Linguistic Approaches to Bilingualism - Volume 4, Issue 2, 2014

The parsing-to-learn hypothesis (Fodor, 1998) identifying the parser as the language acquisition device (PLAD) is discussed for second-language (L2) grammatical acquisition. Parsing is assumed to involve concomitant UG-sanctioned structure generation and licensing by a parameterized lexicon. In this architecture, licensing induces immediate changes to the lexical knowledge base as new feature matrices for categories and their exponents are registered. Stages arise as these matrices are increasingly activated and can be accessed within the narrow window imposed by working-memory limitations. Specifically, the PLAD instantiates feature re-assembly (Lardiere, 2009) in response to licensing failures (Clark & Roberts, 1993), characterizing transitions between grammatical states (Gregg, 1996, 2003) in the Full Transfer/Full Access model (Schwartz & Sprouse, 1994, 1996). The PLAD is examined in light of current research and of evidence showing mandatory engagement of syntax, breakdown reflecting feature organization, and domain-specificity. The PLAD offers potentially fruitful insights about L2 parsing and grammatical development.

This study considers the role of lexical access in the activation and maintenance of referents interacting with syntactic computations during the online processing of wh-dependencies in second-language French by beginning (N = 39), low intermediate (N = 40), and high intermediate (N = 35) learners. Two computer-paced reading tasks involving concurrent picture classification were designed to investigate trace reactivation during sentence processing: The first task targeted sentences that contained indirect object relative clauses, whereas the second task involved indirect object cleft sentences. Response time profiles for sentences containing English-French cognates as antecedents were compared with those for sentences with noncognate vocabulary. All learner participants produced differing response patterns for cognate and noncognate items. Intermediate learners’ response patterns were consistent with trace reactivation for cognate items only; noncognate items induced inhibitions or erratic response patterns. Additionally, a (French-English bilingual) native speaker control group (N = 35) showed the predicted response pattern with the noncognate items only. These findings indicate that the role of lexical access in sentence processing merits further consideration.

The study tests representational and computational accounts of morphological variability in English-speaking learners of Spanish by examining performance on gender and number agreement under different task demands. Second language (L2) learners took either a Speeded grammaticality judgment task (GJT) or an Untimed GJT. The tasks targeted agreement violations of two types: errors in the use of ‘default’ morphology and errors involving ‘feature clashes’ (McCarthy, 2008). In addition, three groups of native speakers took the Speeded GJT at three different presentation rates to examine whether native speakers under a processing burden perform similarly to learners. Natives in the fastest speed performed better with feature clash errors for both gender and number. Learners showed the same pattern for number, but performed better with default errors in gender, suggesting different effects of processing demands for properties unique to the L2. On the Untimed GJT, a subset of advanced learners showed perfect performance with both gender and number.

Lardiere (2009) hypothesizes that second-language (L2) grammar development involves the reassembly of features in the constraints of Universal Grammar. Feature reassembly assumes the domain-specific Agree operation, in which an (interpretable) feature on a probing node values an uninterpretable counterpart feature on a goal node, and spell-out computations, providing morphological expression to these nodes. Because features express class membership (e.g., feminine expressions), Hawkins and Casillas (2008) proposed that agreement in non-advanced L2 acquisition may involve co-occurrence between expressions, computed presumably according to feature compatibility (e.g., Shieber, 1986). These two types of agreement computations predict distinct processing profiles and grammar-processing relations. Results from a self-paced moving-window experiment targeting overt adjective agreement in superlative constructions revealed that L1-English L2-French/Spanish learners’ profiles appear more consistent with feature valuation. Form sensitivity lagged behind computational ability in low-proficiency learners, suggesting that L2 grammar acquisition is subserved by a universal parsing mechanism (e.g., Dekydtspotter, 2001; Schwartz, 1999).

There is ample evidence that native speakers anticipate upcoming information at various levels during sentence comprehension. In contrast, some studies on late second-language (L2) learners support the view that L2 learners do not anticipate information during processing, or at least, not to the same extent as native speakers do. In the current paper, I propose that native and L2 speakers are underlyingly the same as far as sentence processing mechanisms are concerned, and that potential differences in anticipatory behavior can be accounted for by the same factors that drive individual differences in native speakers; in particular, differences in frequency biases, competing information, the accuracy and consistency of the lexical representation, and task-induced effects. Suggestions for future research are provided.