Writing and Revision Strategies of Students With and Without Dyslexia

Abstract

Previous work suggests that written text produced by university students with dyslexia is scored lower than that produced by their peers. The present study used a digital writing tablet to examine the writing process and the quality of text written by university students with dyslexia. Revision behavior during and after writing was also investigated. Thirty-two university students with dyslexia (mean age, 20 years), were compared with 32 typically developing (TD) students matched by age. Students composed a written text in response to an expository essay prompt. In line with previous research, students with dyslexia made a higher number of spelling errors and their essays were rated as poorer than TD students. However, students with dyslexia were comparable to their peers on measures of time spent writing, amount of text produced, and the temporal analyses (handwriting execution, pause times). Students with dyslexia made significantly more revisions to spelling during and after transcription than their peers, although other revision behavior was similar across groups. Explanations for the finding of poor writing quality are explored. Importantly, the findings suggest that continued support with spelling and writing is needed for university students with dyslexia. Instruction directed toward effective revision strategies may also prove useful. Limitations and directions for future research are discussed.

Keywords

dyslexia handwriting revision behavior spelling writing

Writing remains the main method of assessment throughout education. In particular, at university level, students are expected to demonstrate their knowledge of a topic through independent writing. The majority of disciplines require written coursework and, in the United Kingdom, most university students are required to produce handwritten essays in exams at a number of points in their education. While the proportion of United States and other international universities that require handwritten exams is dropping in favor of keyboarded assessments, there are still many university students who choose, or are required, to handwrite under time pressure.

One group of students who self-report long-standing problems with spelling and, specifically, written expression are students with dyslexia (Mortimore & Crozier, 2006). Dyslexia is a specific learning disorder, defined by poor reading (American Psychiatric Association, 2013) and therefore, not surprisingly, persistent difficulties with spelling are frequently noted in this population (Coleman et al., 2009; Peterson & Pennington, 2012). Reading informs writing in terms of providing opportunity to develop procedural and linguistic knowledge (Shanahan, 2016) and models of writing depict both reading and spelling as important foundational components when writing (Hayes, 2012; Shanahan, 2016). Limited exposure to written text coupled with problems with phonology and orthography may therefore hinder the development of written skills.

In 2015/2016, students with specific learning difficulties, which largely includes dyslexia, accounted for 6.2% of the student Higher Education (HE) population (Higher Education Statistics Authority, 2016) in the United Kingdom. Research has shown that students with dyslexia generally leave university with lower grades than their peers without dyslexia (Richardson & Wydell, 2003). One possible explanation for this may be due to their difficulties with meeting the written requirements. Several studies have demonstrated that, in comparison to their same-age peers, written compositions produced by university students with dyslexia are graded lower in quality, while being characterized by a larger proportion of spelling errors (Bogdanowicz et al., 2014; Connelly et al., 2006; Sterling et al., 1998) and that these differences in quality persists even after scripts have been corrected for spelling and handwriting (Galbraith et al., 2012; Gregg et al., 2007; Tops et al., 2013). Another reported finding is that students with dyslexia present with a slower handwriting speed (letters per minute) than their peers (Connelly et al., 2006; Hatcher et al., 2002).

Models of the writing process (Berninger & Swanson, 1994; Hayes, 2012) highlight how transcription skills (handwriting and spelling) need to be automatic to allow working memory resources to be devoted to generating text and higher level planning of the text (including reviewing what has been written). In adulthood, the transcription processes are expected to be proficient (Olive, 2014) and less of a constraint on writing. However, little is known about how poor spelling in adulthood influences the real-time production of text and the possible repercussions on text quality. Research using digital writing tablets has highlighted how spelling interferes with the writing process for children with dyslexia and, in turn, results in less text being composed (Sumner et al., 2013, 2014). Children with dyslexia were found to move the pen across the page at the same speed as their peers, but they paused for longer while writing (i.e., demonstrated a period of inactivity), which was attributed to their difficulty with spelling. It is conceivable that the spelling difficulties experienced by adults with dyslexia could constrain written text production in the same way.

When considering possible explanations for writing problems in dyslexia, spelling and handwriting alone cannot fully explain their textual shortcomings as poorer quality texts remain even after correcting for spelling and handwriting (Galbraith et al., 2012). There may be a cumulative effect of poor spelling and more pausing on the other writing processes that should happen in parallel with transcription. Skilled writers often carry out an evaluation (reviewing/revising) process on the contents of the translation process (converting pre-verbal ideas into surface language structure) as they transcribe their texts (Hayes, 2012). However, given the potentially more limited cognitive resources of students with dyslexia while writing, due to the heavy load of spelling, this may be less likely. Wengelin (2007) reported that, in comparison to a control group, Swedish-speaking adults with dyslexia made more spelling-related revisions when keyboarding. In contrast, a study on Norwegian-speaking adolescents found that within-word revision behavior when keyboarding was similar across weak decoders and those without decoding difficulties (Torrance et al., 2016), suggesting little impact of reading difficulties on the frequency of within-word revisions. Of note, Swedish and Norwegian have a shallower orthography than English which has a deep orthography (Schmalz et al., 2015) and, therefore, we may see a different profile for English-speaking students with dyslexia. Furthermore, revisions while handwriting may differ to keyboarding revisions. It may be that students with dyslexia produce text that has not been evaluated (revised) to the same extent as their peers and this may contribute to the lower quality compositions.

