INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Primary ESL Teachers’ Perceptions of AI Tools for Pupils’ Speaking  
Practice: Opportunities and Limitations  
*Kong Ming Yien., Melor Md Yunus, and Hanita Hanim Ismail  
Faculty of Education, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia  
*Corresponding Author  
DOI: https://dx.doi.org/10.47772/IJRISS.2025.91100171  
Received: 18 November 2025; Accepted: 27 November 2025; Published: 03 December 2025  
ABSTRACT  
This study explores Malaysian primary school English teachers’ perceptions of using Artificial Intelligence (AI)  
tools to support pupils’ speaking practice. Although AI integration is accelerating globally, research focusing  
specifically on AI-supported speaking instruction in primary ESL classrooms remains limited. Grounded in the  
Technology Acceptance Model (TAM), this study examines both the perceived opportunities and challenges of  
AI integration with emphasis on how teachers evaluate the usefulness, ease of use and overall applicability of  
AI tools such as ChatGPT in developing young learners’ speaking skills. A mixed-methods survey design was  
employed involving 60 English teachers from public primary schools in Selangor. Quantitative data were  
analysed using descriptive statistics to identify trends in teachers’ perceptions of AI’s pedagogical benefits and  
limitations while qualitative data from open-ended responses were analysed thematically to provide deeper  
insights into teachers’ experiences. Findings indicate that teachers view AI as a highly valuable resource for  
enhancing oral proficiency, offering instant and personalised feedback, reducing pupils’ speaking anxiety and  
creating engaging, interactive learning environments. Teachers also highlighted that AI expands opportunities  
for self-paced practice beyond the classroom, especially for shy or low-confidence learners. Despite these  
benefits, significant challenges emerged. Teachers expressed concerns about unreliable internet connectivity,  
limited device availability, inaccurate speech recognition especially for young children’s voices, insufficient  
training, data privacy issues and pupils’ potential overreliance on AI-generated responses. These challenges  
highlight the need for stronger infrastructure support, targeted professional development, curriculum-aligned AI  
content and clear ethical guidelines. Overall, the study underscores the promising role of AI as a complementary  
tool for supporting speaking development in Malaysian primary ESL settings while emphasising the systemic  
and pedagogical considerations required for sustainable implementation.  
Keywords: Artificial Intelligence, ESL speaking practice, primary education, teacher perceptions, Technology  
Acceptance Model  
INTRODUCTION  
Speaking is one of the most important yet challenging skills for young ESL learners to master. It requires learners  
to think and respond quickly while maintaining fluency, pronunciation accuracy and communicative competence  
(Tatchakrit, 2024). Although the CEFR-aligned English Language Curriculum aims to strengthen  
communicative abilities in Malaysia, many primary pupils continue to struggle with oral communication,  
especially in developing confidence and fluency (Aziz & Kashinathan, 2021). These difficulties are often linked  
to limited English exposure outside the classroom, speaking anxiety and a lack of meaningful interaction  
opportunities (Kafabih, 2025). The predominance of teacher-centred practices focusing on reading and writing  
further restricts pupils’ speaking practice.  
With rapid technological advancement, Artificial Intelligence (AI) has emerged as a promising tool to enhance  
language learning. AI-powered platforms such as ChatGPT, ELSA Speak and SmallTalk2Me can simulate real  
conversations, provide instant pronunciation feedback and offer adaptive speaking tasks suited to learners’  
proficiency levels (Sangeetha & Jamaludin, 2025). These tools allow pupils to practise at their own pace in low-  
Page 2148  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
pressure environments, reducing fear of mistakes and increasing opportunities for oral communication (Nhan,  
2025). AI can also reduce teachers’ workload by assisting with routine feedback, automated scoring and  
monitoring, enabling teachers to focus more on personalised instruction (Mahmoud, 2024).  
In Malaysian primary ESL classrooms, AI-based speaking tools align well with young learners’ developmental  
needs. Children in the concrete operational stage learn effectively through interaction, exploration and active  
engagement (Pakpaha & Saragih, 2022). Gamified AI features and interactive tasks support these learning  
preferences by offering motivating and age-appropriate experiences. Nonetheless, successful AI integration  
depends heavily on teachers’ acceptance, readiness and perceptions (Zulkarnain & Yunus, 2023), which  
ultimately determine whether AI is used meaningfully or remains underutilised.  
Although interest in AI for language learning is growing worldwide, its adoption in Malaysian primary schools  
remains limited. Barriers such as low digital literacy, limited training in AI-assisted pedagogy, unstable internet  
connectivity and concerns over accuracy, privacy and ethical use of AI-generated content continue to hinder  
implementation (Vincent, 2025). As a result, the potential of AI to enhance speaking proficiency is still  
underexplored in Malaysian primary ESL contexts. Understanding teachers’ perceptions is therefore essential to  
identify both the opportunities such as increased motivation, personalised feedback and expanded speaking  
exposure and the challenges, including technological constraints, accent misrecognition and fears of overreliance  
on AI (Sivanganam, 2025). Without these insights, classroom AI integration may fail to achieve its intended  
outcomes.  
Therefore, this study investigates Malaysian primary ESL teachers’ perceptions of using AI as a tool for speaking  
practice. It examines the opportunities AI offers for developing pupils’ speaking skills and the limitations or  
challenges teachers face when integrating AI-based speaking tools. The study addresses the following research  
questions:  
1. What opportunities do teachers perceive in using AI tools to support pupils’ speaking practice?  
2. What are the perceived limitations or challenges faced by teachers in implementing AI tools for speaking  
instruction?  
