Acknowledgements
We would like to acknowledge the financial support received first from UCLouvain’s Special Research Fund (FSR) and then from the Belgian National Fund for Scientific Research (FNRS; grant reference number: 40018347). Special thanks are also extended to the editors and anonymous reviewers for their valuable comments and suggestions on this article.
1In the university environment, writing, especially in a second/foreign language (L2), holds paramount significance. With the prevalence of English-led programmes in many universities, non‑native English speakers must often engage in English-language academic writing, making L2 writing skills essential. Dyslexic students face particular challenges with such tasks, necessitating an in-depth examination of their L2 writing processes, especially regarding pausing and revision behaviours (Helland & Kaasa, 2005).
2Contrasting with the traditional product-based approach with a focus on final written products, a process-based approach—examining the act of writing in progress—is advocated for its revealing insights into the complexities of composition (Galbraith et al., 2012; Wengelin, 2007). While previous research has probed into dyslexia’s impact on first language (L1) writing processes, dyslexic individuals’ L2 writing experiences have been less explored. Keystroke logging serves as a valuable method in this exploration, offering insights—among others—into the fluency and flow of writing through the analysis of pauses and revisions (Van Waes et al., 2016).
3Addressing the gap around university dyslexics’ L2 writing, our study uses authentic corpus data to examine dyslexic university students’ L2 writing processes, contrasting these with those of non-dyslexic peers. The research focuses on pausing and revision behaviours, seeking to unveil the unique challenges that dyslexic learners encounter in L2 English writing via keystroke logging.
4This article first discusses dyslexia’s impact on L2 acquisition and its significance for university students. It proceeds with some insights into writing processes, examining pausing and revision behaviours through the lens of keystroke logging. This foundation guides our research questions and hypotheses, presented in Section 2. The methodology—detailing the corpus, participants, tasks and data analysis—is laid out in Section 3. Section 4 describes our results, looking at pausing (Section 4.1) and revision behaviours (Section 4.2). The paper concludes by discussing the findings, acknowledging limitations and suggesting directions for future research (Section 5).
5Dyslexia is a specific learning disorder that is highly heterogeneous (Kormos & Smith, 2012). Dyslexic subjects often face challenges with working memory (Jeffries & Everatt, 2004; Szmalec et al., 2011), i.e. the cognitive system responsible for temporarily holding and processing the information needed for performing tasks and making decisions. For dyslexic individuals, the demands on working memory during challenging tasks such as writing may result in cognitive overload (Connelly et al., 2006). This cognitive overload can arise, for example, from the management of spelling, which presents particular difficulties for dyslexic writers. This, in turn, leaves few cognitive resources available for higher‑order processes such as planning content, retrieving ideas and organising ideas (Kellogg, 1996).
6The impact of cognitive overload becomes even more pronounced when considering L2 writing (Galbraith & Vedder, 2019), due to the increased cognitive demands associated with language translation, sentence construction and adherence to grammar and spelling rules which may differ from those in the L1. Past research on L2 writing by dyslexic subjects has supported this view (e.g. Ganschow & Sparks, 2001; Siegel & Smythe, 2004).
7The term “high-functioning dyslexics” has been used to refer to adults who were diagnosed with dyslexia in childhood but who have compensated sufficiently to undertake higher education and whose scores are now within -1 standard deviation of the population average in reading and spelling (Gallagher et al., 1996). Nevertheless, some studies suggest that adult dyslexics rarely compensate completely for their difficulties (e.g. Callens et al., 2012; Tops et al., 2014) and that the degree of compensation varies greatly from one individual to another (Lefly & Pennington, 1991).
8Writing acts as a crucial tool for learning and communication. Yet, for dyslexic students, particularly high‑functioning ones, it poses additional challenges due to complex cognitive and linguistic demands (Connelly et al., 2006). Research on L2 writing has traditionally concentrated on the final product (e.g. Kormos, 2011; Sasaki, 2011), with a focus on the outcome of the writing. However, this product-oriented perspective often misses the intricate process of writing that unfolds in real time—a journey that is especially challenging for dyslexic individuals.
9More recent studies exploring L1 writing in dyslexia acknowledge the relevance of adopting a process-oriented perspective by examining the writing process itself (e.g. Galbraith et al., 2012; Wengelin, 2007), allowing us to observe the composition process as it unfolds in real time, in order to analyse writing from a different but complementary angle. However, most of the literature on high-functioning dyslexics’ writing processes has focused exclusively on L1 writing (see Connelly et al., 2006; Sumner & Connelly, 2020; Wengelin, 2007).
