Funding was provided by NIH grants F32 HD065382 and K99-R00 DC013795 to CQ and NSF grant 0950601 to LAG.
Laboratory Phonology: Journal of the Association for Laboratory Phonology
Infants struggle to apply earlier-demonstrated sound-discrimination abilities to later word learning, attending to non-constrastive acoustic dimensions (e.g., Hay et al., 2015), and not always to contrastive dimensions (e.g., Stager & Werker, 1997). One hint about the nature of infants’ difficulties comes from the observation that input from multiple talkers can improve word learning (Rost & McMurray, 2009). This may be because, when a single talker says both of the to-be-learned words, consistent talker’s-voice characteristics make the acoustics of the two words more overlapping (Apfelbaum & McMurray, 2011). Here, we test that notion. We taught 14-month-old infants two similar-sounding words in the Switch habituation paradigm. The same amount of overall talker variability was present as in prior multiple-talker experiments, but male and female talkers said different words, creating a gender-word correlation. Under an acoustic-similarity account, correlated talker gender should help to separate words acoustically and facilitate learning. Instead, we found that correlated talker gender impaired learning of word-object pairings compared with uncorrelated talker gender—even when gender-word pairings were always maintained in test—casting doubt on one account of the beneficial effects of talker variability. We discuss several alternate potential explanations for this effect.
Quam, C , Knight, S and Gerken, L 2017 The Distribution of Talker Variability Impacts Infants’ Word Learning. Laboratory Phonology: Journal of the Association for Laboratory Phonology 8(1): 1, pp. 1–27, DOI: https://doi.org/10.5334/labphon.25