The Effects of Speaker Variability and Bilingualism on Cross-situational Word Learning

• In Event: Poster Session 14
  In Poster Session: PS 14 Topic - Language, Communication

Abstract

Cross-situational statistical word-learning (XSWL) - the ability to learn words by tracking co-occurrence statistics of words and their referents over time, is a fundamental mechanism underlying lexical acquisition. To date, research examining XSWL has not considered how input variability and individual differences in attention and language experience interact and influence the learning of word-to-referent mappings. In two experiments, we examine the effects of speaker variability, sustained attention, and bilingualism in children’s (Experiment 1) and adults’ (Experiment 2) XSWL performance. In Experiment 1, 40 English monolingual children (MeanAge = 5.73 years; SDAge = 1.07) and 40 Spanish-English bilingual children (MeanAge = 5.85 years; SDAge = 0.83) ages 4-7 completed a XSWL task in a Single Speaker Condition and a Multiple Speaker Condition. Results indicated that speaker variability neither facilitated nor hindered XSWL and that speaker-variability effects did not fluctuate across the two language groups. Notably, exposure to multiple speakers facilitated XSWL in children with poorer sustained attention skills, suggesting that variability in the input may be especially useful to children with poorer cognitive processing abilities.

In Experiment 2, we test whether mature attentional skills are sensitive to speaker variability effects, and whether bilingualism modulates how adults build associations between words and objects. We adopted an online data collection procedure hosted on Gorilla. Data collection is ongoing, and so far 13 English monolinguals and 16 bilinguals between the ages of 18-41 have been recruited. Bilingual speakers spoke different languages such as Arabic, Farsi, Gujarati, Hindi, Spanish, and Vietnamese. Adults completed a XSWL task where they were asked to learn the names of 18 novel objects. Nine word-object pairs were produced by a single female speaker and the other nine word-object pairs were produced by 8 different speakers: four men and four women. Our preliminary results suggest that adults learned object-referent pairings above chance for words produced by a single speaker (Mproportion correct =.39, SD =.23, t(28)=3.35, p < .01 ), and by multiple speakers (Mproportion correct =.42, SD=.22; t(28)=4.24, p < .001). Model results revealed no significant effects of speaker variability or bilingual experience on word-learning accuracy or reaction time (ps > .05) (Figure 1). Future analyses will examine individual differences in sustained attention skills in adults.

Together, the results suggest that while children and adults can successfully accommodate speaker variability, it neither facilitates nor impedes building associations between words and objects over time. Results also indicate that bilingual language experience may not facilitate how listeners accommodate acoustically variable input, either in childhood or in adulthood. It will be important to test whether manipulating perceptual features of objects, and simultaneously learning from multiple speakers and examplars yields differences in word-learning in children and adults.

Authors

• Kimberly Crespo, University of Wisconsin - Madison
  Presenting Author

• Margarita Kaushanskaya, University of Wisconsin-Madison
  Non-Presenting Author