Conversational Implicature

Conversational implicature is a type of inference that emerges from a speech act but not directly through the meaning of the speaker's words, according to Herbert Paul Grice. So, when Bill answers Alice's question in (1), one could draw the implicature that his answer is “no,” even though he never said so outright:

Conversational implicature is to be distinguished from conventional implicature, which refers to pragmatic information that is encoded in language. For instance, consider the conjunction but in George Clooney is famous, but he is nice; while having the same [Page 107]meaning as and, but contrasts the two conjuncts conventionally through language. As can be seen, Grice highlighted the distinction between core linguistic and pragmatically derived parts of meaning and thus broke new ground in philosophy and linguistics. Crucial to the progress of the Gricean approach has been the empirical validation of this distinction and particularly through developmental studies. To appreciate how, consider the utterance in (2), which is said about a turtle that ate an entire set of pears:

This utterance can have either a semantic or pragmatic reading. It is semantically compatible with a stronger, unsaid proposition The turtle ate all of the pears, which also entails (2). At the same time, upon hearing (2), a listener is pragmatically justified to infer (3):

The proposition in (3) is what amounts to a scalar inference (or scalar implicature). Note that it is not linguistically encoded by (2). While accounts for this inference vary somewhat, John Stuart Mill's explanation remains persuasive—that it arises because the speaker could have used a stronger statement and chose not to. To appreciate this in a typical experimental trial, consider a speaker who says (2), for example, when all the pears have been eaten. A child's affirmative (true, right or agree) response is taken as evidence of satisfaction with the semantic reading, while a negative (false, wrong, or disagree) response is taken as evidence that she carried out the pragmatic inference (3).

In the first systematic investigation of scalars, Ira Noveck reported a developmental progression in which pragmatic inference-making increases with age; in other words, he showed that younger children are more likely than older ones to be satisfied with semantic readings. Since then, multiple studies across dozens of languages have replicated this developmental effect while also endeavoring to (1) explain what underlies it and (2) show that young children are more competent at such inference-making than what was initially indicated.