LITERATURE REVIEW  
Teachers’ Perceptions of Using AI Tools in Speaking Instruction  
Teachers’ perceptions play a critical role in determining how effectively AI tools are used in speaking  
instruction. Antonietti (2022) emphasises that teachers are more likely to adopt digital innovations when they  
believe the tools are useful, easy to operate and supported by sufficient training and infrastructure. In language  
education, Sangeetha (2025) found that many teachers acknowledge the potential of AI to support speaking  
development by offering personalised feedback and additional practice opportunities. However, differences in  
digital competence influence these perceptions. An (2022) reported that technologically confident teachers tend  
to view AI positively, while those with weaker digital literacy may feel anxious or uncertain. Firdaus and Nawas  
(2024) similarly observed that some teachers perceive AI as distracting or challenging to integrate effectively.  
In Malaysia, Shakri (2025) highlighted that although teachers recognise AI’s value, many are unsure how to  
implement it purposefully due to limited training, unclear curriculum guidance and uneven access to digital  
resources.  
Perceived Opportunities of AI in Enhancing ESL Speaking Skills  
Research has also highlighted several opportunities associated with using AI to enhance pupils’ speaking skills.  
AI-powered tools provide instant and personalised feedback on pronunciation, fluency and intonation, enabling  
learners to monitor their performance and practise repeatedly (Hidayah, 2025). This benefit is particularly  
relevant in large primary classrooms where teachers may not have time to provide individual oral feedback. AI  
also creates a low-anxiety environment in which pupils can practise speaking privately without fear of  
embarrassment, helping to increase their confidence (Zhang, 2024). Zou (2024) further emphasised that AI  
Page 2149  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
extends speaking exposure beyond the classroom by allowing pupils to practise “anytime and anywhere,”  
supporting learners who may lack English interaction at home. Motivational elements such as games, avatars  
and reward systems add to the appeal of AI, making speaking tasks more engaging and enjoyable (Alexis &  
Becky, 2024). These interactive features align with the needs of young learners and can support cognitive  
development through exploration and creativity. In addition, AI tools are able to adjust speaking tasks to suit  
learners’ proficiency levels, offering meaningful differentiation for both high- and low-achieving pupils  
(Hamayun, 2025).  
Perceived Limitations and Challenges in Using AI For Speaking Practice  
Despite these opportunities, several limitations and challenges have been documented in previous studies. One  
of the most persistent issues concerns the accuracy of AI-generated feedback. Automated speech recognition  
systems often misinterpret children’s voices, developing articulation patterns or diverse accents, which may lead  
to incorrect or unhelpful assessments (Zou, 2024). Mudway (2025) noted that such inaccuracies can undermine  
learners’ confidence or result in misleading corrective input. Technical and infrastructural challenges also remain  
widespread. Junaidi (2024) reported that unstable internet, limited digital devices and insufficient technical  
support frequently disrupt lessons and discourage teachers from using AI. Additionally, many teachers feel  
unprepared to integrate AI effectively because of insufficient training or unfamiliarity with AI-supported  
pedagogy, as highlighted by Sulaiman (2024). Ethical concerns further complicate AI adoption. Kotsis (2025)  
pointed out risks related to data privacy, safety and the potential for pupils to rely excessively on AI-generated  
responses, which may limit creativity and reduce real human interaction.  
Technology Acceptance Model (TAM)  
The Technology Acceptance Model (TAM) developed by Davis (1989) explains that people decide to use a new  
technology based on two main beliefs: Perceived Usefulness (PU) and Perceived Ease of Use (PEOU). PU refers  
to how much teachers believe AI tools such as ChatGPT, ELSA Speak, and SmallTalk2Me can improve pupils’  
speaking skills by giving instant feedback, increasing confidence and supporting personalised learning. PEOU  
refers to how easy teachers feel the tools are to use. When AI is simple to operate, requires little technical skill,  
and fits smoothly into lessons, teachers are more likely to use it. However, difficulties such as unstable internet,  
complex interfaces or low digital literacy can reduce acceptance (Venkatesh & Davis, 2000). These two beliefs  
shape teachers’ attitudes and eventually determine their Behavioural Intention (BI) to use the tools. In this study,  
BI reflects teachers’ willingness to integrate AI into their speaking lessons; positive beliefs increase adoption,  
while concerns about accuracy, privacy or limited devices may reduce it. TAM fits well with this study because  
it closely matches the teachers’ perceptions gathered: the benefits they see relate to PU, the challenges they face  
link to PEOU and their overall willingness to adopt AI corresponds to BI as supported by past educational  
research (Teo, 2011). Since teachers decide how technology is used in the classroom, their acceptance directly  
affects how effectively AI can support pupils’ speaking development in CEFR-aligned primary ESL settings.  
Overall, TAM provides a clear and useful framework for understanding how teachers judge the usefulness and  
ease of AI tools and what influences their intention to use them in English speaking lessons.  
METHODOLOGY  
Research Design  
This study used a mixed-methods survey design to better understand teachers’ perceptions of using Artificial  
Intelligence (AI) tools like ChatGPT to support speaking practice among primary school pupils. A mixed-  
methods approach was chosen because it allowed for both numerical data and detailed explanations within the  
same study. The quantitative part involved a structured questionnaire with Likert-scale items. This revealed  
overall trends in teachers’ perceptions of the opportunities AI offers, the challenges they face and their  
willingness to include AI in speaking lessons. To enhance these findings, the questionnaire included four open-  
ended questions. These invited teachers to share their personal experiences, challenges and suggestions related  
to AI-supported speaking practice. The qualitative insights helped clarify the reasons behind the numerical  
results. Using both types of data improved the study by providing a more complete and validated understanding  
Page 2150  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
of the research issue. This aligns with Creswell’s view, as noted in Chali (2022) that mixed-methods research  
offers complementary evidence that boosts the depth and validity of findings.  
