Improving oral language skills in children with developmental delays: The role of an oral language-oriented instructional program

The development of oral language skills is progressive. Research on vocabulary development examines the influence of environmental and personal factors, which are crucial for both theory building and practice. The language production of young children with special needs is influenced by age and exhibits specific patterns, such as a preference for nouns. More importantly, the vocabulary and grammar development of Chinese-speaking children with autism tends to be synchronized (Li et al., 2023). Children who have received language-related interventions exhibit significant improvement in their language skills, demonstrating that targeted oral language interventions effectively enhance the developmental trajectory of young children at risk of language disadvantage. As oral language skills are foundational to reading, writing, and academic knowledge, this study highlights the critical role of language development investment during the preschool phase in preventing potential academic challenges during subsequent phases (West et al., 2024). Presenting knowledge in the form of images is to strengthen the transmission of information and knowledge because it not only enhances memory through visual reminders but also helps achieve teaching goals (Chang and Chang, 2017). Language is the essential means of interpersonal communication. However, for children, oral expression is more than a demonstration of language ability; it is a means to improve thinking skills, and vocabulary can be regarded as the primary milestone in the language learning process.

For young children, a successful transitions experience from preschool to primary school is critical and has lasting effects on future learning (Sink et al., 2007). Children in preschool and primary school differ in their physical, cognitive, social, emotional, and language development (Liu and Chen, 2010). Effective transitions help children adapt to first grade, shaping learning and growth in the school environment (Liu and Chen, 2010; Sink et al., 2007). Oral language skills enable them to adapt to formal schooling by facilitating communication, comprehension, and participation in classroom activities. Therefore, bridging the gap between these two stages of education is critical (Wong, 2017).

The benefits of integrating technology into education include enabling personalized learning (Chen and Chen, 2017). For example, mobile-assisted systems with audiovisual feedback increase vocabulary retention and pragmatic skills in children with language delays (Luo et al., 2020). Additionally, embedded feedback features, such as visual rewards or auditory cues, in tablet-based interventions promote self-monitoring and reduce dependency on instructor guidance (Xin et al., 2023). Empirical evidence shows that these systems bridge the gap between therapeutic settings and naturalistic environments, allowing for repeated practice and immediate correction, which are key factors for skill generalization (Boyle et al., 2020). However, gaps remain in applying such technologies within inclusive preschool classrooms, particularly for oral language development (Meadan et al., 2020). Therefore, developing corresponding methods has become a focus of attention. In this study, the AT application refers to the teacher using an instant feedback system to monitor students’ learning outcomes rather than the student using it to support their oral communication. The interventions involved integrating an instant feedback system—an assistive technological tool—in teaching.

This study was guided by the following research questions:

RQ1: What are the immediate effects of integrating AT devices (Smartclass + talk) into oral language teaching on improving the oral language skills of preschool children with developmental delays?

RQ2: What are the maintenance effects of integrating AT devices (Smartclass + talk) into oral language teaching on improving the oral language skills of preschool children with developmental delays?

RQ3: To what extent are such interventions socially valid from the perspective of the children's parents?

Four children identified as having developmental delays participated in this study. The screening criteria used to recruit participants included the following. First, children were aged two to four years. Second, they had received or were receiving early intervention (not oral language intervention programs, which are interventions related to oral language comprehension and communication instructions provided by teachers or speech-language pathologists) under an Individualized Family Service Plan (IFSP) after being identified as having an oral language delay. Third, they spoke Mandarin Chinese at home. Fourth, tutors and preschool teachers observed that they required improvement in oral language skills. Fifth, parents provided consent for participation in the study. Sixth, they were all of East Asian (ethnic Chinese) background.

The recruitment flyers were distributed to nonprofit preschools in the northern region of Taiwan. The children had not participated in oral language intervention programs before the study. The researcher holds a doctorate in special education and is a special education professional with approximately 12 years of experience working with children, conducting intervention research, and applying behavioral analysis among children with or at risk for developmental disabilities in school settings. Before the study began, the researcher conducted on-site observations, collected information, and held informal interviews with the parents to gain a better understanding of the participants’ situations. The researcher then used the Revised Preschool Language Impairment Scale (Lin et al., 2009) to measure the percentile ranking of the children's vocabulary development in order to determine the number of words to be included in the research. All names are pseudonyms.

