QUD sensitivity in the computation of scalar implicatures in second language acquisition

ABSTRACT This study aims to investigate adult L2 speakers’ use of different types of information in the comprehension of pragmatic inferences by examining L1-Mandarin Chinese L2-English speakers’ sensitivity to cues all and any in scalar implicature (SI) computation for some. This article and our experimental setup does not seek to understand how the question under discussion (QUD) affects implicature derivation. Rather, we examine whether comprehension of scalar sentences is modulated by QUD. We used an acceptability judgment task in which the felicitousness of the target response containing some in a given picture is manipulated by the QUD containing all or any. The interpretation of some depends on the QUD. For instance, in the question “Are all/any of the squares red?,” all primes an implicature “Some but not all squares are red” reading, whereas any does not. Linear mixed-effects regression analysis conducted on z-score transformed Likert-scale ratings with QUD type (all/any) and Group (native vs. L2 speakers) showed that native speakers differentiated between any and all in their ratings for target sentences with some, but L2 speakers did not. That is, native speakers rated target infelicitous sentences with some in the QUD-any condition significantly higher than in the QUD-all condition. Though we stress that these results must be considered exploratory in nature, our findings are in line with the proposal that native and L2 speakers differ in the type of information they attend to during language comprehension.