Speech-to-text intervention to support text production for students with intellectual disabilities

Abstract

Aim

Writing is a multifaceted skill involving planning, transcription, and revision that is challenging for students with intellectual disabilities. Some studies have examined reading abilities. However, there needs to be more research on writing proficiency in this population. Especially concerning writing with the assistance of technologies such as speech-to-text (STT). To contribute to filling the research gap, this study aimed to investigate whether tailored speech-to-text interventions enhance text production for students with intellectual disabilities.

Methods

The research utilised a single-subject design involving the participation of four students (three girls and one boy) aged 10–13 years with mild intellectual disabilities in a rural municipality in Sweden.

Results

The results of this study revealed significant improvement post-intervention for all four students in word, sentence and text qualities.

Conclusions

The findings suggest that STT offers a valuable tool for students with intellectual disabilities struggling with handwriting, providing new opportunities for self-expression. Pedagogical implications are discussed.

IMPLICATIONS FOR REHABILITATION

• This study investigates speech-to-text technology for students with intellectual disabilities. Previous research on writing among students with intellectual disabilities is very limited and almost missing with the support of assistive technology such as speech-to-text.

• The results show improvements for all participants.

• Speech-to-text technology appears to be valuable for enhancing text production among students with intellectual disabilities.

Keywords:

• Speech-to-text
• students with intellectual disabilities
• writing
• text production
• intervention

Introduction

It has become increasingly common, both in educational and professional settings and in personal life, to conduct written communication via email and text messages rather than making a phone call. To avoid exclusion from this context, possessing a reasonably proficient level of written language skills is important. For many individuals who struggle to read and comprehend written text, writing often poses even greater challenges for students with writing difficulties in general and for students with intellectual disabilities [1,2]. The writing process encompasses various stages, including planning, transcription, and revision [3]. Transcription encompasses both handwriting and spelling. The ability to proficiently engage in traditional pen-and-paper writing necessitates fundamental competencies such as accurately recognising letter forms and their composition, word comprehension, and spelling [4]. Consequently, the revision process is a metacognitive activity that involves interactions between multiple processes during writing [5]. Difficulties in transcription processes such as handwriting and spelling disrupt the quantity and quality of writing [6]. Reading is an integral sub-skill of the writing process, facilitating the ability to read and edit texts and address issues such as spelling errors, grammatical inaccuracies, and organisational deficiencies [7]. For students with intellectual disabilities, these requirements pose a challenge, as research indicates that a significant proportion of students with intellectual disabilities exhibit weak reading skills and a larger percentage struggle with writing tasks [1,2,8]. Moreover, numerous studies have highlighted the necessity of extended time, increased learning opportunities, and better teacher support to consolidate skills for students with intellectual disabilities in contrast to their peers without intellectual disabilities [1,9,10].

Several studies have examined the reading abilities of students with intellectual disabilities [2,8,11–13]. Few studies have investigated the writing proficiency of students with intellectual disabilities. In a literature review aimed at identifying effective interventions for teaching writing to students with intellectual disabilities, Joseph and Konrad [14] identified nine studies involving 31 students who met the established criteria. The methods investigated included visual support, repetition, modelling, and positive reinforcement. The review results indicated that students with intellectual disabilities could benefit from writing instruction and grasp learning strategies that promote writing development, similar to students without intellectual disabilities, albeit with adaptations tailored to their individual needs [14]. Pennington et al. [15] investigated the effects of an intervention in sentence writing for three students aged 10–12 years with intellectual disabilities. The intervention involved using an app and specific teaching strategies to teach participants how to create sentences in response to texts. For each session, the teacher selected books that matched the student’s reading levels. After reading a book, the app selected words and created sentences in response to the story. The authors concluded that a systematic approach to teaching and the use of technology can be employed to develop writing skills in this unique group of students [15]. In an article by Canella-Malone et al. [1], the researchers highlighted six components teachers should consider when instructing students with intellectual disabilities to write. These components (ACCESS) include accommodation and assistive technology, concrete topics, critical skills, explicit instruction, strategic instruction, and systematic evaluation. While the components can be used independently, the authors recommend that teachers consider all components during their planning. Despite tailored and evidence-based instruction, it remains unclear which methods successfully support students with intellectual disabilities in producing text [1]. Bakken et al. [16] conducted a systematic review and meta-analysis to examine writing interventions for students with intellectual disabilities. The interventions used were diverse, with some focusing on cognitive strategy instruction, while others utilised technology-based tools, such as tablets and apps with text-to-speech (TTS) capabilities. This review emphasises the importance of personalised and supportive writing interventions, use of technology, and need for more high-quality research to better support students with intellectual disabilities in developing their writing skills [16].

