The eyes of pre-verbal infants reveal the effects of book reading on word learning

• In Event: Poster Session 05
  In Poster Session: PS 05 Topic - Language, Communication

Abstract

Although there is abundant literature on how reading books to preliterate children relates to their later language abilities (Flack et al., 2018), not much is known about benefits of book reading to pre-verbal infants (Karrass & Braungart-Rieker, 2005). We report a longitudinal study investigating how vocabulary comprehension in 13-15-month-olds, assessed using eye-tracking, is affected by parent-reported book reading 3-7 months prior to testing. Our subjects are 31 Korean mother-child pairs who met our analysis criteria from a larger pool. We conducted a home literacy environment survey and a day-long LENA recording at around 7 months of age, and administered the MCDI-K (Pae, 2003) and an eye-tracking word comprehension test at around 14 months. The survey included questions about book reading, such as frequency of weekly reading (WEEKLY-READING) and age at which parents think children can begin reading (AGE-START-READING). The eye-tracking word comprehension test consisted of 40 target words presented in 20 yoked pairs. Each trial presented a visual stimuli pair with the audio of target word embedded in three sentences. We measured ACCURACY by examining looks to target during the 300-2000ms window post-audio onset on only stimuli where the trials for both yoked pairs were successfully tracked for the participant. Spearman correlation tests revealed a significant correlation between ACCURACY and WEEKLY-READING (r = .51, p < .01, n = 23) and a marginally significant correlation between ACCURACY and AGE-START-READING (r = -.41, p = .06, n = 23). We additionally found a significant correlation between the MCDI-K percentile and LENA’s hourly ADULT-WORD-COUNT (r = .63, p < .01, n = 23), but not between WEEKLY-READING and ADULT-WORD-COUNT (r = .29, p =.18, n = 24). Linear mixed effect models with ACCURACY as the dependent variable, age as a covariate, and random by-item intercepts revealed a significant effect of WEEKLY-READING (β = .04, SE = .01, p < .05) and of ADULT-WORD-COUNT (β = .0001, SE = .00004, p < .05) in separate models. When put into the same model, ADULT-WORD-COUNT lost significance (β = .00004, SE = .0001, p = .41) while WEEKLY-READING became marginally significant (β = .03, SE = .02, p = .08). Our results suggest that children whose parents read with them more often when they begin to segment words (Jusczyk & Aslin, 1995) might have a larger vocabulary than those who begin reading at a later age. Our finding raises the question of whether the activity of reading might be the catalyst of word learning over general language exposure or parent-child interaction. To address this question, it would be advantageous to focus on child-directed speech since simply overheard adult speech is reported to be unrelated to vocabulary outcome (Weisleder & Fernald, 2013). We are currently coding the day-long LENA recordings by manually classifying each one-minute segment as child- or adult-directed so that we can estimate the amount of child-directed speech in input. With these data, we hope to further elaborate on the individual contributions of general child-directed input vs. specific reading time to child vocabulary development.

Authors

• Eon-Suk Ko, Chosun University
  Presenting Author

• Margarethe McDonald, University of Wisconsin-Madison
  Non-Presenting Author