Abstract
Though speaking skills in the mother tongue develop naturally, developing reading skills requires training. Dyslexia is a language-based learning disability that can affect the reading ability in any language. Although many literature is available to address these challenges, the need to teach sentence reading was neglected. This paper addresses problems faced while reading sentences. As computer-based multi-sensory teaching is considered as an effective way to teach children with dyslexia (CwD), the proposed system uses multisensory teaching methods to teach sentence reading in Hindi and Bangla languages. It shows sentences as word packages to teach word sequence and help with comprehension. The platform records the response time of the user to generate a performance report to monitor her performance. For evaluating the system, an expert survey was conducted with 26 special educators in two phases, pre-test and post-test. The questionnaires were prepared following previous literature under the expert’s supervision. The pre-test data helped in understanding the educators’ experience and opinions about assistive technology and the post-test data were used to evaluate the system. In the evaluation, all educators agreed that the system will be very useful for their students and it will give a generalised solution for all the challenges their students face.
Introduction
Education is the basic right of every person and educational technology aims to deliver it to a diverse population with some technical help. It covers a huge area related to education and education-focused technical support including teaching strategy tools, assessment tools etc. It targets a huge group of population from different socioeconomic classes, occupations, ages, abilities or disabilities. Reading is one of the fundamental skills needed for education, but people with dyslexia have difficulty in developing reading skills. They face problems in different aspects of reading such as phonological awareness, spelling skills, word recognition skills etc. [1]. It is a neurobiological disorder [2] that affects the working memory of a person [3] and is not affected by the intelligence [3] of the person. People with dyslexia often have average or above-average intelligence [4]. Having a poor working memory, these children require more time to learn things than their peers. According to the levels and types, the challenges for each person are unique.1
In India, awareness about dyslexia was first raised in 2009 when the RPwD Act was enacted by the Government of India to include specific learning disabilities [8]. It talks about the standard assessment techniques, remediation, and inclusive learning and also about providing facilities [9] that would help children with dyslexia to continue their studies in regular schools. Even though the act was enacted more than a decade ago, very few regular school teachers and individuals without dyslexia are familiar with dyslexia and dyslexia-specific teaching approaches [10, 11]. As a result, implementing inclusive learning in traditional schools is challenging [12]. Additionally, regular schools do not have enough teachers and resources [13] to conduct one-on-one sessions. On the other hand, dyslexia is a language-based learning disability [14] and as the linguistic property for every language differs [15, 16] from the other, the difficulty faced in different languages may differ. Children, for instance, make more orthographic errors while reading Hindi, and more phonological mistakes while reading English [17]. Again, the existence of conjuncts and vowel modifiers in Bangla that are also present in Hindi and other Indian languages makes reading more difficult [18, 19].
The best way to teach a child with dyslexia is by using a multisensory teaching approach [20] that was introduced by Orton-Gillingham [21], which guides to teach a child according to her ability, conducting one-to-one classes and teach using all possible ways employing most of the senses. Again, since English is a foreign language in India, it must be introduced slowly and systematically. Several teaching methods adhere to the Orton-Gillingham approach, but they take time and are rigorous. Furthermore, every student cannot afford a special educator. In these situations, it’s necessary to give a platform that will allow a child to learn without special educators.
The motivation of this work is to provide holistic support to educators in developing sentence reading skills in children with dyslexia. Teaching is a challenging task that needs a good amount of time, energy, understanding and patience. Children with dyslexia take more time to learn anything [22]. thus it is preferable to offer the lesson in a form that will help the child retain the information better [23] and as the difficulties of each child differ from the others, special educators assess each student separately and prepare a customized curriculum for each of them which is a rigorous process. In this situation, a system that supports educators in teaching different language skills at different levels and replicates the teaching method that special educators use was long needed. Moreover, while teaching a student an educator measures the progress through some measures such as response time, reading and writing errors, how many times she took any help etc. Among all of these, the response time is known as the most accurate measure [24] of performance. Even though very few language learning systems are available in Bangla and Hindi, these do not use a dyslexia-specific learning method. As a result, special educators use some of the available systems for some of their lessons but are not able to switch to a technology-based teaching environment [25, 26]. Along with the multisensory teaching approach, another important way to develop and improve skills in dyslexia is to repeat a lesson [21]. Developing a system that regenerates the teaching approach used by special educators can help in practice and repeat the lessons.
