On-line Interpretation of Scalar Quantifiers: Insight into the Semantic-Pragmatics Interface

The relationship between semantic and pragmatic interpretation has been a perennial puzzle in psycholinguistics. General agreement about the existence of these levels of interpretation contrasts with controversy in the field over exact boundaries between these representations and their relationship in real-time processing. Past research has demonstrated that pragmatic processing is rapid, often beginning before phrasal completion (Frisson & Pickering, 1999; Sedivy et al., 1999). However, they leave open the question of whether the pragmatic interpretation is preceded by some degree of semantic interpretation, as most linguistic theories would predict. These current experiments address this question by examining a relatively well-understood test case from linguistics. Horn (1989) noted that scalar quantifiers like some have two distinct readings corresponding to lexical semantics (SOME AND POSSIBLY ALL) and pragmatic inference (SOME BUT NOT ALL). A visual world task was used to track the online interpretation of some, all, two and three in adults and five-year-olds. The context provided two potential referents for the quantified noun phrase (e.g., a girl with 2 of 4 socks and one with 3 of 3 soccer balls). For all, two and three the reference of the quantified noun phrase was disambiguated by the semantics of the quantifier and participants were able to converge on the target shortly after quantifier onset. For some, however, determining the referent required a pragmatic implicature. On these trials looks to the target were substantially delayed, demonstrating a lag between semantic processing and the calculation of the implicature. Nevertheless, adults showed a preference for the target prior to phrasal completion, demonstrating that scalar inferences can occur as an utterance unfolds. Children, however, failed to calculate the implicature, converging on the target only after instructions were completed.