Over the past decades, assistive writing tools such as various speech-to-text (STT) programs have emerged as a complement or alternative to traditional handwriting. This technology enables text to be written by speaking directly to an app or computer program. Utilising this technology, students with writing difficulties can bypass the transcription process and direct their focus to the text content [17]. In a group-based intervention study, speech-to-text (STT) was used to enhance the writing skills of children with Special Educational Needs and Disabilities (SEND). Data were collected over five school years from 30 children (8–16 years) with various speech and writing difficulties. The intervention involved students using STT to produce texts. The main findings suggested that STT was an effective educational tool for enhancing writing skills in students with significant support needs [18]. Furthermore, teachers can establish an inclusive and accessible learning environment to provide students with writing difficulties with the same opportunity to express their creativity like students without such difficulties [18].

To the best of our knowledge, research on the use of STT technology among students with intellectual disabilities is lacking. However, in a review of the use of STT for adolescents with learning difficulties in secondary education conducted by Matre and Cameron [19], eight peer-reviewed studies that met the inclusion criteria (students with disabilities such as dyslexia, dysgraphia, or specific language impairment) were identified. These studies primarily focused on students facing challenges directly related to the development of writing skills. Students with average writing abilities and intellectual disabilities were excluded, as the latter group was considered heterogeneous and thus challenging to compare with students primarily struggling with written language. The review’s findings suggested that STT may enhance students’ ability to produce text with fewer errors and improve reading comprehension and word recognition. However, with only eight peer-reviewed studies from three different countries (Sweden, the USA, and Scotland), there is a substantial need for more robust international research before further insights into the effects of STT on writing-related skills for students with writing difficulties can be provided [19].

While most of these studies have primarily focused on students with learning difficulties, there appears to be a significant gap in research on how STT affects the writing production of students with intellectual disabilities [19–22]. This study aimed to investigate whether students with intellectual disabilities can enhance their text production capabilities by utilising STT as an alternative to traditional pen or keyboard writing. This study poses the following research questions:

• Does individually tailored intervention training in STT increase text production (word and sentence counts) compared to handwriting or keyboard typing for students with intellectual disabilities?

• Does the systematic use of STT result in text containing more concrete and abstract content components than handwriting or keyboard typing for students with intellectual disabilities?

Materials and methods

This study used a single-subject design to investigate the effects of an individualised STT intervention on students with intellectual disabilities. In a single-subject design, a very small number of participants are used. This research methodology focuses on studying and documenting the effects of an intervention at the individual level by observing and recording changes over time, typically before, during, and after an intervention. The detailed follow-up of a few individuals allows for an in-depth analysis of how specific changes affect each participant, which differs from more traditional group-based studies where results from many individuals are combined to find average effects. A single-subject design involves collecting data from each participant over time and comparing them with baseline values [23–25].

The choice of study design was crucial, as previous studies conducted by the research group revealed significant challenges in employing randomised controlled trials or quasi-experimental designs among students with intellectual disabilities. This is due to difficulties in randomising comparable groups, given the heterogeneous nature of the student population and the challenge of making minor differences visible, which could be significant in terms of the daily schoolwork for individual students.

Settings

In Sweden, students with intellectual disabilities can either attend regular mainstream schools and be taught alongside students without intellectual disabilities or attend a specialised school known as a “compulsory school for pupils with learning disabilities”, where teachers are specifically trained to instruct students with intellectual disabilities. Regardless of school type, these students have the right to follow their own curriculum, Lgra22 [26]. The curriculum outlines how students should be taught to write words and sentences, both through handwriting and digitally. They should also receive training in spelling and sentence structure. The writing goal for Grade 6 is for the student to write words and simple sentences and to participate in writing and structuring texts, whereas the goal for Grade 9 is for the student to write multi-sentence texts with explicit content and functional organisation.

