The impact of speech complexity on preschoolers’ attention and learning

• In Event: Enhance cognitive functions across development: factors, strategies or interventions

Abstract

Children routinely hear language spoken directly to them, and also overhear conversations among others. But how do children decide what speech to tune into and learn from? Prior research suggests that infants selectively attend to stimuli that are neither too simple nor too complex (Kidd et al., 2012; 2014; Gerken et al., 2011), consistent with the idea that children attend to stimuli that they can best learn from. Here, we test this idea in the context of language development, by asking whether preschoolers’ attention to spoken language (1) varies as a function of speech complexity, and (2) is directly related to their learning (an untested assumption in previous work).

47 preschoolers (4-6 years, M=4.6) were assigned to two between-subjects conditions, SIMPLE and COMPLEX, where they heard narrations of the same story at two different levels of linguistic complexity, while observing illustrations on an eyetracker (Figure 1). Children in the SIMPLE condition heard a narration that exclusively used familiar words (acquired by 30 months; Fenson, 2007), while children in the COMPLEX condition heard the story narrated using later-acquired words (after ~7 years; Kuperman, 2013). On each of the six pages of the story, an attention-getting GIF competed with the storybook illustration for children’s visual attention. After a first obligatory loop of the narration for each page, the audio for that page continued to repeat for up to five more complete repetitions; however, the next page could be triggered if children fixated for a continuous 1.5s on the distractor GIF, potentially reflecting that they were no longer attending to the story. To quantify children’s attention to the story, we measured the proportion of their gaze to the illustration versus to the GIF, as well as their overall listening time on each page. Finally, at the end of the story, we tested whether children had learned words introduced in the story (i.e., the later-acquired words introduced in the COMPLEX condition; e.g., abscond, flummoxed) and had comprehended the story (e.g., “Who were the boy and the dog looking for?”).

Preschooler’s attention was responsive to linguistic complexity. Compared to children in the COMPLEX condition, children in the SIMPLE condition spent a greater proportion of their time on the story illustration (MSIMPLE=51%; MCOMPLEX=44%; p=.006) and showed longer listening times (MSIMPLE=15.41s/page; MCOMPLEX=10.61s/page; p=0.03). Remarkably, condition (SIMPLE vs. COMPLEX) did not predict differences in learning, but children’s own attentional behavior did. Linear models revealed that children’s total time on the illustrations (in log ms, β=8.87, F(1, 40)=9.64, p=.003) and listening (β=6.28, F(1, 40)=5.64, p=.02) were significant predictors of listening comprehension accuracy, controlling for age and condition (Figure 2).

That learning differences were selectively predicted by differences in children’s online attention, rather than in speech complexity, is consistent with the idea that preschoolers direct their attention to speech that they can learn from. These findings extend prior work with infants and support a view of children as active learners who seek information in their environment that is appropriate given their own level of competence, enhancing their own learning and development.

Authors

• Ruthe Foushee, University of Chicago
  Presenting Author

• Mahesh Srinivasan, University of California - Berkeley
  Non-Presenting Author

• Fei Xu, University of California - Berkeley
  Non-Presenting Author