Investigating the Shape Bias for Word Learning in Children with Autism Spectrum Disorders

Friday, May 16, 2014
Atrium Ballroom (Marriott Marquis Atlanta)
E. Potrzeba1, D. A. Fein1 and L. Naigles2, (1)Psychology, University of Connecticut, Storrs, CT, (2)University of Connecticut, Storrs, CT

Research in word learning strategies for children with autism spectrum disorder (ASD) has shown that children with ASD parallel typically developing (TD) children with word learning strategies such as the noun bias and mutual exclusivity1. However, they have not been shown to display a shape bias, which occurs when children extend a novel word to objects of the same shape, rather than to other physical features such as color or texture2. Previous demonstrations of children with ASD lacking a shape bias included only a small sample (n=15) and restricted age range; moreover, performance correlated with standardized test scores, suggesting that some children with ASD did exhibit a shape bias. We hypothesized that subgroups of children with ASD may have the ability to use the shape bias for word learning.


Extending Tek et al., (2008) we increased the participant total to 32 children with ASD and collected data across 6 visits rather than 4. This allows us to look for subgroups and for a later-emerging shape bias. Error analyses were conducted to investigate alternative looking patterns.


 Children with ASD (n=32, M Mullen EL=17.5, SD=7.2, M-RL=21.1, SD=9.2) were matched with TD children (n=35, M EL=20.5, SD=5.1, M-RL=20.9, SD=3.6) at visit 1 of a longitudinal study. Children were visited every four months. The shape bias was tested with the intermodal preferential looking (IPL) paradigm:  Five novel objects were presented, each was followed by two options; one matched the original in shape while the other matched the original in color.  During the first (NoName) presentation of each set, the baseline audio asked “Which one looks the same?” During the second (Name) presentation, the original object was given a label (e.g., “dax”) and the test audio asked “Where is the dax?” Children who looked longer at the shape match during the Name trials than during the NoName trials demonstrated a shape bias.


TD children showed the shape bias at 20 months (t(22)= -1.787, p<.05, one-tailed)  and at each subsequent visit (t(30)s > 2.0, ps<.05). Children with ASD did not show a significant shape bias, as a group, at any visit (ps>.11). Children with ASD’s errors primarily involved looking longer at the color match (18.8% of all trials) during the test trials across visits; fewer children simply looked at shape (8.8%) or color (11.5%) during both NoName and Name trials. Eighty percent of TD children demonstrated a shape bias in over 60% of their visits. In contrast, the majority of children with ASD showed a shape bias on fewer than 60% of visits; however, three children did show a shape bias at all visits and ten others at more than 60% of visits.


Even at 52 months, most children with ASD did not show a shape bias; however, we did observe several children exhibiting the bias consistently. Further analyses will scrutinize the language and socialization levels of the children who did vs. did not demonstrate a shape bias to investigate the origins of shape bias abilities and difficulties.