Participants  
This study included sixty English language teachers from government primary schools in Selangor, Malaysia.  
Purposive sampling is used to ensure that all participants were currently teaching English at the primary level  
and were familiar with AI tools like ChatGPT, ELSA Speak or other applications. Selangor had been chosen as  
the research location because it has a diverse mix of urban and rural schools and is known for its strong  
technological infrastructure. This made it a good place to examine the use of AI in ESL speaking instruction.  
Participation in the study was voluntary and all teachers provided informed consent before filling out the  
questionnaire.  
Research Instrument  
The study used a structured online questionnaire to collect both quantitative and qualitative data. The instrument  
consisted of four sections: Section A gathered demographic information such as age, teaching experience, school  
location and prior use of AI tools; Section B examined teachers’ perceptions of the opportunities AI offers for  
improving pupils’ speaking fluency, confidence and personalised feedback; Section C explored the limitations  
and challenges of using AI including technological issues, accuracy concerns, privacy risks and potential  
overreliance and Section D assessed teachers’ readiness to adopt AI, focusing on confidence, willingness and  
training needs. Four open-ended questions were also included to obtain deeper insights into teachers’ experiences  
and suggestions. The questionnaire was pilot-tested with ten primary school English teachers to refine unclear  
items. Reliability analysis showed strong internal consistency with Cronbach’s Alpha values of 0.94 (Section  
B), 0.827 (Section C) and 0.977 (Section D). An experienced ESL teacher further reviewed the instrument to  
ensure content validity and alignment with CEFR-speaking expectations.  
Data Collection  
Data collection was carried out through an online questionnaire distributed to participants via Google Forms.  
This method was chosen for its accessibility and convenience which allowed the teachers from various districts  
in Selangor to participate without disruption to their teaching commitments. Before accessing the questionnaire,  
participants were provided with an information sheet explaining the purpose of the study, their rights as  
participants and assurances of confidentiality. The data collection period lasted for approximately two weeks  
during which reminder messages were sent to encourage participation and ensure an adequate response rate.  
Upon submission, responses were securely stored and prepared for quantitative and qualitative analysis.  
Data Analysis  
The data were analysed using both quantitative and qualitative approaches in line with the study’s mixed-  
methods design. Quantitative data from the Likert-scale items were analysed using the Statistical Package for  
the Social Sciences (SPSS). Descriptive statistics including frequencies, percentages, means and standard  
deviations were generated to summarise teachers’ perceptions of AI’s opportunities, limitations and their  
readiness to adopt AI in speaking instruction. Reliability was assessed using Cronbach’s Alpha to confirm the  
internal consistency of the instrument. Qualitative data from the open-ended responses were analysed  
thematically. The analysis involved reading and re-reading the responses to become familiar with the data,  
generating codes and identifying recurring themes that captured teachers’ experiences, concerns and suggestions.  
These qualitative findings were then used to complement and deepen the interpretation of the quantitative results,  
offering a more holistic understanding of the research problem.  
Ethical Considerations  
This study adhered to established ethical standards for research in educational settings. All participants were  
informed about the purpose of the study, the voluntary nature of their involvement and their right to withdraw  
at any time without penalty. Informed consent was obtained prior to data collection. No identifying or sensitive  
personal information was collected to ensure the anonymity and confidentiality of all participants. The data were  
Page 2151  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
stored securely and used solely for academic and research purposes. The study was conducted in accordance  
with the ethical principles outlined in the British Educational Research Association’s Ethical Guidelines for  
Educational Research (BERA, 2024).  
RESULTS  
Findings  
Participants’ Demographic Information  
Table 1 Distribution of Participant Demographics  
Demographics  
Gender  
Description  
Male  
Frequency  
Percentage  
20.0%  
80.0%  
100%  
6.7%  
12  
48  
60  
4
Female  
Total  
Age  
Below 25  
26-35  
35  
18  
3
58.3%  
30.0%  
5.0%  
36-45  
46-55  
Above 55  
Total  
0
0%  
60  
21  
17  
14  
8
100%  
35.0%  
28.3%  
23.3%  
13.3%  
100%  
70.0%  
15.0%  
15.0%  
100%  
85.0%  
15.0%  
100%  
Teacher experience  
Less than 5 years  
5-10 years  
11-15 years  
More than 15 years  
Total  
60  
42  
9
School location  
Urban  
Suburban  
Rural  
9
Total  
60  
51  
9
Have you use AI tools?  
Yes  
No  
Total  
60  
Page 2152  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
A total of 60 primary school English teachers participated in the study. As shown in Table 1, the majority were  
female (80%) while 20% were male. Most participants were between 26 and 35 years old (58.3%) followed by  
those aged 36 to 45 years (30%). Only 6.7% were below 25, and a small group (5%) were aged 46–55. In terms  
of teaching experience, 35% had less than five years of experience while 28.3% had between 5 and 10 years.  
Another 23.3% had 11–15 years of experience and 13.3% had been teaching for more than 15 years. This  
indicates that the sample consisted of teachers with varied levels of professional experience. Most teachers (70%)  
were teaching in urban schools with 15% teaching in suburban and 15% in rural areas. A large majority of the  
participants (85%) reported having used AI tools before which showed a high level of exposure to digital  
technologies among the respondents. These demographic patterns suggest that the participants generally  
belonged to a digitally literate, professionally diverse group mostly located in urban educational environments.  
