Children's Selective Trust in Informants: The Role of Speech Characteristics

Abstract

The main way children learn new information is through other people. However, not all people have information that is valuable and useful to them. Research suggests that children pay attention to many cues to help them gauge an informant’s trustworthiness as an information source. These cues include the speaker’s past behaviour and characteristics, such as their age and confidence. In this thesis, I focus on children’s attention to two speaker characteristics, both involving their speech properties: accent and filled-pause speech disfluencies (e.g., umm and uhh). Past research suggests children prefer to learn from informants who speak like them (Kinzler et al., 2011) and from informants who speak fluently (White et al., 2019). In the present experiments, I aim to better understand the strength and malleability of children’s use of accent and filled-pause speech disfluencies as cues to informant reliability. In Experiment 1, I ask whether 4-7-year-olds adjust their inferences about disfluencies when they are produced by non-native accented speakers. In Experiment 2, I ask which of the two cues (accent or fluency) 4-6-year-olds consider a more reliable indicator of an informant’s knowledge. In Experiment 1, children were presented with two German accented male speakers. Each speaker first presented familiar objects, either fluently or disfluently. In the test phase, each speaker labelled different novel objects with a novel name (e.g., mido), fluently or disfluently. Children were asked which item was labelled correctly. In Experiment 2, children were presented with one native English speaker and one German accented speaker. As in Experiment 1, one speaker produced fluent utterances and one speaker produced disfluent utterances. Results from Experiment 1 demonstrate that, overall, children show a weak preference for endorsing fluent speakers. These results suggest that children do not adjust their interpretation of disfluencies based on a speaker’s nativeness. Preliminary results from Experiment 2 suggest that children consider a speaker’s nativeness to be a stronger cue to their reliability than the speaker’s fluency.