Dyslexics produce inversion errors. Really?

Dyslexics produce inversion errors. Really?

First Author: Daniel Daigle -- Universite de Montreal
Additional authors/chairs: 
Anne Plisson; Rachel Berthiaume
Keywords: Spelling, Dyslexia, Primary education, French, Error Analysis
Abstract / Summary: 

Learning to read and write constitutes a great challenge for most children with dyslexia (Snowling, 2006). A common belief is that dyslexics tend to make a large number of inversion errors. In the context of spelling acquisition, several studies have shown that dyslexics have significant difficulties (Bourassa & Treiman, 2003; Kemp et al., 2009). The most accepted explanation for their difficulties relates to a phonological deficit (Ramus et al., 2003). For some, however, their difficulties are the result of a visual processing impairment (Ans et al., 1998; Krifi, 2004; Stein & Walsh, 1997). The purpose of this study is twofold: to test whether dyslexic students produce inversion errors and, if so, to attempt to explain these errors in terms of phonological or visual processing.

The data presented come from a secondary analysis of the results presented in Daigle et al. (2016). 32 French-speaking dyslexic children aged 11.5 were compared to 24 reading-age matched controls (RA) and to 25 chronological-age matched controls (CA). All subjects produced a written text that was analyzed at the graphemic level. In total, more than 55 000 graphemes were analyzed. Inversion and substitution errors were analyzed more extensively (total number of errors: 7 965).

Results indicate that dyslexic children’s spelling abilities are weaker than those of both the CA, and the RA children. In all groups, almost all errors are substitution ones. These errors are generally attributable to poor phonological representations, especially vowels, and rarely to visual confusions. The difference between these two explanations is statistically highly significant (p <.001).

In the context of writing, dyslexic participants very rarely produce inversion errors. Their errors are generally explainable by weak phonological representations. These data seem to argue in favour of the phonological deficit hypothesis.