Teachers’ Perceptions of Opportunities of AI in Speaking Practice  
Table 2 Teachers’ Perceptions of Opportunities of AI in Speaking Practice  
No. Items  
SD  
D
2
N
0
A
SA  
23  
Mean SD  
0.777  
1
2
3
4
5
6
7
8
AI tools can provide pupils with more 0  
opportunities to practise speaking English.  
28  
4.2  
4.1  
4.2  
4.2  
4.1  
4.1  
4.0  
4.4  
AI enables pupils to practise speaking in a 0  
low-anxiety environment.  
4
1
1
2
2
3
1
7
7
7
9
9
8
2
31  
30  
25  
30  
28  
35  
31  
18  
22  
26  
19  
21  
14  
26  
0.832  
0.715  
0.851  
0.775  
0.791  
0.759  
0.637  
AI-based speaking tools can give instant 0  
feedback on pronunciation and fluency.  
Using AI makes speaking activities more 1  
engaging and interactive.  
AI helps pupils become more confident in 0  
speaking English.  
AI tools support personalised learning 0  
according to pupils’ proficiency levels.  
AI can supplement classroom speaking 0  
practice effectively.  
AI enhances teachers’ ability to design 0  
creative speaking lessons.  
Overall, teachers expressed highly positive perceptions toward the opportunities offered by AI tools for pupils’  
speaking practice. Mean scores for the eight items ranged between 4.0 and 4.4, indicating strong agreement  
across all statements. The highest-rated item was “AI enhances teachers’ ability to design creative speaking  
lessons” (Mean = 4.4, SD = 0.637), suggesting that teachers find AI beneficial in diversifying and enriching  
classroom activities. Teachers also strongly agreed that AI can offer pupils more opportunities to practise  
speaking English (Mean = 4.2), provide instant feedback on pronunciation and fluency (Mean = 4.2), and make  
speaking activities more engaging and interactive (Mean = 4.2). Additionally, teachers agreed that AI tools help  
pupils to build their confidence in speaking, support personalised learning and enable practice in a low-anxiety  
environment (Means between 4.1–4.2). The lowest-rated item, though still positive, was “AI can supplement  
classroom speaking practice effectively” (Mean = 4.0), indicating that some teachers may view AI as a  
complement rather than a replacement for teacher-led activities. These results indicate that teachers perceive AI  
tools as valuable, engaging and supportive resources that can enhance pupils’ oral language development.  
Page 2153  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Teachers’ Perceptions of Limitation and Challenges  
Table 3 Teachers’ Perceptions of Limitation and Challenges  
No. Items  
SD  
D
0
N
4
A
SA  
Mean SD  
0.619  
1
2
3
4
5
6
7
8
AI tools require stable internet and digital devices that 0  
17 39  
4.6  
4.3  
4.2  
4.3  
4.4  
4.2  
4.4  
4.4  
may not be available in all schools.  
Pupils may rely too much on AI and reduce real human 0  
interaction.  
1
4
2
1
1
0
0
10 20 29  
0.804  
0.873  
0.761  
0.736  
0.813  
0.673  
0.645  
AI feedback may not always be accurate or suitable 0  
for young learners.  
6
5
6
25 25  
27 26  
23 30  
Teachers need more training to integrate AI effectively 0  
in speaking lessons.  
Concerns about pupils’ data privacy and safety arise 0  
when using AI tools.  
AI content may not always align with the Malaysian 0  
CEFR curriculum.  
12 22 25  
Excessive screen time when using AI may affect 0  
pupils’ attention or health.  
6
5
22 32  
25 30  
Lack of school support and infrastructure limits AI 0  
implementation.  
Although teachers acknowledged the benefits of using AI, they also shared several real and practical concerns  
about bringing these tools into speaking lessons. The ratings were consistently high (between 4.2 and 4.6),  
showing that many teachers strongly agreed with the challenges raised. The biggest issue was the need for  
reliable internet and sufficient devices for pupils (Mean = 4.6, SD = 0.619), which remains a problem in many  
schools. Teachers also felt that pupils might rely too heavily on AI which will reduce meaningful face-to-face  
interaction (Mean = 4.3). They were concerned about the accuracy of AI feedback (Mean = 4.2), the lack of  
proper training for teachers (Mean = 4.3) and data privacy issues (Mean = 4.4). Some also mentioned that AI-  
generated responses do not always align with the CEFR-aligned curriculum (Mean = 4.2) and that increased  
screen time could affect pupils’ focus and well-being (Mean = 4.4). Many agreed that without strong support  
and adequate infrastructure from the school, effective use of AI is difficult to achieve (Mean = 4.4). Overall,  
while teachers see the promise of AI, they remain cautious because of these practical, pedagogical, and ethical  
concerns.  
Overall Perception and Future Use  
Table 4 Overall Perception and Future Use  
No. Items  
SD  
D
1
N
8
A
SA  
27  
Mean SD  
1
I am interested in learning more about 0  
how to use AI for speaking lessons.  
24  
4.3  
0.761  
2
I believe AI will become an essential tool 0  
in future English teaching.  
0
4
28  
28  
4.4  
0.616  
Page 2154  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
3
4
I would recommend AI tools for 0  
speaking practice to other teachers.  
0
1
9
22  
21  
29  
26  
4.3  
4.2  
0.729  
0.819  
AI should be incorporated into the 0  
school English curriculum.  
12  
Teachers’ overall perceptions toward future AI integration were similarly positive. All items in this section  
scored between 4.2 to 4.4 which reflects a strong willingness to adopt AI tools in English language teaching.  
