Skip to main content
School of the Arts
  1. School of the Arts
  2. Research
  3. Department Research
  4. Linguistics
  5. Linguistics Research Events
  6. Past events
  7. Miscellaneous

Ling Lunch: Dr. Ming Xiang (U Chicago)

Time:
Venue: Linguistics Seminar Room | Arts Research Centre Basement

Recovering syntactic structures

Dr. Ming Xiang (U. Chicago)

 

 

When resolving ellipsis, speakers could either derive semantic interpretations by first constructing unpronounced syntactic representations at the ellipsis site (Merchant, 2009; Chung, Ladusaw, McClosky, 1995), or they could assign an interpretation to the ellipsis site based on the semantic/discourse information present in the antecedent, without the need to construct any syntax at all (Hardt, 1993). In this talk I will present experimental evidence from two studies to show that speakers do recover syntactic structures at the ellipsis site. The first study looked at the syntactic priming effect from ellipsis constructions like 'œFirst Ralph sang a song to Sheila, and then Marcus did'. We show that speakers produced more NP PP structure in their production after they were exposed to a syntactic prime that contained either an ellipsis structure as above, or its full structure counterpart ('¦. and then Marcus sang one to her), but not an intransitive control (... and then Marcus groaned), suggesting syntactic structures were recovered at the ellipsis site.

In the second study, we looked at number mismatch in cases like 'œThe workers here are writers, and the worker there is too'.  Such number mismatch is acceptable in English, despite the fact that number-mismatch in overt syntax is ungrammatical (*'¦and the worker there is WRITERS too). Two general approaches could potentially derive grammatical number-mismatch for VP ellipsis. The first assumes that number agreement is calculated in ellipsis, but when the number feature on the antecedent mismatches with that of the ellipsis site, additional operations are evoked to derive a grammatical representation (e.g. Merchant, 2011 on gender mismatch). This approach could predict extra processing cost for the ellipsis with a number-mismatched antecedent. The second approach, on the other hand, could deny any actual syntactic mismatch, either by assuming agreement is only calculated in morpho-phonology, not in syntax per se (e.g. Bobaljik and Zocca, 2011; Nunes and Zocca, 2005; Bobaljik, 2008); or ellipsis is resolved purely in semantics/discourse, not in syntax (Hardt, 1993). This would predict no processing cost for the seeming number-mismatch in ellipsis, since there is no mismatch to start with. With ERP recordings we show that although number-mismatched VP ellipsis is totally acceptable (as shown by our acceptability judgments), they nevertheless evoked neurophysiological cost, suggesting syntactic number agreement is being calculated at the ellipsis site.

 

 

Back to top