Constraints on Feature Selection in Second Language Acquisition: Processing Evidence from the French Verbal Domain
129-141 (complete paper
or proceedings contents
This paper investigates the constraints on the second-language (L2) acquisition of features that are not instantiated in the first-language (L1) grammar. Some researchers (e.g., Hawkins and Casillas, 2008) argue that learners will rely on context-sensitive strategies, whereas others (e.g., Lardiere, 2009) argue that learners need to select and reassemble the relevant features into new feature matrices. It has also been suggested that learners will be more likely to insert underspecified versus clashing forms (e.g., McCarthy, 2007). Moreover, it has been hypothesized that a universal parser drives (L2) acquisition, assuming that what cannot be represented cannot be learned (see Dekydtspotter, 2001; Fodor, 1998). These proposals were tested with two experiments, which investigated the number feature in subject-auxiliary agreement and the gender feature in past-participle agreement with a moved object clitic in French. Three groups of American learners of French—second-semester, fourth-semester, and advanced learners—and a group of French natives completed a judgment task in a non-cumulative self-paced moving-window format. Final acceptability judgments and reading times (RTs) were analyzed. Although lower proficiency learners seemed to accept all forms independently of contexts, all learners' processing profile exhibited sensitivity to feature clash. This finding suggests not only that L2 learners rely on domain-specific feature checking but also that the parser guides the reassembly of the relevant features before lexical encoding.
Selected Proceedings of the 2009 Second Language Research Forum: Diverse Contributions to SLA
edited by Luke Plonsky and Maren Schierloh
Table of contents