Teachers expressed high interest in learning more about using AI in speaking lessons (Mean = 4.3) and believed  
that AI will become an essential tool in future English teaching (Mean = 4.4). They also indicated a strong  
likelihood of recommending AI tools to other teachers (Mean = 4.3). Furthermore, there was strong agreement  
that AI should be incorporated into the school English curriculum (Mean = 4.2). This suggested teachers’  
readiness for more systematic integration of AI within the educational framework. These findings reflect an  
overall positive and future-oriented mindset with teachers showing openness, interest and confidence in the  
continued use of AI for enhancing speaking practice among primary school learners.  
Open-Ended Questions  
A thematic analysis was conducted to explore teachers’ perceptions regarding the use of AI tools for pupils’  
speaking practice. The analysis generated four overarching themes aligned with the four open-ended questions:  
Table 5 Thematic analysis for open-ended questions  
1. What do you Theme 1: Increased Teachers pointed out that AI gives pupils more chances to practise  
think are the main Opportunities for speaking outside the classroom. AI allows pupils to:  
benefits of using Practice Anytime,  
“Practise anywhere, anytime,” even without a partner  
AI for pupils’ Anywhere  
speaking practice?  
Increase their exposure to English  
Practise at their own pace  
Teachers noted that this flexibility is especially helpful for shy or low-  
confidence pupils.  
Theme 2: Instant Teachers highlighted that one of the biggest strengths ofAI is its ability  
and Personalised to give immediate and personalised feedback. AI can:  
Feedback  
Correct mispronunciations instantly  
Convert pupils’ speech to text for clearer accuracy checks  
Provide individualised feedback that teachers may struggle to  
give in large classes  
Theme 3: Boosting Teachers shared that AI tools help pupils feel less anxious because they  
Confidence and do not feel judged by others. AI creates a low-pressure environment  
Reducing Anxiety that allows pupils to:  
Practise speaking without fear  
Build confidence gradually  
Participate more actively in speaking tasks  
Page 2155  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Teachers also mentioned that AI reduces the “awkwardness” often felt  
by low-proficiency learners.  
Theme  
4: Many responses described AI as:  
Enhanced  
Engagement  
Through  
Interactive  
Creative Learning  
“Fun and interactive”  
Helping to “encourage pupils to speak”  
Making lessons “more interesting”  
and  
Teachers also explained that AI stimulates creativity and helps them  
brainstorm speaking activities, design differentiated tasks, and create  
rubrics for speaking assessments.  
2. What are the Theme 1: Internet The most dominant challenge was unstable or limited internet access,  
biggest challenges Connectivity and especially in rural schools. Teachers frequently mentioned:  
you face or expect Device Limitations  
Poor WiFi connection  
Insufficient devices  
Expensive AI features  
when using AI  
tools?  
Theme  
2: Teachers reported that AI:  
Accuracy,  
Reliability, and AI  
Limitations  
Sometimes provides incorrect or irrelevant information  
May misinterpret accents, especially children’s voices  
Produces “robotic” or unnatural content  
may not always provide feedback that aligns with classroom  
expectations or CEFR requirements.  
Theme  
3
Over- Teachers expressed concern that excessive use of AI may cause pupils  
Reliance on AI and to:  
Reduced Human  
Become dependent on AI-generated content  
Lose creativity  
Interaction  
Hesitate during real conversations  
Show reduced human interaction  
Theme  
4: Teachers highlighted that:  
Insufficient  
Training  
Teacher Readiness  
Some do not know how to use AI effectively  
and  
Many must self-learn due to the lack of formal training  
Managing class discipline while using AI tools is challenging  
Time constraints and heavy workload make it difficult for some  
teachers to explore AI.  
Page 2156  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Theme 5: Ethical Several teachers were concerned about:  
and  
Concerns  
Privacy  
Data privacy  
Safety issues with AI tools  
Younger pupils’ inability to filter information  
3. How can AI Theme 1: Better Many teachers requested AI tools that:  
tools be improved Alignment  
to better support Curriculum  
speaking practice Learning Levels  
in primary ESL  
with  
and  
Align with the Malaysian CEFR curriculum  
Include content based on age-appropriate vocabulary and themes  
Provide tasks suitable for children's linguistic levels  
classrooms?  
Theme 2: Improved Teachers suggested:  
Speech  
More accurate pronunciation recognition for children  
Recognition and  
Child-Friendly  
Interaction  
More natural, human-like voices  
Warm, child-friendly interfaces  
Recognition of Malaysian or local accents  
Theme 3: More Teachers recommended:  
Interactive,  
Role-plays  
Engaging,  
Game-Based  
Features  
and  
Storytelling  
Interactive games  
Real-life simulations  
Theme 4: Better Teachers wanted:  
Teacher Tools and  
Tools for tracking pupils’ progress  
Classroom  
Integration  
Customisable speaking tasks  
Offline modes for low-internet schools  
Free or low-cost AI options  
4. Would you like Theme 1: Strong Many teachers expressed a willingness to attend workshops because  
to receive training Demand for AI training would help them:  
or workshops on Training  
Use AI more effectively  
using  
English teaching?  
Why or why not?  
AI  
for  
Integrate AI into lessons confidently  
Select appropriate tools  
Create engaging speaking activities  
Page 2157  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Interpret AI feedback  
Theme 2: Reasons A minority of teachers said they did not want training due to:  
for Hesitation  
Heavy workload  
Time constraints  
Preference for traditional teaching  
Confidence in self-learning  
Some felt they already understood the basics.  