10Exploring the complexities of writing from a process-oriented perspective can be carried out by means of different methods, used alone or in combination, such as keystroke logging, screencasting, think-aloud protocols and eye‑tracking. For the purposes of this study, we will rely on keystroke logging methods. Keystroke logging involves recording every key struck by the writer during the writing process, including insertions, deletions, revisions and any other changes made by the writer. It provides detailed insight into the temporal dynamics of writing by revealing traces of underlying cognitive processes (Van Waes et al., 2016). This method is ideal for observing and measuring the two aspects of writing processes that will be studied here: pauses and revisions.
11Several keystroke logging studies of pausing and revision behaviours have addressed either the L2 writing processes of non‑dyslexic participants (e.g. Barkaoui, 2019; Spelman Miller, 2000) or the L1 writing processes of high-functioning dyslexics (e.g. Sumner & Connelly, 2020; Wengelin, 2007). Early work by Spelman Miller (2000) analyses pausing patterns in L2 (non‑dyslexic) writing, revealing that duration of pausing increases with increasing levels of textual units, with the longest pauses occurring between sentences. Wengelin (2007) compares the L1 writing of Swedish dyslexic adults to non‑dyslexics, focusing on pauses and revisions at the word level, and discovers that dyslexic subjects display twice as many intra‑word pauses as the control subjects, while no difference was found for the frequency of revisions. Sumner and Connelly (2020) investigate the revision strategies of L1 English dyslexic writers, noting that they revise spellings more than their non‑dyslexic counterparts but still leave numerous errors. The findings of these studies highlight the relevance of using keystroke logging for analysing writing processes and more specifically pausing and revision behaviours.
12As shown above, linguistically-oriented research on the writing of high-functioning dyslexics presents two characteristics: (i) it has paid less attention to the writing process than to the written product and (ii) when tackling the writing process, it has exclusively focused on L1 writing. To our knowledge, no study to date has examined the L2 writing processes of high-functioning dyslexics (but see Álvarez-Cañizo et al., 2023 on L2 dyslexic children). Building on this gap, our research question investigates how the L2 writing processes of high-functioning dyslexic university students compare to those of non‑dyslexic university students, specifically in terms of pauses (pause frequency, pausing time, mean pause duration, pause location) and revisions (revision frequency, revision time, mean revision duration, linearity of revision behaviour). We tackle this comparison by using a learner corpus that, unlike most other learner corpora, comprises not only final texts but also data reflecting the writing process, including keylogging data. Thanks to the analysis of the corpus data, this study offers a valuable contribution to the broader field of writing-process research by improving our understanding of the impact of dyslexia on L2 writing in higher education.
13With respect to pauses, we expect more frequent pauses among dyslexic learners than among non‑dyslexic learners, as observed in previous works on L1 writing (Sumner & Connelly, 2020; Wengelin, 2007). Regarding total pausing time and mean pause duration, we do not expect any significant difference between the groups based on previous literature on L1 (Sumner & Connelly, 2020). Concerning the location of pauses, we expect dyslexic learners to interrupt their writing flow more frequently than the control group particularly at lower textual levels such as within and between words. This hypothesis is based on dyslexic writers’ tendency to experience difficulties when spelling unfamiliar words, leading to a high proportion of pauses within words, as shown by Wengelin (2007) for L1 dyslexic writers.
14Regarding revisions, we expect dyslexic learners to revise their texts as much as their non‑dyslexic peers, as found in some previous works for high-functioning dyslexics writing in L1 (Mazur & Chenu, 2023; Sumner & Connelly, 2020). As for overall revision time and mean revision time, we do not put forward any specific hypotheses due to the lack of previous research addressing these aspects in the context of dyslexic populations. Finally, due to dyslexics’ inability to construct a stable scheme of ideas before writing (Galbraith et al., 2012), we expect dyslexic learners to display an unstructured text composition process, resulting in a less linear revision behaviour than non‑dyslexic learners. We expect dyslexic learners to be constantly going back and forth through the text when revising.
15The study is based on the Process Corpus of English in Education (PROCEED; Gilquin, 2022), a new type of learner corpus that comprises three main data types: written texts, keystroke log files and screencast videos. The keystroke log files and screencast videos, by capturing the keyboard and screen activity, help to keep a record of all the steps involved in the writing process. The data are accompanied by rich metadata including, for each learner, detailed sociolinguistic information as well as some linguistic and psychometric measures (such as language proficiency and working memory).
