CNL Lunch

Sachiko Aoshima

(with Colin Phillips & Amy Weinberg)

The time course of processing long-distance dependencies: evidence from Japanese wh-interrogatives

Thursday December 5th, 12:30pm, 3416 Marie Mount Hall

In this study we present results from three self-paced reading studies of Japanese wh-scrambling structures that address the questions of what drives structure-building in language comprehension, and how incremental structure building is in a head-final language such as Japanese. The results from the first two experiments show that Japanese readers preferentially associate a fronted wh-phrase with the most deeply embedded
clause of a multi-clause sentence. Experiment 1 demonstrates this based on evidence that readers expect to encounter a scope-marking affix on the verb of an embedded clause in wh-fronting constructions. Experiment 2 shows that the wh-phrase is already associated with the embedded clause before the verb is processed, based on a Japanese counterpart of the Filled Gap Effect (Stowe 1986). Given that the first verb in a Japanese sentence appears in the most deeply embedded clause, these findings provide evidence that the time course of processing wh-dependencies is driven by the need to satisfy thematic role assignment requirements of the wh-phrase, rather than by the
need to create a gap as soon as possible (e.g. Frazier & Clifton 1989).

Experiment 2 already indicated that structure-building, in this case
gap-creation, occurs prior to the verb in Japanese. Experiment 3
investigated the stronger claim that grammatical relations between noun phrases are computed in advance of the verb. It did so by testing for sensitivity to constraints on co-reference in Japanese before the verb is processed. The results show that readers actively seek for an antecedent for a pronoun in grammatically sanctioned positions, which occur before a verb. The second and third experiments show that both gap-creation and binding constraint application take place in advance of the verb, against the claim that when critical lexical heads are delayed, structure building should be correspondingly delayed (e.g. Pritchett 1991). Instead, these findings suggest that structures are built incrementally even when critical lexical
heads are delayed.

Taken together, the results of these studies indicate that speakers of Japanese make successive attempts to posit a gap in a position that will allow thematic interpretation to occur as early as possible. Since these successive attempts imply unforced reanalysis, we also explain how our findings can be reconciled with a number of other findings which suggest that unforced reanalysis is generally avoided in parsing (e.g. Kamide & Mitchell 1999; Sturt et al. 2001; Schneider & Phillips 2001). We discuss that all the cases can be accounted for under the present analysis based on the incremental online satisfaction of the lexical requirements.