Theme 3: Practical Teachers suggested that training should be:  
Expectations  
Training  
for  
Short and practical  
Focused on classroom application  
Relevant to primary pupils’ needs  
Supportive of responsible and ethical use  
The thematic analysis revealed four key themes. First, teachers perceived AI as highly beneficial for supporting  
speaking practice, citing increased opportunities for practice beyond the classroom, immediate pronunciation  
and fluency feedback, reduced learner anxiety and more engaging, personalised learning experiences. Second,  
significant challenges were identified, including unstable internet access, limited device availability, concerns  
about the accuracy and reliability of AI feedback, data privacy risks and fears of pupils becoming overly reliant  
on AI at the expense of real human interaction. Third, teachers recommended several improvements such as  
better alignment with the CEFR-aligned Malaysian curriculum, child-friendly and culturally relevant content,  
enhanced speech recognition for young learners, interactive game-based speaking tasks, offline functionality and  
built-in tools for monitoring pupils’ progress. Lastly, most teachers expressed strong interest in receiving AI-  
related training to improve their confidence and pedagogical effectiveness although a minority reported  
constraints such as workload, time limitations or a preference for traditional teaching approaches.  
DISCUSSION  
Table 6 Comparison of Past Studies and Current Study Findings  
Past Study & Key Insight  
Alignment with Current study Findings  
Hidayah (2025) – AI provides instant, personalised Teachers in the current study agreed AI offers instant  
pronunciation feedback  
feedback that helps pupils improve fluency and  
pronunciation.  
Zhang (2024) – AI reduces speaking anxiety  
Teachers observed that AI creates a low-pressure,  
judgment-free environment, especially for shy or  
low-confidence pupils.  
Zou (2024) – AI enables anytime, anywhere practice and Teachers reported that pupils can practise anywhere  
expands speaking exposure and anytime, increasing English exposure.  
Alexis & Becky (2024) – AI’s gamified features enhance Teachers described AI tools as fun, engaging and  
engagement  
interactive, helping pupils participate more actively.  
Page 2158  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Hamayun et al. (2025) – AI supports differentiated Teachers agreed AI provides personalised learning  
learning  
suited to pupils’ individual needs.  
An (2022) and Firdaus & Nawas (2024) – Teachers’ Teachers indicated they need more training and some  
digital literacy affects AI adoption feel unprepared or unfamiliar with AI tools.  
Junaidi (2024) – Technical barriers limit technology use Teachers identified unstable internet and limited  
devices as their biggest challenges.  
Mudway (2025) – ASR misrecognition affects feedback Teachers similarly reported inaccurate or robotic  
accuracy  
feedback, especially for young learners’ voices.  
Kotsis (2025) – Concerns about data privacy and Teachers raised concerns about data privacy, safety  
overreliance and pupils over-relying on AI-generated responses.  
Zulkarnain & Yunus (2023) – Teachers’ acceptance Current findings show teachers are positive overall  
determines success of AI integration  
but implementation remains constrained by readiness  
and infrastructure.  
Sulaiman (2024) – Insufficient training limits effective Teachers strongly expressed the need for practical,  
digital integration hands-on AI training tailored to classroom needs.  
Vincent (2025) – Connectivity and device issues remain This study confirms that internet and device  
major barriers limitations are among the most significant obstacles.  
Sangeetha (2025) and Jamaludin (2025) – AI improves Teachers agreed AI enhances fluency and  
oral fluency and supports speaking practice  
supplements classroom speaking practice effectively.  
Positive Perceptions of AI for Enhancing Speaking Skills  
The findings of this study provide a comprehensive and nuanced understanding of primary ESL teachers’  
perceptions of using Artificial Intelligence (AI) tools to support speaking practice. By integrating quantitative  
data with qualitative insights, the discussion highlights both the strong pedagogical potential of AI and the  
practical realities that influence its successful implementation in the classroom. Overall, teachers expressed  
highly positive views regarding AI’s ability to enhance speaking instruction. The consistently high mean scores  
(4.0–4.4) indicate strong agreement that AI offers meaningful support for oral language development. Teachers  
appreciated that AI tools provide opportunities for repeated speaking practice, instant and personalised feedback,  
and adaptive learning experiences that help accommodate varying proficiency levels. This aligns with the  
Perceived Usefulness (PU) construct in the Technology Acceptance Model (TAM), suggesting that teachers see  
AI as a credible and effective tool for improving pronunciation, fluency and confidence. Qualitative responses  
further strengthened this finding with teachers emphasising that AI enables pupils to practise “anytime and  
anywhere” and offers corrective input that is often difficult to provide consistently in large classrooms. A second  
important finding is AI’s role in reducing speaking anxiety. Teachers noted that pupils feel more comfortable  
practising English in low-pressure, judgment-free environments created by AI applications. This is particularly  
beneficial for shy or lower-proficiency learners who may hesitate to speak in front of peers. These perceptions  
are consistent with previous research showing that AI-based speaking tools foster enjoyment, lower  
communication apprehension, and encourage more active participation. Teachers therefore recognised AI as a  
supportive avenue for building confidence and encouraging risk-taking in oral communication.  
Challenges Affecting AI Integration  
However, despite these advantages, teachers also reported significant challenges that could hinder the effective  
integration of AI in speaking lessons. Technical issues such as unstable internet connection, insufficient digital  
devices and frequent disruptions were among the most commonly cited barriers. These concerns, reflected in the  
high mean scores for limitation items (4.2–4.6), are especially pronounced in rural and under-resourced schools.  