Literature suggests that the computer-assisted multisensory teaching process is the best way to teach children with dyslexia [27]. There are numerous assistive tools available for dyslexia, including ones for phonics, alphabet recognition, spelling, writing, arithmetic, and reader applications. These are available for different platforms such as desktop applications, web applications, android etc. Besides that, game-based assessment tools are also available [28]. Models for predicting student engagement had been developed [29, 30], which are intended to assist in determining the level of difficulty for a given text. And educators may modify the text based on this.
Assistive technology for dyslexia
Despite being very beneficial, the above-mentioned models do not directly aid in reading. A platform called Dyswebxia [31], which operates on a web platform and changes the displayed texts, can assist any user with dyslexia in reading a text. There are some adaptive learning platforms accessible, and Agent-dysl [32] is one of them. It offers children individualised support so they can participate in conventional school classes. Aidodys [33] also functions similarly. Learning apps that aim at improving sentence reading fluency include GraphoLearn [34], Reading Acceleration Program [35], Omega-Interactive Sentence [36] etc. The majority of these tools focus on words to improve word reading fluency with the assumption that doing so will also improve sentence reading fluency. One such tool is Omega-IS, which teaches readers how to read sentences by rearranging words in the proper sequence. Although there is no platform accessible for learning sentences using the multisensory method, there are many applications available for other language skills, such as Bijak Membaka [37], LexiPal [38], Wridy [39] and Discover [40]. Besides these, in recent times two works are available in Bangla for dysgraphia [41] and dyscalculia [42], however for sentence reading in dyslexia no work had been found.
Assistive technologies for dyslexia in hindi and bangla
Even though there is relatively less literature on dyslexia in Bangla and Hindi, there is a lot of work on other disabilities [43, 44, 45, 46]. Additionally, there is hardly any literature available for sentence reading. As a result, children rely on text-to-speech platforms available in Bangla [47, 48] and Hindi [49, 50]. Although these technologies allow students to hear the speech output for a text that is displayed which helps them to avoid reading, these are not specifically designed to address the needs of children with dyslexia. There is a lot of literature on other language skills, such as word predictors [51], virtual keyboards [52], speech recognition software [53] to improve writing ability, learning the alphabet [54], and speech synthesizers to increase word level readability [55], even though there is very little literature for reading sentences. Our search indicates that Avaz,2
The limitations of the past works are discussed below
To the best of our knowledge, no available literature functions at the sentence level in Hindi. And even for the rest of the reading areas such as phonological awareness, word recognition skills, etc. a handful of systems are available in Bangla. Learning to read sentences is just as crucial as learning phonological awareness, word recognition, and spelling. The systems that aim to improve sentence reading abilities focus on the word level in the belief that it will naturally improve sentence reading abilities. There is a very small number of publications that employ a multisensory teaching strategy. There was a clear need for a system that teaches sentence reading through a multisensory approach. When a sentence is divided into successive groups of words depending on word order and count, each group is known as a word package. It is essential in teaching sentence reading as it helps in developing reading comprehension [58]. It was necessary to create a prototype with this feature since the works that are currently available do not offer it. Though a sentence learning system was developed in English which follows the dyslexia-supported teaching method [59], in both Hindi and Bangla, such kinds of works are not available. Moreover, in both languages, the available works are not developed recently.
In this paper, a system has been proposed to teach sentence reading in Bangla and Hindi through a multisensory learning method. To meet the objective some approaches had been made.
Workflow diagram of the system.
As computer-based multisensory teaching is considered to be more effective in teaching a language, this method has been implemented in the proposed system. In this software, four senses or activities are used to help children learn a text, seeing, hearing, speaking and writing. All these activities will be used one after another for the same lesson and after completing a lesson, the user will go to the next text and again repeat all the activities. In Fig. 1 the working principle has been shown. For implementing the multisensory teaching method, the Orton-Gillingham approach has been followed. The senses that are targeted by the system are mentioned below.
Vision: From the very beginning of starting the system, the user will use her sense of vision. In the system, the sentences have been shown in the form of word packets and before start reading the user will see the text and process the graphemes to understand the language and identify the displayed text.
