Eingereichte Manuskripte werden von anonymen Gutachtern beurteilt (»Peer Review«). Das Institutsabonnement (Print- und Onlineausgabe) kostet € 148,- pro Jahr, das Privatabonnement (Print- und Onlineausgabe) € 92,- (jeweils zuzüglich Versa ndspesen: Inland € 7,- 1 Ausland € 16,-). Der Preis für ein Einzelheft betraçt € 45,-. Kündigungsfrist: 6 Wochen zum Jahresende. Eine Auswertung der Linguistischen Berichte erfolgt in: BLLDB (Bibliographie Linguistischer Literatur Datenbank), CSA Arts & Humanities, IBR (Internationale Bibliographie der Rezensionen geistesund sozialwissenschaftlicher Zeitschriftenliteratur), IBZ (Internationale Bibliographie der geistes- und sozialwissenschaftlichen Zeitschriftenliteratur) Neue Abonnements nehmen der Helmut Buske Verlag, Richardstraêe 47, 22081 Hamburg, Tel. 040 1 299 95 80, Fax 040 1 299 36 14, E-Mail: info@buske.de sowie jede Buchhandlung entgegen. Jahrlich erscheinen vier Hefte (Februar, Mai, August und November) mit einem Umfang von je etwa 128 Seiten. Zudem kann jâhrlich ein Sonderheft erscheinen, das den Abonnenten mit einem Nachlass von 15% auf den jeweiligen Ladenpreis geliefert wird. Werkdruckpapier: alterungsbestandig nach ANSI-Norm resp. DIN-ISO 9706, hergestelit aus 100% chlorfrei gebleichtem Zellstoff. Printed in Germany. © Helmut Buske Verlag GmbH, Hamburg 2013. ISSN 0024-3930 www_buske_de 1995,2001,2003) seem to represent "rich pronominal with typical properties languages" ofnon-configurational (cf. Jelinek headmarking 1984, Baker languages, such as free word order and pl'o-drop. 1 will argue, however, that these languages are con figurational and that pre- and post-verbal pronominal-like subject and object morphemes are best analyzed as agreement markers. Evidence is provided from extraction and agreement asymmetries with arguments and from restrictions on the linear order of adjuncts. ln addition, object markers that show up in long dependencies are best analyzed as agreement sive cyclic A' -rnovement. These morphological reflexes can be interpreted ture splitting in the sense of Chomsky (2008) and of the fact that successive-cyclic targets phase-edges. For the most part, the discussion similar phenomena are illustrated for Fijian. Verb Morphology reflexes of succesas evidence of fea- is based on exarnples movement From Kiribati but and Sentence Structure in Kiribati and Fijian ln section l, 1outline the most important morphosyntactic properties of Fijian and Kiribati for the discussion in this paper, especially the morphosyntax of verbs and the pronominal system. Section 2 deals with the status of the subject and object pronominal markings. Illustrating that Kiribati and Fijian are configurational languages, it will be argued that these markings cannot be analyzed as pronominal arguments. Section 3 deals with the analysis of object markers as agreement in the • The data discussed in this paper were collected du ring my stays in Tarawa (Kiribati). Nadi and Suva (Fiji) in 2008-2011. Thanks to my informants Mereisi Kamoe. Meeribwa. Tekarei Russell. Apolonia Tamata, Teeang, and Tereau, as weil as to Paul Geraghty. Shelly Harrison, Marc Richards and the audiences at UCLi\. the University of the South Pacifie (Suva), the University of Leipzig. the University of Frankfurt, and to the participants of the Austronesian Formai Linguistics Association (AFLA XVII) at Stony Brook for comrnents and discussion of the issues raised here. Linguistische Berichte 233/2013 © Helmut Buske Verlag, Hamburg Joachim Sahel 4 context of A' -movement constructions, It will be illustrated that long movement leaves agreeing object agreement markers in embedded and matrix clauses providing evidence of successive cyclic movement through (the embedded and matrix) vP as implied by Chomsky's analysis of phases (2000, 200 1,2007,2008), 1.1 Kiribati ln this section, 1 de scribe some of the central morphosyntactic properties of Kiribati, with particular focus on the pronominal system. Kiribati is an Oceanic (Micronesian) VOS language (Lynch 1998:154f., Lynch et al. 2002, chapter 3) with nominative-accusative Case marking. It is a free word order language without overt Case and number marking on nouns. A syntactic passive construction is realized with the verbal suffix -aki. It is an optional wh-in situ language. ln wh-ex situ structures, the wh-element precedes a marker that is very similar to a demonstrative-pronoun that also introduces relative clauses (Cowell 1951, Trussel 1979). As concems the morphosyntax of verbs, Kiribati, like other Oceanie languages, uses a formai marking for transitive verbs. ln addition, it has a "rich" pronominal system and allows for subject and object pro-drop. It has a set of pre-verbal pronominal subject-c1itic elements/agreement markers, which indicate the person and number of the subject. Furthennore, post-verbal object-c1itics agree in person, number, and, in the third person plural, with respect to animacy. Some Oceanic languages have these subject pronominal-like elements (marking person and number of the subject), which are different from the independent subject pronouns. These subject c1itics are obligatory pre-verbal partic1es. By contrast, independent pronouns are optional and used for emphasis only ("N" is used in place of "1", the first person subject marker, when it precedes the markers na "will" (see the discussion below) and nang "about to"): (l) Kiribati Subject Pronouns & Subject Markers a. Independent Pronouns SINGULAR PLURAL 1. ngai 2. ngkoe 3. ngaia ngaira ngkamii ngaiia b. Subject Agreement Markers SINGULAR PLURAL l. 1-, Nti2. kokam3. ea- Tense aspect markers occur as weil, as for example -na for future/irrealis as in (2a), -a for Past (2b), and a phonetic null morpheme for Present. Assuming that the pre-verbal subject particles are agreement elements (see below), the linear order of the verb affixes in Kiribati (and in the other VOS language Fijian) is the mirror image of verb affixes in SVO/SOV languages, i.e. Agr-Tns- V (Ko-na-mma'akuri ngkoe; 2SG-Future-work you, "You will work.") YS. V-TnsAGR in French/German (Tu dorm-ir-as; you-sleep-Future-2SG, "You will sleep."; Du lern-te-st; you-learn-Pasl-2SG, "You learned."). However, the Agr- Configurationality, Successive Cyclic Movcmcnt and Objcct Agreement 5 Tns complex in Fijian/Kiribati is better analyzed as a kind of auxiliary that is 10cated in P and does not form one complex head with the verb (i.e. "participle") which is located in va. Adverbs, such as lao 'perhaps', kaman 'already' and negation aki 'not' appear between the auxiliary and the main verb and are arguably adjoined to vP (whereas manner adverbs appear in a position following (and not in front of) the verb, see the discussion in section 2 below). Nothing may intervene between the subject pronominal c1itic and the tense aspect marker. (2) a. b. E-na] 3so- Fut./lrreal. "The ship will not go." E-a] tiba 3SG -Past just "The ship has just gene." kaibuke]. ship [ [r aki not nako go te DH u- nako go te DEr kaibuke]. ship [Kiribati] As shown in (2), the subject clitic is analyzed here as an agreement marker that indicates agreement with a phonetically realized (NP/pronoun, CP) subject or a silent pro (see also Jacobs 1984). As discussed in section 2 however, one could altematively imagine it being a full pronoun that receives a 8-role within VP and moves to Spec TP, and that the associated NP, for example te kaibuke "the ship" in (2), is optionally added as a kind of adjunct. ln the following section, 1 will argue against this possibility and for the first analysis, i.e. the agreement marker analysis. 