Jenny, female, is a three-year-old girl diagnosed with a developmental delay by a medical institution. Currently, she is not receiving services from any special education professional team. She obtained a percentile of 36 for the Revised Preschool Language Impairment Scale and a T-score of 47, indicating that her language comprehension ability is consistent with age-appropriate developmental expectations and falls within the average range. She has a limited vocabulary and tends to use single words (e.g. “eat”) rather than complete sentences (e.g. “I want to eat”) during communication.

Bryant, male, is a three-year-old boy diagnosed with a developmental delay by a medical institution. He is not receiving treatment from any professional team. He obtained a percentile of 26 for the Revised Preschool Language Impairment Scale and a T-score of 43, suggesting that his language comprehension ability falls within a normal and average range. The IFSP states that he can use simple language and grammar to communicate about daily necessities and actions. According to the teacher's observation, he expresses needs related to eating and toileting most clearly, using single words rather than complete sentences. He can identify objects at an acceptable level and has begun to use sounds to indicate objects and colors in his environment. In terms of learning, he primarily imitates the actions of others and can understand the meaning of numbers and some emotional words, such as “angry,” “sad,” and “happy.”

David, male, is a four-year-old boy diagnosed with a developmental delay by a medical institution. He receives 2 h of physical therapy and 50 min of oral language therapy every week, during which he receives interventions conducted by speech-language pathologists that target physiological aspects, such as oral massage and articulation correction. He obtained a percentile of 29 for the Revised Preschool Language Impairment Scale and a T-score of 45, indicating that his language comprehension ability falls within a normal and average range. According to his observations, he lacks opportunities and motivation for oral expression because he rarely communicates with peers and family members. He cannot fully use correct vocabulary to express his needs and mostly expresses himself with gestures and sounds. However, he possesses a limited vocabulary for interpersonal interaction and general use, such as describing everyday life and common objects from his IFSP.

Jessica, female, is a three-year-old girl diagnosed with a developmental delay by a medical institution. She is not receiving treatment from any professional team. She obtained a percentile of 22 for the Revised Preschool Language Impairment Scale and a T-score of 42, suggesting that her language comprehension ability is consistent with age-appropriate developmental expectations and falls within the average range. She demonstrates appropriate communication intentions; however, she primarily uses actions and crying to express and communicate her needs and goals. When identifying objects in daily life, she mainly imitates others and does not understand what the objects are. She rarely expresses her emotions and exhibits limited reactions to emotions.

This study adopted the topics, materials, and vocabulary from the textbook “100 Lessons of Direct Instruction of Oral Language to Preschool Children” developed by Tzeng (2018) and used the direct instruction method for the classes. The classes were divided into nine units. Specifically, vocabulary related to actions, emotions, numbers, and connectives was taught in two in-class sessions each because these are abstract and complex concepts that students rarely experience during their development. Vocabulary related to common terms, information, everyday objects, parts and whole, acronyms, and position and orientation in space was taught in one in-class session each. The four participants were evaluated 21 times during the research period using the “Oral Language Skill Evaluation Scale” developed by the present research team. The evaluation began with the first participant, and the researcher performed subsequent evaluations in accordance with the general principles of research design.

We designed the teaching process and content in line with the developmental level of the participants’ age group (three to four years old). Specifically, we removed Chinese characters and presented all content using images. As mentioned above, we adopted direct instruction as the teaching method, in which the teacher used direct questions and expected the children to respond orally. The teacher also employed multimedia technology in the classroom, using Microsoft PowerPoint to present photos and animated images corresponding to the topic and vocabulary list.

To examine the effectiveness of the AT teaching method for oral language skills among preschool students with developmental delays, this study used a single-case research method with a multiple-probe design replicated across subjects because the effect was not reversible after affecting the target behavior (Ledford and Gast, 2024). The researchers plotted data points for each teaching intervention and participant in a staggered manner and analyzed the relationship between the intervention and learning outcomes. A relationship was present if (a) the target behavior was at a stable level before the intervention was implemented; (b) the target behavior should increase after the intervention was introduced; and (c) the improved target behavior should remain stable for a set period (two consecutive sessions) for all children.

