Most pragmatic responses to underinformative some-statements are associated with scalar implicatures

A common method for investigating scalar implicatures is to use underinformative sentences, such as Some X are Y, when evidence indicates that All X are Y. Underinformative sentences can have a logical (some and possibly all) or pragmatic (some but not all) interpretation. Recently, Kissine and De Brabanter (2023; Cognition) presented experiments that question whether false responses to underinformative sentences indicate explicit derivation of the implicature. Their findings cast doubt on the conclusions of much recent research in experimental pragmatics. Here, we present three experiments that build on their findings using a similar method while incorporating design differences. In a two-phase paradigm, participants evaluated underinformative sentences (Some elephants are mammals) in Phase 1 before selecting a sentence interpretation (logical, pragmatic, or neither) in Phase 2. In all three experiments (N = 52; N = 103; N = 100), participants were congruent (with the explicit derivation hypothesis) significantly more than predictions based on chance, all p's < 0.05, with effects that were more pronounced when considering a subset of participants who were consistent in their Phase 1 interpretations (N = 22; N = 62; N = 67), all p's < 0.05. Overall, the results support the explicit derivation hypothesis, contrary to Kissine and De Brabanter, and are consistent with assumptions in the pragmatics literature. Nonetheless, there are limitations to the paradigm (e.g. the influence of interpretation paraphrasing), and individual differences in consistency with explicit derivation predictions. It is therefore possible that while most participants were deriving enrichments of the sort not all to underinformative sentences, some participants were not.