1 leave aside the question of whether the sentence-final subject in V(O)S sentences is located in a right-peripheral vP-/TP-specifier position or whether it is in a left-peripheral specifier (assuming a base-generated SVO order) and the predicate undergoes leftwards movement into this functional projection (assuming multiple specifiers) or adjoins to it ([TI' [PTedP ... ] lr Subject lr T tPTedP]]] or hl' [PTedP ... ] hl' Subject lr T tPTedP]]]) but 1 tentatively adopt the rightwards specifier analysis ([TI' lr T [PTedP ... ]] Subject]) (see Sabel2011 for discussion). The subject agreement marker is obligatory in finite clauses (3), whether or not a nominal (3a), an overt pronominal (3b), no overt subject (3c), or, as in extraction contexts (see the discussion below), a covert copy (trace) is present. Sentences with null subjects such as (3c) involve (small) pro. They receive a pronominal interpretation. (3) a. b. c. *(E) nakonako te 3sG-0(Pres) walks DFI "That man is walking." *(Ti- )na nako ngaira. 1PL-Fut. go I.PL "We will go." *(1) nako Tarawa 1sG-0(Pres) go Tarawa "1 go to Tarawa." uaarei. man pro. 1SG [Kiribati] (4) Configurationality, Joachim Sabel 6 1 nako Tarawa 1so go Tarawa "1 go to Tarawa." ngai ISG 7 ferent object markers are used depending on whether the object is animate (7a) or not (7b) (see (Sb».2 /*ngaiia. 3PL [Kiribati] The overt independent subject pro no un in (3b) has much the same function and distribution as an overt subject pronoun in a pro-drop language like Spanish or ltalian. These pronouns can be topicalized and focused. When the optional independent subject pronoun is realized, emphasis is placed on the subject, as in (4) ys. (3c). (4) shows in addition that the pre-verbal subject morpheme has to agree with the post-verbal pronominal subject. Kiribati, Fijian and other Oceanic languages have the same sets of elements for objects, i.e. independent object pronouns as weil as formally distinct object morphemes/agreement markers. The latter are obligatory (Iike the subject marker). ln contrast to the subject markers, the object markers (see (Sb» are suffixed to the main verb as illustrated in (6)-(7) below (see Harrison 1978 for relevant arguments and a (comparative) diachronie analysis): (5) Kiribati Object Pronouns & Object Markers a. Independent Pronouns SINGULAR PLURAL b. Object Agreement Markers SINGULAR PLURAL 1. ngai 2. ngkoe 3. ngaia 1. -ai 2. -ko 3. -a ngaira ngkamii ngaiia Successive Cyclic Movcment and Object Agreement -ira -ngkamii -ia (animate), -i (inanimate) The 2PL object marker -ngkamii is identical to the 2PL independent pronoun. Other object markers are derived from independent pronouns by deletion of the prefix ng(a/ai)-. ln addition to subject and object markers, Oceanic languages have suffixes that mark a verb as transitive (i.e. as having an object). The transitive verb form of the verb noor "see" in Kiribati, for example, is marked with -i-. To this is suffixed the object marker. With a 3SG nominal object, an active verb has two transitive forms, as illustrated in (6) and in the 3PL it has also two forms (7). The object marker is realized in ail four cases, but the transitive marker does not attach to ail persons and verbs in the same way. As illustrated in (6a) (where we have a subjectpro), it is not overtly realized with a 3SG post-verbal overt nominal object, but overt when the object is extracted or a non-overt (pro) (6b; where we have a subject and an object pro). ln the plural, the transitive marker is added whether the object is overt or non-overt, as illustrated in (7).1 However, two dif- 1 The verbal object marker -ia agrees in o-fcaturcs with a post-verbal nominal object (ia) as weil as with a focused pre-verbal object (iia) ln (ib) and (iib), the verb has the wrong object agreement marker (FM = focus marker). (i) a. E tangir-i-ia Meeiri ao Tien Rui. 3SG loves-Tr.-3PLlanim. Mary and Tien Rui b. * E tangir-i-a Meeiri ao Tien Rui. 3SG loves-Tr.-3SG Mary and Tien Rui "Rui loves Mary and Tien." (6) a. i. ii. (7) * a. l. ii. * noora 'see hirn/her/it' noor-a te 1 kameapro. ISG see DET dog noor-i-a te kameapro. 1 ISG see DET dog "1 see/saw a/the dog" nooriia 'see them (animate)' 1 noor-i-ia+anim (aomata) pro. "1 see people/them." 1 noor-i-i.anlm aomata pro. "1 see people." b. i. nooria 'see him/her/it' 1 noor-i-a pro pro. "1 saw him/her/it." ii. * 1noor-a pro pro "1 saw hirn/her/it." b. i. noorii 'see them (inanimate)' 1 noor-i-i.anim (booki) pro. "1 see books/them." ii. * 1 noor-i-ia+anim books pro. "1 see books." ln addition, ditransitive verbs do not make use of the transitivity marker. The same holds for verbs in passive and no un incorporation structures.' If passive or no un incorporation is involved, the object marker is absent as weil, whereas in ditransitives, i.e. in double object constructions the otherwise obligatory object agreement marker is realized only with a silent nominal object (pro or trace), and in passives and prepositional ditransitives the object marker is generally excluded (cf. the discussion in section 2). ln Kiribati, whatever object the verb agrees with is the NP/CP selected by the verb, i.e., usually the theme, but for double object constructions with verbs like 'give' or 'tell' it is the recipient. NP/CP arguments trigger (subject or object) verb agreement not adverbs or PPs. Only one object marker may appear on the verb. 1.2 Fijian Fijian is another Oceanic VOS language with nominative-accusative Case marking (Lynch et al. 2002, chapter 3). There are over thirty "Fijian" dialects (see Geraghty 1983 for details). ln this paper, 1 will discuss examples from Standard (ii) tangir-i-ia Rui. Mcciri ao Tien aika e loves-Tr. -3 PLian im. Rui Mary and Tien FM 3SG tangir-i-a Rui. b. * Meeiri ao Tien aika e loves-Tr.-3SG Rui Mary and Tien FM 3SG ··It is Mary and Tien that Rui loves." 2 With sorne verbs, the transitive marker is never overtly realized (kune-O-a "find-3SG"). whereas another verb c1ass uses -e- as transitive marker with SG (for ex .. taboraa-e-ai "insult-Trme") Furthermore, when the verb stem ends in a nasal. the transitive marker in 3SG takes the forrn ofa doubled consonant b 'ain-n-a "use-Tr.-3PS.SG. ..• kanim-m -a "attach to-Trans.-3PS.SG." 3 (i) is an example with incorporation of a theme argument m 'ane "rnoney" in the prepositional ditransitive construction. The verb shows up in ils intransitive forrn, i.e. anga. (i) 1 anga-m •ane [nakon teuaarei]. 1SG give-rnoney to DET-man '1 give money to this man.' a. 8 Joachim Sahel Fijian, i.e. (Bauan Fijian, the eastern dialect of the island Bau), and from the North-West Viti Levu variant (spoken in the area ofNadi). Fijian has free word order without overt Case and number marking on nouns like Kiribati. Like other Oceanic languages, Fijian uses a formaI marking for transitive verbs, and has a set of pre- and post-verbal clitic pronouns/agreement markers, which indicate the person and number of the subject and object. The verbal markers in Fijian behave as in Kiribati, and variations between overt and non-overt transitivity markers are also found in Fijian. Finally, Fijian like Kiribati allows for noun incorporation. ln this case, the verb appears in its intransitive form. Fijian is also an optional wh-in situ language. Wh-ex situ is realized as wh-preposing into the pre-verbal position. 