Statistical learning in children with DLD

Lukács, Á., Lukics, K., & Dobó, D. (2021). Online Statistical Learning in Developmental Language Disorder. Frontiers In Human Neuroscience, 15.

Aim of the paper:

Children learn language by recognising patterns and regularities in language. They learn words based on reoccurring syllables; this process is referred to as Statistical Learning (SL). For example, in the phrase “Pretty Baby” the syllables “pre” and “ty” co-occur more regularly than “ty” and “Ba”. This tells children “pre-ty” is a word, and “ty-ba” is not a word. Examples of tasks used to measure SL include visual non-verbal measures (e.g. shape patterns) and acoustic verbal measures (e.g. audible syllable patterns). Usually researchers make up their own language to test this type of learning

Previous studies have focused on offline testing when comparing SL between children with and without DLD. This study tests SL ability of children with DLD and without DLD, using both online and offline testing. In doing so, it aims to understand whether children with DLD show impairments in SL and compare the sensitivity of different tests (i.e. online vs. offline) in identifying these differences.

Online tasks were part of a listening task during the training phase of the study. Participants needed to press a different button when they heard a specific "word" in the training language. How accurate and quick they were to this particular word were the online measures. The offline tasks required more reflection of the training language, for instance, deciding between two items which was most similar to the training language. Also part of the offline task was completing a partially presented "word" from the training language.

What was found:

  • Online testing found children with DLD were as efficient at SL as typically developing children.

  • However, with the offline tests it was clear that children with DLD struggled with SL in both visual and acoustic measures.

  • In these offline tests, children with DLD showed no SL for the acoustic verbal tasks (whilst typically developing children showed significant learning).

What does this mean?

These findings suggest that offline testing alone is does not give an accurate understanding of a child’s SL potential or ability. Instead, a combination of testing both online and offline should be used to give a more accurate understanding of SL in DLD and non-DLD populations. Online measures provide evidence that children with DLD are as efficient at extracting patterns for learning as typically developing children. It is possible that there are other skills required for the offline tasks that interfere with the child’s ability to learn in this way (e.g., requiring short-term memory).

Where can I read this paper?

This paper is open access, which means everyone can read it. Please click here to find the full paper.