The aim of the present study was to examine the writing product and process of English-speaking university students with dyslexia and to investigate how students approach revising their written compositions. Performance was compared with a typically developing (TD) group of students matched for age. The writing process was assessed using a digital writing tablet, which enables the observation of revisions made during transcription and has the added benefit of not interrupting the composing process, as in think-aloud protocols. Posttranscript revisions were also analyzed to determine whether students pick up on changes to be made (or mistakes) when they proof-read their text after composition.

Based on the existing literature, it was predicted that texts written by students with dyslexia would contain a higher number of spelling errors and be rated as poorer in quality than their peers (Connelly et al., 2006; Galbraith et al., 2012). Based on findings from younger populations with dyslexia (Sumner et al., 2014), it was expected that students with dyslexia would spend more time pausing than their peers, reflecting a breakdown in text-making. Thus, the amount of time pausing and the number of pauses were measured. In addition, considering revision behavior, Chenoweth and Hayes (2001) reported that poorer adult writers revise frequently, but Torrance et al. (2016) found no evidence that weak decoders were more likely to make changes to their text. As students with dyslexia have both reading and spelling difficulties, we predicted that they would make spelling mistakes while writing and, as a result, we expected their revision behavior to reflect this difficulty by observing more frequent crossings out or returning to spellings and consequently less changing of the meaning of their texts compared with their peers. Thus, given the significant spelling challenges for students with dyslexia, we would predict that any revisions made would focus largely on spelling. However, we also expected that students with dyslexia would revise their spellings less effectively (during and posttranscription): successfully correcting proportionally fewer spelling errors than their peers. In contrast, we would expect the comparison group that do not present with spelling difficulties to show a range of revision behaviors (e.g., editing spelling, adding in or removing text or punctuation).

Method

Participants

Participants had English as their first language. Inclusion criteria for both groups were performance within the average range (±1.5 SDs) on measures of verbal (similarities task) and nonverbal (matrices task) cognitive ability (Wechsler Adult Intelligence Scale-3 [WAIS-III], Wechsler, 1997). Table 1 reports group characteristics. No significant group differences in age (p = .78) or cognitive ability (p > .18) were found, and both groups scored close to the population mean (10) on the cognitive tasks. All but three students with dyslexia were students studying psychology as their major (the other subjects consisted of early childhood studies, law, and occupational therapy), and all but four of the TD students were also students studying psychology (the remaining were business studies, architecture, sport and exercise, and geography students). Therefore, both samples of participants would likely be equally well versed in writing essays to social science essay topics. The majority of the sample were first year undergraduate students, apart from seven second-year students (four with dyslexia; three TD students). Year of study did not correlate with any of the writing quality measures and, therefore, was not included as a covariate in subsequent analyses.

Table 1.

Means and Standard Deviations for Scores Obtained on the Background Measures by Students With and Without Dyslexia.

Background measures	Dyslexia (n = 32)M (SD)	TD (n = 32)M (SD)	F	df	p
Age
Years; months	20;0 (1.82)	19;8 (1.68)	0.36	1.62	.55
Nonverbal^a
Scaled	10.44 (2.04)	11.23 (2.20)	2.52	1.62	.12
Verbal ability^a
Scaled	10.28 (1.92)	10.19 (1.71)	0.61	1.62	.44
Reading^b
Standard	99.77 (11.72)	113.46 (4.97)	36.86	1.62	<.001
Spelling^b
Standard**	92.94 (9.87)	109.34 (5.52)	67.41	1.62	<.001
Reading fluency^c
Standard**	82.78 (10.87)	98.34 (11.02)	32.35	1.62	<.001
Nonword fluency^c
Standard**	82.25 (12.72)	105.41 (12.44)	54.21	1.62	<.001

Note. WAIS-III = Wechsler Adult Intelligence Scale-3; WRAT-3 = Wide Range Achievement Test-3; TOWRE = Test of Word Reading.

Nonverbal (Matrices), verbal (Similarities) scaled scores M = 10, SD = 3 (WAIS-III; Wechsler, 1997). ^bReading, spelling (both WRAT-3; Wilkinson, 1993). ^cReading fluency measures (TOWRE; Torgesen et al. 1999) standard scores M = 100, SD = 15.

p < .05. **p < .001.

