Early verbal working memory as a predictor of later language outcomes in late talkers Jayne Newbury, Thomas Klee, Stephanie Stokes and Catherine Moran University of Canterbury, New Zealand Disclosure statement The authors declare they have no relevant financial or nonfinancial relationships to disclose 1
Does this child have a language impairment? Used with permission Multiple risk factor models Reilly et al. (2010) - environmental, child and family factors Significant predictors were: gender; birth weight; non-english speaking background; socio-economic disadvantage; family history; mothers education level / vocab; late talker status at two Developmental factors (e.g. receptive language, gesture, symbolic play, non-verbal IQ etc.) Best models are moderately successful Individual prediction less accurate Few studies have considered underlying processing variables 2
Baddeley s model of working memory Central Executive Verbal working memory Visuospatial Sketchpad Visual Semantics Episodic buffer Episodic Long term memory Phonological loop Language Phonological short term memory Why working memory? Central Executive Verbal working memory Visuospatial Sketchpad Episodic buffer Phonological loop Visual Semantics Episodic Long term memory Language Phonological short term memory 3
Study Timeline Time 1 Ages 24-30 months January June 2012 N = 79 Language Cognition Working Memory Child and parent demographics Time 2 Ages 41-49 months July-December 2013 N = 78 Test Protocol at Time 1 MacArthur-Bates Communicative Development Inventory (Words and Sentences) Preschool Language Scales 4 Mullen Visual Reception Scale Key Word Working Memory Test Parent questionnaire Otoacoustic emission hearing screen Time 2 Outcome Measure Preschool Language Scales -4 4
Key Word Working Memory Task 5
Description of Sample 79 children at Time 1; 78 at Time 2 English exposure > 80% of the time Mean age at Time 1: 26.6 months 68.4% male Aimed for 50:50 late talkers / typically developing children None had any developmental diagnoses at the time Skewed towards higher parent education Late Talker Criterion Late talkers were defined as those scoring 1 standard deviation below the mean (or lower) in expressive vocabulary OR no two word combinations 24 late talkers / 55 typically developing at 24-30 months* *The child who left the study before T2 was typically developing. 6
Research Question #1 What group differences were observed in verbal working memory (VWM) between the late talkers and typically developing children at age 24-30 months? Hypothesis: Late talkers will score more poorly on VWM than typically developing children, even when controlling for receptive language differences. LT and TD group mean comparison at 24-30 months Variable T-test Significance Effect size (d) Receptive Language t=6.55 (df=77) p<.001 1.54 Receptive Vocabulary t=6.65 (df=77) p<.001 1.54 Visual Cognition t=2.57 (df=74) p<.01 0.66 7
Verbal Working Memory Typically developing Late talkers This difference was significant, t= 3.90, df = 77, p<.001, d=0.93 Controlling for language The late talkers and typically developing groups were compared on VWM using an analysis of covariance. When controlling for receptive language, the late talkers no longer scored significantly lower in VWM than the typically developing group. F(1,76) =.74, p =.39, partial eta squared =.01 8
Summary No evidence to suggest that low VWM is a particular characteristic of late talkers over and above lower receptive language VWM could be useful as a predictor for later outcomes as a range of scores are present in the late talker group Research question #2 What are the bivariate and multivariate associations between the behavioural measures at 24-30 months and the expressive language score on the PLS-4 at 41-49 months? Hypotheses: Early VWM will strongly predict later expressive language on a bivariate level Early VWM will account for unique variance in expressive language outcomes in a multivariate regression model 9
Does early VWM predict later expressive language? Bivariate correlations (one tailed) T1 assessment variables (24-30 months) T2 PLS-4 expressive language raw scores (41-49 months) Verbal working memory.78*** Receptive language.74*** Expressive language.72*** Expressive vocabulary.66*** Visual cognition.54*** Note. ***p <.001 Does early VWM predict unique variance in later expressive language? Multivariate linear regression model This model was significant, F(5,68) = 32.01, p<.001, and accounted for 70% of the variance in PLS-4 expressive language raw scores Model B SE B Std. β Significant unique variance Age at initial assessment -0.71 0.28 -.20* 3% Expressive vocabulary 0.01 0.00.21 VWM 0.41 0.11.42*** 7% Receptive language 0.24 0.14.22 Visual cognition 0.25 0.17.15 Note. p<.10 *p<0.05 **p<0.01 ***p<.001 10
Summary and conclusion VWM predicted unique variance in language outcomes on a group level VWM may play a role in language acquisition Only 4/78 children had low expressive language at outcome unable to comment on whether the VWM task would be useful clinically to improve prediction of late talker outcomes Larger sample size needed! 11