Evidence for an Own-Age Advantage in Children’s Voice Identification

• In Event: Poster Session 14
  In Poster Session: PS 14 Topic - Language, Communication

Abstract

Adults find it difficult to identify child voices compared to adult voices (e.g., Cooper et al., 2020; Creel & Jimenez, 2012). One explanation is that child voices are more acoustically variable than adult voices (e.g., Lee et al., 1999), making them difficult to identify. Alternatively, voice identification may be influenced by shared similarities between the talker and listener. For example, adults might be best at identifying adult voices while children might be best at identifying child voices (akin to the own-age advantage reported in face recognition; Anastasi & Rhodes, 2005). Indeed, people tend to have greater experience with similar aged individuals (e.g., children attend preschools with own-age peers), stronger motivation to attend to age-matched peers, and greater ease perceiving speech attributes of those with similar vocal tract anatomy (e.g., Schuerman et al., 2015). In the current study, we examine the possibility of an own-age advantage in children’s voice identification.
29 5-year-old children (target N = 48; 17 female) participated in an AX “same-different” talker discrimination task that featured child and adult talkers. All children were recruited from the Greater Toronto Area and were English dominant, with at least one locally accented parent. Auditory stimuli consisted of 32 single word recordings by the same 12 Canadian-English toddlers (6 female) who were longitudinally recorded at 2.5-, 3.5-, and 5.5-years-old and 12 adults (6 female). During the task, participants were presented with pairs of words (e.g., strawberry-boat) by gender-matched talkers and were asked to indicate whether the words were produced by the same talker or two different talkers. Each participant completed a total of 16 randomized trials (8 “same” and 8 “different” trials) that featured 16 different talkers, with 4 talkers from each age group. If an own-age advantage in voice identification exists, we expected listeners to perform best with speech by their age-matched peers, the 5.5-year-olds.
Our results indicate that while the effect of talker age on performance was only marginally significant with the current sample size (p = 0.09), 5-year-old listeners, on average, performed the best and significantly above chance level (0.5), with 5.5-year-old talkers (M = 0.65; p < 0.001) compared to the all other talker ages (Figure 1). Their performance was also significantly above chance with 2.5-year-old talkers (M = 0.58, p < 0.01), but was not above chance with 3.5-year-old talkers (M = 0.57, p = 0.052), or adult talkers (M = 0.51, p = 0.39).
Thus far, our findings suggest that 5-year-olds demonstrate an own-age advantage, as they identified the voices of their age-matched peers best. However, we plan on investigating this further by conducting an online version of this task in order to collect additional data. Given the efficient nature of online data collection, we will increase our target sample size to 60 children, as well as compare performances by younger children and adults who vary in their experience with children (e.g., preschool teachers, university students) to examine if this apparent own-age advantage in voice identification is impacted by frequent exposure to individuals of particular ages.

Authors

• Madeleine Yu, University of Toronto
  Presenting Author

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