Scalar diversity in alternative activation: Lexical decision with adjectives

When one hears The soup was warm, one may compute the implicature that the soup was not hot. Most standard accounts suggest that the stronger alternative hot is needed for the implicature’s derivation (see Sauerland, 2012 for an overview of different accounts). This suggests that during online comprehension, these alternatives should be activated, which has received support in recent experimental data (De Carvalho et al., 2016; Ronai & Xiang, 2023). In the current study, we test this alternative-based approach through scalar diversity, the finding that scales vary substantially in the degree to which they give rise to implicatures (van Tiel et al., 2016). This could be due to the availability of alternatives in the mind (Rees & Bott, 2018) and/or the underlying semantic structure of the scale (e.g., Gotzner et al., 2018). We tested two predictions of the alternative-based view: (a) the more activated a particular strong term of a scale by its weaker scale-mate is, the higher the likelihood of it giving rise to an implicature, and (b) that the strength of activation would depend on the structure of the scale. We ran a lexical decision experiment (N=150) with the set of scalar adjectives (N=64) from the study of Gotzner et al. (2018) to test this. We found that while the strength of priming correlated with the implicature endorsement rates found by Gotzner et al. (2018), the relationship was of the opposite direction—those scales that were less primed were the ones associated with higher rates. As for semantic structure, only the boundedness of scales was found to be predictive of priming, but again in the opposite than predicted direction—bounded scales were less primed. The talk will discuss some ways in which the semantic structure of adjectives can support pragmatic reasoning without lexical alternatives (the Measurement Mechanism, Gotzner, 2022).

Link to session