Variation in Bangla complementizer order at the syntax-prosody interface Brian Hsu (University of Southern Calfornia) Introduction A long-standing question in the analysis of second-position effects is the degree to which they result from syntactic versus phonological factors (for recent discussion, see Werle 2009; Holmberg to app.). I argue for an interactionist view à la Bošković (2001) where word order is determined in syntax but constrained by prosodic conditions. I present data from Bangla, where the finite complementizer je is obligatorily non-initial in certain embedded clauses. Although the pattern resembles traditional secondposition effects, it poses challenges to purely syntactic or phonological accounts. While there are clear prosodic but not syntactic motivations for non-initial complementizer placement, interpretational restrictions on phrases preceding non-initial je indicate that the order is generated in syntax. I make two principle claims: [1] Both initial and non-initial complementizer orders are derived within the extended CP domain (Rizzi 1997), and depend on whether je is pronounced in Force or Fin; [2] Lower copy spellout in Fin applies to avoid placing the complementizer at an intonational phrase edge. The Bangla je puzzle Finite embedded clauses are optionally introduced by je. In a neutral context, the embedded clause is postverbal with respect to the main verb, and je is obligatorily initial within its clause. (1) John jane [(je) ma (*je) oSudh John knows [(that) mother (that) medicine 'John knows that mother took medicine.' (*je) (that) khey-eche] eat-PERF] As noted by Bhattacharya (2001) and Simpson & Bhattacharya (2003), focused or topicalized embedded clauses are preverbal. In these contexts, je is obligatorily preceded by at least one phrase within its clause. (2) Jon [(*je) ma (je) oSudh John [(that) mother (that) medicine 'John knows that mother took medicine.' (je) (that) khey-eche] eat-PERF] (3) [(*je) ma (je) oSudh (je) [(that) mother (that) medicine (that) 'John knows that mother took medicine.' khey-eche] Jon eat-PERF] John jane knows jane knows While preverbal je has been viewed as a kind of topic or focus marker (Bhattacharya 2001), I argue that it is better analyzed as a non-initial complementizer since it is associated uniquely with declarative and finite clauses. Furthermore, I show that embedded clause subjects that precede je are not obligatorily topicalized or focused. The question is then why initial-je orders are required for postverbal CPs, while only non-initial-je orders are permitted preverbally. Topic, focus, and non-initial je In preverbal clauses, fronted object arguments that precede je are subject to interpretational restrictions. Possible topics like definite DPs readily precede je (4). However, indefinite or quantified objects can only appear in this position if clearly focused (5-6). (4) John [chatro du-to -ke je dadubhai dekh-eche] John student 2-CL -ACC that grandfather see-PERF 'John said that grandfather saw the two students' bol-lo say-PST (5) *John [kau-ke John anyone-ACC bol-lo say-PST (6) je that dadubhai dekh-e grandfather see-PERF ni] John [KAU-KE je dadubhai dekh-e John anyone-ACC that grandfather see-PERF 'John said that grandfather didn't see ANYONE.' ni] NEG NEG bol-lo say-PST In a pattern that mirrors some V2 phenomena (Holmberg to app.), embedded clause subjects do not need to be topicalized or focused, since non-focused, quantified subject DPs freely precede je. (7) Jon [kew je ase-ni] bol-lo John [anyone that come-NEG] say-PST 'John said that no one came' Proposal I claim that all embedded clauses share a single derivation within the split-CP structure of Rizzi (1997); The complementizer je is first merged in Fin, and a copy is merged in Force. This derives the fact that je is in all occurrences specified as both [+finite] and [+declarative]. By default, only the highest copy is pronounced, generating the je-initial order of postverbal clauses. Word order in preverbal clauses is derived by spell-out of je's lower copy in Fin, with at least one phrase present in TopicP, FocusP, or FinP. (8) [ForceP (jei) [TopicP (XPtop) [FocusP (XPfoc) [FinP (XPsubj) (jei) [IP ... The lower copy spell-out analysis correctly predicts that je is separated from the clause edge by more than one phrase if there are multiple topics or if TopicP and FocusP are simultaneously filled. It also accounts for their ordering restrictions; topics precede focus, and the focused phrase precedes je (9). Such patterns are crucially not predicted by a more local operation like Morphological Merger (Embick & Noyer 2001). (9) John [Meri Borders-e kal HÆMLET je kin-eche] John Mary Borders-LOC yesterday HAMLET that buy-PERF 'John knows that it was HAMLET that Mary bought yesterday at Borders' jane knows (S&B 2003) The role of prosody I argue that lower copy spell-out takes place uniquely to avoid the pronunciation of je at the edge of an intonational phrase. Following the diagnostics of Khan (2008), differences in attested boundary tones and the availability of pauses show that preverbal CPs are parsed as separate intonational phrases (IntPs) whereas postverbal CPs correspond to a smaller intermediate phrase. The attested intonational phrasing is as in (10): (10) a. Postverbal CP [S V [CP ... ]] (IntP ) b. Sentence-medial CP [S [CP ... ] V] (IntP (IntP ) ) c. Sentence-initial CP [ [CP ... ] S V] (IntP(IntP ) ) Thus, je is obligatorily non-initial uniquely when its CP is a separate intonational phrase. I propose that lower copy spell-out occurs to satisfy a phonological restriction against small function words at the left edges of intonational phrases, mirroring patterns observed in Slavic clitic placement. In the constraintbased framework of Selkirk (1986), this results from a constraint that requires the left edges of intonational phrases to be aligned with minimally bimoraic prosodic words, ALIGN-L(INTP, PWD); vowel lengthening patterns show that je can not form a minimal word, and as such can not be initial within a preverbal CP. Conclusion This paper presents a novel analysis of preverbal Bangla je as a non-initial complementizer whose position is derived through lower copy spell-out. The data expands the typology of secondposition-like effects; it provides further evidence that such orders arise in the extended CP-domain (cf. Roberts 2012) through the influence of phonological constraints on prosodic organization. Selected references Bhattacharya, T. 2001. Peripheral and clause-internal complementizers in Bangla : A case for remnant movement. Proceedings of WECOL 2000: 1–14. Bošković, Ž. 2001. On the nature of the syntaxphonology interface: Cliticization and related phenomena. Oxford: Elsevier. Embick, D., and R. Noyer. 2001. Movement Operations after Syntax. Linguistic Inquiry 32: 555–595. Holmberg, A. Verb second. To appear in Syntax – an International Handbook of Contemporary Syntactic Research, ed. T. Kiss and A. Alexiadou. Walter de Gruyter Verlag. Khan, S. ud D. 2008. Intonational Phonology and Focus Prosody of Bengali. Ph.D diss., University of California, Los Angeles. Rizzi, L. 1997. The fine structure of the left periphery. In Elements of Grammar, ed. L. Haegeman, 281–337. Dordrecht: Kluwer. Roberts, I. G. 2012. Phases, head movement and second-position effects. In Phases: Developing the Framework, ed. A. Gallego, 385–440. Berlin: De Gruyter Mouton. Selkirk, E. 1986. On derived domains in sentence phonology. Phonology Yearbook 3: 371–405. Simpson, A., and T. Bhattacharya. 2003. Obligatory overt Wh-movement in a Wh-in-situ language. Linguistic Inquiry 34: 127–142. Werle, A. 2009. Word, Phrase, and Clitic Prosody in Bosnian, Serbian, and Croatian. Ph.D diss., University of Masschusetts, Amherst.
© Copyright 2024 ExpyDoc