Two-year-olds Recognize Known Words in Unfamiliar Accents but Cannot use Them to Learn new Ones

• In Event: Poster Session 05
  In Poster Session: PS 05 Topic - Language, Communication

Abstract

By their second birthdays, infants process speech efficiently in their native language and are expert word-learners. They also successfully recognize familiar words and learn new words spoken in unfamiliar accents (Best et al., 2009; Schmale et al., 2011).

Here, we delve deeper into 2-year-olds’ processing of unfamiliar accents. Prior evidence reveals that in native-accented speech, 2-year-olds efficiently use known words to infer the meaning of novel words: if infants hear “The dax is sleeping,” they infer “dax” refers to an animal, not an inanimate object (Ferguson et al., 2017). However, if processing unfamiliar accents remains a challenge for 2-year-olds, as for older children (Bent, 2014), infants may struggle to infer meanings solely from linguistic context.

To test this question, we adopted the eye-tracking paradigm established by Ferguson et al. (2017) with native-accented speech. However, we presented the sentences in Spanish-accented speech, an unfamiliar accent for our participants. See Figure 1. Infants (n=48) heard dialogues featuring familiar nouns (6 trials) and then novel nouns (6 trials). No referents were shown during dialogues. We varied whether the novel nouns were presented in an Informative linguistic context (e.g., “The dax is sleeping”) or a Neutral linguistic context (“The dax is clean”). At test, infants viewed an animate and inanimate object and were prompted to look to the target noun’s referent. If infants used the linguistically Informative dialogue to infer the referent of the novel noun, they should look more to the animate referent. If the unfamiliar accent posed too great a processing challenge, then performance should resemble the Neutral condition.

In Experiment 1, although 2-year-olds successfully identified the referents of familiar nouns in Spanish-accented speech, p<.01, they did not use familiar words to learn novel nouns. A cluster-based permutation test revealed that 2-year-olds failed to use the Informative context to learn novel nouns: unlike previous native-accent conditions, looking patterns in the Informative and Neutral conditions did not significantly diverge, p>.5 (Figure 2).

To assess whether infants’ difficulties with the unfamiliar accent truly stemmed from the challenge of using the linguistic context, not simply learning words, we conducted Experiment 2. The task was identical to Experiment 1’s Informative condition, except the referent for each novel noun was present during the dialogue. Two-year-olds (n=24) in this Co-present Referent condition successfully learned novel words: performance diverged from the Neutral control condition, p=.01, with infants looking more to the animate referent 550ms to 1250ms after noun onset. Thus, infants successfully learned novel words in an unfamiliar accent when a co-present referent was available. Additionally, infants’ performance was predicted by their preferences for the target object on familiar noun trials, r(20)=.46, p=.032. Success in comprehending familiar words across accents is thus associated with success in learning new ones.

These findings reveal a nuanced developmental trajectory for processing unfamiliar accents. While 2-year-olds recognize and learn words in unfamiliar accents—and these skills are inter-related—they still struggle in using known words to learn new ones. This coheres well with older children and adults’ continued difficulties in processing unfamiliar accents.

Authors

• Alexander LaTourrette, University of Pennsylvania
  Presenting Author

• Sandra R Waxman, Northwestern University
  Non-Presenting Author