Hearing: The sense of hearing gets used when the second activity, i.e., listening to the displayed text is performed. The system produces speech output with an incorporated text-to-speech engine. After seeing the displayed text the user can listen to the text with the help of a button on the screen and the user can use it many times if she needs it. Speaking: Here speaking the text implies reading the displayed text. Reading includes many cognitive skills, such as letter recognition, orthography-to-phonology mapping, and finally connecting the sounds to speak. As so many cognitive processes occur together, it affects the working memory. Moreover, reading, listening, and writing all together cause relearning of lessons which helps in remembering the maximum of the lesson [21]. After listening to the sound of the displayed text, the user reads the sentence and there is a speech-to-text converter connected in the background, it converts the speech input into text and then through string matching the system checks if the user has read the text correctly or not.
Writing: Writing helps in remembering lessons better, hence it was important to include a writing activity in the system. In this system for writing the sentence the users use a virtual keyboard to type the text, writing in this way will be as effective as writing using pen-paper as it incorporates the retrieval practice. In addition, typing is another complex cognitive method that includes eye movement and finger movement. While typing the users tap on the letters that activate the sense of touch along with the kinaesthetic sense. After the student writes the sentence, the system again checks if she writes it correctly or not and on a mistake, the system gives a picture hint to try again
As mentioned before, special educators measure students’ progress through some parameters, based on the suggestions received in the previous evaluation [60] the parameters used in generating the performance report were reconsidered and along with response time, number of reading errors, number of writing errors, total count of audio assistance utilization and total count of hint use are taken.
Hindi and Bangla both have alphasyllabic orthography, which has some characteristic differences from alphabetic orthography for example English. As the targeted group will learn these languages as a native language we can assume that they will have good speaking fluency in these languages as children develop speaking skills in their mother tongue naturally.3
Apart from orthography, the sentence structure in Hindi and Bangla is also different from that of English. English follows a subject-verb-object structure whereas Hindi and Bangla follow a subject-object-verb structure [16]. All the mentioned differences have been addressed in the proposed system.
To develop reading skills, the proposed system has been designed in a way that students can learn and revise different spelling rules from the tasks available in the system. The system consists of five tasks for each of the spelling rules and each of the tasks has five sentences. After starting the system the users will be able to select a task to perform and after that, each sentence listed in that task will start to appear in the form of word packets. For each of the word packets, the users will be encouraged to perform three activities listening, reading, and writing. When all these activities are complete the system will move on to the next sentence of a task. According to some research, response time is the accurate measure of performance evaluation [24], hence at the end of a task, the system produces a performance report.
The figure shows how the system displays a sentence in Bangla UI in the form of word packages.
While developing the Hindi and Bangla versions of the previous sentence learning system [59] which is available for English, the main challenge was to create an appropriate lesson plan, selection of spelling rules and the distribution of it among different tasks. There was another challenge in developing the keyboard which is discussed in Subsection 4.2. As English is a much-researched language, a defined curriculum exists which had been used as an example while creating the curriculum. However, Bangla and Hindi do not have any predefined lesson plans for dyslexia. The challenges of dyslexia in these two languages are also not explored properly. Hence before implementing the systems, all the linguistic features of these languages were explored thoroughly and with previous research and mainly special educators’ guidance, the difficulties of dyslexia in these two languages were understood to prepare the curriculum. After designing the initial curriculum, some suggestions were given by special educators in the first evaluation of the system which was also conducted with special educators using a system-related questionnaire [60]. Based on this, the curriculum was redesigned after thorough research on the linguistic features of Bangla and Hindi as mentioned before. Based on their suggestions some textbooks were referred to while redesigning the curriculum.
For teaching sentences at a primary level, educators follow different books and depending on their suggestions tasks were created following the spelling rule from books, “Barnaparichay” and “Kishalay” for Bangla and “Vasant” and “Rimjhim” for Hindi. In these books, the spelling rule division is explained clearly and while distributing the spelling rules among the tasks the authors have arranged them according to the difficulty level of the rules. Another challenge after distributing the spelling rules was selecting the sentences to cluster under each of the tasks. While selecting the sentences, many parameters were considered such as average length of sentence, average length of words in the sentence, average number of syllables present in a word, presence of conjuncts in words, and presence of vowel modifiers which can be used to measure readability of sentences [61].
The flow of sentence in the Hindi interface is shown. The system shows every sentence in the form of word packages.
