Do children extend pragmatic principles to non-linguistic symbols?

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

Integrative Statement

A foundational principle of communication is that speakers should offer as much information as required during conversation (Grice, 1989). When a speaker offers a weak statement (“Some ladybugs have spots”), the listener often infers that the speaker doesn’t know whether a stronger statement (“All ladybugs have spots”) is true, since a fully knowledgeable speaker should provide fully informative statements (Bergen & Grodner, 2012). Classic theories of communication assume that the principle of informativeness extends beyond linguistic interactions (Grice, 1989; Sperber & Wilson, 1986) but relevant evidence so far is limited (Papfragou, Friedberg & Cohen, 2017). Thus we ask whether 4- and 5-year-olds expect drawings to be informative in accord with the creator’s knowledge. We adopt a simple paradigm (Kampa & Papafragou, 2017) in which 4- and 5-year-olds can successfully link scalar utterances to speaker knowledge and extend it to non-linguistic representations.

In Experiment 1, 4-year-olds, 5-year-olds, and adults (N=25 per group) saw a pair of photos depicting a girl (“Danielle”) facing the participant behind a two-compartment box (Fig.1). In one photo, Danielle and the participant had full visual access to both compartments of the box (full knowledge). In the other, the participant had full access, but Danielle could only see one compartment (limited knowledge). Participants were told Danielle would look at each box, then draw what she sees in one of them. A new photo showed Danielle’s drawing that either depicted both objects (Strong) or only one object (Weak; Fig.1). Participants were asked to select which box she drew. A pragmatic responder should link the strong drawing to the full-access box and the weak drawing to the limited-access box, in accordance with levels of informativeness. A logistic mixed effects model predicting accuracy with Trial (Strong,Weak) and Age (Adults, 5s, 4s) as fixed predictors (random effects included intercepts for subjects, items and a slope for Trial) revealed an effect of Age: adults performed better than children (β=-1.6094, p=.008), but 4s and 5s did not differ (p=.24). Children’s performance was not different from chance on weak trials (Fig.2).

To test whether the detail in the drawings impeded their symbolic function (Uttal et al., 2009), Exp.2 modified Exp.1 by using schematic drawings (Fig.1, N=25 per group). Results replicated the Age effect of Exp.1 (Fig.2). However, comparisons across experiments revealed that participants performed better in Exp.2 than Exp.1 (β=-1.6022, z=-2.284, p=.022). Adults performed better than children overall (β=-1.6022, z=-3.752, p<.001), but 5-year-olds performed better than 4-year-olds (β=1.2361, z=2.773, p=.006). In Exp.2, performance differed from chance except for weak trials in the youngest group.

We conclude that the principle of informativeness extends to non-linguistic symbols: 5-year-olds and adults expect drawings, like utterances, to be informative in accordance with creator knowledge. However, the drawings’ visual detail affects their success. Furthermore, 4-year-olds do not successfully link incomplete drawings to partial knowledge. Given that under similar conditions 4-year-olds can use weak utterances to infer lack of speaker knowledge, we attribute their present failure to the developing understanding of the symbolic (and selective) representational function of drawings.

Authors

• Alyssa Kampa, University of Delaware
  Non-Presenting Author

• Anna Papafragou, University of Delaware
  Presenting Author