16This study relies on keystroke log files from PROCEED, which were collected by means of the Inputlog software (Leijten & Van Waes, 2013). Keystroke logging makes it possible to carry out more advanced analyses of fluency (number, location and duration of pauses, linearity, number of revisions, etc.) than in conventional approaches in which fluency is limited to measuring the total number of words produced and the average number of words per minute.
- 1 Ethical approval for data collection was granted by the Hospital faculty Saint‑Luc – UCLouvain ethi (...)
17The corpus sample used here comprises data produced by four dyslexic learners and four non-dyslexic learners from a Belgian university, who were matched for L2 proficiency level, with an even distribution between genders.1 Dyslexic participants were diagnosed with dyslexia by a speech therapist in primary school. Participants’ English proficiency levels, measured by means of the LexTALE test (Lemhöfer & Broersma, 2012), ranged from 63.75% to 77.5%, with a mean of 72.08% (SD = 5.10). All participants had French as their L1 and learned English as an L2. They had all taken at least one English course as part of their university curriculum and were enrolled in a Master’s programme in non-literary fields. The two groups exhibited comparable mean performances on the Corsi test, a standardised assessment of visuospatial working memory (Kessels et al., 2000), suggesting no difference in working memory capacity between them.
18All students performed two narrative tasks2 in English on a laptop, using an AZERTY keyboard. The first task required students to write about one of three familiar themes based on a text prompt (their dream job, their biggest achievement or the best day of their life). The second task involved a picture prompt, where students chose among four images (holiday styles) the one that appealed to them the most. Given the small sample size, no distinction will be made between the two tasks in this study. Yet, it can be pointed out that the “dream job” theme was the most common choice for the text prompt, while there was no particular preference among the holiday styles in the picture prompt. For each task, students were asked to write a text of 150 to 350 words related to their personal experience(s), based on the prompt they chose. Participants were allowed to use external resources that facilitate writing such as dictionaries and grammar tools. They were given an indicative 15‑minute time limit per task, with their writing times ranging from 12 to 18 minutes for each text. Participants were tested individually.
19Based on the metadata, we first ran a typing skills analysis using Inputlog’s copy task.3 The results indicated that both groups exhibit comparable scores in terms of typing speed, correctness and overall accuracy. Only one dyslexic student notably demonstrated a higher frequency of typographical errors. We then performed a pause summary analysis for each participant using Inputlog to identify pauses in the keystroke logs. We adopted a pause threshold of two seconds, following a common convention in writing research (e.g. Wengelin, 2007). With the help of Inputlog, we calculated the frequency of pauses and their duration, analysed their location and compared them between the two groups of students. Through a revision analysis in Inputlog, we calculated the frequency of revisions and their duration. We also considered the linearity of the revision behaviour using S‑notation analysis in Inputlog. In terms of statistics, t‑tests were used to assess the significance of the differences between the two groups after verifying that the data were normally distributed and that the variances were equally balanced. These tests were conducted with the help of R, and the analyses also took into account individual variation through standard deviation (SD) measures.
20In what follows, we present the main results for pausing behaviour (Section 4.1) and revision behaviour (Section 4.2). For each feature analysed, we provide the means (M) and SDs to highlight the potential degree of variation among learners. The results for pausing/revision time and pause/revision frequency are reported per 100 words to accommodate variations in text length and p‑values are provided in the tables.
21Table 1 presents the results for three aspects related to pausing behaviour: pausing time, i.e. the cumulative duration of pauses, pause frequency, i.e. the number of pauses observed, and mean pause duration, i.e. the average duration of each pause. The results are based on a minimum pause threshold of two seconds. In this table and the next, the asterisk indicates values that are statistically significant.
Table 1. – Means and standard deviations (SD) for results obtained on the pausing behaviour of dyslexics (n = 4) and non‑dyslexics (n = 4).
|
Pausing behaviour
|
Dyslexics
Mean (SD)
|
Non‑dyslexics
Mean (SD)
|
p
|
|
Pausing time in sec.