The system has two separate but similar modules for two different languages. To implement mention the multisensory teaching approach some components have been added to the system
Display screen: As it is more useful to teach sentences using word packages [58] the proposed software shows sentences in a word-by-word way, which is shown in Figs 2 and 3, And this way all the sentences are shown and when a whole task is completed the performance chart is generated. Speech synthesis: It is used to listen to the text after seeing it, for generating the speech output the Google text-to-speech converter had been used. A button is available on the screen that is associated with the component. Reading error detector: We have used Google speech-to-text API for converting the speech input of the user, it converts the speech into text and then through string matching it compares the generated text with the displayed text and generates a similarity score. When the similarity score comes as more than 95%, it is concluded that the user read the text properly. As the Google speech-to-text API does not follow the rule of capitalization and punctuation while converting the speech to text, the similarity threshold is set to 95%. Alphabet-picture hint: the user may get stuck while reading hence the software has an alphabet-picture list as a hint for both languages. Each list has 10–12 alphabet-picture pairs and there are 5 lists used in both the interfaces, in Fig. 6 one alphabet-picture list for both interfaces can be found. Writing field and keyboard: As the physical keyboard of any desktop or laptop does not have a Bangla or Hindi keyboard, it was essential to add keyboards for each language. Though Hindi and Bangla phonetic keyboards are available, it is complicated for a child to use. Moreover, the use of an external virtual keyboard while performing a task will cause distraction. The keyboard design was another challenge of Hindi and Bangla systems which was different from the English sentence learning system [59]. Both languages have a widespread alphabet list with vowels, consonants and a huge variety of conjuncts which is absent in English. The consonant clusters in English use the regular orthographic representation, they use the traditional alphabets to implement conjuncts, but in Bangla and Hindi every cluster has a different representation and some of them have more than one representation, a transparent representation and a non-transparent representation [18]. After developing the first model, it was evaluated by thirty special educators to evaluate the design [60] where the educators suggested to redesign the keyboards for both Hindi and Bangla systems. As a solution, all the alphabets had been clustered first into groups like vowels, vowel modifiers, consonants, diphthongs and triphthongs and different areas had been allotted to different groups that can be treated as sub-keyboards. In Fig. 7 all the sub-keyboards are shown. When a user taps on a conjunct, the system shows its separated form.
Performance report generated by Bangla UI after completion of a task.
Performance report generated by Hindi UI after completion of a task.
As mentioned before, response time is the most accurate measure for performance evaluation. The secondary objective of the system is to help educators in teaching by decreasing their workload. Hence a detailed analysis of the student’s performance is done by the system and the report is delivered in the form of a document. As the educators mentioned in the previous evaluation phase which was discussed in the authors’ previous work [60] that only capturing the response time is enough to assess the performance of students, the proposed work records the reading time for each word, the number of errors while reading, the number of times the hints were used, the number of errors while writing, number of times the audio help is taken. When the user starts to read a sentence a timer starts in the background and while she performs all the activities the timer runs. When the user completes reading the whole sentence it stops and stores that time. And again when the next sentence starts the timer starts automatically from zero and this way the reading time for each sentence is calculated, and in between when the user moves from one-word packet to the other, the time intervals are captured by the system.
Reading each sentence with the system can be considered as a small task as it includes many activities even at the word level. Hence completing a task takes a good amount of time and when the user finishes the last sentence the reading times along with the sentences and the other details are displayed in a form of a report, called the performance report. In Figs 4 and 5 the system generated performance reports are shown. A user can store her performance record in a PDF format using a given button at the end if wanted. When the report gets stored the data gets deleted automatically from the system memory and this way the memory space used by this system will remain fixed. The purpose of the chart is to give the educators an overview of the performance of the student, with the help of it the teacher will still get to assess the performance of the user even if she performs the task in the educator’s absence.
Bangla and Hindi system screenshots showing alphabet-picture charts used as hint. (a) Screenshot of Bangla interface showing one of the five alphabet-picture charts; (b) Screenshot of Hindi interface showing one of the five alphabet-picture charts.
The diphthong-keyboard and triphthong-keyboard are shown which can be accessed by two specific buttons.
Literature shows that the content and efficiency of an educational system can only be evaluated with experts, not with students [62]. Hence for evaluation, an expert survey was conducted. In the experiment, 26 special educators, who have been working with dyslexia for more than 5 years, participated. Of these 26 special educators, two teach mathematics, five are associated with vocational training, three teach language in higher grade, and sixteen teach in the primary grades. Among these sixteen educators fourteen teach bangla or hindi only and other two educators teach all subject including language. The experiment was conducted in three parts, a pre-test questionnaire fill-up, a demonstration of the system followed by a post-test questionnaire fill-up. To get a more system-specific response, a questionnaire was designed and validated with the help of three special educators and psychologists.