2 Configurationality in Kiribati and Fijian There has been sorne discussion as to whether the obligatory pronominal marking in pro-drop languages should be treated as agreement or as an argument bearing a B-role. The issue can only be decided by appealing to a particular anaIytical framework. With respect to the pronominal argument hypothesis (Jelinek 1984, among others), 1 will argue in the following two sections that the analysis of the subject and object c1itics as pronouns in KiribatiiFijian is not adequate and that these elements are best treated as agreement markers. It has been claimed that for a range of languages, the use of pronominals has consequences for the structural positions of nominal expressions and may provoke a c1uster of syntactic properties characteristic of a certain type of free word order language, i.e. the "pronominal argument languages" (Jelinek 1984; Baker 1995, 200 1, 2003). The idea is that free word order is not a homogeneous phenomenon and that there is no single macro-parameter that is responsible for the free word order phenornenon, (see Hale 1983; Baker 1995,2001, and Pensalfini 2004 for analyses and discussions of different types of free word order languages, i.e. configurational and non-configurational). As regards nonconfigurational free word order languages, the assumption of subject/objectc1itics acting as pronominal arguments to which O-roles are assigned has led to an analysis of the corresponding full NPs as adjuncts, for example, in languages such as Mohawk (Baker 2001). Being adjoined to the predicate phrase (or TP), the NPs rnay be generated in different linear order (scrambling), rnay not (in contrast to the pronouns) appear in VP, and they may be freely omitted (prodrop). Extraction asymmetries among NPs do not exist. Furtherrnore, because the adjuncts are ail dominated by the same nodes, the absence of Case rnarking on full NPs is explained. Case is usually assigned under a Spec-Head or within a head complement (domain) relationship (assuming the operation Agree) that is missing in the non-configurationallanguages. Kiribati and Fijian also have a rich pronominal system, free constituent order, absence of overtly Case marked nominal constituents and the possibility of Configurationality, Successive Cyclic Movernent and Object Agreement 9 allowing freely for pro-drop. Therefore, Fijian, for example, has been claimed to be a nonconfigurational pronominal argument language (Alderete 1998, Aranovich 2004). This claim can be extended to Kiribati, which has similar characteristics and constituency ordering to that of Fijian. ln this section, 1 will argue that the Oceanic languages under discussion are not "pronominal argument languages" by illustrating that Fijian and Kiribati display important configurational properties. This then implies that Kiribati and Fijian contradict another correlation that has been proposed in the literature and that concerns the free word order and rich nominal Case inflection in configurational free word order languages (Sapir 1921, Haider 1988, Alexander 1990, Boskovié 2005, among others). Kiribati and Fijian represent counterexamples to the claim that rich nominal Case intlection makes scrambling possible or that rich Case inflection is a necessary condition for a language to have scrambling (see the discussion in section 4). The following examples illustrate the free word order phenomenon in Kiribati, a language without overt nominal Case marking. The variants of ordering of nominal objects in (8) are found in the post-verbal domain with the prepositional ditransitive construction (The language allows for clause-internaI scrambling but, like scrambling languages such as German and Dutch, not for scrambling out offinite clauses.):" (8) a. b. c. d. e. f. A anga te boki [nakonRui] taan 3PLgive OH book to Rui DET Ali: 'The teachers give the book to Rui.' A anga te boki taan reirei [nakon Rui]. A anga [nakon Rui] taan reirei te boki. A anga taan reirei [nakon Rui] te boki. A anga [nakon Rui] te boki taan reirei. A anga ta an reirei te boki [nakon Rui]. reirei. (V TH PO S) teachers (VTH Spa) (V PO S TH) (V S PO TH) (V PO TH S) (V S TH PO) [Kiribati] Nevertheless, Kiribati displays word order restrictions typical of a configurational language. For example, we observe asymmetries between the ordering of therne and recipient arguments in the double object construction. The order in (9a) is the unmarked/basic word order, and (9b) is derived by scrambling. (9) a. b. A anga-n Rui te boki taan reirei. (V REC TH S) 3PLgive-APPI. Rui DEI book DET teachers Ail (9)-{ 10): 'The teachers give Rui the book.' A anga-n Rui taan reirei te boki. (V REC S TH) 3PLgive-APPL Rui DEl' teachers DEl book [Kiribati] 4 Given that Fijian and Kiribati are headmarking languages. subject and object agreement markers may potentially disambiguate the grammatical function or arguments in the sentence. r Joachim Sabel \0 Other orders not adjacent (10) a. b. c. d. th an in (9) in which the recipient to the verb are impossible, *A * * * 3pl A A A anga-n give-appl anga-n anga-n anga-n Configurationality, Rui is in post-verbal position te boki det book taan reirei te boki taan reirei Rui taan Rui det Rui taan reirei te boki retrei. (V TH REC S) teachers te boki. (V S TH REC) Rui. (V TH S REC) Rui. (V S TH REC) This word order pattern resembles the restnction that we know from configurational language English, see (Il). The double object construction lows only for the order recipient (II) a. b. but cf. (l0). the al- > theme: John gave the children the books. gave the books the children. * John Successive Cyclic Movement and Object Agreement \1 Subj] (which also explains the position of the applicative affix -n as a result of "incorporation," i.e. as an effect of the intermediate movement step of verb movement to v in (9), (10), and (12» the theme cannot be extracted over the recipient. Note that the proposed structure contains a low applicative affix in the sense of Pylkkânen (2002) and McGinnis (2001). The Superiority Condition effect (Chomsky 1973) has been later explained in terms of the Minimal Link Condition or as a Relativized Min imality/ Intervention Effect (Rizzi 1990; Chomsky 1995, 2000, 2001).5 If two elements a, p can potentially be attracted by a head K because they ail bear a matching feature, only the one which is "closer" to K can move, where p is closer to K than a if p c-commands a. Assuming that the recipient asymmetrically c-commands the theme, the indirect object is closer to the attracting C head than the theme in (I2)--{ 13). Within the analysis of Chomsky (2000, 2001) the extraction asym- Consider next wh-extraction phenomena. When an object XP is preposed for information al reasons in Kiribati, the clause takes on a subordinate look (and the (di)transitive verb appears with an object marker agreeing with the fronted object, compare also the examples (ii) with focus-fronting in fn. 1 with a simple transitive verb and the discussion in section 3). A wh-element can either appear in sentence-initial position (J 2a) or it remains in situ. ln the