Date: Fri, 11 Mar 2005 15:18:56 -0600 From: Brent Henderson Subject: Linearization of Chains and Sideward Movement
AUTHOR: Nunes, Jairo TITLE: Linearization of Chains and Sideward Movement SERIES: Linguistic Inquiry Monograph PUBLISHER: MIT Press YEAR: 2004
Brent Henderson, Department of Linguistics, University of Illinois, Urbana- Champaign
Jairo Nunes' book "Linearization of Chains and Sideward Movement" takes seriously two important ideas in recent minimalist theories. One is Chomsky's (1993) notion that movement can be decomposed into the operations Copy and Merge. In this view, a syntactic element to be moved is first copied by the computational system. Then the copy of the item is merged at the root of the derivation in its appropriate place. The second idea, embodied in Takahashi's (1994) Form Chain, is the notion that chains are representational syntactic objects rather than just notational conventions. Under this view chains do not license movement. In fact, the derivational history of an element and its copies is unimportant for the purposes of chain formation. As long as all of the members of a chain are in the proper relationship (c-command) at the end of a derivation, the chain is legitimate.
Nunes' most original idea is recognizing that the adoption of Copy, Merge, and Form Chain as independent operations predicts that certain types of derivations not possible under more traditional notions of movement and chain formation should be attested. In particular, there is nothing in the system to prevent movement relations between parallel, unconnected derivations from occurring as long as the conditions of chain formation are met at the end of the derivation. Such kinds of movement are what Nunes refers to as 'sideward movement,' and they allow an elegant and novel account of problematic constructions such as parasitic gap constructions and across the board movement. Take, for example, a parasitic gap construction such as (1):
(1) Which report did you file __ [without reading __ ]
The two most salient properties of parasitic gap constructions have been widely discussed. First, (1) involves extraction from an adjunct, an extraction generally disallowed across languages. Second, this extraction is disallowed if the argument gap is not also present:
(2) *Which report did you file a book [without reading __ ]
Thus, two questions arise. First, why should extraction out of adjuncts ever be allowed? And second, why should it be allowed dependent upon extraction of an argument from the main clause? Nunes points out that the Copy theory of movement and Form Chain offer a natural answer that does not resort to construction-specific rules or ad hoc solutions.
First, imagine the derivation of the adjunct in (1) has taken place as in (3):
(3) [without reading [which report]]
Now, in parallel, we begin construction of the main clause. However, rather than merging a distinct argument as the object of the verb 'file,' we make a copy of [which report] and allow the copy to be merged as the argument. We will now have two parallel structures as in (4)
(4) a. [without reading [which report]k] b. [you file [which report]j]
Finally, wh-movement of the object of the verb [file] and the two structures in (4) are merged. The final derivation is as in (5). It is at this point that Form Chain applies, constructing two independent chains. One is between the highest copy of [which report] and the copy of [which report] in argument position. The other chain is between the highest copy of [which report] and the copy in the adjunct.
(5) [[which report]j/k did you file [which report]j [without reading [which report]k]]
Since only the highest copy of each chain is pronounced, the result is that only the topmost copy of [which report] gets pronounced while both lower copies are not, yielding the familiar double-gap construction.
With the adoption of Form Chain and the Copy theory of movement, the second major question that arises has to do with linearization and pronunciation. For any non-trivial chain, why do we pronounce only one particular chain link? Why does that link always seem to be the head of the chain, rather than its tail or some intermediate link? Nunes' answer to this comes from the independently-motivated Linear Correspondence Axiom (Kayne 1994) which he takes to apply at the level of linearization only. Since copies are non-distinct elements, he argues, any non-trivial chain cannot be linearized since it will require that the same item both precede and follow some other element. Therefore, before the LCA can apply, an operation called Chain Reduction must apply, deleting all but one link of the chain. Only then can linearization according to the LCA take place. But what decides which links are deleted and which one is preserved? Nunes argues that the only way in which copies in a chain differ is with regard to their formal features. Movement, being subject to Last Resort, only takes place for purposes of feature checking. Thus, lower copies in a chain will have formal features unchecked that the topmost copy in a chain will not. Since the computational system requires that formal features be eliminated by the interface level (to meet the requirement of Full Interpretation), Chain Reduction will always choose to preserve the highest copy and delete the lower ones. Nunes also discusses some marginal exceptions to these principles, showing that they follow from independent assumptions and do not contradict the system.
