Black, A. & Hudson Kam, C.
The University of British Columbia
The present experiment looks for developmental differences in the stability of learners? memories for trisyllabic sequences extracted during an auditory statistical learning (SL) task. Eighty adults and 88 child learners (ages 5-10, data collection ongoing) were exposed to a 2-min continuous speech stream. Participants? memories were tested through 2-AFC comparisons of words, part-words, and fake-words. Fake-words were constructed by manipulating the initial, medial, or final syllables of words. This manipulation allows us to examine the relative strength of memories across syllabic positions.
Adults segment the language as expected, and show no positional asymmetries in their memory for word units. The children reveal an unexpected pattern: only 7-yos (n=18) successfully segment the language. Eight to 10-yos (n=57) demonstrate successful segmentation at the beginning of the testing phase - but the subsequent exposure to part-words and fake-words during testing interferes with this knowledge. Finally, 5- and 6-yos (n=32), who fail to distinguish words from part-words, demonstrate successful learning of the correct initial and final syllable transitions in word~fake-word tests. In sum: between the ages of 5-10, units extracted from a SL task do not have the same representational status as those found in adulthood.