Page 2159  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Such barriers directly affect teachers’ Perceived Ease of Use (PEOU) and in turn reduce their overall intention  
to adopt AI tools. Teachers also raised pedagogical concerns related to the accuracy and reliability of AI-  
generated feedback. Many noted that AI tools occasionally misinterpret young children’s voices, dialects or  
accent features, resulting in inaccurate or overly robotic feedback. These issues align with earlier research  
indicating that AI speech recognition systems often struggle with the diverse linguistic characteristics of younger  
learners. Ethical and safety-related concerns also emerged strongly in the teachers’ responses. Participants  
worried about data privacy, exposure to inappropriate content, and the possibility that pupils may rely too heavily  
on AI-generated responses, compromising their creativity and real communicative interactions. These concerns  
highlight the need for clear guidelines and responsible implementation practices when integrating AI into  
primary education.  
Teachers’ Suggestions for Improving AI Tools  
Teachers also provided practical suggestions for improving AI tools to better support speaking practice. Many  
emphasised the need for tools that align more closely with Malaysian CEFR-speaking descriptors and include  
culturally relevant, age-appropriate content. They also recommended enhancing speech recognition accuracy for  
children, incorporating child-friendly voices, and expanding interactive features such as storytelling, games, and  
role-play activities. Requests for offline functionality further highlight the importance of ensuring equitable  
access in low-bandwidth environments. Finally, the strong willingness among teachers to attend AI-related  
training underscores the need for sustained professional development. Teachers recognised that training could  
help them integrate AI more effectively, interpret feedback appropriately, and design creative speaking activities.  
However, a minority expressed reluctance due to workload and time constraints, suggesting that future training  
should be flexible, practical, and directly applicable to classroom realities.  
Summary  
Overall, the findings illustrate a dual reality: while teachers are optimistic about the transformative potential of  
AI in supporting speaking development, they remain cautious about the structural, instructional and ethical  
challenges that accompany its use. A balanced and strategic approach focused on infrastructure readiness, teacher  
training, curriculum alignment and clear guidelines is therefore crucial for ensuring meaningful and sustainable  
AI integration in Malaysian primary ESL classrooms.  
CONCLUSION  
This study showed that primary ESL teachers generally view AI tools as valuable for enhancing pupils’ speaking  
practice particularly through instant feedback, increased speaking exposure, reduced anxiety and more engaging,  
personalised learning experiences. At the same time, teachers identified notable challenges such as unstable  
internet connectivity, limited device availability, inaccurate speech recognition for young learners, insufficient  
training and concerns about privacy and overreliance on AI-generated responses. These findings highlight that  
while AI holds strong potential to support speaking development in Malaysian primary schools, its successful  
implementation depends on overcoming the technological, pedagogical and ethical barriers that teachers  
continue to face.  
RECOMMENDATIONS  
To ensure effective integration of AI in primary ESL classrooms, schools should strengthen digital infrastructure  
by improving internet stability and increasing access to devices especially in underserved areas. Developers  
should refine AI tools to better recognise children’s voices, incorporate Malaysian English accents and align  
content with CEFR requirements while offering offline modes for low-connectivity settings. Teachers would  
benefit from practical, hands-on professional development focused on AI-supported speaking instruction, data  
privacy and responsible use. Clear school-level guidelines should also be established to balance technology use  
with meaningful human interaction and prevent overreliance on AI.  
Page 2160  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
Future Research  
Future research could broaden the sample to include schools across different Malaysian states to enhance the  
generalisability of findings, particularly in rural areas with varying digital readiness. Experimental and  
longitudinal studies are recommended to measure AI’s impact on pupils’ oral fluency, pronunciation accuracy  
and speaking confidence over time. Additionally, studies exploring pupils’ perspectives, teachers’ digital  
readiness and the design of CEFR-aligned AI speaking tasks would provide deeper insights into effective AI  
integration. Research on ethical and privacy issues is also needed to ensure safe and responsible use of AI in  
primary education.  
REFERENCES  
1. Antonietti, C., Cattaneo, A. & Amenduni, F. (2022). Can teachers’ digital competence influence  
technology acceptance in vocational education?. Computers in Human Behavior. 132 (2): 1-9.  
DOI:10.1016/j.chb.2022.107266  
2. An, X., Chai, C. S., Li, Y., Zhou, Y., Shen, X., Zheng, C., & Chen, M. (2022). Modeling English teachers’  
behavioral intention to use artificial intelligence in middlechools. Education and Information  
Technologies, 28(5), 5187–5208. https://doi.org/10.1007/s10639-022-11286-z  
3. Aziz, A. A., & Kashinathan, S. (2021). ESL Learners’ Challenges in Speaking English in Malaysian  
Classroom. International Journal of Academic Research in Progressive Education and Development,  
10(2), 983–991.  
4. British Educational Research Association [BERA]. (2024). Ethical Guidelines for Educational Research  
(5th ed.). www.bera.ac.uk/publication/ethicalguidelines-for-educational- research-2024  
5. Chali, M. T., Eshete, S. K. & Debela, K. L. 2022. Learning How Research Design Methods Work: A  
Review of Creswell’s Research Design: Qualitative, Quantitative and Mixed  
The Qualitative Report. 27(12): 2956-2960.  
Method Approaches.  
6. Davis, F. D. (1989). Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information  
Technology. MIS Quarterly, 13(3), 319-340.  
