Chain formation as inclusion relations between copies Carlos Muñoz Pérez Universidad de Buenos Aires & Georg-August-Universität Göttingen The Copy Theory of Movement (Chomsky 1995) aims to explain the so-called displacement phenomena in natural language under the assumption that more than one instance of the same constituent may be present in a syntactic representation. The more complete implementation of this theory is due to Nunes (2004), who assumes that the deletion of links in non-trivial chains follows from the LCA (Kayne 1994), since the phonological realization or two (or more) versions of the same syntactic object (SO) would carry a violation of the antisymmetry requirement on linear orderings (i.e., if X precedes Y, then Y cannot precede X). According to Nunes, the copy surviving the deletion operation (Chain Reduction) is the one that has checked more uninterpretable features (uFF), since this choice would allow fewer application of the post-syntactic mechanism that erases those features (FF-Elimination). Phonetic Realization of Traces is one of the phenomena that Nunes’ implementation of Copy Theory explains the best. One of such cases involves multiple copy pronunciation, as it may be attested in Romani (data from McDaniel 1986). (1) Kas misline kas o Demìri dikhlâ? Whom you-think whom Demir saw Nunes argues that this type of pattern may be explained if it is assumed that an intermediate copy may be morphologically reanalyzed as part of some functional head (e.g., a complementizer). Therefore, assuming Chomsky’s (1995) proposal that the LCA does not apply word-internally, the phonetic realization of those copies is explained in terms of economy: since they are not going to violate the antisymmetry condition, it is not necessary to erase them. According to Bošković’s (2002) analysis, another case of pronunciation of traces is attested in Serbo-Croatian, a multiple wh-fronting language in which two adjacent homophone interrogative pronouns cannot be pronounced. (2) a. Ko šta kupuje? b. *Ko kupuje šta Who what buys Who buys what (3)

a.

*Šta šta uslovljava? b. Šta uslovljava šta? What what conditions what conditions what (4) a. Šta neprestano šta uslovljava? b. *Šta neprestano uslovljava šta? What constantly what conditions What constantly conditions what Following a proposal in Franks (1998), Nunes suggests that Chain Reduction applies on the higher copy in a non-trivial chain when it is necessary to satisfy some PF language specific requirement; in this case, Antihomophony (Golston 1995). Thus, a lower copy in the chain should receive phonetic realization instead. Three problems regarding Nunes’ theory. First, if the choice of the copy that is going to survive Chain Reduction is based on the presence of uFF in the lower copies, the standard assumption on the last resort nature of movement cannot be maintained: if movement does not delete the uFF, its application is vacuous in terms of interpretability. (5) a. [TP T [vPDPuFF [v’… ]]] [uninterpretable representation] b. [TPDPuFF [T’ T [vPDPuFF [v’… ]]]] [interpretable representation (e.g., Chomsky 1995)] b’. [TPDPuFF [T’ T [vPDPuFF [v’… ]]]] [uninterpretable representation (Nunes 2004)] Moreover, if the elimination of uFF does not really motivate Move, then their only reason to be is to signalize what copies should be erased. In the second place, the kind of identity relation that Nunes’ system assumes, Non-Distinctiveness, requires the introduction of indexes (or any other formal marking) to identify those SO that are copies and form a single chain, therefore violating the Inclusiveness Condition.

Third, without additional assumptions (e.g., Trace Conversion; see Fox 2002), copy theory predicts erroneously that two copies should be interpreted in the same way. This is clearly false for those cases where the higher copy is interpreted as an operator and the lower copy as a variable (e.g., whmovement). Set-theoretic chain formation. This paper proposes a system based on three assumptions: (i) syntax operates with sets; (ii) phonological exponents are introduced at PF; (iii) there are no uFF, particularly those that Preminger (2011) calls derivational time-bombs. Since the operation Agree (Chomsky 2001) assigns a feature to any Goal that relates with a non-defective Probe (e.g., Case), the new SOs generated by the Copy operation will have systematically more features than their original versions. (6) [XP α{X, Y, Z, …} X … [YP α {Y, Z, …} Y … [ZP Z … α {Z, …} … ]]] It is possible to capitalize this inclusion relation in order to abandon Non-Distinctiveness as the underlying relation to non-trivial chains and to redefine the Conditions on Chain Formation. (7) Two constituents α and β form a chain if: (i) α is a superset of β; (ii) α c-commands β; (iii) there is no constituent γ between α and β such that β is a subset of γ and γ is not a subset of α. Furthermore, this system allows for an elegant explanation of the semantic asymmetries between copies: a higher copy may be interpreted differently from the lower one since it received additional features that changed its meaning. (8) [CP [Which book] [C’CQ+does [TP John read [which book]]]]? Q + D = Operator

D = Definite Description (≈ variable; see Fox 2002)

Finally, the inclusion relation allows deciding which link in the movement chain is going to be pronounced according with the principle in (9). Such a principle follows both from general conditions on the recoverability of information and the Subset Principle (Halle 1997), which presupposes that syntactic terminals are richly specified with features. (9)

Given a Chain CH, apply Vocabulary Insertion to the link which is the superset of all the remaining links. This system captures in a very simple and minimalist fashion the “sameness” relation that holds between the links of a nontrivial chain without losing Nunes’ empirical coverage regarding phonetic realization of traces. First, while in Nunes’ system it is necessary to resort to globalist conditions on PF (Embick 2010) to motivate the erasure of the higher copy in cases as (2b), a system based on Late Insertion can resort to language specific local conditions on Vocabulary Insertion to capture the same patterns (e.g., in Serbo-Croatian, Vocabulary Insertion cannot apply two consecutive times for the same Lexical Item). Second, if as Nunes assumes, the reanalysis responsible for multiple copy realization relies in an application of Fusion, an operation that combines the features of two syntactic terminals, then patterns as (4) receive a straightforward explanation: according to (7), the intermediate copy of kas will be interpreted as the head of a new movement chain in PF: (10) [CP kas{Q, Acc, D, φ,} [C’ CQ …[CP [C + kas]{C, Acc, D, φ,} [TP … [VP V kas{Acc, D, φ,}]]]]] SELECTED REFERENCES. Bošković, Ž. 2002. On multiple wh-fronting. Linguistic Inquiry 33: 351-383 ║ Chomsky, N. 1995. The Minimalist Program. Cambridge, MA: MIT Press ║Chomsky, N. 2001. Derivation by Phase. Michael Kenstowicz (ed.), Ken Hale. A Life in Language. Cambridge, MA: MIT Press ║ Embick, D. 2010. Localism versus globalism in morphology and phonology. Cambridge, MA: MIT Press ║ Fox, D. 2002. Antecedent-Contained Deletion and the Copy Theory of Movement. Linguistic Inquiry 33: 63–96 ║ Franks, S. 1998. Clitics in Slavic. Paper presented at the Comparative Morphosyntax Workshop, Bloomington, Ind. ║ Golston, C. 1995. Syntax outranks phonology: Evidence from Ancient Greek. Phonology 12: 343–368 ║ Halle, M. 1997. Distributed Morphology: Impoverishment and Fission. MIT Working Papers in Linguistics 30: 425-449 ║Kayne, R. 1994. The Antisymmetry of Syntax. Cambridge, MA: MIT Press ║ McDaniel, D. 1986. Conditions on wh-chains. Doctoral Dissertation, City University of New York ║ Nunes, J. 2004.Linearization of chains and sideward movement. Cambridge, MA: MIT Press ║ Preminger, O. 2011. Agreement as a fallible operation. Doctoral Dissertation, MIT.