Abstract
Scalar implicatures, the phenomena where a sentence like "The pianist played some Mozart sonatas" is interpreted, as "The pianist did not play all Mozart sonatas" have been given two different analyses. Neo-Griceans (NG) claim that this interpretation is based on lexical scales (e.g., ), where the stronger term (e.g., all) implies the weaker term (e.g., some), but the weaker term (e.g., some) implicates the negation of the stronger term (i.e., some = not all). Post-Griceans (PG) deny that this is the case and offer a context-based inferential account for scalar implicatures. While scalar implicatures have been extensively investigated, with results apparently in favor of PG accounts, the psychological reality of lexical scales has not been put to the test. This is what we have done in the present experiment, with a lexical decision task using lexical scales in a masked priming paradigm. While PG accounts do not attribute any role for lexical scales in the computation of scalar implicatures, NG accounts suggest that lexical scales are the core mechanism behind the computation of scalar implicatures, and predict that weaker terms in a scale should prime stronger terms more than the reverse because stronger words are necessary to the interpretation of weaker words, while stronger words can be interpreted independently of weaker words. Our results provided evidence in favor of the psychological existence of scales, leading to the first clear experimental support for the NG account.