MODERN EDUCATION AND DEVELOPMENT
Выпуск журнала №-29
Часть–1_Июль –2025
242
USING AI-POWERED CHATBOTS TO ENHANCE SPEAKING SKILLS
IN ENGLISH AS A FOREIGN LANGUAGE CLASSROOMS
Qodirqulova Maftunaxon Muhiddin qizi
Chirchiq Davlat Pedagogika Universiteti
ingliz tili
Abstract: The challenge of improving speaking skills in English as a Foreign
Language (EFL) classrooms is long-standing, especially in contexts with limited
teacher time and peer interaction. This research explores the integration of AI-
driven chatbots designed specifically for conversational practice and its impact on
learners’ fluency, pronunciation, and confidence. Over a fourteen-week
intervention, a cohort of eighty intermediate-level learners engaged in structured
weekly sessions with chatbots capable of adapting to individual accuracy and style.
Performance data, learner reflections, and instructor observations were collected
to evaluate change. The findings reveal substantial improvements in learner
willingness to communicate, speaking fluency, and pronunciation accuracy, coupled
with enhanced confidence and motivation. However, challenges surface around the
chatbot’s contextual awareness, occasional misunderstanding of learner input, and
learners’ preference for human feedback in nuanced conversations. The study
concludes with design recommendations to optimize chatbot roles in EFL speaking
development and suggests best practices for classroom implementation.
Keywords: AI chatbot, speaking fluency, EFL classroom, pronunciation
feedback, adaptive conversation, learner autonomy, oral competence
Introduction
Developing oral competence in English as a Foreign Language has
always been an essential yet challenging objective. Classroom limitations, such as
large class sizes, limited teacher-student speaking time, and shy learners, exacerbate
the difficulty of fostering fluent conversations. Advances in artificial intelligence,
MODERN EDUCATION AND DEVELOPMENT
Выпуск журнала №-29
Часть–1_Июль –2025
243
particularly conversational chatbots powered by natural language processing, offer
new possibilities for practice outside human interaction constraints. Such chatbots
can engage learners in realistic dialogue, adapt to individual proficiency levels, and
offer instant feedback on pronunciation and fluency. These characteristics align with
communicative approaches and autonomy-supportive teaching principles.
Despite the surge in interest, there remains limited empirical evidence
regarding effectiveness of chatbots for speaking development. Questions abound
about sufficiency of AI feedback, effects on learner confidence, and best practices
for integration into curricula. This study addresses the gap by examining how
structured chatbot use influences speaking fluency, pronunciation accuracy, and
learner confidence in an intermediate EFL context.
Methodology
This mixed-method study took place at a university language center over
fourteen weeks and involved eighty intermediate-level English learners divided
randomly into experimental and control groups. The experimental cohort accessed
an AI chatbot platform twice weekly—each session designed to stimulate 15–20
minutes of spoken conversation on varied task prompts, including role-plays,
problem-solving scenarios, and open-ended discussions.
The chatbot employed speech-recognition and pronunciation scoring
algorithms, offering real-time corrective feedback on stress, intonation, and
articulation. It also scaffolded learners by rephrasing prompts according to their
proficiency and adjusting difficulty dynamically. Learners in the control group
continued standard classroom oral practice activities without chatbot support.
Data collection included pre- and post-intervention speaking assessments
(analyzed by independent raters using fluency and pronunciation rubrics), weekly
reflection journals, and end-of-study focus-group interviews with experimental
group participants and teacher reflections.
Qu antitative analyses explored changes in fluency rates (words per
minute and pause durations) and pronunciation error rates. Thematic analysis of
MODERN EDUCATION AND DEVELOPMENT
Выпуск журнала №-29
Часть–1_Июль –2025
244
qualitative data sought to uncover shifts in confidence, engagement, and learner
attitudes toward AI-produced feedback.
