Pozzan, L.
CUNY-The Graduate Center
Second language learners of English often fail to perform subject-auxiliary inversion in main questions but over-apply it in embedded questions. Despite data reported in the literature (Lightbown & Spada, 2006; Lee, 2008), the full extent and causes of these non-target productions remain largely unknown. They might be due to L1 transfer, input characteristics, or innate bias. Alternatively, they might constitute performance errors, e.g. due to time pressure while speaking or writing. Two experiments shed light on these issues.
In an elicited production task, we examined inversion rates in main and embedded questions in advanced learners of English whose L1s (Spanish and Chinese) differ from English, and each other, in terms of inversion and wh-movement. Results were similar for the two L2 groups, indicating that L1 transfer is not the major source of non-target inversion patterns. L2 learners differed significantly from native speaker controls only on wh-questions: they often failed to invert in main questions for wh (11% vs. 0%, p<.01) but not yes/no (2% vs. 0%, p>.1), and over-applied inversion in embedded questions for wh (29% vs. 0%, p<.001) but not yes/no (1% vs. 0%, p>.05).
A Magnitude Estimation acceptability judgment task used the same materials to investigate whether the production results were due to non-target L2 grammars or performance factors. Acceptance patterns did not differ between learners and native controls (p>.1). Both groups rated target main and embedded questions higher than non-target ones. Both rated uninverted main questions higher for wh than for yes/no (p<.001), and inverted embedded questions higher for wh than for yes/no (p<.001).
In sum, while advanced L2 learners’ production patterns are influenced by factors that play no role in native speakers’ productions (e.g. question type; also specific wh-words), this difference disappears in acceptability judgments, a task traditionally assumed to tap into grammatical competence.