This study included three experimental phases: baseline (prior to the intervention), intervention (during the implementation of the intervention), and maintenance (after the intervention). Each intervention session lasted 25–30 min, with five sessions per week for approximately three weeks. Given the urgency of oral language development expressed by participants and their parents, this intervention was implemented at a substantially higher intensity than those reported in the reference studies. Because the objective was to improve children's oral language skills through direct instruction and the comprehensive effect of the textbook, the teacher delivered a 10-min PowerPoint presentation covering the vocabulary to be taught in the session and the corresponding scenarios to stimulate learning motivation. Next, the teacher (a) introduced the vocabulary in the unit (approximately 10 min) and (b) used the materials to engage the children in question-and-answer games (5 min) to evaluate how the knowledge was received. The sessions typically lasted for 25 min.

During initial probing, the researcher conducted at least five to seven evaluations with each participant before the intervention to ensure that their oral language skills were both low and stable for at least three consecutive periods. During the intervention phase, the teacher conducted nine evaluations until the children met the mastery criteria. Evaluations during the maintenance phase ensured that participants’ improved performance remained stable over time without intervention. This phase included (a) one week following the last intervention, and (b) one week of maintenance observation when other participants were assessed. Therefore, data for the maintenance phase were collected two weeks after postintervention was assessed.

The study consisted of three stages: baseline, intervention, and maintenance. (1) Baseline: the criterion for moving into intervention was variability of no less than 20%. During the baseline phase, oral language skills were evaluated using the “Oral Language Skill Evaluation Scale” repeatedly until the results remained stable for three consecutive assessments for all participants. (2) Intervention: the acquisition criterion was a score of no less than 70 on the scale. In the intervention phase, the teacher used a PowerPoint presentation to introduce the teaching content from the textbook, along with instructions, gestures, corresponding text, and images to convey the knowledge. Afterward, the children participated in a question-and-answer game. The teacher provided instant feedback to the students through the “Smartclass + talk” software. Specifically, after the instruction of each section, the teacher used the platform to evaluate the children's mastery of the words. After completing the entirety of the oral language teaching course, participants moved into the maintenance phase. (3) Maintenance: assessments were conducted two weeks postintervention. In the maintenance phase, the teacher evaluated their oral language skills using the “Oral Language Skill Evaluation Scale.” The evaluation of oral language skills was conducted in a one-on-one format between the researcher and the participant. Given that the participants lacked the necessary literacy skills for a written test, a question-and-answer format was used. All items were presented to the participants in the form of pictures to provide visual prompts. For example, in an item with three pictures, each picture contained two objects: a balloon and a desk. In the three pictures, the balloon was placed in different locations relative to the desk (on the desk, in the drawer, and under the desk). The researcher provided the instruction: “Please point to the picture in which the balloon is on the desk,” and scored the participant's answer accordingly.

The independent variable in this study was the oral language-oriented instructional program (intervention), which included the instant feedback system. This method was tailored to meet the needs of the participants, with teachers using the instant feedback system to monitor learning outcomes and provide practice and feedback. The teaching material used in this study was “100 Lessons of Direct Instruction of Oral Language to Preschool Children,” developed by Tzeng (2018) under the Ministry of Science and Technology research project.

The instant feedback system “Smartclass + talk” is a campus-oriented digital content and teaching software approved by the Ministry of Education of Taiwan (2024). It was developed by a Canadian company, Robotel, specifically for language learning. In this study, it was mainly applied to instantly evaluate learning performance at each stage. If the accuracy rate was lower than 90%, the teacher re-taught the relevant material until accuracy exceeded 90%, and then proceeded to the next topic. During classroom activities, the software was accessed via mobile phone interfaces, while postclass assessments were conducted using a laptop screen. Each assessment page displayed four images, one of which showed the correct answer (e.g. a police officer). The examiner (a teacher assigned to provide in-class support) posed questions such as, “Please point to the picture of the police officer.” The student responded orally, and accuracy was determined based on the number of correct verbal responses. These were comprehension questions. By recording student performance using audio and video functions, this approach provided a more precise measurement of vocabulary compared to previous methods that relied on interrater reliability.