This study was conducted in a compulsory school for pupils with learning disabilities in a Swedish rural municipality.

Participants

The study included four students (three girls and one boy) aged 10–13 years with mild intellectual disabilities. All students attended a compulsory school for pupils with learning disabilities and followed a curriculum designed for students with intellectual disabilities [26]. Inclusion criteria for participation were that the students had a mild intellectual disability and speech that teachers assessed as being sufficiently clear for dictation. Teachers at the involved school chose students they thought would gain from using STT technology for text production.

Julia, a 10-year-old girl, did not have Swedish as her native language, but she had been living in Sweden since the age of four and could communicate in Swedish with a good understanding. She was diagnosed with a mild intellectual disability at the age of 9. In addition to regular classroom instruction, Julia received weekly individual literacy instruction from a special education teacher.

Hannah, an 11-year-old girl with Swedish as her native language, spent her first four school years in a regular compulsory school class and received literacy instruction both in the classroom and from a special education teacher several times a week. The diagnosis of a mild intellectual disability was made before starting the 4th grade, and that’s when she began attending the compulsory school for pupils with learning disabilities.

Emma, a 13-year-old girl with Swedish as her native language, had been attending the compulsory school for pupils with learning disabilities since the 4th grade and, in addition to a mild intellectual disability, also had ADHD. Throughout her school years, Emma regularly received individual support from a special education teacher several times weekly to aid her literacy development.

John, a 12-year-old boy with Swedish as his native language, like the other students in the study, spent several years in the regular compulsory school before being diagnosed with a mild intellectual disability. He participated in classroom instruction but received individual instruction from a special education teacher to strengthen his literacy skills.

All four students mastered all letter sounds, but their decoding abilities varied. Hannah could read simpler sentences with some fluency, while Julia, Emma, and John, despite extensive training, had weaker decoding abilities.

None of the students had prior experience producing texts using STT technology.

Both oral and written information was provided to the students and their guardians, and written informed consent was a prerequisite for participation in the study. This study was approved by the Ethical Review Authority of X (2022-04925-02).

Measurements

As previous research has indicated that the ability to write a text depends on functional reading skills [3,4,7], a special education teacher at the school administered a word-decoding test before and immediately after the intervention. LäSt [27] is a commonly used test to measure word decoding skills among students in Sweden. It consists of two parts, each containing lists of words and non-words. However, because the purpose of the test was to assess word-decoding ability, only the two lists of words were used in the study. The task for the student was to read as many words as possible on each list within 45 s. The numbers of correctly read words in each part were then combined. The potential results in the word-decoding section ranged from zero to 200 correctly read words. The test-retest reliability was .93, and norm data were available for students in Grades 1–6 of compulsory school. The validity of the test I supported by the correlation between decoding words (A + B) and other commonly used assessments, which ranges from .59 to .88.

Text quality

The definition of text quality is not unequivocal, and there is a lack of consensus within research and practice regarding how overall text quality should be assessed [28]. Two overarching approaches are to employ analytical or holistic methods. The analytical method involves an evaluator assessing various components such as vocabulary, grammar, spelling, and readability, which are subsequently summarised into an average score. A holistic method involves evaluating a text on a scale and comparing it with sample texts at different levels [10,20]. Advantages and disadvantages exist with these distinct methods; however, what is common to both is that previous research has employed them to assess the level of texts written by students without intellectual disabilities but with or without learning difficulties. For a student with an intellectual disability, writing difficulties are often not confined to a specific area such as spelling; the disability also affects the students’ ability to translate ideas into writing, the structure and organisation of the text, and the ability to use nuanced language [29]. This impacts text quantity and quality, necessitating the utilisation of elements from both methods mentioned above to create tailored conditions for assessing student texts. In this study, we chose to score the texts based on three aspects: the number of words and sentences produced and the quality of the text’s content. The first author and one of the co-authors separately analyzed the quality of the texts and then synchronised their assessments. The scoring method used to measure text quality based on the content is shown in Table 1.

Table 1. Scoring of content in student texts.

