On ellipsis: A processing solution to the undergeneration problem?
Lyn Frazier, University of Massachusetts, Amherst
Ellipsis poses interesting problems for the language processor because phonologically empty constituents must be syntactically structured and interpreted. My colleague Chuck Clifton and I have been developing a theory of processing forms of ellipsis that may cross sentence boundaries, such as VP ellipsis and Sluicing (Frazier and Clifton 1998, 2001). On this account structure from the antecedent is copied into the ellipsis site by a cost-free copying mechanism at LF (alternatively, one may think of this as structure-sharing (Frazier and Clifton 2005, Martin and McElree submitted). Antecedents which are part of the main assertion of an utterance are favored, especially across sentence boundaries.
Syntactic accounts of ellipsis require a syntactically matching antecedent at LF (either as a direct consequence of grammatical requirements, or in order to identify the syntactic structure in the elided constituent in a processing implementation of a hybrid account like Merchant's (1999, 2004, to appear)). Such accounts undergenerate: many attested examples of ellipsis do not include a syntactically matching antecedent for the elided constituent. In the talk, the focus will be on this undergeneration problem.
One elegant approach to the undergeneration problem was proposed by Kehler (2002), based on a theory of discourse coherence relations. On this view, elided constituents require a matching antecedent only if they enter into certain kinds of discourse coherence relations (Kehler 2002). Despite the explanatory potential of this approach, there are empirical problems (Frazier and Clifton 2006). Instead of pursuing a discourse coherence solution, a processing approach will be proposed: when the discourse does not contain a matching antecedent for an elided constituent, the processor builds one at LF using the materials at hand (Arregui, Clifton, Frazier, and Moulton, 2006). On this approach, some instances of ellipsis will be acceptable even though there is no pre-existing matching antecedent simply because the processor can create one at LF using its normal mechanisms, e.g., those needed for performing syntactic reanalysis of garden path sentences. In addition to these 'recycled' antecedents, I will discuss ongoing research on another class of antecedents which, by hypothesis, are created as a by-product of drawing a certain class of implicatures.
References
Arregui, A., Clifton, C., Frazier, L., and Moulton, K. 2006. Processing elided VPs with flawed antecedents: The recycling hypothesis. Journal of Memory and Language, 55 (2), 232-246.
Chung, S. Ladusaw, W. and McCloskey, J. 1995. Sluicing and Logical Form. Natural Language Semantics 3, 239-282.
Frazier, L. and Clifton, C. 1998. Comprehension of sluiced constituents. Language and Cognitive Processes, 13, 499-520.
Frazier, L. and Clifton, C. 2000. On bound variable interpretations: The LF-only hypothesis. Journal of Psycholinguistic Research, 29, 125-139.
Frazier, L. and Clifton, C. 2001. Parsing coordinates and ellipsis. Copy ". Syntax 4, 1-22.
Frazier, L. and Clifton, C. 2005. The syntax-discourse divide: Processing ellipsis. Syntax, 8, 121-174.
Frazier, L. and Clifton, C. 2006. Ellipsis and discourse coherence, Linguistics and Philosophy, 29 (3), 315-346.
Kehler, A. 2002. Coherence, Reference and the Theory of Grammar, CSLI Publications. Stanford University.
Martin, A. and McElree, B. Submitted. A content-addressable pointer mechanism underlies comprehension of verb phrase ellipsis.
Merchant, J. 1999. The syntax of silence: Sluicing, islands and identifying in ellipsis. University of California doctoral dissertation.
Merchant, J. 2004. Fragments and ellipsis. Linguistics and Philosophy, 27, 661-638.
Merchant, J. To appear. Variable island repair under ellipsis. To appear in K. Johnson (ed.) Topics in ellipsis. Cambridge University Press: Cambridge.