Early language variation and working memory: A longitudinal study of late talkers and typically developing children
Thesis DisciplineSpeech and Language Sciences
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
This research explored whether variation in working memory ability helps account for the wide variation in toddlers' language skills and improves predictive models of language outcomes over time. A cohort of typically developing (TD) (n = 55) and late talking children (n = 24) were assessed at two time points. The initial assessment took place at ages 24-30 months and the outcome assessment occurred 18 months later, when the children were aged 41-49 months. The assessment battery included standardised tests of language and visual cognition; assessments representing aspects of Baddeley's model of working memory: phonological short term memory (PSTM), a measure of processing speed, verbal working memory (VWM), visual spatial working memory (VSWM), and a parent report questionnaire of executive functioning (EF). Study 1 explored the associations between these aspects of working memory and concurrent expressive vocabulary at ages 24-30 months and examined group differences in the measures between TD and late talking children. Study 2 explored associations between aspects of working memory and concurrent expressive language in the same cohort at 41-49 months of age. Group differences in the measures between resolved late talkers (RLTs) and TD children were explored. Finally Study 3 explored the ability of the measures used at 24-30 months to predict language outcomes at 41-49 months. These results were considered in relation to the prediction of language outcomes on group and individual levels. Overall the results indicated a strong relationship between early PSTM and early language measures. A novel finding was that PSTM was significantly lower in the late talking and RLT groups compared with the TD groups, even after controlling for group differences in language and phonology at both time points. This confirms previous research that PSTM plays a role in early expressive vocabulary acquisition, and suggests that early PSTM deficits may be a causal factor for some cases of late talking. For the whole group, three working memory variables (VWM, Emotional Control and Shift) measured at 24-30 months added unique variance to predictive models in total language scores at 41-49 months after previously established early predictors (receptive language and parent education) had been entered into the hierarchical regression model (receptive language R²Δ = 59%; parent education R²Δ = 2%; VWM R²Δ = 8%; Emotional Control R²Δ = 1% and Shift R²Δ = 2%). This is another novel finding which supports the concept of working memory playing a unique role in language acquisition between the ages two and four years. Processing speed did not contribute unique variance to regression models predicting language when other working memory measures were included. The A not B task (measuring VSWM) did not correlate with language. There were concerns with construct validity with the EF parent report measure (Behaviour Rating Inventory of Executive Function – Preschool Version), which meant that the results from this assessment were interpreted with caution. In terms of clinical outcomes, 83% of the late talkers resolved their language delays over the 18 month period, but as a group showed a seven-fold increase in being identified for clinical concerns at the outcome assessment than children who were not late talkers. The majority of these concerns were for poor phonology. While early VWM, Shift and Emotional Control added unique variance to outcome total language scores on a group level, they did not improve prediction of individual outcomes in language impairment status at 41-49 months. Early receptive language delay was a more powerful predictor of later language impairment than late talking in this cohort, as these children (n = 9) showed only a 44% rate of resolution.