Each detail mentioned by the student:	1 point	e.g.	A dog (1 point)
about the image:			The tree (1 point)
Each new independent detail observed in the image:	1 point	e.g.	The aunt is stroking the dog (3 points)
			I see a yellow lamp and a green chair (4 points)
			I notice a woman feeding a dog (3 points)
In addition to the above scoring criteria,	2 points	e.g.	The dog will probably be happy to meet its owner (2 points)
2 points are awarded for personal reflections			Perhaps there is water in the bottle (2 points)
Not explicitly depicted in the image but			The rabbit probably wants to run on the grass (2 points)
Relevant to the context:

Procedure

The study design consisted of baseline, intervention, and maintenance. Each baseline and intervention session measured the student’s ability to write or dictate a text. Students received the same instructions, whether writing by hand or dictating the text, which involved describing what they could see in a picture. No instructions regarding the quantity of output were given, nor were any instructions provided concerning spelling, sentence structure, or other writing rules. The students were encouraged to describe the image freely. An online LIX calculator was used to determine the number of words and sentences produced by the students. The research team compiled images that were used as writing prompts in all sessions. A random image was selected for each student and session. The study was conducted over 10–12 weeks, with a maintenance assessment conducted three months later. The teacher maintained a logbook for each session throughout the study period.

Before the study commenced, participating teachers received a comprehensive overview of the study’s design. Initially, they were provided with a written guide outlining the study’s planning. Subsequently, the study’s third author reviewed the content during a digital meeting, providing an opportunity for questions. Teachers were also given a logbook detailing the content of each baseline, intervention, and maintenance session. Instructions were given that, after each session, teachers should describe the image used and provide a comment on how the particular session worked for the student.

The teachers were then trained on two occasions on the app (Claro Speak) intended for dictation. Claro Speak is an app that has many similarities with a standard word processing program. However, it also allows for scanning text and having it read aloud via text-to-speech and producing written text through dictation. The Claro Speak app was installed on the students’ iPads.

During the first session, the first author demonstrated how the app worked and explained the available settings. In the second session, teachers had the opportunity to practice using the app, experimenting with different methods of dictation and editing for satisfactory results. If there were any uncertainties about the implementation after the training, teachers were informed that the first author could return for further clarification of instructions and hands-on practice with dictation.

Baseline phase

This phase was conducted three times over two weeks. In each session, the teacher presented a randomised image and asked the students to describe what they saw in the image. The students wrote their responses independently using a pen or keyboard. The teacher then read aloud what the students had written and asked if any revisions were desired. After making any changes, the teacher re-read the text. The teacher noted the image that was used for the session in the logbook, and provided comments and responses to a few questions about the session’s implementation.

During two weeks after completing the baseline, the students practised using the STT tool (ClaroSpeak Plus) in six sessions to familiarise themselves with it before the intervention started. The practice phase included students giving responses to questions (e.g., “What is your name? What is your favorite color?”) to observe how the text was generated on the screen. Students also practised dictating words or sentences related to the images. The teacher asked questions using the STT tool, allowing students to see the questions on the screen. When errors occurred, the students practised correcting the text using the STT tool. The students who quickly learned to use the tool did not need to complete all six sessions, while there was the option for additional training sessions if students needed more practice to use the tool independently.

Intervention phase

This phase lasted six weeks, comprising 15 intervention sessions for each student. All intervention sessions were conducted individually in a separate room outside the classroom with one student and one teacher. Each session lasted approximately 30 min and commenced with the teacher prompting students with questions to practice utilising the STT technology. Subsequently, the teacher presented an image designated for the intervention session and asked the students to describe what they had seen. Students dictated their responses using STT. The teacher read aloud what the student had spoken into the tool and then asked if the student wanted to make any corrections. Once the students corrected their text and were satisfied with the results, the teacher reread the text, and the session concluded. All 15 intervention sessions were conducted similarly. The intervention aimed to investigate whether the students’ text production increased with STT. If there was an improvement in text quality, it occurred solely through student influence.

Maintenance phase

Three months after the completion of the intervention, two maintenance sessions were conducted. The first session mirrored the baseline, in which students wrote or typed their image descriptions manually. Students used the STT tool in the second session to dictate their image descriptions. Images previously used at baseline were randomised for handwriting and STT sessions to compare the text production of the same image before and after the intervention.

