Two steps are involved in the resolution of pronouns as co-referent with R-expressions: (a) the identification of referents in discourse representations (Kamp, 1981) and (b) the syntactic licensing by the binding theory (Chomsky, 1981). Additionally, in French, pronoun resolution implicates the checking of the uninterpretable gender feature, a feature not selected in the English grammar. The present study sought to determine at which level--the entities, the discourse referents or the lexical entries, or the features deployed--the relevant information is encoded by examining pronoun co-reference with synonyms with different gender specifications (with five experimental quadruples). High-intermediate (n = 7) and advanced (n = 13) English learners of French, and a native control group (n = 9), completed the same judgment task in a self-paced moving window format and in a PowerPoint presentation using the eye-tracking technology. Each experimental item included a short context with a noun marked for gender and a follow-up sentence that contained the pronoun, whose gender was manipulated. Grammaticality judgments were examined. In the moving-window format, reading times and, in the eye-tracking methodology, gaze durations (i.e., total reading time) and fixations on the pronoun and on the following segments were analyzed. Results indicate that L2 learners relied on all three levels to resolve pronouns, including the level of the features. The findings also provide further support for a feature-geometric account (e.g., Harley & Ritter, 2002).
Proceedings of the 12th Generative Approaches to Second Language Acquisition Conference (GASLA 2013)
edited by Jennifer Cabrelli Amaro, Tiffany Judy, and Diego Pascual y Cabo
Table of contents