Clustered bar chart – table 1 question 1 – Diagrammatic representation to show educator’s experience in terms of years in the field of dyslexia and it shows most of the experts have experience of more than ten years.
Clustered bar chart – table 1 question 2 – A figure to show different levels of dyslexia and the number of educators work with those levels of dyslexia; most of the participants work with moderate, mild and borderline dyslexia.
The questionnaire was developed with the help of some previous literature [63, 64, 65, 66]. The questions from these paper were selected and then modified or specified based on the query of the authors, the system. Some dyslexia specific terms were also used while modifying the questions. After designing the questionnaire, it was verified and revised by three field experts who work with dyslexia. As mentioned, the evaluation was done in three parts, where two different questionnaires were used and in the post-test part, a categorical evaluation was performed, the system was evaluated based on three categories, relevance of content, design of the interface and usability. Each of the categories had five questions and a mix of multiple choice questions, short answer type questions and a Likert scale was used.
Pre-test questionnaire
Pre-test questionnaire
Post-test questionnaire to evaluate the relevance of content
Post-test questionnaire to evaluate the system in terms of instruction design
Post-test questionnaire to evaluate the system in terms of usability
Pre-test questionnaire: Table 1 shows the questionnaire used in this part, it had three multiple-choice questions and two seven-pointer Likert scale, the options of the multiple choice questions are shown in Table 1. In this segment, the educators’ expertise was evaluated to understand how much experience they have in the field of dyslexia and also their experience with assistive technology. Post-test questionnaire for content evaluation: The questionnaire used to evaluate the content designed for this system is shown in Table 2. It evaluated the content in terms of spelling rule selection, spelling rule distribution, sentence selection. It had five questions where seven-pointer Likert scales were used. Here one represents strongly negative feedback and seven represents strongly positive feedback. Post-test questionnaire to evaluate design of interface: Questions used for this category are shown in Table 3. It had two short answer type question and three seven-pointer Likert scales. For questions 3 and 4, educators were allowed to mention more than one feature to remove or add. Post-test questionnaire to evaluate usability: Questions used in this category were designed to examine the efficiency of the system, there were two multiple choice questions and three seven-pointer Likert scales. The questions are shown in Table 4 and the options for the multiple choice questions are shown in that table. Through the first three questions of this category, the authors tried to understand the participant’s opinions i.e., how much did the like the system.
Bar chart – table 1 question 3 – A graphical implementation of the opinion related to the importance of teaching sentence reading.
Bar chart – table 1 question 4 – A graphical representation of educators’ opinion about the importance of multisensory teaching approach.
Clusterd bar chart – table 1 question 5 – A figure that shows the number of educators who use technical help while teaching and how frequently they use.
Bar chart – table 2 question 1 – A graphical representation of educators’ answers about the effectiveness of the curriculum design in teaching sentence reading.
The evaluation was conducted in one-to-one sessions. Before the evaluation started, the content of the educators was taken. In the beginning, the educators were asked to fill up the pre-test questionnaire which is shown in 1, as the questions were clear and to the point, there was no need to explain the questions except the Likert scale grading system. After pre-test questionnaire fill-up the educators were given a break of five minutes before starting the demonstration, where all the educators were introduced to both the modules developed for Bangla and Hindi. The demonstration was twenty minutes long, at this stage the educators had the opportunity to use the system by themselves, though it was not mandatory, some of the educators used and checked different features of the system for better understanding. After the demonstration educators were asked to rate the system out of seven using the post-test questionnaire. The questionnaires were presented in google forms where the experts marked their answers or wrote them. After collecting the answers it was converted into a Microsoft Excel sheet using Google Forms inbuilt feature. Then all the scores were implemented into charts and analysed and compared.
Bar chart – table 2 question 2 – The figure represents the educators’ perspective about the relevance of the curriculum for the targeted user group.
Bar chart – table 2 question 3 – The figure represents educators’ opinion about including conjuncts in the primary level.
For analysing the result different charts were used. Though Google Forms generates graphs for every question, Microsoft Excel was used to generate the most suitable chart to evaluate and compare the feedback. All the results concluded after analysing the feedback are discussed below.