Social validity

Throughout the study, teachers wrote daily comments on how students experienced the intervention sessions and how the technology performed. In the middle and end of the intervention, the first author gathered all participating teachers to discuss how teachers and students perceived the training. Feedback regarding user-friendliness, meaningfulness, and the potential for transferability to daily work was discussed during these sessions.

Data analysis

The data were analyzed using visual analysis and a statistical test of non-overlap of all pairs (NAP). Visual analysis methods involve examining data patterns and trends by visualising them in graphs or charts. Presenting data graphically enhances the comprehensibility of the results [30]. In this study, the NAP assessed the degree of data overlap between the baseline and intervention phases [31]. An online single-case effect size calculator was used for the NAP calculations [32].

Results

All four participating students completed the baseline, intervention, and maintenance phases. The students’ teachers reported that the students found it enjoyable to use speech recognition to produce text. With few exceptions, they completed all intervention sessions to the best of their ability. Visual inspection of the data indicated a stable baseline for the four students (Figure 1) but also revealed variability in their ability to write text by hand. The number of written words, number of sentences, and text content scores increased for all students during the intervention compared to the baseline. However, there were individual differences among the students; for some students, there were individual variations in the results during the intervention. Similarly, the word-decoding test conducted before and after the intervention showed variations in decoding ability among the students (Table 2).

Figure 1. Number of words, sentences, and scores in text production at baseline, intervention, and maintenance for each student.

B = Baseline, I = Intervention, M = Maintenance.

*

= Handwriting or keyboard typing.

= Speech-to-text (STT).

Table 2. The number of decoded words at the baseline and maintenance phases.

	Baseline	Maintenance
Julia	8	6
Hannah	58	53
Emma	23	28
John	18	20

Julia’s results indicated weak decoding ability in the initial testing of the number of decoded words (Table 2). Similarly, the baseline measurements indicated significant challenges in writing and composing text for her (Table 3). Julia’s results varied across measurements, as evidenced by the relatively high standard deviations in Table 3. Similarly, a visual inspection of the data revealed that Julia’s results during the intervention were inconsistent across measurements. At the same time, all measured outcomes were higher than any of the baseline measurements for the number of written words and sentences and the content scores (Figure 1). The NAP showed strong effects (1.00) for the number of written words and sentences and content scores. The maintenance phase exhibited a similar pattern to the baseline and intervention, indicating that Julia’s capacity to generate text increased when using STT technology.

Table 3. Participants’ mean scores regarding written words and sentences and scores on the text content during the baseline and intervention phases.

	Julia	Hannah	Emma	John
	M (SD)	M (SD)	M (SD)	M (SD)
Written words
Baseline scores	0.3 (0.6)	18.3 (2.3)	1.3 (1.5)	0.0 (0.0)
Intervention scores	16.1 (6.0)	40.9 (9.0)	16.5 (7.2)	7.8 (4.3)
Written sentences
Baseline scores	0.0 (0.0)	1.0 (0.0)	0.7 (0.6)	0.0 (0.0)
Intervention scores	2.3 (1.0)	5.0 (1.2)	3.0 (1.6)	1.1 (0.4)
Text content points
Baseline scores	0.3 (0.6)	5.7 (2.1)	1.0 (1.0)	0.0 (0.0)
Intervention scores	6.9 (1.8)	13.6 (2.6)	7.1 (3.1)	3.9 (1.2)

M: Mean; SD: Standard deviation.

Hannah was the most vigorous writer of the four participants during the baseline phase (Table 3). However, in the intervention phase, she used more words and sentences and obtained higher content scores than at baseline (Table 3 and Figure 1). Visual inspection of the data in Figure 1 showed a substantial initial increase in Hannah’s results at the beginning of the intervention phase, followed by a more varied pattern. The mean number of written sentences increased from one at baseline to five in the intervention phase. The intervention results demonstrated a strong effect of NAP (1.00) relative to the baseline in all measured aspects. The maintenance phase revealed that the effect of the intervention persisted in terms of both the number of words and sentences produced and text content.