Students with dyslexia

Thirty-two students with dyslexia (23 female; 18–25 years) were recruited from Oxford Brookes University in the United Kingdom, via poster advertisements and through the psychology department to gain course credits. About 88% of the sample had received a dyslexia diagnosis from an Educational Psychologist, while the remaining 12% reported previously being identified by teachers as having dyslexia. No students reported any additional diagnoses (such as attention-deficit/hyperactivity disorder [ADHD] or language impairment). Single-word reading and spelling ability (both from the Wide Range Achievement Test-3 [WRAT3], Wilkinson, 1993) as well as word and nonword reading fluency (Test of Word Reading [TOWRE]; Torgesen et al., 1999) were assessed. Analyses revealed that students with dyslexia scored significantly below the TD group on spelling and reading performance (ps < .001).

TD control group

Thirty-two TD students (24 female; 18–26 years) were recruited from Oxford Brookes University, using the same method and assessments as above. Students did not report any developmental diagnoses such as dyslexia, language impairments, ADHD, autism, nor medical conditions; and performed age-appropriately on the cognitive, reading and spelling assessments.

Measures

Written composition

Participants were given 20 minutes in which to write an essay. The following writing prompt was taken from the Graduate Record Examinations (GRE) Analytical Writing Measure (Educational Testing Service, 2012), and was appropriate for social science–based university students (Connelly et al., 2006): “Present your perspective on the issue below, using relevant reasons and/or examples to support your views: “The media (e.g., books/film/music/television, for example) tend to create rather than reflect the values of society.” This prompt was read aloud to the participant and typed above the writing paper. All participants wrote in English.

Participants wrote their essay with an inking pen, on to lined paper that was placed on the surface of a digital writing tablet. On completion of the writing task, all students were given the opportunity to read over their writing and could make any amendments that they felt were necessary. Students made any posttranscript changes with a different color pen so that edits could be identified by the researcher.

Procedure

This study was approved by Oxford Brookes University Research Ethics Committee. Participants were tested individually by the first author, beginning with the selection measures and then the writing task. The writing task was recorded on a digital writing tablet (Wacom, Intuos 4) and analyzed using Eye and Pen software (100 Hz sampling frequency). The tablet angle could be adjusted to a comfortable writing position.

Data Analysis

Product measures

Spelling errors were counted from each composition. As a measure of productivity, the numbers of words written and crossed out were also counted, but do not include words added or deleted in the later revision stage. Prior to rating text quality, compositions were typed and spelling errors were corrected to eliminate scoring bias. The Wechsler Objective Language Dimensions (WOLD; Rust, 1996) analytical marking criteria were used, which comprises 6 subcomponents: Ideas and Development, Organization and Coherence, Vocabulary, Sentence Structure, Grammar and Usage, and Capitalization and Punctuation. Each component is given a score between 1 and 4, generating a total raw score out of 24 (higher scores indicating better performance). This scoring scheme was chosen for the high level of interrater reliability reported in previous research across a wide age range, from adolescents to adults (e.g., Connelly et al., 2006; Dockrell et al., 2007). This was confirmed when interrater reliability was measured for 20% of the samples (randomly selected) and ranged from κ = .72 to .84 (p < .001) for the six component scores, while the Pearson’s correlation for the overall raw score demonstrated an agreement of .94.

Process measures

Eye and Pen software enables the analysis of temporal characteristics of the written text. The tablet surface records the XY coordinates of the pen position to a laptop. Time spent on the writing task was identified from the time when the pen first started to write until it was last lifted. Pauses were identified using a 30 ms threshold (as used in Sumner et al., 2013), which is a baseline value set by the software according to the sampling frequency (Alamargot et al., 2006). A pause represents a period of inactivity, either off the paper (in-air) or when the pen is held stationary on the paper. Both these types of pauses were combined to provide an overall picture of the duration of time spent pausing while on task (in comparison to “writing time” which represents physically writing on the paper). Mean pause duration and the overall number of pauses made were also calculated.

Execution speed (cm/s) is the physical distance covered by the pen divided by the time spent writing. Time spent pausing is excluded from the calculation and thus provides a clear indication of the speeded production of the motor processes required when handwriting.

Online revisions

Compositions were played back in real-time and, when observed, the type of revisions were recorded and sorted into the following six categories: crossing out text, adding in text, correcting spelling, improving legibility (i.e., going over handwriting), punctuation and capitalization, and correcting grammar (e.g., correcting tense, changing to the plural form). The number of correct spelling changes was also noted. Revisions were watched/coded by two raters and reliability for the revision categories ranged from κ = .82 to .87 (p < .001).

