Adjectival scales and three types of implicature

Abstract

In this work, we explore the relationship between three different inferences triggered by gradable adjectives. In particular, we look at scalar implicature and two competing inferences occurring under negation - scale reversal (indirect scalar implicature) and a type of manner implicature called negative strengthening. In a series of experiments, we test a variety of adjectival scales and explore correlations between different inferences. Our results show that some scales are more likely to generate scalar implicature while others lean more towards generating negative strengthening. The extent to which scalar implicature and scale reversal correlate for the same scales, in turn, is lower than expected. We discuss our findings with respect to the mechanisms underlying the three types of inferences and factors accounting for differences across scales, with a focus on semantic distance, boundedness, the type of standard of comparison and adjectival extremeness.