Emma’s baseline results suggest a weak ability to produce text through handwriting (Figure 1). Visual inspection in Figure 1 displayed varied results, primarily during the intervention phase. The means of the respective measured components in Table 3 indicated that Emma’s results increased during the intervention phase. Emma’s mean results more than doubled between the baseline and intervention. This was applied to all measured components: number of written words, sentences, and content scores. During the intervention, the number of written words, sentences, and content scores showed a strong effect on NAP (0.93–1.00). The maintenance phase demonstrated that the ability to produce words and sentences increased when using STT technology.

According to Table 2, John had a limited decoding ability. Moreover, in the baseline phase, he displayed significant difficulty producing text by handwriting or keyboard typing (Table 3 and Figure 1). Visual inspection of the data during the intervention revealed substantial individual differences between the baseline and intervention phases. John increased his ability to produce text at both the word and sentence levels and improved the text quality. NAP showed an effect size of 1.00. The same pattern was observed during the maintenance phase.

Discussion

This study aimed to investigate whether the use of STT enhances the ability of individual students with intellectual disabilities in Sweden to produce text. After the completion of the intervention, the results demonstrated a significant increase in the ability of each student to produce text. The NAP values indicated strong effects (NAP = 0.93–1.00) on the number of written words and sentences and text quality. STT appeared to be a successful alternative for text production among students with intellectual disabilities.

In this study, all students showed significant improvements in their text production during the intervention compared with their baseline performance. Julia, Emma, and John had weak decoding abilities, which may have led to challenges in handwriting-based text production [6,7]. However, they exceeded expectations by increasing their text production when using STT. Surprisingly, Hannah, despite having stronger decoding abilities and not being expected to face the same handwriting difficulties, exhibited a similar pattern during the intervention compared to the baseline. She demonstrated an immediate increase at the start of the intervention, followed by consistent progress. Previous research has highlighted the need for students with intellectual disabilities to receive more time and teacher support to consolidate new knowledge [1,9]. However, in this study, teachers reported that students quickly embraced and adapted to the technology, which was further supported by the maintenance results, where the difference between handwriting and dictation in STT remained consistent even during the maintenance phase.

The significant increase in the production of text by each student when using STT raises questions about whether students with intellectual disabilities have the opportunity to express their actual knowledge, thoughts, and emotions through traditional text production methods, such as handwriting and keyboard typing. Even though it is challenging to determine what constitutes low or high scores owing to the lack of standardised tests for this group, there is a risk that students with intellectual disabilities may be mistakenly perceived as facing more significant challenges than they actually do because they are not given opportunities to demonstrate their abilities in text production. While STT may not be a solution for all students with intellectual disabilities, it can be a valuable tool for those who struggle with handwriting. The difference between the ability and inability to produce coherent texts is substantial, making it a topic worthy of further investigation.

The present study’s results suggest that all four participating students benefited from using STT for text production. Based on their individual capacities, all students significantly increased their number of written words and sentences, thus addressing the first research question.

The second research question pertained to text quality. To the best of our knowledge, there is limited research on how text quality can be assessed in texts written by young students with intellectual disabilities. These disabilities typically entail difficulties across all domains that influence writing proficiency, that is, planning, transcription, and revision [3], making it complicated to evaluate text quality using the aforementioned holistic or analytical methods [20]. This study primarily aimed to investigate if and how STT affected text production for students with intellectual disabilities. No instructions regarding formal aspects were provided to the students, and the research team chose to maintain a simple concept of text quality. Thus, we scored the number of concrete details provided by each student and the number of abstract assumptions each student could make regarding a given image.

This study’s results are significant from several perspectives. First, for students with intellectual disabilities who lack or have limited ability to express themselves through handwriting, STT provides them with entirely new opportunities to convey themselves through writing. Consider John as an example: Unlike his peers, he did not experience the same improvement in the number of written words or sentences. He could not compose coherent texts when he attempted to write them manually. However, when he utilised STT, his performance improved compared with his baseline results. It is worth noting that even minor improvements can hold substantial clinical significance for individual students. In John’s case, the use of STT not only enabled him to write but also allowed him to express himself in writing. Second, as STT positively impacts text quantity [6,17], it provides a basis for work on text quality; that is, teachers can instruct students on how to produce text in sentences and paragraphs. Third, STT allows teachers to create an inclusive and accessible learning environment, ensuring that students with writing difficulties have the same opportunities to express their thoughts as students without such difficulties [18]. Fourth, students with intellectual disabilities face challenges in reading and producing text in traditional ways [1,2]; nevertheless, this study’s results indicate that their capacity is more significant than what is typically expected when they are provided with the right tools to showcase it.