Pre-test findings
Question 1: All of the educators who participated in the evaluation had an experience of more than five years. 5 of the educators have experience of for than twenty years, 7 educators have experience of six to ten years, 9 educators have experience of eleven to fifteen years, and other five educators have experience of sixteen to twenty years. Figure 8 displays a bar chart obtained by the answers of this question. Question 2: Based on the training and experience educators are assigned to any level to work with and they can work with one or more dyslexia levels. Fifteen participants work with borderline dyslexia, eighteen of the total twenty-six participants work with mild level of dyslexia, twenty-five of the total twenty-six educators work with moderate level of dyslexia, and, nineteen and seven participants among the total twenty-six educators work with severe and profound levels of dyslexia respectively. The bar chart, generated from the answers of this question is shown in Fig. 9. Question 3: From the feedback of the special educators, it can be concluded that teaching sentence reading is important even if the learner can read words fluently. The finding of this question is also supported by previous literature [67]. The bar chart, shown in Fig. 10 represents the result of this question which shows a total of twenty-two positive responses with ratings five, six, and seven. Question 4: All of the educators commented that multisensory teaching is important in teaching children with dyslexia, the Fig. 11 shows the rating given by the educators on a seven pointer Likert scale for this question. twenty-three participants rate this teaching method a seven out of seven. Question 5: Four educators, who are associated with vocational training do not use any assistive technology. Figure 12 shows the bar chart generated from this answer. Some of those nine educators, who use assistive technology regularly as shown in the figure, use Kathamala [68]. Other educators use Gupshup,6
Bar chart – table 2 question 4 – Graphical implementation of the answers about the effectiveness of the curriculum design in teaching sentence reading.
Bar chart – table 2 question 5 – Graphical implementation of the ratings given to the selection of sentences.
About the design of the system we received mixed opinions, After analysing all the answers, it can be concluded that the design can only be assessed after conducting an experiment with children. However, one opinion was received nearly from everyone that, adding audio instruction at every level will increase the ease and will gather more engagement. On the other hand, from the perspective of efficacy, nearly every educator agreed that this system will be useful for children with dyslexia. All educators agreed that it will be helpful for children having borderline dyslexia. Twenty-three and seventeen educators out of twenty-eight respectively told that it will be useful for mild and moderate levels of dyslexia. After the evaluation of the system, it can be said that the proposed system will successfully give a generalised solution for dyslexia, From the comments we received, which are shown in the Fig. 27, it is clear that the system had received positive feedback and the educators are eager to use it in their classes. In the pre-test 9 educators responded that they always take technical support and 5 educators mentioned that they never take the help of technology. In the post-test only two educators denied using the software as they teach mathematics, other educators show interest in using it. In Fig. 26, the pie-chart representation of both pre-test response and post-test response is shown which will help in comparing the responses.
Bar chart – table 3 question 1 – Graphical implementation of the ratings given by educators to the design of interface.
Question 1: Figure 13 shows the bar chart generated from values obtained for this question. nine experts gave a six-point rating and three educators gave a seven-point rating. Eight expects gave and average rating and rest of the experts rate a five to effectiveness of the curriculum. The curriculum design indicates the selection of spelling rules, distribution of spelling rules among tasks, selection of sentences under each spelling rule and distribution of sentences among tasks based on difficulty levels. Based on the answers to this question, as no one gave below-average rating, i.e., less than four-point, it can be concluded that the designed curriculum will be effective in teaching sentence reading. Question 2: The answer to this question is represented in the form of a bar chart in Fig. 14 which shows that nine experts rate five or six and three and four experts rate the system four and seven respectively. Compared with the answer of the previous question, this question received a better rating. Hence it can be concluded that the content of the proposed system is more relevant for dyslexia than regular children. The reason was not addressed tough can be assumed that the system has repetation of spelling rules and sentences among tasks which can cause frustration in regular children as they do not need so much time in learning a skill. Question 3: The use of conjuncts was supported by some of the expert whereas some experts suggested to implement different modules and keep the conjuncts in higher level modules as a comment, shown in Fig. 27 suggests. The rating given by educators to the use of conjuncts is shown in Fig. 15. As it can be seen that it received both low and high ratings and maximum experts gave an average rating, hence this question needs to be addressed in a thorough experiment with the primary user group. Question 4: Figure 16 shows the chart generated from the answers obtained for this question. Based on those answers it can be concluded that the selection of sentences is appropriate as it received a five-point rating from five educators, a six-point rating from fourteen educators and a seven-point rating from two educators. Question 5: As there were conflicts of opinion about the use of conjuncts, about the distribution of spelling rule we received mixed opinions too. Figure 17 shows a graphical representation of the answers received for this question. In total, twelve experts rate the distribution of spelling rules with a five or six. Nine educators rate with a four-point and the rest of the educators rate with a two or three. Though the analysis of the answer to this question gives a better result than Question 3, the distribution of spelling rules still needs to be revised.
