Acquisition of Relevance implicatures: A case against a Rationality-based account of conversational implicatures

In our L1 acquisition experiment, we tested between the Language-based and Rationality-based accounts of how Relevance implicatures are computed, and found support for the Language-based account but not the Rationality-based account. While on the Language-based account (e.g., Horn, 1984 and Levinson, 2000), Relevance implicatures are situated within the language module, on the Rationality-based account (Kasher, 1991), they are not situated within the language module, and are interpreted as being derived by Rationality-based reasoning, which is also instrumental in deriving non-linguistic inferences.We tested children aged 5;1–8;1 on computing Relevance implicatures and non-linguistic inferences that were parallel in nature. On the Language-based account, children were predicted to perform better on computing non-linguistic inferences than Relevance implicatures that are parallel in nature because in order to compute Relevance implicatures children need to master additional linguistic prerequisites. On the Rationality-based account, children were not predicted to perform better on computing non-linguistic inferences than Relevance implicatures.We found that children performed significantly better on computing non-linguistic inferences than Relevance implicatures, which provided evidence for the Language-based account. We argue that reasoning about language, and specifically, about the role of seemingly irrelevant utterances in discourse, constitutes the main acquisition challenge with respect to Relevance implicatures.