Clahsen and Felser (CF) have written a fairly comprehensive review of the current literature on on-line second language (L2) processing, presenting data from eye movement, self-paced reading, and event-related potential (ERP) studies with the aim of evidencing possible differences between native language (L1) and L2 processing. The thrust of the article, in regard to adult L2 processing, is apparently an attempt to gather evidence to bolster their argument about “shallow processing” in adult L2 learners. Although the authors provide the reader with a generally good overview of the current literature, their argumentation seems to be flawed at times. Consider, first, the authors' presentation of recent ERP evidence. The authors claim that L2 adult learners may lack automaticity in comparison to native speakers in regard to syntactic processing. This is based upon a delayed N400 response, often found in L2 learners compared to native speakers, as well as by the pattern of anterior negativities to morphosyntactic violations. Later, however, this line of argumentation is seemingly undermined. First, as CF rightly underline, the range of variability in anterior negativities found in L2 learners falls within the range of variation observed in native speakers. As such, variability in this response cannot be taken as a marker of differential processing specific to (shallow) syntactic processing in the second language (see also Frenck-Mestre, 2005; Osterhout et al., 2004). Second, as CF later note, the N400 (as well as P600) is systematically observed in adult L2 learners, and is often highly similar to that found for native speakers. Consider, next, the behavioral evidence cited by CF on adult L2 syntactic processing. The authors cite work on various structures, notably relative clause attachment (which has received a great deal of attention in both monolingual and L2 studies). Concerning this structure, although CF cite studies, which show both clear L1 influence on L2 processing and differential effects as a result of experience with the L1, they favor studies that fail to show such effects and reject Mitchell et al.'s (2000) tuning hypothesis as an explanatory model. (Note Mitchell and colleagues have indeed produced evidence of their own showing limitations of their model.) It is also noteworthy that CF's argumentation about the sensitivity of the measure they used to test for immediate preferences for this structure is not as strong as it could be. Indeed, where they report L2 preferences (for low attachment following thematic prepositions), the literature shows the same systematic preference independent of the language tested (cf. Mitchell et al., 2000). As such, the sensitivity of their measure may not be adequately demonstrated. In sum, although CF provide the reader with an impressive collection of current L2 studies, the viewpoint that they espouse does not seem to be as substantiated as they wish to claim.
A META-ANALYSIS OF SENSITIVITY TO GRAMMATICAL INFORMATION DURING SELF-PACED READING: TOWARDS A FRAMEWORK OF REFERENCE FOR READING TIME EFFECT SIZES
