Frost, R. L. . 1 , Monaghan, P. . 1 & Christiansen, M. H. 2
1 Lancaster University
2 Cornell University
Learners can extract transitional information from speech and use it to help identify word boundaries and linguistic regularities. Critically, studies suggest statistical language learning may benefit from the presence of high-frequency marker words, which may act as anchors for segmentation to occur around, while also assisting with grammatical categorisation.
To test this, we familiarised adults with continuous speech comprising repetitions of bisyllabic target words, and compared learning to the same language but with high-frequency monosyllabic marker words preceding target words, and distinguishing them into two otherwise unidentifiable categories. Participants completed a 2AFC test of segmentation, and a similarity judgement categorisation test. We tested transfer to a cross-situational word-action/object learning task, where target word categories were either consistent or inconsistent with the action/object distinction. This was followed by a vocabulary test, which assessed learning of names for actions and objects.
Participants segmented the speech better than chance, but only the marker word condition demonstrated categorisation on the transfer task. For the vocabulary test, scores for the marker word condition were significantly higher for participants receiving consistent, rather than inconsistent, word-action/object pairings. Results indicate that high-frequency marker words may assist grammatical categorisation at the point where speech segmentation is just being learnt.