Overall, Nunes' book is a brilliant theoretical exercise and embodies the spirit of the minimalist program. He makes the simplest and most natural assumptions possible about the nature of movement and chain formation, and then explores the implications of his assumptions, demonstrating that they increase the understanding of how movement works in otherwise difficult to analyze multiple gap constructions. The explanatory power he achieves is impressive.
Missing from the work, however, is any detailed discussion of alternative accounts of the empirical puzzles his system solves. Especially surprising is a complete lack of reference to Nissenbaum's (2002) influential account of parasitic gap constructions. One would like to know if Nissenbaum's account is compatible with Nunes', or if there are reasons to choose one over the other. Discussion of alternative minimalist accounts of Across The Board constructions, such as Citko's (2001) Parallel Merge account, are also missing.
The basic ideas from the book, discussed above, were first put forth in Nunes' 1995 Ph.D. dissertation at the University of Maryland, College Park. Since that time, many linguists have found these ideas to be useful in many syntactic domains. Roehrs and Labelle (2003), for instance, adopt Nunes' compositional view of movement to explain performance errors in children learning French. Boeckx (2003) demonstrates that sideward movement captures the basic properties of donkey anaphora. Nunes and Uriagereka (2000) show that sideward movement also captures many of the conditions on extraction domains.
Finally, Kiguchi (2003) develops Nunes' assumptions to argue that the traditional notion of 'chain' has no place in a minimalist theory of syntax. Indeed, one wonders if the reductions Nunes achieves could be taken further, eliminating the need for copying and external merge altogether. Such an approach has been taken by some researchers, who suggest that movement should be described in terms of remerger (see Starke 2001, among several others). However, Nunes does not discuss the possibility of multi-dominance structures as a solution to the non- minimalist flavor of the tradition Movement operation.
Chomsky, Naom. 1993. A minimalist program for linguistic theory. In The View from Building 20: Essays in linguistics in honor of Sylvain Bromberger. Ken Hale and Samuel Keyser (eds). Cambridge, MA: MIT Press, pp 1-52.
Citko, Barbara. 2001. Parallel Merge and the Syntax of Free Relatives. Ph.D. Dissertation, University of New York, Stony Brook.
Kayne, Richard. 1994. The Antisymmetry of Syntax. Cambridge, MA: MIT Press.
Kiguchi, Hirohisa. 2003. Syntax Unchained. Ph.D. dissertation, University of Maryland, College Park.
Nunes, Jairo and Juan Uriagereka. Cyclicity and Extraction Domains. Syntax, 3, 1, 20-43.
Roehrs, Dorian and Marie Labelle. 2001. The Left Periphery in Child French: Evidence for a Simply-Split CP. In Romance Languages and Linguistic Theory 2001: Selected Papers from 'Going Romance,' Amsterdam 6- 8 December 2001. Jan Schroten, Mauro Scorretti, Petra Sleeman & Els Verheugd (eds). Amsterdam: John Benjamins, pp 279-294.
Takahashi, Daiko. 1994. Minimality of Movement. Doctoral dissertation, University of Connecticut, Storrs.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Brent Henderson is a Ph.D. candidate in linguistics at the University of Illinois, Urbana-Champaign. His interests include the syntax of relative clauses, phrase structure, and minimalist theory. He is currently writing his dissertation on Bantu relative clauses and their theoretical consequences.