2018 Friday Session A 0900

Friday, November 2, 2018 | Session A, East Balcony | 9am

Early availability and phonological bias in the acquisition of noun classes
J. Culbertson, H. Jarvinen, F. Haggarty, K. Smith

Previous research on acquisition of noun class systems, like grammatical gender, has found that child learners rely disproportionately on phonological cues (e.g., noun endings) to determine the class of new nouns (Karmiloff-Smith 1981, Perez-Pereira 1991, Demuth 2000, Rodina & Westergaard 2012), even when competing semantic cues (e.g., natural gender, animacy) are demonstrably more reliable in their language (Gagliardi & Lidz 2014).

Here we test two hypothesized explanations for this surprising finding. One explanation appeals to the ​intake​; phonological information is available to children from the earliest stages of acquisition, before noun meanings have been learned (Polinsky & Jackson 1999, Demuth 2000, Gagliardi et al. 2017). Initial representations of class may therefore be based on phonological features of nouns, only later integrating semantics. This is supported by artificial language learning studies showing that adults are more likely to use an earlier learned cue – whether phonological ​or​ semantic – to assign class to novel nouns (Culbertson et al. 2017). However, another possible explanation is that children have an a priori preference for phonology because it is local or intra-linguistic, while semantic information is in some sense non-local or external (Perez-Pereira 1991, Gagliardi & Lidz 2014, Gagliardi et al. 2017). Here, we report the first tests of these two hypotheses using artificial language learning with children.

In Experiment 1, we trained 6-7 year-olds (N=39) on an artificial noun class system with a perfectly reliable semantic cue (animacy-based; nouns referring to aliens in one class, nouns referring to planets in another) ​or​ phonological cue (vowel-reduplication; nouns with the form CiCi in one class, nouns with the form CaCa in another). Evidence for the existence of two noun classes came from a plural marker which differed by class. After training, children were shown singular forms and asked to produce the plural. We found that children (and adult controls, N=40) are equally successful at learning either cue (Figure 1). In Experiment 2, we took those two cues and trained children (N=20) on a system in which they were confounded: both cues were present, and predicted class equally well. During testing we presented children with novel items in which the cues conflicted (the noun meaning was compatible with one class, but the noun phonology was compatible with the other). When cues conflicted, a bias became apparent: child learners were more likely to use the phonological cue (while adult controls, N=20, consistently used the semantic cue, Figure 2). Finally, in Experiment 3 we investigated the effect of staging by presenting children with either the phonological cue (no pictures, N=20) or the semantic cue (no nouns, N=20) ​first, following by training with both cues. Here, as with adults in Culbertson et al. (2017), children’s reliance on phonology was strengthened by having the phonological cue available first, and weakened by having the semantic cue available first (Figure 3).

These results ​shed light on how representations of noun class develop during learning, ​suggesting that both early availability ​and​ a bias favoring phonological cues may contribute to children’s over-reliance on phonology.

References

Culbertson, J., Gagliardi, A., & Smith, K. (2017). Competition between phonology and semantics in noun class learning. ​Journal of Memory and Language​, ​92​, 343–358.

Demuth, K. (2000). Bantu noun class systems: loanword and acquisition evidence of semantic productivity. In ​Systems of Nominal Classification ​(pp. 270–292). Cambridge, UK: Cambridge University Press.

Gagliardi, A., Feldman, N., & Lidz, J. (2017). Modeling statistical insensitivity: Sources of suboptimal behavior. ​Cognitive Science​, ​41​, 188–217.

Gagliardi & Lidz. (2014) Statistical insensitivity in the acquisition of Tsez noun classes. ​Language​, ​90​, 58–89.

Karmiloff-Smith, A. (1981). ​A functional approach to child language: A study of determiners and reference​. New York, NY: Cambridge University Press.

Mills, A. E. (1985). The acquisition of German. In D. Slobin (Ed.), ​The crosslinguistic study of language acquisition, Vol. 1 ​(pp. 141–254). Hillsdale, NJ: Lawrence Erlbaum.

Perez-Pereira, M. (1991). The acquisition of gender: What Spanish children tell us. ​Journal of Child Language​, ​18​, 571–590.

Polinsky, M., & Jackson, D. (1999). Noun classes: Language change and learning. In B. A. Fox, D. Jurafsky, & L. Michaelis (Eds.), ​Cognition and Function in Language ​(pp. 29–50). Stanford, CA: CSLI.