A deep look into the developing lexicon: Revelations from covert picture-naming

Ngon, C. & Peperkamp, S.

Laboratoire de Sciences Cognitives et Psycholinguistique

Towards the end of their second year of life, infants have stored word-forms with phonological detail but often articulate those words incorrectly or do not attempt to produce them at all. We focus on comprehended words that are not yet produced overtly, and show that infants can 1. covertly produce these words, and 2. categorize them according to their length. Two groups of 21-month-old French-learning infants are tested in an anticipatory eye-movement procedure (McMurray & Aslin, 2004); one group is presented with mono- vs. trisyllabic words and the other with mono- vs. disyllabic words. The experiment starts with a learning phase, where infants are trained to anticipate the appearance of an object to the left or right side of a screen depending on the length of the object's label (mono- vs. tri- or disyllabic). The anticipation window occurs after the central presentation of the image of the object accompanied by its auditory label. In the test phase, new objects are presented, but without their auditory label. Thus, the only way for infants to anticipate the object's reappearance is to internally generate its name themselves and inspect its length. Crucially, each infant is tested on words s/he comprehends but does not yet pronounce according to parental report. So far we have tested 45 infants divided over the two groups, and found that in the test phase, infants orient significantly more often towards the correct side than the incorrect side for the mono- vs. trisyllable group (t(22) = 3.29; p = 0.003) but only marginally so for the mono vs. disyllable group (t(21) = 1.48; p = 0.077). These results show that infants can covertly generate unpronounced word-forms, and categorize them as mono- vs. trisyllabic. Testing additional infants will determine whether successful categorization holds for words with a smaller phonological difference.