The dependent variable in this study was children's oral language skills. We used the “Oral Language Skill Evaluation Scale” developed by our research team to measure the dependent variable. The scale was based on Tzeng's (2018) teaching materials for children's oral language skills, which is a curriculum-centered measure based on the intervention program implemented in this study. The teacher asked the children to answer questions according to the images presented to them, and the collected data were used to analyze immediate and sustained results. The scale included 25 questions covering nine units. Each complete answer received four points, each partial answer received two points, and each incorrect answer received zero points. The maximum score was 100.

To mitigate the negative effect of repeated testing in a single case study, it is generally recommended to design alternate versions of achievement assessment tests to reduce participants’ sensitivity to the test and minimize the practice effect (Fraenkel and Wallen, 2006). We used a two-way specification to evaluate the content validity of the evaluation scale. In line with the research purpose and design, we used the scale multiple times during the study period. To minimize the practice effect arising from deliberate vocabulary practice for correct answers under repeated testing conditions, we also designed alternative versions. Each version maintained consistent types of items (all in question-and-answer format), evaluation objectives, difficulty levels, and total number of items. Additionally, the direction, scope, and distribution were identical in all versions. We followed the principles of replica test design, which refers to multiple measurements and assessments of the participant's target behavior under the same conditions. Five versions of the scale were developed (replicas 1 to 5), which were then presented sequentially, from 1 to 5, in the designated evaluation order.

We assessed the participants’ oral language skills and reviewed the data collected from all sessions using the recorded videos. Interobserver agreement (IOA), which covered the baseline, intervention, and maintenance phases, was calculated by dividing the number of agreements between the two evaluators by the total number of agreements and disagreements. The average IOA was 92.5% for Jenny (range = 77–100%), 94.5% for Bryant (range = 85–100%), 96.2% for David (range = 88–100%), and 93.7% for Jessica (range = 90–100%). Before data collection, both researchers completed reliability training by independently reviewing the same video and rating the children's performance. This training video was not included in the study. Additionally, they independently coded the evaluation criteria and the frequency of spontaneous and prompted responses. The researchers then met to compare their coding and resolve any discrepancies by discussing operational definitions and providing examples of dependent variables.

Visual analysis was conducted for all intervention, baseline, and maintenance probe data independently by the primary researcher. Six features guided the visual analysis of graph data: trend, variability, immediacy of effect, nonoverlap, consistency of data patterns, and level of performance. It is essential to clearly describe the data analysis plan, including the protocol used, any noteworthy formative decisions (such as modifications and phase changes), summative analysis, and conclusions regarding the presence or absence of a functional relation (Ledford and Gast, 2024).

We computed Tau-U by comparing the raw data points of the baseline and intervention phases and the intervention and maintenance phases (Parker et al., 2011). Tau-U is a valuable metric for assessing effect sizes in single-case design studies because it considers the overlap between baseline and intervention phases, as well as the trends and levels within each phase. Furthermore, it offers greater statistical power compared to other nonoverlap indices (Parker et al., 2011). A Tau-U result between 0.93 and 1.00 indicates a large effect size, between 0.66 and 0.92 indicates a moderate effect size, and below 0.65 indicates a small effect size (Ok et al., 2021). Tau-U was calculated using an online calculator (www.singlecaseresearch.org).

This study used one procedural fidelity checklist (Kim et al., 2020). The fidelity checks were conducted by the primary researcher. The checklist, applied during the baseline and maintenance phases, included four items: (a) the child and the researcher completed questions on the evaluation scale, with no environmental arrangement; (b) the researcher refrained from following the child's lead or narrating or imitating the child's play, indicating no responsive interaction; (c) when the child showed interest in an object representing the target word but did not say it, the researcher did not provide the word, indicating no language modeling; and (d) when the child showed interest in the object but did not say the target word, the researcher did not use prompts to introduce it, indicating no milieu teaching. This procedural fidelity was periodically examined by the primary researcher over the course of the study, and all checks met the required standards.