first case, a particle (ae), glossed here as focus marker (FM), is obIigatory (this particle cannot appear with wh-in situ). The following pair shows that the goal can be extracted in the double object construction. The theme, however, cannot be extracted over metry can be derived from the hypothesis that vappl? is not a phase head that can bear an EPP feature (the theme may therefore not be moved to Spec ApplP from where it otherwise could move further on to Spec CP). However, an EPP feature can appear on the higher v. Only the closer recipient argument can be attracted by this feature and can move throught Spec vP. The recipient blocks whmovement of the theme to Spec CP in (l2b) and (l3b). Given the applicative structure illustrated below example (13), the absence of the transitivity marker in ditransitive constructions is a consequence of the fact that both nominal objects are arguments of the applicative head. Consider next passivization in Kiribati double object constructions. Recipi- the recipient: ents can be passivized (12) [te boki Antai ae ko anga-n-a DET book who FM 2SG give-APPL-3sG 'Who do you give which book?' b. -/?'[Te boki ra] ae ko anga-n( -a) give-APPL( -3sG) DEr book which FM 2SG a. 'Which book did you give who?' ra]? which antai? who [Kiribati] A similar restriction holds in English: Wh-movement in double object constructions is more restricted for themes than for recipients. (13b) is only grammatical with a special (non-information question) echo-reading (cf. Barss and Lasnik 1986). (ln movement structures, a tracy/copy/occurrence of the moved element X is written as <X>.) (13) a. b. (14) a. b. Assuming * in contrast to themes: A anga-n-aki te boki taan reirei (irouia 3PL give-APPL-PASS the book the teachers (by 'The teachers were given the book (by the children).' E anga-n-aki taan reirei te boki (irouia 3sGgive-APPL-PASS the teachers the book (by 'The book was given the teachers (by the children).' that the recipient asymmetrically c-commands the theme, ataei). children) ataei). children) [Kiribati] the indirect object is closer to the attracting head T than the theme. Therefore, the recipient blocks NP-movement of the theme to Spec TP in (14b). Furthermore, passivization of the theme becomes acceptable when the recipient is extracted: Who did you give <who> which book? book did you give who <which book>? * Which When two wh-phrases are present, the structurally higher ("superior") one must move. Assuming that the recipient is in a structurally higher position, ... V [vP Recipient [y. <V> Theme]] (where <V> marks the base position of the raised verb), or with a more detailed structure for Kiribati (assuming, however, that Vto-vappl-to-v takes place) [vp v LApplP Recipient lvappl' vappl -n [vp V Theme]]] 5 Superiority effects are also auested in Kiribati with respect to extraction of subjects and (referential) adjuncts iai anrai? nako (il a.' la ae e RP who go where Foc Past ia? nako b. Antai ae e where who Foc Past go "Who goes where?" 0= Configurationality, Joachim Sahel 12 Antai ae e anga-n-aki te boki who FM 3SCî give-ApI'L-PASS the book 'Whom was given the book (by the children)?' (15) (irouia (by ataei)? children) [Kiribati] a. The teachers were given books (by the children). b. ??The books were given teachers (by the children). (17) Whom were the books given (by the children)? 13 guage without overt Case and number marking on nouns, It is also an optional wh-in situ language where wh-ex situ and focus-fronting are realized as XPpreposing into the pre-verbal position. ln Fij ian, the same extraction asymmetry exists in double object constructions as in Kiribati (12) and English (13); i.e. a goal can easily be extracted but not a the me (see Sabel 2011). Furthermore, restrictions for the licensing of focused arguments in the post-verbal domain in Fijian can be explained if we assume a low structural focus position (Jayaseelan 2008, among others), i.e. configurational structure. On the basis of two assumptions, i.e, that focus precedes background in VOS languages (Sabel 2011), and that recipient arguments precede theme arguments in the base-generated structure, we can derive that only answer (19b) is marked. Only in this case, both relevant preference constraints on word order are violated (as illustrated by both asterisks, reflecting a violation of both word order constraints *Rec > Th and *Foc> Background). (l9a) is a well-formed answer, No word order constraint is violated at ail, and the answers in (18) are also well-formed because the violation of a single constraint is tolerable. Similar facts are observed in the VOS language Kiribati (see Sabel 2011 for the Kiribati data and more discussion on information structure in VOS languages): The contrast between (I4b) and (15) is expected if we follow Chomsky (1995, 2001) who argues that traces (i.e. copies without PF-features) do not count as interveners. ln these respects, the Kiribati double object construction again resembles configurationallanguages such as English, i.e. (16)=(14) and (17)~(15): (16) Successive Cyclic Moverncnt and Object Agreement Given the proposed structure for ditransitives, (9a) represents the canonical order. (9b) involves PF-extraposition ofVP, whereas (lOa-d) cannot be derived. ln the prepositional ditransitive (8), 1 assume that the theme is promoted and moved to Spec vapplP and the recipient is a demoted PP that is right-adjoined to Spec vapplP giving rise to (8a).6 ln addition to this, (8b) involves PP-extraposition over the subject, (8c) vapplP-Extraposition of the lower vapplPsegment (stranding PP), (8d) PP-extraposition over the subject followed by vappIP-extraposition, (8e) "short" vapplP-Extraposition over PP, and (8t) is derived by extraposition ofvapplP. This is the explanation for the scrambling data on the basis of a configurational sentence structure. Let us next consider sorne extraction facts from Fijian. The latter, like Kiribati, is a language with a ri ch pronominal system and a free word order lan- (18) Q: a. b. As already mcntioned in the tcxt, the morphology of ditransitivc verbs is exccptional cornto transitive verbs such as noora "see' in (6)-(7). No transitivity but an applicative marker -n (anga-n in (9)) is realized even with 3sc; overt post-verbal objects as in (9) If -n would be a transitivity marker it should, for exarnplc, show up with extraction of the thème in the prcpositional ditransitive construction. This. however, is not the case. The verb in (i) appcars as intransitive (sec also (8)) as in noun incorporation cxamples ((i) in fn. 4)) rcirci' (i) Tera ac a anga nakon Rui taan teachcrs what foc 3pl. give to Rui DET "What did the teachers give to Rui?" The passive (see (14a). ( 15) and (24b)) and the prepositional ditransitive construction (i) are the onlv extraction contexts in Kiribati that 1 am aware of in which an extractcd object has not a correlatcd agreement marker on the transitive vcrb. (i) could be analvzcd as a passivizcd applicativc (with a demoted (optional) recipient argument. Marc Richards. pcrsonal comrmuucatiom. providing the hasis for a unificd analysis of both cases. The sarnc tacts as with anga "givc" are obscrved with the vcrb "tel!" IU(/ (intrans.) that appcars as a root in the prepositional ditransitives and with the suffix -ng as tua-ng in the double object construction (tuang-ng-a "tell- APPL-3SCî")Note that only anga "give" and tua "tel!" allow for the double object construction in Kiribati. Benefactivc ditransitives. for example. appear onlv in the prepositional ditransitive construction. Recall that the object agreement marker in double object constructions behaves likc the transitivity marker in 3SG in Kiribati (cf. the discussion related to (6.ii)) The object marker and the phonetically realized NP can nat co-occur, except the object NP is covcrt (pro) or extractcd (cf the discussion of examples (12a) YS. (25a)). 