Investigating Predictors of Individual Differences in Productive Vocabulary and Their Ability to Identify Late Talking Toddlers.

Jago, L (2017). Investigating Predictors of Individual Differences in Productive Vocabulary and Their Ability to Identify Late Talking Toddlers. Presented at the 3rd Annual LuCiD Language and Communicative Development Conference

Abstract:

Understanding individual differences in language acquisition is fundamental to developing our knowledge of language impairments. Currently, late talkers are identified on the basis of productive vocabulary, usually, at 24 months. However, earlier late talking status has been shown to be a poor predictor of later language abilities (Fernald & Marchman, 2012; Dale, Price, Bishop & Plomin, 2003).
In this study, we investigate predictors of individual differences in productive vocabulary at 24 months as measured by the Communicative Development Inventory (CDI). These predictors include earlier measures of receptive and productive vocabulary, non-word repetition, quality of input, gender, family history of speech and language impairments, and speed of processing familiar items. Following this, we investigate the strength of these predictors in identifying late talking status at 24 months as determined by productive vocabulary scores using the CDI. 

The first analysis will investigate the association between these measures and productive vocabulary scores. A regression analysis will investigate the ability of these predictors to explain the variance in productive vocabulary at 24 months.
A second analysis will investigate the strength of these predictors in identifying late talking children from typically developing children. The receiver operating characteristic (ROC) analysis will measure the overlap in scores obtained by children identified as late talking compared to children with typically developing language. This analysis will also provide details on scores with optimal sensitivity and specificity for classifying children as Late Talkers.

The preliminary results for the sensitivity and specificity analysis suggest that a number of the predictors successfully identify late talking children, yielding acceptable levels of sensitivity and specificity. In addition there was little overlap between scores achieved by late talkers and typically developing children, as shown by the area under the ROC curves. Successful predictors were the number of words understood at 18 months, the number of words understood and spoken at 18 months, mean length of utterance at 24 months and non-word repetition at 25 months. However, input averaged across 18-21 months, as measured by adult word count using LENA technology, did not distinguish between late talking and typically developing children. Analyses for the predicting individual differences in vocabulary at 24 months is ongoing and will be presented.

Implications for theories of individual differences and criteria for identifying children at risk of late talking will be discussed.