Context effects in children's calculation of scalar implicatures: the case of disjunction

• In Event: 1-042 - Epistemic reasoning in pragmatic interpretation

Integrative Statement

Introduction: The disjunction ‘or’, has two interpretations, inclusive as in "John can play flute or clarinet" (meaning he can play either one or both) and an exclusive, as in "Yesterday John played flute or clarinet." The exclusive interpretation is achieved by means of a scalar implicature (SI), which negates the stronger conjunctive interpretation. Singh et al. (2016), found that children interpret disjunction conjunctively even in contexts where such an interpretation is impossible for adults. We show that when children are given a pragmatically reasonable context for the disjunction, they perform more adult-like. Children interpret disjunctions exclusively in contexts where the character uttering the disjunctive sentence has some level of ignorance of the situation being described, a context in which exclusive disjunction is pragmatically appropriate.

A new experiment: Singh et al. (2016) used a Truth Value Judgement (TVJ) task with a puppet who made statements about a picture. Their child responses showed a strong bias to the conjunctive interpretation of 'or', but their adult responses showed a failure to calculate SIs, since adults failed to reject the conjunctive interpretation. This casts some doubt on the experiment. In the context of a picture of a boy holding a banana and an apple, it’s very pragmatically odd to say “The boy is holding an apple or a banana” as in Singh et al.’s (2016) task. Since the picture is the only context, the sentences violate the principle of relevance. Instead of a TVJ task, we used a version of the “Magic Box” technique (Huang et al. 2013) to access alternatives, and a blindfolded puppet who “couldn’t remember exactly what happened”. This improves pragmatic support for the disjunction. Participants chose from three pictures, one of which was partially covered by “Wally the Worm” to create the Magic Box effect. The worm covers just the crucial components of the scene, and so the covered alternative is more visibly a possible alternative. Figure 1. To show that children calculated implicatures correctly, a set of some/all sentences was used. Children (N=15,
ages 2;6–6;11, mean 4;8) saw 9 experimental items and 6 control items. Adults (N=16) saw 9 experimental items, 12 control items and 41 fillers.

Results: On control some/all trials, adults performed at ceiling, and children chose the expected picture 80% of the time. The disjunction results are in Figure 2. Responses varied depending on which alternatives were visible. When 'both' and 'neither' alternatives are visible (A), adults mainly chose the covered picture. When one item and either 'both' or 'neither' were visible (B/C), adults split between the 'one' and the covered picture. This shows that they are treating the covered picture as the other 'one' alternative. If children showed a conjunctive interpretation bias, we would expect them to reliably choose the 'both' picture in A and C, but they did not. They chose equally among the three pictures. In B and C they chose equally between the covered and the 'one' picture, while in A they chose equally between the covered and the 'both' picture.

Authors

• Alan Munn, Michigan State University
  Presenting Author

• Rachel Stacey, Michigan State University
  Non-Presenting Author

• Robert Felster, Michigan State University
  Non-Presenting Author

• Adam Smolinski, Michigan State University
  Non-Presenting Author

• Adam Liter, University of Maryland - College Park
  Non-Presenting Author

• Cristina Schmitt, Michigan State University
  Non-Presenting Author