A bidirectional study in L2 acquisition of pragmatics: The case of (un-)bounded adjectival scales

This study tests whether the underlying semantic properties of gradable adjectives influence second language (L2) learners’ pragmatic inferencing behavior, and further examines the extent to which awareness of scale structure applies cross-linguistically. Using an inference evaluation task, native and L2 speakers of both English and French were asked to decide whether the speaker of an utterance meant to negate the stronger term in the scale. Our results demonstrate that native speakers display sensitivity to bounded vs. unbounded adjective scales whereas L2 learners do not. Native participants derived implicatures more often in bounded scales that have endpoint denoting adjectives vs. unbounded scales containing relative adjectives where boundaries between scalemates are less clearly delineated. Learners derived implicatures at similar rates across both scale types and significantly more often than native speakers. However, an effect of proficiency emerged among learners with more exposure to the L2, suggesting that semantic information begins to be incorporated into the implicature derivation routine with greater command of language. These findings are evaluated in terms of hypotheses on the types of information that are accessible to the L2 parser at early stages of acquisition. We also interpret these results on the basis of a default view on implicature generation.