Priming of early closure: evidence for the lexical boost during sentence comprehension

Abstract

Two self-paced reading experiments investigated priming in sentences containing ‘early’ vs. ‘late closure’ ambiguities. Early closure sentences impose relatively large processing costs at the point of syntactic disambiguation. The current study investigated a possible way to reduce processing costs. Target sentences were temporarily ambiguous and were disambiguated towards either the preferred ‘late’ closure analysis or the dispreferred ‘early’ closure analysis. Each target sentence was preceded by a prime that was either structurally identical or that required a different syntactic analysis. In Experiment 1, all the prime sentences shared the same critical verb as the target. In Experiment 2, verb repetition was eliminated by reorganising the stimuli from Experiment 1. In Experiment 1, processing of the disambiguating verb was facilitated when an ‘early’ closure target sentence followed an ‘early’ closure prime. In Experiment 2, there were no significant priming effects, although an overall difference in processing time favoured ‘late closure’ targets. Combined analyses verified that the pattern of results in Experiment 1 differed significantly from Experiment 2. These experiments provide the first indication that ‘early’ closure analyses can be primed and that such priming is more robust when a critical verb appears in both the prime and the target sentence. The results add to the body of data indicating a ‘lexical boost’ for syntactic priming effects during comprehension. They have implications for theories of syntactic representation and processing.

Keywords:

• syntax
• parsing
• syntactic priming
• syntactic ambiguity
• unification grammar
• usage-based grammar
• head-driven phrase-structure grammar
• lexical boost

Notes

1. Recent studies have also demonstrated that exposure to a prime sentence that has a given structure can influence the final interpretation assigned to a subsequent globally ambiguous sentence (Branigan et al., 2005; Raffray & Pickering, 2010).

2. Trueswell & Kim's study involved a fast priming manipulation, in which a subliminally presented verb either did or did not match the subcategory preference of a fully visible verb. Reading times were shorter when the two verbs had congruent subcategory preferences.

3. Semantic repetition, in this case, refers to the simple fact that when two adjacent sentences contain the same verb, they are referring to similar actions. It is possible that when two sentences refer to the same action (e.g., ‘examining’), the second could be easier to process than the first because the conceptual knowledge associated with the referenced event is already activated.

4. The terms ‘early’ and ‘late’ are not used here to imply or endorse any particular representation and processing theory. They are used because they are common terms for the kinds of sentences that were tested.

5. The argument structure hypothesis predicts the lexical boost in double-object dative vs. prepositional-object dative priming. The evidence regarding comprehension priming for this type is mixed, as noted previously.

Additional information

Funding

This project was supported by grants from the National Science Foundation (NSF#1024003, 1355776, SBE-0541953) and the National Institutes of Health (MH099327, 1R01HD073948). Thanks to Liv Hoversten for assistance in data collection. Thanks also to Don Donaldson for writing the self-paced reading software.