6 pared (19) Q: A vagan-i-a '0 qasenivuli [na cava] '0 3sGgive-TR-3sG DET teacher DET what DET 'What does John give to the teacher?' A vagania '0 qasenivuli [na ivola] '0 Jone. 3sG give- TR-3sG DETteacher DH book DET John (Rec > Th, *Foc > Background) A vagania [na ivola] '0 qasenivuli '0 Jone. 3sG give-TR-3sG OH book OH teacher DFT John (*Rec > Th, Foc> Background) A 3sG Jone? John [Fijian] vagan-i-a ['0 cei] na ivola '0 Jone? DH whom DET book DET John 'Whorn does John give the book?' a. A vagania ['0 qasenivuli] na ivola '0 3sG give- TR-3sG DH teacher DEl book DH Jone, (Rec > Th, Foc> Background) John b. # A vagania na ivola ['0 qasenivuli] '0 3sc; give-rn-Jso DET book DFT teacher OH Jone. (*Rec > Th, *Foc > Background) John give-rn-Jsc The following examples illustrate the order of adjuncts in Kiribati and Fijian, This order is fixed, as in other configurational languages. The data suggests again that sentences and DPs have the full set of functional categories and a fulIy-fledged hierarchical structure. Consider, for example, the order of manner (MA) and frequency adverbs (FA). The only possible order in English is FA> L IE==- Configurationality, Joachim Sabel 14 a. John has always washed the clothes weil. b. * John has weil washed always the clothes. (21) a. (22) kunnikai raoi uatii E weil clothes washes 3SG Ali: 'Rui always washes the clothes n tai b. E uatii kunnikai raoi c. * E uatii n tai nako kunnikai d. * E uatii kunnikai n tai nako a. b. c. d. e. (24) '0 DET na liku vakaviti (draunikau) damudamu DET skirt Fijian leaf red "the red Fijian skirt made from leaves" aomata]. [[1' a] [vp <aomata> tangira te anene] 3PL loves DET music people "The people love the music." [[1'e/*a/*0] tangir-aki te anene (irouia aomata)]. 3SG/3PL/0 was-Ioved DET music (by people) "The music was loved by the people." [Kiribati] Tuming now to the object markers, a first argument for their nature as agreement morphemes cornes from the fact that in many languages a verb can show object agreement with the NP bearing the goal/benefactive role but not with the NP bearing the theme role (see, for example, Baker 2008: 43 for Chichewa). As already mentioned at the end of section 1.l,the same effect can be observed in Kiribati. (25a) is excluded because the object marker and the phonetically realized NP can not co-occur in post-verbal position, but only if the object is realized as pro or extracted (cf. (12a), the last paragraph of section 1.1 and the discussion in fn. 7). (25b) is grammatical as it does not contain the object marker. The interesting case involving agreement is (25c) compared to (25d). Given that in each of these examples one of the objects is silent, the condition for realizing the object marker is met. ln (25c--d) the probe v and the objects match in pers on and number features. Both objects are c-commanded by the probe v. However, only (25c) is possible, i.e. the object marker may only agree with the recipient. This asymmetry follows from the fact that the recipient is structurally higher than the theme and asymmetrically c-commands the theme. The recipient intervenes between v and the theme, rendering agreement with it impossible. Given the pattern in (25) it seems natural to analyze the object marker as an agreement marker. Jone. John [Fijian] ln Kiribati, there is no difference between attributive adjectives and relative constructions. Fijian, however, has attributive adjectives and they do not appear in free but only in a certain linear order. (23) a. b. nako Rui. n tai in time every Rui weil' nako Rui. Rui. raoi [Kiribati] Rui. raoi vakacau sulu sava leya i Sa often TNS/PROGwash wrong DET clothes Ail: 'John often washes the clothes wrong.' '0 Jone. Sa sava i sulu leya vakacau '0 Jone. * Sa sava vakacau i sulu leya '0 Jone. * Sa sava i sulu vakacau leya '0 Jone. i sulu * Sa sava vakacau leya 15 thematic but syntactic) subject in the passive (24b). This is expected under the agreement marker analysis but unexpected under the analysis of the subject pronominal clitic as B-marked pronoun.? MA (20a), but MA> FA in Kiribati (Z l a-b) and Fijian (22a-b). Other orders are excluded (cf. (20b), (21c--d), (22c-d), see Sabel2011 for more discussion). The examples in (21 }-(22) show that the linear order of adverb(ial)s is strictly constrained in Fijian and Kiribati, as in English (20). (20) Successive Cyclic Movement and Object Agreement *(draunikau) leaf [Fijian] The discussion in this section has shown that the order of arguments and adjuncts in Fijian and Kiribati is restricted as in configurational languages such as English which means that sentences and DPs have the full set of functional categories and a fully-fledged hierarchical structure. Kiribati and Fijian are not "pronominal argument languages" and the subject and object clitic elements are not pronominal arguments but agreement markers. An independent reason for analyzing the subject pronominal clitic as an agreement marker is suggested by agreement facts in passive constructions. Following the traditional analysis of passive sentences, the derived syntactic subject is base-generated as complement of the verb. If the subject clitic marker was 0marked by the verb in (24b) and (15) above we would expect that (i) it should either show up as a 3PS plural marker in active and passive sentences and block NP-movement of the theme object in the passive construction, or that (ii) the subject marker should be absent in the passive construction. Both predictions are not borne out. The subject marker agrees in the active sentence (24a) with the (thematic and syntactic) plural subject but with the derived singu1ar (non- (25) a. * [rp 1 [vp angan-n-a [vp teuaarei [v' <verb> 1SG give-APPL-3sG DET-man <pro>2] pro2]. ISG "1 give the man the moncy." b. [TP 1 [vp anga-n [vp teuaarei [y <verb> te ISG give-APPL DET-man DET <pra>2] pra2]. ISG "1 give the man the money." te DET m'ane]] money m'ane]] money 7 On indcpendent grounds, Hale (1998) reaches a similar conclusion for subject markers in Marshallese, the most closely related language to Kiribati. Marshallese. however, is an SYO language. like ail other Micronesian languages. The verb-initial character of Kiribati has been argued to be a result oflanguagc contact with the verb-initial Polynesian language Tuvaluan. L ------~ Configurationality, Joachim Sabel 16 c. [vp pro 1 [IP 1 [vp angan-n-a give-APPL-3SG 3SG ISG <pro>2] [\. <verb> te DIT (27) Prol]] 3SG <Pro>2] [Kiribati] agreement elements. (28) 3 Verbal abject agreement and long extraction Object agreement in Kiribati resembles object agreement in active sentences with participles in French that appears with overt object movement as in (26b). a. b. Paul a repeint(-*es) Paul has repainted(-3PLlFEM) Quelles chaises Paul a-t-il which chairs Paul has-t-he Ko noor-t-ra aomata 2SG see-Tr-Jri. people "Which people did you see?" Aomata ra aika ko which FM people 2sG ra? which noor-i-ia <aomata ra> ? see-Tr.