The author asked one parent of each participant (two mothers and two fathers) to complete the social validity questionnaire after the last maintenance session to rate the session. The questionnaire included seven items. Six items were rated using a five-point Likert scale (1 = low social validity, 5 = high social validity), whereas the last item was open-ended: “What other comments or feedback would you like to share about the intervention and the research study?” The six items were designed to examine parents’ perceptions of (a) children's improvements in the targeted skills, (b) long-term positive effect, (c) negative effect, (d) expectations about learning and intervention plans in the future, (e) satisfaction with the interaction between the researcher and their child, and (f) satisfaction with the intervention procedures. The researcher calculated the means and ranges of the results.

Overall, during the baseline phase, the children's oral language skills remained consistently low, exhibiting a trend of zero-celeration. In addition, all participants’ oral language skills improved between the third and fifth intervention sessions, indicating a functional relation between the intervention and oral language skills. Probe data from the maintenance phase showed the general effects of the intervention on oral language skills for all participants. Figure 1 shows the results for each participant.

Jenny: During the baseline phase, Jenny's oral language skills were at a low level compared to the intervention and maintenance phases and showed a stable, with zero-celeration trend. After the intervention, her oral language skills immediately increased to 44, and this trend gradually continued. During the intervention phase, her oral language skills showed an accelerating trend with considerable variability. The level of answer accuracy rate produced was 42.8 (range = 40–46) at baseline and 63.9 (range = 44–80) during intervention probes. During the two-week maintenance probes, her oral language skills remained at a high level and showed an accelerated trend (M = 62, range = 58–67). From the baseline phase to the intervention phase, the average performance level increased from 42.8 to 63.9, representing a gain of 21.1. The percentage of nonoverlapping data was 89%, suggesting immediate effects of the interventions. From the intervention phase to the maintenance phase, the average performance level decreased from 63.9 to 62, a reduction of 1.9. The percentage of nonoverlapping data was 100%, indicating that the skill level remained good even after the completion of the intervention, which demonstrates strong maintenance effects.

The Tau-U for the baseline and intervention phases was 0.933 (p = .005; 95% confidence interval (CI) [0.385, 1]). The differences between the baseline phase and the intervention phase demonstrated large effects, suggesting significant immediate effects of the interventions.

Bryant: During the baseline phase, Bryant's oral language skills were at a low level compared to the intervention and maintenance phases, showing a stable trend with a slight increase. After the seventh intervention, his oral language skills increased noticeably to 77. During the intervention phase, his oral language skills showed an accelerating trend with small variability. The produced answer accuracy rate level was 46.5 (range = 37–53) at baseline and 60.9 (range = 49–77) in intervention probes. During the two-week maintenance probes, his oral language skills remained at a high level and showed an accelerating trend (M = 56.8, range = 51–61). From the baseline phase to the intervention phase, the average performance level increased from 46.5 to 60.9, representing a gain of 14.4. The percentage of nonoverlapping data was 67%, suggesting immediate effects of the interventions. From the intervention phase to the maintenance phase, the average performance level decreased from 60.9 to 56.8, a reduction of 4.1. The percentage of nonoverlapping data was 100%, indicating that the skill level remained good even after the completion of the intervention, which demonstrates strong maintenance effects.

The Tau-U for the baseline and intervention phases was 0.815 (p = .009; 95% CI [0.298, 1]). The differences between the baseline phase and the intervention phase showed high effects, suggesting significant immediate effects of the interventions.

David: During the baseline phase, David's oral language skills were at a low level compared to the intervention and maintenance phases and showed a stable trend with zero-celeration. Following the intervention, his oral language skills began to gradually increase. During the intervention phase, his oral language skills showed an accelerating trend with considerable variability. David demonstrated high learning motivation during the intervention phases and was satisfied with the feedback provided by the researcher. The produced answer accuracy rate level was 41.4 (range = 36–45) at baseline and 65.8 (range = 43–81) during intervention probes. During the two-week maintenance probes, his oral language skills remained at a high level and showed an accelerating trend (M = 61.2, range = 58–65). From the baseline phase to the intervention phase, the average performance level increased from 41.3 to 65.8, representing a gain of 24.5. The percentage of non-overlapping data was 89%, suggesting immediate effects of the interventions. From the intervention phase to the maintenance phase, the average performance level decreased from 65.8 to 61.2, a reduction of 4.6. The percentage of nonoverlapping data was 100%, indicating that the skill level remained good even after the completion of the intervention, which demonstrates strong maintenance effects.