Bar chart – table 3 question 2 – A figure to represent the ratings given by educators to the difficulty level of instruction.
Bar chart – table 3 question 5 – Graphical implementation of the opinion on the system’s ability to increase student engagement.
Bar chart – table 4 question 1 – A figure showing the opinions about the effectiveness of the system in teaching sentence reading.
Bar chart – table 4 question 2 – Figure shows the number of educators rate the system with different numbers on Likert scale based on their liking.
Bar chart – table 4 question 3 – Figure shows the number of educators who are willing to used the system.
Bar chart – table 4 question 4 – Figure shows the educators’ opinion about the effectiveness of the system for different levels of dyslexia.
Bar chart – table 4 question 5 – Figure shows the educators’ opinion about the effectiveness of the system for different levels of dyslexia.
A comparative representation of how many educators decided to switch from traditional teaching method to the proposed system.
Some comments given by the educators in education.
Question 1: Figure 18 shows a graphical representation of the answers received for this question. This question received mostly five-point and six-point ratings, along with five sour-point ratings and four seven-point ratings. It can be concluded from its answer that the special educators liked the design of the interface. Question 2: This question used a seven-pointer Likert scale where one indicates very difficult and seven indicates very easy. The rating given by the experts are represented in Fig. 19 in a form of a chart. Experts rate the difficulty as average to very easy. Eight experts gave a five-point and six-point rating. Seven and three experts gave the ratings four and seven respectively. Question 3: This question was a short answer type question where the experts needed to write the answer in the form and writing more than one facility was allowed. It receives different types of answers, The most received suggestions are, addition of different pictorial cues for each of the sentences which will help to remember any word (12 suggestions), addition of audio instruction (9 suggestions) and animated pictures for hints (7 suggestions). Question 4: This question was a short answer type question too. Similar to the previous question, educators needed to write their opinions on this question and were free to give more than one suggestion. For those questions, the most received answer was that no feature is needed to be removed, another most received statement was that the use of word package can be removed as it includes the repetition of a sequence of activities. Question 5: Though this question received different types of ratings, most of the experts gave five or six-point ratings. Figure 20 shows the graph generated from the ratings received for this question. As the computer-based teaching process is more effective in dyslexia [69], this system could be expected to increase the engagement of the students, which is also indicated by the feedback from the educators.
Question 1: Figure 21, the bar chart generated with the answers to this question, shows a total of fifteen educators gave usability a six-point or seven-point rating. Though this question received two three-point and four-point rating, it is negligible compared to the positive rating, hence it can be concluded that according to the experts it will be effective in teaching sentence reading. Question 2: To clarify the answers to the previous question, this question was designed, according to the special educators and psychologists who participated in the design of the questionnaire, this question will justify if the experts find the system to be effective. Figure 22 shows the graphical representation of answers to this question. It received a similar type of rating from the experts, as an example, twelve and three experts gave six-point and seven-point ratings respectively to the previous question, in this question thirteen and four experts rated a six and a seven. Question 3: This question was a multiple choice question, the options for this question is shown in Table 4. Figure 23 shows the bar implemented by the answers to this question, fourteen educators said that they will use this system for some of their students as not all of their students read Bangla or Hindi as their first language. Two mathematics teachers and three vocational training teachers said that the proposed system will not be effective for their lessons. Two of the vocational training teachers who teach recitation, drama and three language teachers who teach in higher grades as discussed previously, marked that they need to use the system before commenting as some of their students might find this system interesting and with some degradation in the system, they can use it. Two educators who teach Bangla said they would like to use this system. Question 4: As a comment shown in Fig. 27, some of the experts commented that this system is mainly a generalised version, and some of their students having severe or profound levels of dyslexia need more assistance, hence for those children this system will not be as effective. Their answer to this question is in Fig. 24, all of the educators said that it will be useful for borderline dyslexia, twenty-three experts out of twenty-six found it useful for borderline and mild levels of dyslexia. And, seventeen experts found it useful for borderline, mild and moderate levels of dyslexia. Question 5: As the multisensory teaching approach is the most effective way to teach in dyslexia, it was tried to be implemented in the system, Fig. 25, the graphical representation of the answers to this question, shows the experts opinion on the efficacy of this proposed system in implementing multisensory teaching approach. Twelve and eight experts gave the highest rate, from which it can be concluded that the proposed system has implemented multisensory teaching methods properly, the educators verbally illustrated their feedback that mostly use the same activities in implementing a multisensory teaching approach.
