Page 603 - ISC PROCEEDINGS 21.4

controlled experimental setting which restricts their applicability to the classroom. Third,
empirical studies involving the effectiveness of AI-based language learning tools in actual
ESL classes at universities based on a mixed-method design are limited. Thus, the study
fills these gaps by concurrently studying the vocabulary learning and pronunciation
acquisition between ESL learners in an actual classroom environment through
quantitative and qualitative approaches in the university.
Research Questions:
1. How well AI-based language learning systems can enhance vocabulary learning in
ESL learners?
2. What does AI-based pronunciation aids do to the pronunciation of ESL students?
3. How do learners perceive the use of AI technologies in learning vocabulary and
practicing pronunciation?
4. What is the effect of incorporation of AI tools in regard to engagement and
independent learning in ESL classrooms?
5. Methodology
5.1. Research design
The research design adopted this work was a mixed-method study research design,
which was used to explore the efficacy of AI-based tools in improving vocabulary learning
and pronunciation growth among ESL students. The quantitative part took a pre-test and
post-test study to identify the difference in the performance of the students prior to and
after the intervention. Simultaneously, the qualitative aspect involved the concept of
student surveys and classroom monitoring to gain a more profound insight into the
student experiences, engagement, and perception of AI-based tools of language learning
usage.
5.2. Participants
The sample size was 50 university-level ESL students who were taking an English
course in the university. The sampling used was convenience sampling since the students
were easily available in the teaching environment of the researcher. All the participants
had intermediate English proficiency based on their placement test scores and previous
academic achievements in English language courses.
5.3. Instruments
A number of research tools were also utilized in order to gather quantitative and
qualitative data. The vocabulary test had 30 items comprising multiple choice questions,
matching and sentence completion exercises, to gauge the knowledge of the students
regarding vocabulary, both at the start of the intervention and the end. Word-reading and
sentence-reading tests were considered as a part of the pronunciation test and were
assessed with the help of AI-based speech recognition programs along with a teacher
scoring rubric to balance the scores. Besides that, a 15-item questionnaire was designed
based on the Likert scale to assess the engagement, motivation, perceived usefulness,
and perceived autonomy of using AI-based tools. Participation by the learners, the
frequency and pattern of their interaction, and the pattern of using the tools in the
instructional sessions were also recorded through classroom observation checklists.
5.4. Reliability and validity
The research instruments were well determined in terms of reliability and validity.
The internal consistency of the questionnaire was satisfactory and Cronbach alpha
coefficient of the questionnaire was 0.87, which indicates high reliability. The reliability of
the pronunciation assessment procedure was achieved by performing the inter-rater

602