Posttranscription revisions

The number of revisions made in each category (i.e., changes to the text after the full text had been composed) were calculated. The same categorization was used as above, but with the addition of: Inserting arrows to change the order of text. Identifying and coding posttranscript revisions that were clearly marked in a different color pen was particularly easy and is reflected by the 100% interrater reliability.

Results

Table 2 presents product characteristics of the written text produced by the two groups. The word count includes all words, even those crossed out. Speed of productivity was calculated as the total number of words written per minute. For normally distributed data, independent samples t tests revealed no significant group differences between students with dyslexia and the TD group on the measures of the overall writing time, t(62) = −.14, p = .88, d = 0.04, word count, t(62) = −.78, p = .44, d = 0.28, and words per minute, t(62) = −1.12 p = .27, d = 0.15.

Table 2.

Means and Standard Deviations for Scores Obtained on the Writing Product Measures.

Productivity	Dyslexia (n = 32)M (SD)	TD (n = 32)M (SD)	p
Writing time (min)	15.20 (3.70)	15.34 (3.35)	.88
Word count	279.75 (89.39)	297.93 (96.22)	.44
Words per min	18.56 (4.55)	19.78 (4.01)	.27
Number of spelling errors**	9.56 (8.30)	1.25 (2.21)	<.001
Number of crossings out	5.03 (5.06)	3.93 (4.59)	.12

Note. TD = typically developing group; word count = all words written on the digital tablet. Words per min = word count/writing time.

p < .05. **p < .001.

For the two remaining measures in Table 2 that were not normally distributed, Mann–Whitney U tests revealed that students with dyslexia made significantly more spelling errors in their final texts than TD students, U = 104.50, Z = −5.54, p < .001, d = 1.81, but no group differences were found for the number of crossings out made, U = 396.50, Z = −1.56, p = .12, d = 0.30. In proportion to text length, 4% of the text produced by students with dyslexia contained spelling errors, in comparison to only .5% for the TD group.

Product Findings

Table 3 presents the WOLD quality ratings of the written compositions. Overall (raw score), the compositions of students with dyslexia were rated as significantly poorer than the TD students, t(62) = −4.73, p < .001, d = 1.18. To control for multiple comparisons of each WOLD component, the significance level was Bonferroni-corrected to .008. Only those measures that met these criteria are flagged in the table. Students with dyslexia were rated as significantly lower than their peers in Organization and Coherence, U = 298.00, Z = −3.12, p = .002, d = 0.69, Sentence Structure, U = 256.00, Z = −3.91, p < .001, d = 0.88, Grammar, U = 249.00, Z = −3.92, p < .001, d = 0.91, and Capitalization and Punctuation, U = 248.50, Z = −3.87, p < .001, d = 0.91. However, there were no group differences in Vocabulary, U = 412.00, Z = −1.53, p = .13, d = 0.25, or Ideas and Development, U = 350.50, Z = −2.41, p = .02, d = 0.48 (albeit there is a trend toward significance for the latter).

Table 3.

Means and Standard Deviations for Scores Obtained on the WOLD Writing Quality Ratings.

Text quality	Dyslexia (n = 32)M (SD)	TD (n = 32)M (SD)	p
WOLD raw^a*	17.03 (2.66)	19.18 (2.68)	<.001
Ideas & development	2.78 (.71)	3.08 (.59)	.02
Organization & coherence*	2.41 (.66)	2.97 (.69)	.002
Vocabulary	3.00 (.67)	3.25 (.56)	.13
Sentence structure*	2.66 (.61)	3.28 (.52)	<.001
Grammar and usage*	2.59 (.61)	3.25 (.57)	<.001
Capitalization & punctuation*	2.53 (.72)	3.22 (.55)	<.001

Note. WOLD = Wechsler Objective Language Dimensions; TD = typically developing group.

Writing scoring criteria (WOLD; Rust, 1996). Raw score out of 24; each component out of 4.

Bonferroni-corrected significance, p < .008.

Process Findings

Table 4 reports the temporal writing measures. Pause time was calculated as a percentage of overall time. Writing time represents only the time spent making a mark on the paper (excludes pause time). Execution speed (cm/s) also excludes pause time and examines the distance covered by the pen divided by the writing time. No significant group differences were found for any of the measures in Table 4: mean pause duration, t(62) = .13, p = .89, d = 0.04, total number of pauses, t(62) = −.77, p = .45, d = 0.19, pause time percentage, t(62) = 1.12, p = .26, d = 0.29, writing time percentage, t(62) = −1.15, p = .12, d = 0.18, or execution speed, t(62) = .39, p = .69, d = 0.10.

Table 4.

Means and Standard Deviations for Scores Obtained on the Writing Process Measures.