(/100 words)
|
209.89 (68.49)
|
106.13 (39.44)
|
0.0033*
|
|
Pause frequency
(/100 words)
|
35.74 (11.32)
|
18.66 (4.98)
|
0.0031*
|
|
Mean pause duration in sec.
|
5.52 (1.71)
|
5.45 (1.49)
|
0.9379
|
22When considering the total pausing time per 100 words, it appears that dyslexic writers exhibit a significantly higher mean (M = 209.89, SD = 68.49) than their non‑dyslexic counterparts (M = 106.13, SD =39.44), with a statistically significant p‑value of 0.0033. This finding suggests that dyslexics tend to pause more in their writing, with an average of nearly four minutes per 100 words, to be compared with non‑dyslexic writers who average nearly two minutes per 100 words, for an average task duration of 15 minutes.
23Moving on to the frequency of pauses per 100 words, we see that the dyslexic group exhibits a higher mean (M = 35.74, SD = 11.32) compared to the non‑dyslexic group (M = 18.66, SD = 4.98). On average, dyslexic learners make nearly twice as many pauses per 100 words as non‑dyslexic learners. This difference is statistically significant (p = 0.0031) and indicates a trend where dyslexic writers tend to make more frequent pauses when writing.
24Concerning the mean pause duration, dyslexics (M = 5.52, SD = 1.71) show a slightly higher value than non‑dyslexics (M = 5.45, SD = 1.49). This difference does not reach statistical significance (p = 0.9379), however. This suggests that the average duration of individual pauses is not fundamentally different between the two groups.
25Alongside these pausing aspects, it is also interesting to consider pause location, since Wengelin’s (2007) findings for L1 indicate that dyslexic students produce a higher proportion of pauses within words and between words than non‑dyslexic students. Table 2 shows the frequency of pauses in different locations, along with p‑values, indicating the frequency of pauses (minimum 2 seconds) per 100 words.
Table 2. – Means and standard deviations (SD) for pause location frequency per 100 words in dyslexics (n = 4) and non‑dyslexics (n = 4).
|
Pause location
|
Dyslexics
Mean (SD)
|
Non‑dyslexics
Mean (SD)
|
p
|
|
Pauses within words
|
4.88 (2.70)
|
2.13 (1.55)
|
0.0291*
|
|
Pauses before words
|
17.38 (8.00)
|
6.88 (2.30)
|
0.0070*
|
|
Pauses before sentences
|
0.50 (0.76)
|
0.63 (0.52)
|
0.7061
|
|
Pauses before paragraphs
|
1 (1.31)
|
0.50 (0.53)
|
0.3427
|
|
Pauses after words
|
3 (2.00)
|
4.13 (5.54)
|
0.6025
|
|
Pauses after sentences
|
0.13 (0.35)
|
0.63 (1.41)
|
0.3589
|
|
Pauses after paragraphs
|
0
|
0.25 (0.71)
|
0.3506
|
|
Miscellaneous pauses
|
0
|
0
|
0
|
26Both dyslexic and non‑dyslexic learners show a tendency to mainly pause within words, before words, and after words. The mean pause frequency within words among dyslexic learners is 4.88 pauses per 100 words (SD = 2.70), compared to 2.13 pauses per 100 words (SD = 1.55) for non‑dyslexic learners. Although these numbers are relatively low, the difference is statistically significant with a p‑value of 0.0291. Similarly, dyslexic writers display a higher mean pause frequency before words, with an average of 17.38 pauses per 100 words (SD = 8.00), compared to non-dyslexic learners, who exhibit 6.88 pauses per 100 words (SD = 2.30). This difference is also statistically highly significant, with a p‑value of 0.0070. While dyslexic learners show a mean pause frequency after words (3 pauses per 100 words) that is lower than that of non‑dyslexics (4.13 pauses per 100 words), this disparity does not reach statistical significance (p = 0.6025). Differences observed between groups with respect to pauses before sentences, before paragraphs, after sentences, after paragraphs and miscellaneous pauses are not statistically significant either.
27Table 3 presents the results for three aspects related to revision behaviour: revision time, i.e. the cumulative duration of revisions, revision frequency, i.e. the number of revisions observed, and mean revision duration, i.e. the average duration of each revision.
Table 3. – Means and standard deviations (SD) for results obtained on the revision behaviour of dyslexics (n = 4) and non‑dyslexics (n = 4).
|
Revision behaviour
|
Dyslexics
Mean (SD)
|
Non‑dyslexics
Mean (SD)
|
p
|
|
Revision time in sec.
