Cross-Situational Word-Face Learning in Children with ASD

Background: It was reported that children with autism spectrum disorder (ASD) have difficulty in learning words via social cues (e.g., speaker’s eye gaze; Baron-Cohen et al., 1997). However, some children with ASD acquire vocabularies as rich as typically developing (TD) children. One of potential efficient strategies is a cross-situational learning, which is a mechanism for learning the words across multiple trials even when there is no definite cue for the word-object correspondence in one trial. We previously demonstrated that the cross-situational learning is an efficient strategy for children with ASD (Akechi et al., 2012, IMFAR). Children with ASD also reportedly have difficulty in face memory (Weigelt et al., 2011). However, whether cross-situational learning is also efficient to learn associations between words and faces in ASD was not investigated yet.

Objectives: To investigate whether children with ASD learn words-faces association effectively using cross-situational learning.

Methods: Participants consisted of 20 children with ASD (mean age 9.1; range 6-12) and 20 TD children (mean age 9.1; range 7-12), who were matched on verbal mental age (VMA). There were 6 novel words and 6 novel faces. In the training phase, two novel faces were presented on the monitor and two corresponding novel words were presented via the loudspeaker in each trial. There is no definite cue for the word-face correspondence in one trial. Each word-face pairs was presented 10 times. In the test trials, two faces and one word were presented and the participant was asked which face is the referent. Each face was presented twice as a target and twice as a non-target. Thus, there were 12 test trials in total.　The procedure was same as the previous study (Akechi et al., 2012), except that the objects were replaced with faces.

Results: The performance in the test trials in both the ASD (p < .005) and the TD group (p < .001) were above chance level (6/12 = 50%). However, there was a significant difference between groups in the performance (p < .05). Additionally, the analysis across experiment showed that the performance of children with ASD in the present study was lower than that of those in the previous study (p < .005) but there was no significant difference in the TD children (p > .05).

Conclusions: Results suggest that children with ASD can learn novel face-word associations using cross-situational learning, but compared to TD children, they have difficulty learning the name of human face cross-situationally.