Temporal performance	Dyslexia (n = 32)M (SD)	TD (n = 32)M (SD)	p
Mean pause duration (ms)	707.02 (391.25)	692.61 (403.93)	.89
Total number of pauses	897.06 (601.06)	1016.06 (639.22)	.45
Pause time (min)	7.95 (2.25)	7.53 (2.21)	.46
Pause %	53%	49%	.26
Writing time (min)	7.17 (2.09)	7.83 (2.63)	.28
Writing %	47%	51%	.12
Execution speed (cm/s)	3.67 (.95)	3.58 (.91)	.39

Note. TD = typically developing group.

Revisions: Online and Posttranscription

One student with dyslexia and two TD students did not make any online revisions. Of those who did revise their texts, no group differences (dyslexia: M = 9.87, SD 5.78; TD: M = 7.53, SD 5.21) were found for the mean number of online revisions made, t(59) = 1.64, p = .11, d = 0.31. In contrast, six students with dyslexia did not make any posttranscription revisions to their text, as well as 10 TD students. Again, of those students who did revise posttranscription, no group differences (dyslexia: M = 8.65, SD 7.78; TD: M = 6.41, SD 6.56) were found in the number of revisions made, U = 224.00, Z = −1.29, p = .19, d = 0.31.

Table 5 details the types of online and posttranscription revisions made by both groups, shown as a frequency distribution across the various revision categories (each column amounting to 100%) (see Note 1). Online + posttranscription frequencies for each group were compared statistically (the final column detailing the p values). Students with dyslexia made significantly more spelling revisions than the TD group, U = 308.00, Z = −2.84, p = .005, d = 0.65, suggesting an awareness of their spelling mistakes. However, all other group comparisons of revisions were nonsignificant (ps > .12). It is clear that, for both groups of students, crossing out text revisions were most common, and then adding text. However, comparisons from the word count composed on the tablet (shown in Table 2) to the final word count after the revisions actually made very little difference. In fact, as a whole, the mean number of words written by students with dyslexia and the TD group decreased by 5 words each. The data also suggest that students with dyslexia go back to add in punctuation more frequently than the TD group, although this was not significant.

Table 5.

Frequency Distribution of Revisions Made by Students With and Without Dyslexia.

Revision type	Dyslexia		TD		p
Revision type	Online	Posttranscript	Online	Posttranscript	p
Correcting spelling*	10%	26%	6%	14%	.005
Crossing out text	48%	19%	52%	19%	.84
Adding in text	8%	25%	6%	40%	.53
Punctuation	17%	13%	9%	10%	.12
Legibility	13%	11%	23%	9%	.75
Correcting grammar	4%	5%	4%	7%	.58
Inserting arrows	—	1%	—	1%	.96

Note. Number of students making online revisions, D = 31, TD = 30; and posttranscription revisions, D = 26, TD = 22. Values of p—revisions (online + posttranscription) combined. TD = typically developing group.

p < .05.

Of the total number of spelling errors made by students with dyslexia (including those made initially and later corrected online or posttranscription) only 20% of the errors were corrected either online or posttranscription, meaning that 80% of spelling errors were not corrected. Moreover, when students with dyslexia did edit their spellings, 85% of the edits were made to the correct version of the spelling, indicating that even when students with dyslexia worked on a spelling error it was not always corrected properly. In contrast, 34% of the total spelling errors were corrected by the TD group either online or posttranscription and each time they were corrected properly. Typically a spelling error went unnoticed in the TD group (no attempt to revise 64% of errors made), although as seen in Table 2 their average number of errors was only 1.25 (SD 2.21).

Predictors of Writing Quality

The final stage was to determine the proportion of variance in writing quality accounted for by those variables where we noted significant group differences. WOLD overall score was the outcome measure in a regression analysis. Reading and spelling ability (WRAT-3), and the proportion of spelling revisions made, were entered as predictors in Step 1. The Group variable (Dyslexia vs. TD) was entered into Step 2. The regression results are shown in Table 6. The overall model was significant, F(3, 59) = 8.19, p < .001, and predicted 30% of the variance overall. Only spelling ability was a significant predictor (in Step 1) of writing quality. Including the group comparison at Step 2 did not result in a better model fit overall (p = .08) indicating that group differences no longer existed after accounting for the role of spelling ability on writing quality, thus suggesting any impact of group was already accounted for by the variables entered in the previous step (spelling ability).

Table 6.

Regression Analysis: Results Predicting Writing Quality.

Regression predictors	β	t	R ²	∆R²	p
Step 1			.25	.21	.001
Reading ability^a	.12	0.86			.39
Spelling ability^a	.42	2.84			.006
Spelling revisions^b	.03	0.19			.85
Step 2			.30	.24	.08
Reading ability^a	.03	0.21			.84
Spelling ability^a	.27	1.57			.12
Spelling revisions^b	.05	0.41			.68
Group (D vs. TD)	.31	1.82			.08

Note. Standardized coefficients and t values are shown. ∆R² represents the change in R² with the addition of Step 2 (the Group variable). D = Dyslexia group; TD = typically developing group.