-3PL [Kiribati] The analysis of the object marker as a pronoun wrongly predicts (27b) to be ungrammatical, because it would represent a weak crossover effect (Rizzi 1986). ln Kinande, for example, where the object marker is a pronoun, the example corresponding to (27b) is impossible (Baker 2003). Again, the agreement marker analysis for Kiribati (and Fijian) makes the right prediction. 1 assume that the trigger for wh-ex situ or focus-fronting is an (optional) selected strong [+focus]-feature or edge feature in a focus projection (associated with the focus marker), and that this feature attracts the moved wh-phrase, which ends up in the specifier of a focus phrase headed by FM and being selected by the C-phase head (see Sabel 2000b for this analysis). C may be filled with the subordinator bwa "that" in embedded sentences.? With complement sentences, the matrix verb agrees with the embedded sentence: The discussion in this section has provided evidence that NPs in Kiribati/Fijian show restrictions similar to NPs in configurational languages. The argument positions in vPIVP are occupied by the full NPs as in languages such as English, French or German. It is possible to have a post-verbal object together with an object marker, or a preverbal object without one (as in the PP-ditransitive and passive construction), i.e, both elements are not in complementary distribution. NPs are not adjuncts that can be freely ordered, and this holds for an analysis in which the subject and object marker is assumed to occupy an argument position and to bear a B-role or for an analysis in which it is assumed that the argument positions can only be occupied by empty argumentai pro and the markers are (26) a. b. ISG "1 give the man it." 17 Kiribati is an optional wh-in-situ language. (27) shows wh-in and wh-ex situ of an object. m'ane)) money Pro2]' 1 SG "1 give him the money." d. * [rp 1 [vp angan-n-a [vp teuaarei [v <verb> 1SG give-APPL-3SG OH-man pro2). Successive Cyclic Movcmcnt and Object Agreement quelles chaises? which chairs repeint*( -es) <quelles chaises>? repainted(-3pLlrEM) [French] Ti ata-i-a (bwa) e 1PLknow- Tr.-3SG (that) 3SG Rui. Rui "We know that Rui loves Mary b. * Ti ata-i-ia (bwa) e l rt.know-Tr-Spt, (that) 3SG Rui. Rui "We know that Rui loves Mary a. Meeir ao Mary and Tien Tien and Tien." tangir-i-ia Meeiri ao loves- Tr-3PL Mary and Tien Tien tangir-i-ia loves-Tr-Jrt, and Tien." [Kiribati] Now consider the following sentences with long foc us movement. ln this section, 1 discuss agreement effects with A' -rnovernent in Kiribati and Fijian. This phenomenon provides independent evidence for the agreement marker analysis, as weil as for successive cyclic movement through phase-edges. It will be illustrated that the object agreement marker acts differently from the verbal marking found in Western Austronesian languages. ln the case of long extraction in Fijian/Kiribati the object marker on higher verbs is interpreted as co-referential with the extracted element (and not with the clause it is extracted from). This difference is due to the fact that Kiribati and Fijian verbal object markers are not voice but agreement markers.» sentence initial position and its base position. With clause-internai extraction. verb morphology encodes the type of argument that is extracted. With long extraction. the morphology appears on ail verbs intervening bctween the wh-element in the highest clause and its base position. With sentence internai extraction the WA languages differ with respect to the number of different verbal morphcmes they use for different cxtractees (agent, recipient. theme. instrument ... ) Turning to long extraction. it can be observed that verbs in higher clauses do not mark the type of the extracted element hut uniformly the argumenthood of the clause from which extraction has taken place. This phenomenon IS obscrved in long wh-movernent. rclativization. topicalization and focus preposing and has bccn analyzed as "wh-agreement" (Chung 1998. among ethers). as an effect of "voice-marking," or as a reflex of Case-agreement (Rackowski and Richards 2005). Q (i) is the structure for a matrix or embeddcd wh-question with wh-ex situ. Focused XPs targe! the same position as wh-phrases. (i) .. (V) [cPI'QI (bwa) [FocP wh [F"" aika l'FPpd [TP [.p <wh> V-OM <wh> subjectllll 8 ln Western Austronesian (WA) languages such as, for example, Chamorro, Indonesian, Malagasy, and Tagalog a special voice marking appears on verbs on the path between a wh-element in L Configurationality, Joachim Sabel 18 ln (29a-b) the FM agrees correctly with the extracted focus constituent and also the embedded verb correctly agrees in person and number. ln (29a) even the matrix verb agrees with the long extracted DP. (29b) is ungrammatical because the matrix verb does not bear the agreeing object marker. Similar verbal object marker effects can be observed with focus fronting and Dual-marking in Fijian. (30) a. b. *E kilai-rau 0 John ni e lornani-rau 3SCiknow-DUAL DET John that 3SG loves-DUAL Bill kei Bob 0 Mary. Bill and Bob DH Mary 0 Bill kei Bob e kilai-rau 0 John ni lornani-rau < > 0 Mary 0 DEr [Fijian] The examples illustrate a morphological effect of long successive cyclic movement in Kiribati/Fijian. The matrix verb has to agree with the long extracted element. Kiribati/Fijian differ in this respect from object agreement with active verbs in French where object/participle agreement is local (clause-internaI) and may not cross CP, cf. (26) ys. (31):10 (31) Quelles chaises which chairs Marie a-t-elle Mary has-t-she dit/*-es said-SG/-3pLlI'EM que that Cyclic Movement and Object Agreement 19 We can derive the object agreement phenomena by assuming feature-movement, i.e. e-feature splitting as assumed in Sabel (2000a) for the analysis of thereexpletive constructions, in Chomsky (2008) for an account of CED extraction asymrnetries, and as in Obata (2010) to account for improper movement and special agreement effects in Bantu. ln ail these works, it is assumed that o-features may move out of a OP to check an EPP feature. ln Kiribati, for example, o-feature movement checks an EPP feature on v and the movement operation results in the phonetic realisation of the object marker on the verb in standard examples such as (6i) or (7i). Now consider (27b). Here, after o-feature movement, the remnant DP in base position still contains an [i(interpretable)Foc!Q] that is attracted together with the phonetic features of the wh-phrase by an edge feature in Foc, as in (27b). ln (29a), (30b) the remnant focus-phrase moves from the base position to the phase edge of C. Being located in Spec of CP, the [i<p]in OP can undergo <pprobing by the matrix v and then the wh-phrase moves to Spec vP and being equipped with an [iFoc/Q] feature it is further extracted to check the edge feature in FocP. To explain the difference between French and KiribatiiFijian, 1 adopt Obata's (20 JO) idea that, as a parametric property the [uCase] on D P can be separated from the [<p[-set in sorne languages. ln contrast to French, Kiribati allows for this option, leaving the DP "active" after o-checking in the embedded sentence. Note that under this