Figure 26 shows two pie charts representing the number of educators who use any assistive technology in their daily lesson and the number of educators who decided to which to this system after the demonstration. Figure 27 shows some of the comments given by the experts, after analysing all the charts and comments, it can be concluded that the proposed system will be helpful in teaching sentence reading to children having borderline, mild or moderate level or dyslexia. This system will reduce the workload of educators, if an educator set the student specific curriculum in this system, it will carry out further processes and produce the performance report which the educator will use to assess the student’s progress.
In the survey, a conflict of opinion had been observed about some features. It can be clarified after conducting an experiment with children. An experiment will also validate the feedback of the educators. According to the evaluation of the system, the limitation of the work is that it can give a generalised solution for the challenges of dyslexia. In future, after conducting an experiment more specified modules of the system can be developed to target children who have mild to moderate or severe dyslexia. Furthermore, system modules for other Indian languages can also be developed in future.
Conclusion
The goal of the proposed work was to provide some support for children with dyslexia to learn sentence reading and practice in the absence of special educators and also help the educators in teaching sentence reading. Though the research and developmental works on dyslexia in English have been going on for a long period of time, in Hindi and Bangla we found a shortage of both research and developmental works. This proposed system can be used in schools or special education centres to conduct one-to-one sessions with a large number of students as it does not save any user data. The available systems are not dyslexia-focused and most of them do not use a multisensory teaching approach. As the solution, this system offers a teaching approach which includes three activities reading, listening and writing which trigger the working memory of the user. The content used in the system was created especially for developing sentence reading skills in Bangla and Hindi. Moreover, the available systems discussed in Section 2 were developed nearly three to five years back, and no recent work was found in this area. These systems mainly work for phonological awareness and word recognition, whereas the proposed system works for sentence reading. The available systems teach conjuncts in a way different from the traditional representation, these use an easier representation to learn, however, it will cause difficulty for students in higher grades when they will read from the regular books. As a solution, the proposed system teaches the conjuncts using their traditional representation along with a feature to show its separated form to decrease the difficulty for the students. Another advantage of the proposed system over the available works is that, it produces the performance report. As this feature was evaluated in the previous evaluation [60], no question on this feature is discussed in this paper. However, this feature increases its uniqueness compared to the available systems.
In the situation of COVID-19, all the institutes remain suspended for an indefinite time and therefore many children with disability failed to carry on their studies during that period. As a result of this, their education got sacrificed a lot as parents do not have the knowledge of how to teach them. In this scenario, the proposed work will help a lot to practise the previous lessons and learn new lessons in the absence of special educators. In the future, the modules can be extended to other languages as well.
Author contribution
CONCEPTION: Dipshikha Podder, Manjira Sinha, Anupam Basu.
PERFORMANCE OF WORK: Dipshikha Podder.
INTERPRETATION OR ANALYSIS OF DATA: Dipshikha Podder.
PREPARATION OF THE MANUSCRIPT: Dipshikha Podder.
REVISION FOR IMPORTANT INTELLECTUAL CONTENT: Manjira Sinha, Anupam Basu and Tirthan-kar Dashgupta.
SUPERVISION: Manjira Sinha and Anupam Basu.
Ethical consideration
This paper discusses a survey that was conducted with special educators. As no experiment was conducted, this paper is exempt from Institutional Review Board approval.
Footnotes
Acknowledgments
The authors are grateful to the institutes, Breaking Through Dyslexia(BTD), Behala Bodhayan and Indian Institute of Cerebral Palsy(IICP), for supporting the work and giving permission to talk and collect knowledge from special educators. We are thankful to the special educators for giving their valuable time and opinion about the work which will help to improve it in future. And lastly, the authors want to express their gratitude to the Ministry of Human Resouce Development(MHRD) for supporting and sponsoring the work.
Conflict of interest
The authors have no conflict of interest to report.