(/100 words)
|
40.31 (17.31)
|
25.16 (19.64)
|
0.1243
|
|
Revision frequency
(/100 words)
|
26.59 (6.63)
|
33.21 (12.68)
|
0.2183
|
|
Mean revision duration in sec.
|
1.51 (0.46)
|
0.69 (0.43)
|
0.0024*
|
28When considering the revision time per 100 words, it turns out that dyslexic writers exhibit a higher mean (M = 40.31, SD = 17.31) than non‑dyslexics (M = 25.16, SD = 19.64). For dyslexics, the average total revision time per 100 words is about 40 seconds and for non‑dyslexics, it is about 25 seconds. The observed difference is not statistically significant (p = 0.1243), indicating that there is no strong evidence that dyslexic writers spend more time revising than non-dyslexic writers.
29As regards the average frequency of revisions per 100 words, the results are the opposite of the other findings, being slightly lower among dyslexics (M = 26.59, SD = 6.63) than among non‑dyslexics (M = 33.21, SD = 12.68). However, this difference fails to reach statistical significance (p = 0.2183).
30The mean revision duration is twice as high for dyslexics (M = 1.51, SD = 0.46) as for non‑dyslexics (M = 0.69, SD = 0.43), with dyslexics spending around 1.5 seconds per revision and non‑dyslexics spending around two‑thirds of a second on average. The observed difference is statistically significant (p = 0.0024) and implies that dyslexics typically allocate more time per revision compared to their non‑dyslexic counterparts.
31In addition to these revision aspects, we examined the linearity of the revision process. We used the S‑notation function of Inputlog that generates a file representing the writing process. By numbering the revisions, S‑notation reveals the sequential order in which text editing actions occur during the writing process. We used this information to qualitatively compare the linearity of revisions in dyslexic and non‑dyslexic learners. We present only one example from each group here, but our qualitative analysis suggests that these examples are representative of the broader patterns observed within each group.
32S‑notation relies on various symbols to represent different editing actions. A vertical bar labelled with a sequential number (e.g. “|6”) signals a break in the text, a point in the writing process where the writer deviates from linear text production. Text written between curly brackets (e.g. “{by house}19”) represents a text insertion that occurs in the text produced so far, labelled with a sequential number. Text written between square brackets (e.g. “[and not]21”) corresponds to a deletion, labelled with a sequential number. Each sequential number indicates the order in which these activities—breaks, insertions or deletions—are executed during the writing process.
33Figure 1 illustrates the S‑notation analysis of a dyslexic student’s writing process for an entire writing task. It shows that the editing actions in the text do not display a strictly linear order, with insertions and deletions occurring all over the place throughout the text. The non‑linear order of the figures indicates that the dyslexic writer continually revises the text by going back and forth, suggesting a non‑linear writing and revision process.
Figure 1. – S‑notation analysis of a dyslexic student’s writing process.
34Figure 2 illustrates the S‑notation analysis of a non‑dyslexic student’s writing process for an entire writing task. In contrast to Figure 1, it depicts a markedly sequential series of editing actions, as appears from the succession of the sequential numbers. This suggests that the student is revising the text concurrently with the writing process, maintaining a consistent pace throughout. Unlike the non‑dyslexic writer’s pattern, revisions in this case are integrated seamlessly into the writing flow, reflecting a more synchronised approach to composition and editing.
Figure 2. – S‑notation analysis of a non‑dyslexic student’s writing process.
35It is important to note that the analysis of the copy task using Inputlog reveals that the dyslexic student exhibits a higher frequency of typos when typing compared to their non‑dyslexic counterpart. This increased occurrence of typos indicates lower typing accuracy and speed, which could impact the overall efficiency and fluidity of the writing process in the writing task represented in Figure 1.
36Our study aimed to investigate the differences in L2 writing processes between high-functioning dyslexic and non‑dyslexic university students, focusing on pausing and revision behaviours. The findings reveal several differences in the writing processes of these two groups and largely confirm our hypotheses.
37In our analysis of pausing behaviour, we hypothesised that dyslexic individuals would exhibit more frequent pauses compared to the control group, aligning with findings from previous research on L1 writing (Sumner & Connelly, 2020; Wengelin, 2007). Confirming this expectation, our results reveal that dyslexic learners produce on average nearly twice as many pauses per 100 words as non‑dyslexic learners. Regarding total pausing time, we did not expect any significant difference between the groups, consistent with previous research in L1 (Sumner & Connelly, 2020). However, our findings indicate otherwise, showing a statistically significant difference in overall pausing time, with dyslexic learners pausing longer than the control group. Concerning mean pause duration, we did not expect any significant difference between the groups based on previous literature on L1 and we did not find any in the study, although dyslexic learners show a slightly higher mean than the control group.
