Samara, A. 1 , Smith, K. 2 , Brown, H. 3 , Fehér, O. 2 & Wonnacott, E. 1
1 Division of Psychology and Language Sciences, University College London
2 School of Philosophy, Psychology and Language Sciences, University of Edinburgh
3 School of Psychology, Warwick University
Languages exhibit variation which may be conditioned/ partially conditioned on sociolinguistic cues such as social context or gender (e.g., Smith et al., 2013). Is this driven by child or adult language learners/users? As a first step, in two artificial language learning experiments we investigate whether 6-year-olds and adults are sensitive to speaker identity cues. In the exposure languages, nouns were followed by meaningless variable particles (e.g., either bup or kem). Two speakers produced the language: in the fully predictive condition speaker1 used 100% kem, speaker2 100% bup; in the partially predictive condition speaker1 used 75% bup (25% kem), speaker2 75% kem (25% bup). Four exposure sessions took place over four days. Production and judgment tests probed learning of the association between speaker and particle usage. Tests used both trained and untrained nouns. Preliminary analyses (n = 36) suggest both children and adults learned and generalized in the fully predictive condition. Learning in the partially predictive condition posed greater difficulty to both groups and was reliable only in judgments with trained nouns. This suggests sensitivity to statistical sociolinguistic cues, but this learning may be tied to specific items. Data collection is ongoing and we will also compare monolingual and bilingual participants.