analysis agreement patterns can be analyzed as an effect of successive cyclic movement. The examples (29), (30b) provide evidence for movement through vP and hence for phase theory. Recall that the passivized verb (participle) in (14a), (15) and (24b) does not have an object agreement marker (for an analysis of object agreement phenomena in passives in Indo-European languages such as Swedish, Norwegian, and French, see Holmberg 2002 and Richards 2009). We can explain this fact as follows: in passives and prepositional ditransitives, v is defective, i.e. it is occupied by the passive morpheme and does not value the [uCase] on the object OP which excludes raising of the e-features to v triggering object agreement. The multiple occurrences of agreement in (29a), (30b) provide a further argument against the analysis of the object marker as a pronoun. Furthermore, the phenomenon observed with object agreement in this section also shows up with extracted subjects: it is therefore reasonable to assume that the subject agreement marker in Kiribati/Fijian is derived by feature-splitting in the same way as the object agreement marker. Meeiri ao Tien aika ti ata-i-ia bwa e Mary and Tien FM 1PL know-Tr.-3PL that 3sg tangir-i-ia < >Rui. loves- Tr.-3 PL Rui "lt is Meeri and Tien that we know that Rui loves." b. * Meeir ao Tien aika ti ata-i-a bwa e Mary and Tien FM PL know- Tr.-3SG that 3sG tangir-i-ia < >Rui. [Kiribati] loves-Tr.-3PL Rui a. (29) Successive Paul Paul a repeint-es? has repainted-3PLlFEM Given the analysis of Case and agreement checking in Chomsky (2000, 2001, 2007, 2008), (29a) and (30b) should be impossible. The lower v c-probes the object and values as a reflex the [u(interpretable) Case] on the object OP. I assume here that probing by v precedes feature-inheritance from v-to-V Given that the [uF] on the object is eliminated, the object is rendered inactive and inaccessible for further operations such as œ-probing by the matrix v. (29a) and (30b) should be as impossible as the French example (31) with overt object agreement on the matrix verb. The data suggest that o-feature checking and Case checking do not correlate in Kiribati and Fijian. 4 Concluding Scrambling Remarks: Fijian and Kiribati as Configurational Languages To sum up, several phenomena have shown that Fijian and Kiribati are configurational languages. The verbal markers in these languages are agreement morphemes, as in Indo-European languages. Consequently, the free word order A comparable phenomenon as in Kiribati is found, for example. in the form of tonal reflexes in languages such as Kikuyu. ln Kikuyu. a long A' -moved element triggers special tonal patterns on every predicate that appears in between the base position of the extracted element and its goal position (see, for example, Clements 1984. Sabel 2000b. Lahne 2009). 10 L -, ,. Joachim Sabel 20 Configurationality, Scrambling Generalization If a language has scrambling, then it also has morphological Alderete. J. 1998. Canonical types and noun phrase configuration in Fijian. ln: Matt Pearson (cd.), ('(LA Occasional Papers in Linguistics 21, Proceedings of the third and fourth meetings of the Austronesian FormaI Linguistic Association (AFLA III-IV), 19-44. Los Angeles, CA: UCLA Department of Linguistics Alexander, M. 1990. Agreement configurations: grammatical relations in modular grammar. Ph.D. thesis, MIT Aranovich, Raul 2004. Incorporation, Pronominal Arguments. and Configurationality. Ms. University of California, Davis. Baker. Mark 1995. The Polysynthesis Parameter. Oxford: Oxford University Press Baker. Mark 2001. The Natures of Nonconfigurationality. ln: Mark Baltin and Chris Collins (eds.), The handbook of contemporary Syntactic Theory, 407-438. Oxford: Blackwell Publishers. Baker. Mark 2003. Agreement, Dislocation and Partial Configurationality. ln: Andrew Carnie, Heidi Harley and MaryAnn Willie (cds.). Forma! Approaches to Function in Grammar: ln Honor of Eloise Jelinek, 107-134. Amsterdam: Benjamins. Baker, Mark 2008. The Syntax of Agreement and Concord. Cambridge: CUP. Barss, Andrew and Howard Lasnik 1986. A note on anaphora and double objects. Linguistic Inquiry 17: 347-354. Boskovié. Z. 2005. Left branch extraction. structure of NP. and scrambling. ln: J. Sabel and M. Saito (eds.), Thefree word order phenomenon: Its syntactic sources and diversity, 13-73. Berlin: Mouton de Gruyter. Chomsky, Noam 1973. Conditions on Transformations. ln: S. R. Anderson and P. Kiparsky (eds.). A Festschriftfor Morris Halle, 232-286. New York: Holt. Chomsky. Noam 1995. The Minimalist Program. Cambridge, MA: MIT Press Chomsky. Noam 2000. Minimalist Inquiries: the Framework. ln: R. Martin, D. Michaels and J. Uriagereka (eds.). Step by step, 89-156. Cambridge, MA: MIT Press. Chomsky. Noam 2001. Derivation by Phase. ln: M. Kenstowicz (ed.), Ken Hale: A Life in Language. 1-52. Cambridge. MA: MIT Press. Chomsky. Noam 2007. Approaching UG from below. ln: U. Sauerland and Il.-M. Gârtner (eds.), Interfaces' Recursion = Language", 1-30. Berlin: Mouton de Gruyter, Chomsky. Noam 2008. On Phases. ln: R. Freidin, C. P. Otero and M.-L. Zubizarreta (eds.), Foundational Issues in Linguistic Theory, 133-166. Cambridge, MA: MIT Press. Chung, S. 1998. The design of 'agreement ': evidence from Chamorro. The University of Chicago Press. Clements. George 1984. Binding Domains in Kikuyu, Studies in the Linguistic Sciences 14, 37-56. Cowell. Reid 1951. The Structure of Gilbertese. Rongorongo Press. Fukui. Naoki 1993. Parameters and Optionality. Linguistic lnquiry 24: 399-420. Geraghty, Paul 1983. The history of the Fijian languages. Ol. special publication No. 19. Honolulu: University of Hawaii Press. Haider, Hubert 1988. 8-Tracking Systems - Evidence from German. ln: L. Maràcz and P. Muysken (eds.), Configurationality. 185-206, Dordrecht: Foris. Haider, Hubert 1997. Scrambling - Locality, Economy, and Directionality. ln: Shigeo Tonoike (ed.), Scrambling, 61-91. Tokyo: Kurosio (Linguistics Workshop Series). Hale, Kenneth L. 1983. Warlpiri and the Gramrnar of Non-configurational Languages. Natura! Language and Linguistic Theory L 5-47. Hale. Mark. 1998. Diachronie aspects of Micronesian clause structure. Canadian Journal of Linguistics 43: 341-357. Case on NPs. Obviously, this generalization is not fulfilled in Fijian and Kiribati. But we know already that other languages such as Bulgarian, for example, behave like Fijian and Kiribati and thus are also in conflict with the generalization (32). Therefore, the correlation between overt Case morphology on NPs and the scrambling property has to be stated as follows: (33) Scrambling Generalization: If a language has morphological 21 References phenomenon in these languages has to be seen as the result of a scrambling operation that applies to a configurational base structure. That this conclusion is on the right track is confirmed by the fact that the discussed languages fit weil into the typological classifications of the scrambling phenomenon. First, scrambling languages show different locality restrictions on the scrambling operation. For example, scrambling out of finite complement clauses is possible in languages such as Hindi, Japanese, Korean, Mohawk, Persian and Russian, but not in Dutch, German, Polish, Warlpiri, Fijian and Kiribati, where scrambling is only possible from complement