ABSTRACT
Skilled, typically developing readers and children with dyslexia read correct sentences and sentences that contained verb errors that were pseudo-homophones, morphological over-regularisations or syntactic errors. All errors increased looking time but the nature of the error and participant group influenced the time course of the effects. The pseudo-homophone effect was significant in all eye-movement measures for adults (N = 26), intermediate (N = 37) and novice typically developing readers (N = 38). This effect was larger for intermediate readers than other groups in total duration. In contrast, morphological over-regularisations increased gaze and total duration (but not first fixation) for intermediate and novice readers, and only total duration for adult readers. Syntactic errors only increased total duration. Children with dyslexia (N = 19) demonstrated smaller effects of pseudo-homophones and over-regularisations than controls, but their processing of syntactic errors was similar. We conclude that dyslexic children's difficulties with reading are linked to overreliance on phonological decoding and underspecified morphological processing, which impacts on word level reading. We highlight that the findings fit well within the grain-size model of word reading [Grainger, J., & Ziegler, J. C. (2011). A dual-route approach to orthographic processing. Frontiers in Psychology, 2, 54. doi:10.3389/fpsyg.2011.00054].
Acknowledgements
We would like to thank all of the children, parents, teachers and research assistants involved in the research.
Disclosure statement
No potential conflict of interest was reported by the authors.
ORCID
Helen L. Breadmore http://orcid.org/0000-0003-3050-8908
Julia M. Carroll http://orcid.org/0000-0002-3614-6883
Notes
* Both authors are now at Centre for Advances in Behavioural Science, Coventry University, Coventry, UK.
1 We investigated whether adults’ gaze durations in response to over-regularisations changed strategically over the course of the study by including the additional fixed factor order (first half of experiment, second half of experiment). Neither the main effect (p = .1599) nor the interaction (p = .323) were significant.
2 Although our criteria of a standard score below 90 may seem lenient in the general population, it is important to recognise that these participants’ classmates were performing better than the standardisation sample – the standardisation sample have a mean of 100 and an SD of 15, compared to 104 and 7 for typically developing peers sampled across the RA and CA groups, and 85 and 4 for children with dyslexia
3 Main effect of participant group (dyslexia, RA) was significant only for syntactic errors in log fixation duration; pseudohomophones χ2(1) = 0.31, p = .6; over-regularisations χ2(1) = 1.60, p = .21; syntactic errors χ2(1) = 3.90, p = .0482. The difference was significant in all conditions for log gaze duration; pseudohomophones χ2(1) = 3.96, p = .0466; over-regularisations χ2(1) = 7.58, p = .0059; syntactic errors χ2(1) = 5.52, p = .0188. The effect of participant group was not significant in log total duration; pseudohomophones χ2(1) = 0.71, p = .40; over-regularisations χ2(1) = 2.64, p = .1039; syntactic error χ2(1) = 1.14, p = .3.