Toddlers’ Interpretation of Pronunciation Variation Depends on Context

• In Event: 3-089 - Children’s navigation of linguistically diverse environments: Accommodating accented speakers

Integrative Statement

Listeners are often faced with speech patterns that deviate from their past experiences. Some of these patterns result from enduring factors, such as a speaker’s different regional accent. Other patterns are more transient, such as a temporary change in a speaker’s pronunciations because they have something in their mouth. Adults treat these two types of variation differently; they are less likely to learn a pattern if there are signs that the variation is transient. For example, after hearing a speaker who has an atypical pronunciation of s, adults modify their s category, but not if the speaker has a pen in her mouth (Kraljic et al., 2008). This suggests that the speech processing system is efficient, and focuses on learning only the variation that is likely to persist in order to facilitate future processing.

Adults have had a lifetime of experience with the types of cues that signal enduring or transient variation. Being able to distinguish between these two types of variation would be particularly useful for younger listeners, who are in the process of building their native language knowledge. Although there is work demonstrating that toddlers can adapt to novel accents, to our knowledge there is none exploring whether there are constraints on this learning process.

We ask whether extra-linguistic visual context influences toddlers’ learning of a novel pronunciation pattern. Across four conditions, 19-month-olds (N=96) were exposed to familiar words, either pronounced typically (e.g., “balloon!”) or in a novel accent involving a consonant place change (e.g., “dalloon!”). They heard these pronunciations while viewing images of the named objects. Previous work has demonstrated that toddlers can learn a similar vowel-shifting accent at this age (White & Aslin, 2011). In the No Face/Accent condition, toddlers heard the accent during exposure, but did not see the speaker. In the Control condition, toddlers saw the speaker during exposure, but heard typical (rather than accented) pronunciations. In the two critical conditions, toddlers heard the accented pronunciations and saw the speaker. In the Lollipop-IN condition, she produced the accented words with a lollipop in her mouth; in the Lollipop-OUT condition, she produced them with a lollipop against her cheek.

At test, toddlers’ recognition of the accented words was compared to their recognition of typical pronunciations. If toddlers learned the accent, there should be no difference between these two types of pronunciations. Indeed, there was no significant difference between the accented and typical pronunciations in the No Face/Accent (p=.85) and Lollipop-OUT conditions (p=.17), indicating that toddlers learned the consonant accent in these conditions. However, there were significant differences in both the Lollipop-IN (p=.006) and Control conditions (p=.005). Therefore, despite exposure to the accent, toddlers who saw the speaker with a lollipop in her mouth did not learn the accent.

These results suggest that toddlers treated the lollipop as an explanation for the atypical pronunciations, which blocked their learning of the accent. This is the first demonstration that toddlers’ learning of a novel speech pattern is affected by extra-linguistic contextual cues to the pattern’s stability.

Authors

• Katherine S. White, University of Waterloo
  Presenting Author

• Olivia Daub, University of Waterloo
  Non-Presenting Author