Accent Distance and Perceptual Adaptation in Toddlers and Adults

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

Integrative Statement

Accents vary in how difficult they are to understand, arising from how distant they are from listeners’ native accent (e.g., a Canadian listener will have an easier time understanding an American over a Scottish accent). While there is evidence that listeners are capable of adapting to unfamiliar accents (e.g., White & Aslin, 2011), little is known about how accent distance influences adaptation. We sought to address this issue by examining whether listeners utilize different adaptation strategies depending on the type of accent exposure they receive (near or distant from their native accent) and their accumulated linguistic knowledge (young children vs. adults). An accent that is far from a listener’s own may induce a general expansion strategy (Schmale et al., 2015), whereby listeners become more tolerant of pronunciation deviations overall, even ones to which they have never been exposed. When encountering a near accent, on the other hand, a linguistically-based strategy may be employed, with listeners making targeted phonetic adjustments to only the specific categories affected by the accent.

Two phases were administered: exposure and test. During exposure, 27-month-old toddlers (n=72) and adults (n=72) were presented with individual images on a screen along with concurrent auditory labeling. Items were labeled in one of 3 artificial accents: Near, Far or Farther. In the Near Accent, only one target pronunciation change was present, (e.g., “æ” pronounced as “aw”). In addition, fillers produced with a standard Canadian English accent were included. The Far Accent provided the same target pronunciation change, but now the fillers contained a range of pronunciation changes (e.g., milk as “melk”, horse as “horsh”). The Farther Accent contained the same number of pronunciation changes as the Far Accent but larger phonetic deviations (e.g., milk as “mulk”, horse as “horp”). The test phase consisted of an eye-tracking procedure. The speaker instructed listeners to look at one of two objects on the screen. Listeners heard unaccented real words (e.g., duck), nonwords (e.g., chone), items with the trained pronunciation change (“æ” ⇒ “aw”), and an untrained pronunciation change (“b” ⇒ “s”) to test their willingness to accept pronunciation deviations. If hearing the Far or Farther accent induces a general expansion strategy, then listeners would be predicted to accept items with both trained and untrained pronunciation changes as being words.

Results revealed all groups of children (Fig. 1A) and adults (Fig. 1B) successfully recognized real words as well as items with the trained pronunciation change. For adults, listeners in the Farther Accent group were more likely to accept untrained pronunciation changes, suggesting that hearing numerous, large phonetic deviations induced a general expansion strategy. Interestingly, young children exposed to Far and Farther Accents were less accurate at recognizing real words relative to the Near Accent group. This suggests that exposure to a highly variable accent with multiple pronunciation changes enhances children’s uncertainty about the identity of unaccented pronunciations. Overall, these findings indicate that accent distance can impact how listeners adapt to accented segments but can also have ramifications for how they interpret canonical pronunciations.

Authors

• Angela Cooper, University of Toronto
  Presenting Author

• Caterina Bordignon, University of Toronto
  Non-Presenting Author

• Elizabeth K Johnson, University of Toronto Mississauga
  Non-Presenting Author