Wednesday, October 24, 2012

Grammatical Difficulties in Children with SLI: Is Learning Deficient?

Grammatical Difficulties in Children with SLI: Is Learning Deficient? Hsu, H. J., & Bishop, D. (2011). Hum Dev. 53(5): 264-277.

One view of language learning suggests that children (and humans generally) are sensitive to the patterns of sounds that they hear in language. This statistical learning is based on the frequencies or probabilities of sounds in the language. First, young children store heard sentences in an exemplar-by-exemplar fashion (Ellis, 2002), memorizing individual items. In time, children develop more  abstract (syntactic) patterns of different types, and these frames allow them to produce correct grammatical sentences (Ellis, 2002).

Hsu and Bishop discuss the viewpoint that the grammatical deficits in SLI are due to an impairment in the ability to extract these statistical patterns from the input. As a result, children with SLI may require more exposure to learn abstract syntactic patterns. Indeed, Bishop, Adams, and Rosen (2006) found that even after daily training with a particular construction type, children with SLI did not achieve fluent automatic comprehension. It may be that the poor statistical learning in SLI results from other deficits, however. Limitations in short-term memory (Coady & Evans, 2008) or speech perception (Joanisse et al., 2000) may negatively impact statistical learning abilities in SLI.

Children with SLI may have difficulty identifying patterns in the language they hear. Isolating particular patterns to increase the frequency of exposure and highlight the form are approaches that would be consistent with this viewpoint.

Blogger: Areej Balilah. 

1 comment:

  1. With respect to statistical learning in SLI, a recent paper by Evans is also highly relevant:

    Evans, Saffran & Robe-Torres (2009). Statistical Learning in Children With Specific Language Impairment. J. Speech Lang. Hear. Res.

    That study found that kids with SLI show much slower implicit learning of transitional probabilities in a novel language. It strikes me that typically developing infants are able to track probabilities in a similar task, which supports the view that statistical learning of this type plays an important role in learning grammar.