infinitivais. Second, it has been claimed that the licensing of scrambling is connected with the head parameter (Fukui 1993, Haider 1997, Saito and Fukui 1998) and third, that pro-drop is a necessary condition for a language for scrambling to be possible. Fijian and Kiribati fulfil both licensing conditions. ln addition, however, some authors have claimed that overt Case morphology on NPs is a nec essary property relevant for the availability of scrambling (Sapir 1921, Haider 1988, Boskovié 2005, among others): (32) Successive Cyclic Movernent and Object Agreement Case on NPs, then if also has scram- bling. The implication (32) is compatible with the situation in Bulgarian, Fijian, Kiribati, ... as weil as with the situation in non-scrambling languages such as English. However, it raises the question as to why a language such as English does not have scrambling. The difference between ail the mentioned languages without overt Case morphology on NPs needs to be explained diachronically. Old Bulgarian, Old English and even Proto-Austronesian languages had rich Case morphology on NPs and scrambling. The morphological markers were lost (for idiosyncratic reasons) in the history of Bulgarian, English, Fijian and Kiribati, but for some independent reason the scrambling property was lost only in English. L 22 Joachim Sabel Hinweise für Autorinnen und Autoren Harrison, Shelly 1978. Transitive marking in Micronesian languages. Pacifie Linguistics, Series C-61: 1067-1127. Holmberg, Anders 2002. Expletives and Agreement in Scandinavian Passives. Journal of Comparative Germanie Linguistics 4: 85-128. Jacobs, Roderick A. 1984. Sorne Syntactic Pro cesses in Kiribati. ln: Bender, Byron W. (ed.), Studies in Micronesian Linguistics, Pacifie Linguistics C-80, 1984, 467--490. Dept. of Linguistics, Reasearch School of Pacifie Studies, ANU, Canberra. Jayaseelan, K. A. 2008. Topic, Focus and Adverb Positions in Clause Structure. Nanzan Linguistics 4, 43--68. Jelinek, Eloise 1984. Empty Categories, Case, and Configurationality, Natural Language and Linguistic Theory 2,39-76. Lahne, Antje 2009. Where There is Fire There is Smoke: Local Modelling of Successive-Cyclic Movement. Ph. Diss.: Univ. of Leipzig. Lynch, John 1998. Pacifie languages: An introduction. Honolulu: University of Hawaii Press. Lynch, John, Malcom Ross and Terry Crowley 2002. The Oceanic Languages. Richmond: Curzon. McGinnis, Martha. 2001. Phases and the syntax of applicatives. ln Proceedings of HELS 3 L ed. Min-Joo Kim and Uri Strauss. GLSA, University of Massachusetts, Amherst, 333-349. Obata, Miki 2010. Root, Successive-Cye/ic and Feature-Splitting Internai Merge: Implications for Feature-Inheritance and Transfer. Ph. Diss.: Univ. of Michigan. Pensai fini, Robert. 2004. Towards a typology of configurationality. Natural Language and Linguistic Theory 22: 359--408 Pylkkânen, Liina. 2002. Introducing Arguments. Ph. Diss. MIT Rackowski, Andrea and Norvin Richards 2005. Phase Edge and Extraction: A Tagalog Case Study, Linguistic lnquiry 36(4),565-599. Richards, Marc 2009. On Feature Inheritance, Defective Phases, and the MovementMorphology Connection, ms. Univ. of Leipzig. Rizzi, Luigi. 1986. On the status of subject clitics in Romance. ln Studies in Romance Linguistics, ed. O. Jaeggli and C. Silva-Corvalan, 391--420. Dordrecht: Foris. Rizzi , Luigi 1990. Relativized Minimality. Cambridge, MA: MIT Press. Sabel, Joachim 2000a. Expletives as Features. WCCFL 19 Proceedings, R. Billerey and B. Lillehaugen (eds.), 411--424. Somerville: Cascadilla Press. Sabel, Joachim. 2000b. Partial Wh-Movernent and the Typology of Wh-Questions. ln: Uli Lutz et al. (eds.), Wh-Scope Marking, 409--446. Amsterdam: John Benjamins. Sabel, Joachim 2011. Deriving Linear Order in OVIVO Languages: Evidence from Oceanic Languages. ln: C. Moyse-Faurie and J. Sabel (eds.), Topics in Oceanic Morphosyntax, 2764, Berlin: Mouton de Gruyter. Saito, Mamoru and Naoki Fukui 1998. Order in Phrase Structure and Movement. Linguistic lnquiry 29: 439--474. Sapir, Edward 1921. Language. An introduction ta the study of speech. New York: Harcourt. Trussel, Stephen 1979. Kiribati (Gilbertese) Grammar Handbook. Peace Corps Language Handbook Series. Louvain-la-Neuve Alle redaktionellen Zuschriften und Sendungen erbitten wir an die Redaktion der Linguistischen Berichte: Linguistische Berichte z. Hd. Herm Prof. Dr. Günther Grewendorf Johann Wolfgang Goethe-Universitât Fachbereich Neuere Philologien, Institut für Linguistik Grüneburgplatz 1 D-60629 Frankfurt am Main E-Mail: Ib@lingua.uni-frankfurt.de Wir bitten darum, jedes Manuskript in dreifacher Ausfertigung einzureichen; die Kopien sind für die Begutachtung erforderlich, der jedes Manuskriptangebot unterzogen wird. Bevorzugte Sprache der Beitrâge ist Deutsch; englisch- und franzôsischsprachige Beitrâge werden akzeptiert, sofem sie stilistischen Standards genügen. Stellen Sie Ihrem Aufsatz eine kurze Zusammenfassung scher Sprache voran. (abstract) in engli- Namen und Vomamen aller (Ko-)Autor(inn)en sollen auf dem Manuskript vermerkt sein, einschlieJ31ich der Adresse (mit Telefonnummer), an die die Korrekturfahnen geschickt werden sollen. Die Autor(inn)en erhalten 20 Sonderdrucke ihres Beitrags. Autor(inn)en, deren Beitrâge zur Verôffentlichung vorgesehen sind, erhalten von der Redaktion ein Merkblatt mit Richtlinien zur Manuskriptformatierung, deren Berücksichtigung unbedingt erforderlich ist. Der Beitrag ist anschlieBend in elektronischer Form (E-Mail-Anhang) im Word-Format oder ais rtf-Datei an die Redaktion zu senden. Zudem werden Autor(inn)en gebeten, der Redaktion einen inhaltsgleichen Ausdruck des Beitrags per Post zukommen zu lassen. Alle Besprechungsexemplare von neu erschienenen Werken der Sprachwissenschaft und eng verwandter Disziplinen schicken Sie bitte an die Redaktion. Die Auswahl behâlt sich die Redaktion vor. Rücksendungen unverlangt eingesandter Bücher kônnen nicht vorgenommen werden. Joachim Sabel Université catholique de Louvain, Faculté de philosophie, arts et lettres, 1, Place Blaise Pascal, 1348 Louvain-la-Neuve, BelgiquelBelgium E-Mail: sabel@lige.ucl.ac.be l Mitteilungen für die LB-Info schicken Sie bitte an Klaus Müllner, Postfach 21 51,65779 Kelkheim, kmuellne@uni-mainz.de. LB 233/2013 Inhalt Linguistische Berichte Heft 233 Beitrâqe aus Forschung und Anwendung Herausgegeben von Günther Grewendorf und Arnim von Stechow Sprachtypologie Joachim Sabel Configurationality, Successive Cyclic Movement and Object Agreement in Kiribati and Fijan 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ;; 0 0 3 ~ a., h, t~ Sprachprodu ktion Laura Di Venanzio Ein Dilemma - Der funktionale Kopf in deutschen Selbstreparaturen 23 000000 Semantik Jennifer Rau Wie faktiv sind faktive Prâdikate? 00000000000000000000000000000000000000 •• 0 0 0 0 0 0 00 0 0 0 0 0 0 51 Syntax John Rote Velde Meeting Interface Requirements on Coordinate Structures 69 00000000000000000000 Rezension Andreas Pankau Paul Mo Postal: Edge-Based Clausal Syntax 109 00000000000000000000000000000000000000000 Informationen und Hinweise LB-Info 0000000000000000000000000000000000000000000000000000 Hinweise für Autorinnen und Autoren 0 000000000000000000000000000000000000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 •• 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 121 125 fl3 BUSKE View publication stats