Limitations

The study was based on one-on-one teaching, that is, individual instruction from a teacher to a student. Although all involved educators received thorough training before the study on how the technology worked and how the different sessions were to be conducted, there was still a risk of varying conditions for the students when different teachers conducted interventions. Moreover, it was important that everything was as normal as possible, that is, that the student was taught by someone they were familiar with and trusted.

One limitation of this study is that we did not use standardised tests for the target group. To the best of our knowledge, standardised reading and writing tests designed for students with intellectual disabilities are unavailable, at least in Sweden.

A setup with only three baseline measurements may be seen as limited. Using a multiple baseline design would have been methodologically preferable, but students with intellectual disabilities face challenges in reading and writing, and conducting additional measurements would likely impose a greater burden on them rather than providing additional information for the study. Moreover, it is questionable whether carrying out a larger number of tests is ethically correct when students cannot read. Furthermore, each student reached a saturation point during baseline measurements, indicating that conducting further measurements would not have affected the results.

Another factor that could have influenced the intervention results was that some images might not have motivated or interested the students enough for them to describe them, while other images might have been more motivating. Students with intellectual disabilities often struggle to maintain focus and concentration for extended periods, which could also negatively impact the intervention results. However, these aspects form part of intellectual disabilities and cannot be ignored.

Technology such as STT has limitations even though advancements are constantly being made. Students must be precise in pronunciation and speak at a slightly slower pace than usual. One risk with technology is that if the teacher fails to maintain the student’s motivation, they might lose interest when the dictated content does not appear in writing as expected.

The results of the study might be generalisable to other contexts, as several studies conducted among students without intellectual disabilities [19–22] have shown that speech recognition is useful for text production.

Practical implications

Several studies have indicated that most students with intellectual disabilities struggle with reading and writing [1,2]. One study suggested that handwriting and spelling difficulties negatively impact text quantity and quality [6]. The results of the current study support teachers of students with intellectual disabilities in using STT in their instruction, benefiting students with and without writing difficulties. Education for students with intellectual disabilities in Sweden has been extensively discussed in recent years. There is a lack of information about students’ knowledge levels and how their progression is ensured. The Swedish National Agency for Education has formulated grading criteria in the new curriculum, Lgra22 [26], regarding which students should be challenged. The ability to express oneself in writing is essential for meeting these criteria. Students should be taught to write words and sentences, practice spelling and sentence structure, and strive to structure texts effectively [26]. One risk is that students may not move beyond being taught to write words and sentences because many students with intellectual disabilities face transcription difficulties [17]. By using STT, attention is directed toward the content of the text, and students with writing difficulties receive support in bypassing the transcription process. To work on text structure, an appropriate text is required, and if STT, as shown in this study, increases text quantity, new opportunities emerge to improve text quality for each student with intellectual disabilities.

Conclusions

This study aimed to focus on the use of technology and investigate whether STT could be an alternative for students with intellectual disabilities to produce texts. The students were instructed to describe what they knew about a presented picture rather than concentrate on formulating correct sentences or using a specific number of words in a nuanced language. The results indicated increased text production when students did not need to think about the transcription process [6,17] and the need for different challenges in future text production. This study highlighted each student’s individual development by employing a single-subject design. The choice of this approach also showed heterogeneity among students with intellectual disabilities. Although there were only four participants, all of whom had mild intellectual disabilities and were roughly of the same age, they displayed unique conditions and different requirements for support in enhancing their text production abilities. However, using STT for text production proved successful for all students in this study. In future studies, it would be interesting to expand STT with personalised interventions focusing on text quality to enhance students’ opportunities to produce texts with nuanced content.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by The Swedish Institute for Educational Research.