Reading, spelling (both Wide Range Achievement Test-3; Wilkinson, 1993). ^bTaken from the writing task (Graduate Record Examinations; Educational Testing Service, 2012).

Discussion

This study provides a novel contribution to the literature by examining both product and process characteristics of writing, as well as the types of revisions made during and after transcription. In doing so, we gain insight into the transcription process for students with and without dyslexia and the higher level self-regulating behavior of reviewing written compositions, both of which have been shown to be important components of the writing process (Hayes, 2012; Olive, 2014).

As expected, students with dyslexia made a higher proportion of spelling errors within their text than the TD group (Connelly et al., 2006; Galbraith et al., 2012; Tops et al., 2013). Further supporting existing studies, compared with their age-matched peers, and after correcting for spelling errors, the written texts composed by students with dyslexia were rated as lower in quality and text quality was predicted by spelling ability. Consideration of the WOLD subcomponents revealed that students with dyslexia and their age-peers actually performed similarly on the ratings of vocabulary, but below their peers on the assessment of organization/coherence, punctuation, grammar, and sentence structure, and a trend toward significance was noted when comparing the two groups on the ideas and development rating. Another study on adolescents with writing difficulties (Dockrell et al., 2007) found that the WOLD ideas and vocabulary components load on to the same factor in a factor analysis, which they interpreted as relating to semantics/meaning of the text, while the remaining components were related and classed as rule-based factors of written language. This fits nicely with our data, as reliable group differences were found to suggest that students with dyslexia have difficulty with the rule-based factors and not with written vocabulary (relating to semantics). However, as noted above, it is unclear whether the semantic aspects of developing ideas in writing is an area of difficulty for students with dyslexia and this could be usefully investigated further with larger samples.

In contrast to child studies (Sumner et al., 2013), overall lower quality ratings cannot be attributed to less text being produced, less time spent on the task, or more time spent pausing and not “text-making.” Students with dyslexia wrote a similar amount of text to their peers—a finding also found by Galbraith et al. (2012)—and, interestingly, these groups were comparable on handwriting execution, pause time, and the number of pauses made. This pattern of results suggests that the spelling problems of university students with dyslexia do not hinder the transcription process (fluency of handwriting/writing). Moreover, spelling difficulties do not appear to influence students’ ability to self-regulate the executive process of reviewing/revising text. Students with dyslexia engaged in the reviewing process during and after transcription a similar amount to those without dyslexia. However, students with dyslexia did spend a higher proportion of their revisions on correcting spellings, in comparison with their peers; although 80% of spelling errors were still not identified.

Although our initial predictions about spelling ability directly constraining the resources available for the transcription and revision processes do not appear to be fully confirmed, it may be that the influence of spelling ability on working memory and executive resources happens in a more indirect way as spelling ability predicts writing quality. Given the lower quality ratings, it is clear that students with dyslexia have more to respond to/revise and, although they revise a similar amount to the control group, they may not have the resources to identify all of the sections that need working on, and to correct them appropriately, and so more difficulties with organization/coherence, punctuation, grammar, and sentence structure remain. In addition, as suggested by Torrance et al. (2016), lower quality ratings in adulthood may be partly attributed to poorer knowledge of written language conventions, as a result of less reading exposure over the years. It is also likely that students with dyslexia will have had less writing practice. This may, in part, explain why students with dyslexia demonstrated a specific difficulty with rule-based aspects of writing: having limited text exposure linking to an underdeveloped awareness of how to use written language, even though they can use their verbal skills to compensate in other ways (idea generation and vocabulary).

Limitations of the present study should be addressed. Although comparable to existing studies, the sample may be considered small and was taken from one university. Furthermore, it is recognized that students with dyslexia in the present sample are considered “high-functioning” in comparison with the wider population of adults with dyslexia (Mapou, 2008). They may have developed other compensatory strategies to writing, given that they are often required to submit written coursework, including choosing to return to errors after production ceases, but with a higher likelihood that they will miss the errors (Van Waes et al., 2010). In addition, many students now use word processors where spelling errors are highlighted and so pausing performance in the present handwritten study may be related to the lack of awareness of every spelling error being produced. Students with dyslexia report that a spell check error notification causes them to pause more than other writers (O’Rourke et al., 2017). Moreover, allowances are now made in Higher Education when students with dyslexia make spelling errors in assignments (Conway & Turner, 2011), which may add to students not registering these mistakes. Future research considering compositions that have been typed by U.K. students with and without dyslexia would be an interesting comparison with the present data. In addition, relating to the present writing measure, reliability measures were unfortunately not available for the writing prompt, as it was not a standardized assessment. However, other researchers have used prompts from the GRE database with similar populations (Connelly et al., 2006) and interrater reliability for the scoring of the product and process measures was considered.