38In terms of pause location, we expected dyslexic writers to produce more pauses within and between words, based on L1 dyslexic writers’ tendency to experience difficulties when spelling unfamiliar words (Wengelin, 2007). Confirming this hypothesis, our study reveals that dyslexic learners pause on average twice as frequently within words as non-dyslexic learners and also pause more often before words than non‑dyslexic students. Our findings suggest that the patterns observed by Wengelin (2007) in L1 writing by high-functioning dyslexics can also, to some extent, be reflected in L2 writing and that dyslexic learners experience potential challenges related to word‑level processing and spelling in L2. However, it is important to nuance these findings by considering that increased pause frequency among dyslexic students may not solely be attributed to spelling difficulties. These pauses could also result from the need to locate the correct key, reflecting a broader challenge with the mechanical aspects of typing. This is further supported by the typing skills analysis, which demonstrated that although both groups exhibit comparable scores in typing speed, correctness and overall accuracy, one dyslexic student had a higher frequency of typographical errors.
39For the analysis of revision behaviour, we hypothesised that the dyslexic and non‑dyslexic groups would exhibit similar patterns of revision frequency, as observed in previous studies on L1 writing (Mazur & Chenu, 2023; Sumner & Connelly, 2020). Despite a slight tendency for non‑dyslexic learners to produce a higher proportion of revisions per 100 words, our findings indicate no significant difference between the groups. As for overall revision time and mean revision time, we did not formulate any specific hypotheses due to the lack of previous research addressing these aspects in the context of dyslexic populations. In our study, we did not find any statistical difference in terms of overall revision time, but our results reveal a statistically significant difference between the groups concerning the mean revision duration, which is twice as long in dyslexics as in non‑dyslexics. This discrepancy may also be influenced by typing skills, as suggested by the typing skills analysis.
40Furthermore, we hypothesised that dyslexic learners would display an unstructured text composition process, reflecting their challenge in constructing a stable scheme of ideas before writing (Galbraith et al., 2012). Qualitative analysis based on S‑notation confirms this hypothesis, revealing that dyslexic learners generally adopt a non‑linear approach to composition and editing, characterised by frequent revisiting and editing of earlier text throughout the writing process. In the case of the dyslexic student whose writing process is represented in Figure 1, the revisions primarily involved correcting typos. By contrast, non‑dyslexic learners adopt a more sequential and integrated approach to editing.
41Overall, the standard deviation values reveal a greater dispersion and consequently a more pronounced heterogeneity among dyslexic individuals compared to their non-dyslexic counterparts. This is particularly noticeable in the measures related to pausing. Conversely, for revision measures, the pattern tends to be reversed, with non‑dyslexic individuals mostly demonstrating higher variability. These results underscore the substantial heterogeneity within the dyslexic group regarding pausing behaviours, while also acknowledging the presence of variability in the revision behaviour of the non‑dyslexic population. They also point to the possible ambivalence of revisions, which may be a sign of disfluency but also a characteristic of good writers seeking to fine‑tune their texts.
42Despite these promising results, our study faces certain limitations. Firstly, the small sample size limits the extent to which our findings can be generalised to broader populations. Secondly, the small sample and the high individual variation in the dyslexic group may not fully capture the diversity and heterogeneity that is typical of dyslexic students (see Kormos & Smith, 2012). This aspect might overlook variations that exist within the group, with another sample potentially revealing different trends.
43Future research endeavours should seek to address these constraints by incorporating larger and more diverse samples of dyslexic learners. It would also be interesting to consider different pause thresholds, and in particular shorter thresholds than the one used in this study (two seconds), as shorter pauses have been claimed to correspond to lower‑order processes (Barkaoui, 2019). Further insights into the writing processes of dyslexic and non‑dyslexic students could additionally be gained by conducting a detailed analysis of the S‑notations for types of revision and by considering comparative output assessment between the two groups in terms of language complexity and meaningful content. Thus, while the findings of this research on L2 writing by high-functioning dyslexic learners of English represent a step forward in our understanding of writing processes and their interaction with dyslexia, there is still room for further research and discoveries in this area.
Flux RSS