The Tau-U for the baseline and intervention periods was 0.905 (p = .0026; 95% CI [0.411, 1]). The differences between the baseline phase and the intervention phase showed large effects, indicating significant immediate effects of the interventions.

Jessica: During the baseline phase, Jessica's oral language skills were at a low level compared to the intervention and maintenance phases and showed a stable trend with zero-celeration. Following the intervention, her oral language skills immediately increased to 58, and the trend remained stable. During the intervention phase, her oral language skills showed an accelerating trend with considerable variability. Jessica received frequent feedback and had positive interactions with intervention. The produced answer accuracy rate level was 43 (range = 39–46) at baseline and 68.8 (range = 56–82) during intervention probes. During the two-week maintenance probes, her oral language skills remained at a high level and showed an accelerating trend (M = 67.6, range = 64–70). From the baseline phase to the intervention phase, the average performance level increased from 43 to 68.8, representing a gain of 25.8. The percentage of nonoverlapping data was 100%, suggesting immediate effects of the interventions. From the intervention phase to the maintenance phase, the average performance level changed from 68.8 to 67.6, an increase of 1.2. The percentage of nonoverlapping data was 100%, indicating that the skill level remained high even after the completion of the intervention, which demonstrates strong maintenance effects.

The Tau-U for the baseline and intervention periods was 1 (p = .001; 90% CI [0.507, 1]). The differences between the baseline phase and the intervention phase demonstrated large effects, suggesting significant immediate effects of the interventions.

The mean rating on the social validity questionnaire was 4.85 (range = 4–5). The ratings provided by the four parents ranged from 1 to 5, indicating a high level of acceptability, generalizability, satisfaction, preference, and effectiveness for this intervention. No negative effects of the intervention were observed. The open-ended question aimed to collect additional feedback; the parents’ responses were positive and general. Examples of feedback include “This program is also a great learning opportunity for me” (the caregiver) and “I really enjoyed how the interventionist developed a great relationship with my child, and my child really enjoyed the play sessions with the interventionist.” One parent rated all items 5, except one. In addition, two parents reported that the intervention procedure was humane (4) rather than very humane (5). Two parents completed the questionnaire after being informed about goal setting, intervention, and goal setting procedures of the study by the other parent in their household. The mean rating of the four parents who observed the sessions was 4.85 (range = 4–5), and the mean rating of the two parents who were informed of the procedures of the sessions was 4.91 (range = 4–5).

This study has several limitations. First, it was conducted using a single-case research design. Although four children participated, the coverage of the study was limited. Future studies should adopt a quasi-experimental design to include larger samples. Second, this study focused on children's oral Chinese skills. Because of differences between disciplines, the findings may not be generalizable to other subjects. Moreover, the vocabulary was limited to nine categories (action concepts, emotion words, counting, conjunctions, common vocabulary and information, familiar objects, part—whole relationships, antonyms, and spatial/positional terms). Therefore, the inferences and reference value of the findings are restricted to this scope. Third, as the intervention was conducted during the preschool phase, adjustments are needed to transfer the intervention mode and evaluation methods to other educational phases. Fourth, the intensity of the intervention in this study was significantly higher than those in previous studies. This should be considered when interpreting the learning outcomes. Future research should investigate the effect of the intervention style on improving skills related to other disciplines.

The findings of this study have several theoretical and practical implications. Drawing on these results, we propose the following recommendations. First, in teaching, it is necessary to provide children with additional practice opportunities in units that cover abstract concepts, as well as more real-life scenarios, to further enhance both immediate and sustained improvement. An instant feedback system should also be included to monitor learning effects. Second, in assessment, image-based question-and-answer evaluation methods and technology-driven assessments should be considered for preschool students to capture learning performance accurately.

Future research could replicate this study with modified independent or dependent variables, incorporate other communication aids or sound-generating devices, and use evaluation methods and alternative treatments tailored to the abilities of each child. For example, researchers could compare the outcomes of using oral instruction intervention alone versus combining these with AAC (Chang et al., 2018) to assess children's learning performance on a set of target vocabulary.