Results
Speaking fluency exhibited statistically significant improvement in the
experimental group, with average speech rate increasing by 11% and pausing
frequency decreasing notably. Pronunciation scoring illustrated a 9% reduction in
segmental and suprasegmental errors among chatbot users. Control group
performance remained largely unchanged over the same period.
Qualitative findings underscored chatbot-driven behavioral change:
learners highlighted a newfound willingness to speak without fear of embarrassment,
noting feelings of freedom to make errors. They appreciated the chatbot’s
nonjudgmental, on-demand environment, which allowed repeated practice without
shame. Several remarked on the conversational experience's realism, expressing
greater spontaneity compared to scripted classroom dialogues.
However, learners also reported occasional frustration when the chatbot
misinterpreted phrasing, or seemed unable to respond meaningfully to subtle
emotional or cultural content. Many still preferred human interaction for complex
communicative nuances or open-ended discussions. Teachers noted that chatbot logs
revealed recurring pronunciation patterns—elements that guided in-class emphasis
and remedial exercises.
Discussion
Results demonstrate that AI chatbots can effectively supplement
speaking practice in EFL settings. Gains in fluency and pronunciation suggest
chatbots fill a gap in opportunities for oral rehearsal, particularly valuable for
learners with limited access to native speakers. The autonomous, non-threatening
atmosphere engenders higher motivation and self-correction habit, contributing to
long-term learning agency.
But limitations surface when it comes to deeper dialogical nuances. These
are currently outside the scope of generic chatbot frameworks. The mismatch of
responses or contextual misunderstandings highlights the need for hybrid learning
MODERN EDUCATION AND DEVELOPMENT
Выпуск журнала №-29
Часть–1_Июль –2025
245
pedals—wherein chatbots are integrated but not exclusively entrusted with speaking
practice. Teacher involvement remains essential for scaffolding cultural and
pragmatic language aspects.
Pedagogically, chatbots appear most effective when implemented with
clear structure. When learners knew session goals and received directed reflection
prompts, speaking outcomes improved. This outcome aligns with broader
technology-enhanced language learning theory that emphasizes guided integration
rather than mere novelty.
Conclusion
AI-powered chatbots show strong potential to enhance speaking skills in
EFL classrooms, especially by providing quantity of practice that is often
unachievable in traditional settings. These tools support fluency development,
pronunciation accuracy, learner autonomy, and confidence.
To adapt for wider classroom integration, instructional designers should
prioritize chatbots with high speech-quality input, carefully scaffolded lesson
design, and combined use with human instructor guidance. Future research should
examine long-term retention effects, multilingual learning settings, and evolving AI
capabilities for socio-pragmatic competence.
REFERENCE
1. Godwin-Jones, R. (2019).
Emerging Technologies: AI and Language Learning
.
Language
Learning
&
Technology,
23(3),
3–11.
2. Xu, Y., & Wang, Y. (2021).
AI-Powered Chatbots in EFL Speaking Instruction:
Pedagogical Benefits and Challenges
. Computer Assisted Language Learning,
34(7),
867–889.
3. Winkler, R., & Söllner, M. (2018).
Unleashing the Potential of Chatbots in
Education: A State-Of-The-Art Analysis
. In ECIS 2018 Proceedings.
4. Lee, L. (2020).
Chatbots and Language Learning: Investigating Learners’
Perceptions
and
Usage
Behavior
.
ReCALL,
32(3),
272–287.
5. Fryer, L. K., & Carpenter, R. (2006).
Bots as Language Learning Tools
. Language
Learning
&
Technology,
10(3),
8–14.
MODERN EDUCATION AND DEVELOPMENT
Выпуск журнала №-29
Часть–1_Июль –2025
246
6. Huang, Y.-M., & Hew, K. F. (2018).
Implementing an AI Conversational Agent
to Facilitate Real-Time Practice in Language Learning
. Educational Technology
Research
and
Development,
66(3),
575–589.