Although beyond the scope of the present paper, other research could examine revision strategies in more detail by either coupling handwriting (or keyboarding) recordings/analysis with eye tracking or think-aloud protocols to explore the cognitive strategies that students are using when reviewing their work; or investigating whether students with dyslexia are effective at proofreading written texts to determine whether they can identify and correct rule-based errors (sentence structure, punctuation and grammar). Furthermore, examination of executive functions in students with dyslexia and the link to self-regulatory behavior when writing could also be explored.

Practical implications that can be raised from the present findings point toward support still being required in spelling for university students with dyslexia, as well as writing more generally (i.e., organization/coherence, punctuation, grammar, and sentence structure). Less of a focus is given to teaching the mechanics of writing in secondary and further education (Kiuhara et al., 2009), but students with dyslexia are being shown to require such support (as found here and in Connelly et al., 2006). Furthermore, instruction directed toward revision strategies may also prove useful for this group, yet further research is needed in this area to determine how this may be targeted.

In summary, students with dyslexia compose handwritten texts that, overall, are rated as poorer quality (demonstrating a difficulty with rule-based conventions of writing) and contain a higher proportion of spelling errors than their peers. However, they demonstrate a similar temporal profile to their peers and adopt similar revision strategies, aside from correcting spelling more. It may be that students with dyslexia need to dedicate time to reviewing punctuation, grammar, and the organization of their texts, as these are the areas that they performed worse on in comparison with their peers. Poorer quality written text may be a product of less exposure to print or less writing practice in students with dyslexia. Further research is warranted to determine the factors that constrain written text quality in this population, especially given that written assessments form a large part of university assessment.

Footnotes

Acknowledgements

Special thanks go to all students who participated in this study.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was joint funded by The Waterloo Foundation and Oxford Brookes University.

ORCID iD

Emma Sumner

Notes

References

Alamargot

Chesnet

Dansac

Ros

(2006). Eye and Pen: A new device for studying reading during writing. Behavior Research Methods, 38(2), 287–299. https://doi.org/10.3758/BF03192780

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). American Psychiatric Publishing.

Berninger

V. W.

Swanson

H. L.

(1994). Modifying Hayes and Flower’s model of skilled writing. In Butterfield

(Ed.), Children’s writing; toward a process theory of development of skilled writing (pp. 57–81). JAI Press.

Bogdanowicz

K. M.

Lockiewicz

Bogdanowicz

Pachalska

(2014). Characteristics of cognitive deficits and writing skills of Polish adults with developmental dyslexia. International Journal of Psychophysiology, 93, 78–83. https://doi.org/10.1016/j.ijpsycho.2013.03.005

Chenoweth

Hayes

(2001). Fluency in writing: Generating text in L1 and L2. Written Communication, 18, 80–90. https://doi.org/10.1177/0741088301018001004

Coleman

Gregg

McLain

Bellair

L. W.

(2009). A comparison of spelling performance across young adults with and without dyslexia. Assessment for Effective Intervention, 34, 94–105. https://doi.org/10.1177/1534508408318808

Connelly

Campbell

MacLean

Barnes

(2006). Contribution of lower order skills to the written composition of older students with and without dyslexia. Developmental Neuropsychology, 29, 175–196. https://doi.org/10.1207/s15326942dn2901_9

Conway

Turner

(2011). Guidance for good practice: Institutional marking practice for dyslexic students (Association of Dyslexia Specialist in Higher Education, UK). Retrieved from https://adshe.org.uk/wp-content/uploads/marking_guidelines_for_good_practice.docx

Dockrell

J. E.

Lindsay

Connelly

Mackie

(2007). Constraints in the production of written text in children with specific language impairments. Exceptional Children, 73, 147–164. https://doi.org/10.1177/001440290707300202

10.

Educational Testing Service. (2012). The GRE analytical writing measure. https://www.ets.org/

11.

Galbraith

Baaijen

Smith-Spark

Torrance

(2012). The effects of dyslexia on the writing processes of students in higher education. In Torrance

Alamargot

Castello

Ganier

Kruse

Mangen

Tolchinsky

van Waes

(Eds.), Learning to write effectively: Current trends in European research (pp. 195–198). Emerald Group. https://doi.org/10.1163/9781780529295_044.

12.

Gregg

Coleman

Davis

Chalk

J. C.

(2007). Timed essay writing: Implications for high-stakes tests. Journal of Learning Disabilities, 40, 306–318. https://doi.org/10.1177/00222194070400040201

13.