7. Firdaus, A., & Nawaz, S. (2024). Viewpoints of teachers about the usage of artificial intelligence  
in ELT: Advantages and obstacles.University of Chitral Journal of Linguistics & Literature, 8(1), 82-93  
8. Hamayun, M., Saeed, A. Amin, I. & Chapra, L. (2025). AI Supported Differentiated Instruction Cognitive  
and Emotional Outcomes for Diverse Learners . The Critical Review of Social Sciences Studies. 3(3):  
469-486. DOI: 10.59075/crsss  
9. Hidayah, A. N., Riyati, D. & Gres, E. (2025). AI-powered pronunciation tools: supporting english  
learners in multilingual classrooms who dislike english in pontianak. Wiralodra English Journal 9(1):13-  
26. DOI:10.31943/wej.v9i1.350  
10. Junaidi, N. H. & Jamaludin, K. A. (2024). The Challenges of Digitization of Education Among Primary  
School Teachers: Through Systematic Literature Review. International Journal of Academic Research in  
Progressive Education & Development. 13(3): 2364-2376. DOI:10.6007/IJARPED/v13-i3/22144  
11. Kafabih, A., Herda, R.K. & Monteza, A. M. M. (2025). Classroom Praxis of Applied Oral  
Communication Strategies in Enhancing English Speaking Skills in Indonesian Secondary School.  
Research  
and  
Innovation  
in Applied  
Linguistics-Electronic  
Journal  
3(2):195-214  
DOI:10.31963/rial.v3i2.5526  
12. Kotsis, K. T. (2025). Issues between Artificial Intelligence and Personal Data in Education. International  
Research in Education. 13(1): 45-65. DOI:10.5296/ire.v13i1.22850  
13. Mahmoud, C. F. & Sorensen, J. T. (2024). Artificial Intelligence in Personalized Learning with a Focus  
on Current Developments and Future Prospects. Research and Advances in Education 3(8):25-31.  
DOI:10.56397/RAE.2024.08.04  
14. Mudaway, A. M. A. (2025). Exploring EFLLearners’Perceptions on the Use ofAI-Powered  
Conversational Tools to Improve Speaking Fluency: ACase Study at Majmaah University. Forum for  
Linguistic Studies. 7(1): 590-598. https://journals.bilpubgroup.com/index.php/fls  
15. Nhan, L. K., Hao, N. T. M. & Quang, Luong, V, Nhat. (2025). Revolutionizing speaking skills  
improvement: AI’s role in personalized language learning. International Journal of  
Research and Scientific Studies 8(2):4637-4649. DOI:10.53894/ijirss.v8i2.6408  
Innovative  
Page 2161  
www.rsisinternational.org  
INTERNATIONAL JOURNAL OF RESEARCH AND INNOVATION IN SOCIAL SCIENCE (IJRISS)  
ISSN No. 2454-6186 | DOI: 10.47772/IJRISS | Volume IX Issue XI November 2025  
16. Pakpaha, F. H. & Saragih, M. (2022). Theory Of Cognitive Development By Jean Piaget. Journal of  
Applied Linguistics 2(2):55-60. DOI:10.52622/joal.v2i2.79  
17. Sangeetha, L. & Jamaludin, K.A. (2025). Integration of Artificial Intelligence (AI) Tools in Empowering  
the Teaching of English-Speaking Skills: Benefits and Challenges. International Journal of Academic  
Research in Progressive Education and Development. 14(4): 1-12. DOI:10.6007/IJARPED/v14-  
i4/26466  
18. Shakri, M. A. M., Ngoi, G. P. Saari, K. A. & Marliana, N. L. (2025). The Application of Generative  
Artificial Intelligence TechnologyAmong Language Teachers in Facing Current Educational Challenges.  
Journal of ICT in Education. 12(2): 138-148. DOI: https://doi.org/10.37134/jictie.vol12.2.9.2025  
19. Sivanganam, J., Yunus, M.M. & Said, N. E. M. (2025). Teachers’ Perceptions in Using  
ArtificialIntelligence (AI) in ESL Classrooms. International Journal of Academic Research in  
Progressive Education and Development. 14(1): 2409-2427. DOI:10.6007/IJARPED/v14-i1/25034  
20. Sulaiman, R. B. A., Bin Hassan, H., Kemanusiaan, S. P. (2024). E-Learning Readiness From the Aspects  
of Attitude, Literacy, Facilities and Access to Technology Among Religious Secondary School Students  
in the State of Pahang. Al-Sirat Journal, 24(1). https://ejournal.unipsas.edu.my/  
21. Tatchakrit, M., Ruedeerath, C. & Joey, A. L. S. (2024). “I think I speak without hesitation”: Learning-  
Oriented Assessment to Enhance English Oral Communication of Thai Pre-Service Teachers.  
rEFLections. 31(3). https://files.eric.ed.gov/fulltext/EJ1459980.pdf  
22. Teo, T. (2011). Technology Acceptance Research in Education: A Review of the Literature.  
23. Venkatesh, V., & Davis, F. D. (2000). A Theoretical Extension of the Technology Acceptance Model:  
Four Longitudinal Field Studies. Management Science, 46(2), 186-204.  
24. 24. Vincent, J., Yunus, M. M. & Said, N. E. M. (2025). Using AI Platforms to Improve Listening and  
Speaking Skills in ESL Primary Students. International Journal of Academic Research in Prohressive  
Education and Development. 14(1): 1829-1843. DOI:10.6007/IJARPED/v14-  
i1/24848  
25. Zhang, C., Meng, Y. & Ma, X. Artificial intelligence in EFL speaking: Impact on enjoyment, anxiety, and  
willingness to communicate. 121(3). DOI:10.1016/j.system.2024.103259  
26. Zou, B. et al. (2024). Exploring EFL learners’ perceived promise and limitations of using an artificial  
intelligence speech evaluation system for speaking practice. System. 126(103497): 1-24.  
10.1016/j.system.2024.103497  
27. Zulkarnain, N. S., & Yunus, M. M. (2023). Primary Teachers’ Perspectives on Using Artificial  
Intelligence Technology in English as a Second Language Teaching and Learning: A Systematic Review.  
International Journal of Academic Research in Progressive Education and Development, 12(2), 861–875.  
Page 2162  
www.rsisinternational.org