The main goal of this paper is to analyze written texts produced by monolingual French university students, with and without dyslexia. More specifically, we were interested in the linguistic characteristics of the words used during a written production and of the type of word errors. Previous studies showed that students with dyslexia have difficulties in written production, whether in terms of the number of spelling errors, some syntactic aspects, identification of errors, confusion of monosyllabic words, omissions of words in sentences, or utilization of unexpected or inappropriate vocabulary. For this present study, students with dyslexia and control students were asked to produce written and spoken narrative and expository texts. The written texts (N = 86) were collected using Eye and Pen© software with digitizing tablets. Results reveal that students with dyslexia do not censor themselves as regards the choice of words in their written productions. They use the same types of words as the control students. Nevertheless, they make many more errors than the control students on all types of words, regardless of their linguistic characteristics (length, frequency, grammatical classes, etc.). Finally, these quantitative analyses help to target a rather unexpected subset of errors: short words, and in particular determiners and prepositions.

Une des difficultés mentionnées par les étudiants dyslexiques est la production de textes (Mazur-Palandre, Abadie et Bedoin, 2016). Chez les sujets neurotypiques, les processus orthographiques s’automatisent progressivement avec l’expérience (Berninger et Swanson, 1994), ce qui permettrait – si on se place dans le cadre de la théorie de la capacité pour la production écrite (McCutchen, 1996) - d'affecter davantage de ressources à des processus de niveau supérieur, tels que la révision. Chez les individus dyslexiques, les processus orthographiques sont peu automatisés, et ce même à l’âge adulte ce qui peut donc impacter la production textuelle (Fayol et Miret, 2005). Cette étude vise à analyser les productions écrites d’étudiants dyslexiques et contrôles afin de voir de quelle manière la révision diffère entre les deux groupes d’étudiants. Les étudiants ont été invités à produire des textes écrits, collectés par le biais d’une tablette graphique et du logiciel Eye and Pen©, permettant d'observer la temporalité de l’activité d’écriture, dont les révisions. Les résultats indiquent que les étudiants dyslexiques et contrôles ne diffèrent guère quant à la quantité et les types de révision, ce qui suggère que les étudiants dyslexiques ont mis en place des stratégies. Néanmoins, des analyses complémentaires nuancent les résultats en montrant des différences entre les deux groupes. The revision process during written text production in French dyslexic students? One of the difficulties mentioned by dyslexic students is text production (Mazur-Palandre, Abadie et Bedoin, 2016). In subjects with normal written language development low level processes, such as spelling, become growingly automatized (Berninger & Swanson, 1994) allowing for allocation of more cognitive resources to higher level processes such as those required for revision. In dyslexic individuals spelling remains poorly automatized even in adults and this may impact compositional performance (Fayol & Miret, 2005). Thus, this study aims at analyzing the written productions of dyslexic and controls students in order to observe how revision differs between to the two groups. The students were asked to produce written texts collected using a digitizing tablet and the Eye and Pen© software which allows for the observation of the temporality of the writing activity, including revision. The results indicate that dyslexic and control students do not differ much in the amount and types of revisions, suggesting that dyslexic students have developed strategies. Nevertheless, the additional analyzes nuance the quantitative results by showing differences between the two groups.