Hatcher

Snowling

M. J.

Griffiths

Y. M.

(2002). Cognitive assessment of dyslexic students in higher education. British Journal of Educational Psychology, 72, 119–133. https://doi.org/10.1348/000709902158801

14.

Hayes

J. R.

(2012). Modeling and remodeling writing. Written Communication, 29, 369–388. https://doi.org/10.1177/0741088312451260

15.

Higher Education Statistics Authority. (2016). Students in higher education 2015/16. https://www.hesa.ac.uk/

16.

Kiuhara

Graham

Hawken

(2009). Teaching writing to high school students: A national survey. Journal of Educational Psychology, 101, 136–160. https://doi.org/10.1037/a0013097

17.

Mapou

R. L.

(2008). Comprehensive evaluation of adults with learning disabilities. In Wolf

L.E.

Schreiber

H.E.

Wasserstein

(Eds.), Adult learning disorders: Contemporary issues (pp. 248–273). Psychology Press.

18.

Mortimore

Crozier

W. R.

(2006). Dyslexia and difficulties with study skills in higher education. Studies in Higher Education, 31, 235–251. https://doi.org/10.1080/03075070600572173

19.

Olive

(2014). Toward an incremental and cascading model of writing: A review of research on writing processes coordination. Journal of Writing Research, 6, 173–194. https://doi.org/10.17239/jowr-2014.06.02.4

20.

O’Rourke

Connelly

Barnett

A. L.

(2017). Investigating the impact of spellcheck on text production for writers with and without dyslexia [Paper presentation]. Writing Research Across Borders (WRAB) IV, Bogota, Columbia.

21.

Peterson

R. L.

Pennington

B. F.

(2012). Developmental dyslexia. The Lancet, 379, 1997–2007. https://doi.org/10.1016/S0140-6736(12)60198-6

22.

Richardson

J. T. E.

Wydell

T. N.

(2003). The representation and attainment of students with dyslexia in UK higher education. Reading and Writing: An Interdisciplinary Journal, 16, 475–503. https://doi.org/10.1023/A:1024261927214

23.

Rust

(1996). The Manual of the Wechsler Objective Language Dimensions (WOLD) UK edition. The Psychological Corporation.

24.

Schmalz

Marinus

Coltheart

Castles

(2015). Getting to the bottom of orthographic depth. Psychonomic Bulletin and Review, 22, 1614–1629. https://doi.org/10.3758/s13423-015-0835-2

25.

Shanahan

(2016). Relationships between reading and writing development. In MacArthur

C. A.

Graham

Fitzgerald

(Eds.), Handbook of writing research (pp. 194-207). Guilford Press.

26.

Sterling

Farmer

Riddick

Morgan

Matthews

(1998). Adult dyslexic writing. Dyslexia, 4, 1–15. https://doi.org/10.1002/(SICI)1099-0909(199803)4:1<1::AID-DYS87>3.0.CO;2-F

27.

Sumner

Connelly

Barnett

A. L.

(2013). Children with dyslexia are slow writers because they pause more often and not because they are slow at handwriting execution. Reading and Writing, 26, 991–1008. https://doi.org/10.1007/s11145-012-9403-6

28.

Sumner

Connelly

Barnett

A. L.

(2014). The influence of spelling ability on handwriting production: Children with and without dyslexia. Journal of Experimental Psychology: Learning, Memory and Cognition, 40, 1441–1447. https://doi.org/10.1037/a0035785

29.

Tops

Callens

van Cauwenberghe

Adriaens

Brysbaert

(2013). Beyond spelling: The writing skills of students with dyslexia in higher education. Reading and Writing, 26(5), 705–720. https://doi.org/10.1007/s11145-012-9387-2

30.

Torgesen

J. K.

Wagner

R. K.

Rashotte

C. A.

(1999). Test of word reading efficiency. Pro-Ed.

31.

Torrance

Rønneberg

Johansson

Uppstad

P. H.

(2016). Adolescent weak decoders writing in a shallow orthography: Process and product. Scientific Studies of Reading, 20, 375–388. https://doi.org/10.1080/10888438.2016.1205071

32.

Van Waes

Leijten

Quinlan

(2010). Reading during sentence composing and error correction: A multilevel analysis of the influences of task complexity. Reading and Writing, 23, 803–834. https://doi.org/10.1007/s11145-009-9190-x

33.

Wechsler

(1997). Wechsler Adult Intelligence Scale-3, UK edition. The Psychological Corporation.

34.

Wengelin

(2007). The word-level focus in text production by adults with reading and writing difficulties. In Rijlaarsdam

(Ed.), Writing and cognition: Research and applications (Studies in writing) (Vol. 20, pp. 68–82). Elsevier.

35.

Wilkinson

G. S.

(1993). Wide range achievement test (3rd ed.). Wide Range.