Mr. Caveman: The dog painted the triangle c. Experimenter to participant:

Is that right? Full-size table Table options View in workspace Download as CSV Again, simply recognising that Mr. Caveman only said ‘the triangle’, having witnessed the dog painting FG-4592 the triangle and the heart, is sufficient reason to object to the utterance, without further requiring the computation of the implicature that the dog did not paint the heart. It is therefore not clear whether binary judgment tasks test participants’ sensitivity to informativeness or their actual derivation of implicatures. This observation is also potentially critical for other paradigms used to study implicature, including the Felicity Judgment task (Reinhart, 2004; Foppolo et al., submitted for publication; among others), sentence-to-picture matching tasks (Hurewitz et al., 2006) and visual world eye-tracking studies. To take an example of the latter, Huang and Snedeker, 2009a and Huang and Snedeker, 2009b investigate

whether children aged 5½ and adults use a scalar implicature to select the appropriate referent from a display of four pictures. In an example of their critical condition, two of the pictures are of girls, one Wnt inhibitor of whom has some of the socks (there being other socks in the display), while the other has all of the soccer balls (there being no other soccer balls in the display). Participants are instructed to ‘point to the girl with some of the socks’. The critical issue is whether participants will fixate on the target referent (the girl with some of the socks) before the onset of ‘socks’, which is the semantic disambiguation point. To succeed in this task, we argue that participants do not need to draw an implicature, but simply have to be sensitive aminophylline to the fact that ‘the girl with some of the…’ would be underinformative if it referred the girl with (all of) the soccer balls.

As in the binary judgment paradigm, participants will also succeed in the task if they draw the implicature (‘some but not all of the…’), but once again they do not need to do so. Sensitivity to informativeness is a precondition for implicature derivation in the Gricean approach and all its major reformulations (e.g. Chierchia, 2004, Geurts, 2010, Levinson, 2000 and Sperber and Wilson, 1986/1995; among others). Our interim conclusion is that the literature so far has relied upon paradigms that test the former without necessarily also testing the latter. The third observation we wish to make is that pragmatic infelicity in the widely used paradigms does not give rise to the same kind of violations as logical falsity. As a result, the pragmatically appropriate response to underinformative utterances in these paradigms is not clear. First let us suppose that participants are resolving judgement tasks by being sensitive to informativeness (rather than deriving implicatures). Underinformative utterances are strictly speaking true, but sub-optimal.