Predicting future poor readers from pre-reading visual skills: A longitudinal study

Abstract

Reading is essential for learning, from literature to physics, from paper to screens on e-readers and smart phones. Even if it is well known that learning to read implies good language skills, children also need to develop good oculomotor and visual-perception skills. Thereby, any deficits in visual processing may affect learning. The possible impact of visual deficits is rarely considered, especially with regard to eye movements and visual perception. Hence, these deficits are usually discovered much later or remain undiagnosed. The present study aimed at assessing the usefulness of visual processing related measures in the early detection of reading difficulties. Visual skill differences that are apparent early in kindergarten might provide predictive insights into risk for learning difficulties at school entry. We used a prospective, longitudinal approach where visual processes (assessed with the Developmental Eye Movement (DEM) test) were measured in 51 preschoolers, and the impact of these processes on future reading development was explored one year later, in Grade 1. Results showed that (1) 31% of our sample of preschoolers showed visual processing impairments (without any clinical complaints) and (2) reading accuracy and speed in first graders were significantly correlated with visual skills assessed in kindergarten, thus confirming the significant role of oculomotor and visual-perception processes in the acquisition of reading skills. These suggests the potential for these measures to be used clinically for identifying children at risk for low academic achievement, enabling appropriate targeting of early interventions.

Keywords:

• DEM-test
• eye movements
• Kindergarten
• longitudinal study
• preliteracy skills
• reading acquisition
• visual information processing

Notes

1 By visual skills or visual processing, we do not mean visual acuity but rather a complex visual and/or visuo-attentional mechanism specific to the reading activity.

2 It should be noted that other linguistic skills play important roles in successful reading acquisition, as knowledge of alphabetic letters (Lonigan et al., 2000; Parrila et al., 2004; Storch & Whitehurst, 2002) and word forms (Goodman, 1986; Lonigan et al., 2000).

3 Note that only two issues are still being discussed regarding the DEM-test, i.e. (1) is it secure to use the ratio measure (Medland et al., 2010; Raghuram et al., 2018 vs. Ayton et al., 2009; Portnoy & Gilaie-Dotan, 2020; Webber et al., 2011; Wood et al., 2018) and (2) is it possible to use the DEM test as a specific measure of saccadic control (Ayton et al., 2009; Webber et al., 2011 vs. Moiroud et al., 2018; Raghuram et al., 2018). We do not feel concerned here, since investigating eye movement control in children was clearly not the focus of interest of this paper.

4 Adjusted Horizontal Time = Test C time × [80/(80 - number of omission + number of addition)]

5 In reading or neuropsychological research, it is more often 1.5 or 1.65 (corresponds to 5%), to be able to identify disorders. But when the question of interest is to track difficulties or delay in the developmental trajectory, the threshold used traditionally is 1 SD below the average (16%), since this is the threshold beyond which the assessed skill can be considered as poor. This cut-off allowed us to identify all poor readers (reading test) and all children with visuo-perceptual difficulties (DEM test) (Lefavrais, 2005; Muneaux, 2018).

6 Only 30 of the 51 children evaluated in kindergarten could be re-assessed and followed longitudinally to the 1st grade.

7 The possibility that this effect was linked to the small sample of children was runed out since we obtained similar results, i.e., a superiority of the VT index, in another longitudinal study following 137 preschool children (Ducrot & Vernet, 2021, see also Footnote 8). We assume that this effect is linked to the young age of our participants; as children move through school, the role of language and other higher level cognitive skills become more significant in a child’s reading performance (Flax, 2006; Goldstand et al., 2005).

8 Note that the data we collected in the interim with a computerized version of the DEM-test made us confident that the effects obtained are replicable and generalizable. This experiment was conducted in a larger sample composed by 137 preschoolers, using the same procedure and longitudinal design, with a group of 137 preschool children. The data we obtained support those first obtained in this study, with the paper-based version of the DEM test, that is., the VT and HTaj measures (1) are significantly correlated with reading outcomes one year later and (2) significantly predict children's reading achievement in Grade 1, i.e., 12 months before reading training (Ducrot & Vernet, 2021).

9 The creation of an easy-to-use software for clinical practice is in progress in our laboratory, with the aim to assess visuo-perceptual and attentional abilities, taking the children school level into account. This software, called Diaglect, would allow us to obtain an automatic diagnosis and several remedial modules will be proposed according to the child's visual profile.

Additional information

Funding

Support for this research was partly provided by an APEX Research Fellowship from the “Conseil Régional PACA” to the last author [APEX VISALECT, DEB-17-1055] and by a PhD Fellowship from the “Fondation de France” to the first author [VISALECT_NF1_00099576].