Vol. 12/ Núm. 4 2025 pág. 1767
https://doi.org/10.69639/arandu.v12i4.1780
Using Grok to improve english speaking fluency in B1 English
as a Foreign Language Learners
El Uso de Grok para mejorar la fluidez en la expresión oral en estudiantes B1 del
inglés como lengua extranjera
Franklin Absalon Mora Preciado
franklinmora942@gmail.com
famorap@ube.edu.ec
https://orcid.org/0009-0008-6988-2845
Universidad Bolivariana del Ecuador
Durán-Ecuador
Isabel Batista Medina
isabelbm1956@gmail.com
https://orcid.org/0000-0002-1701-4957
Universidad de Oriente.Cuba
Josué Reinaldo Bonilla Tenesaca
jrbonillat@ube.edu.ec
https://orcid.org/0000-0002-6748-2345
Universidad Bolivariana del Ecuador
Durán-Ecuador
Artículo recibido: 10 octubre 2025 -Aceptado para publicación: 18 noviembre2025
Conflictos de intereses: Ninguno que declarar.
ABSTRACT
Many students of the B1 English level from the “Hispanoamericano” High School of Guayaquil-
Ecuador have difficulties speaking in English with fluency due to fear of making pronunciation
mistakes and because of lack of practice outside the school. This investigation was carried out
with 50 B1 EFL learners, who attended classes on Saturdays. This research was based on an action
research design supported with the Communicative Language Teaching approach. Additionally,
Grok, an Artificial Intelligence (AI) app was used as the main tool to improve fluency by applying
the Analysis- Design -Development -Implementation and Evaluation (ADDIE) Model.
Furthermore, data was collected through the use of survey, analysis of chats generated with the
voice input of students and the app. Moreover, the results showed students felt more relaxed and
motivated when practicing speaking with the app and showed they made fewer pauses, and their
answers were longer and more coherent. The survey indicated that the improvement in the trust
and motivation average increased from 3.2 to 4.2. Finally, the conclusions and implications
showed that the use of this app in a classroom improved fluency, lowered anxiety, and improved
metacognitive strategies. Therefore, this research shows that the use of Artificial Intelligence tools
Vol. 12/ Núm. 4 2025 pág. 1768
like Grok can be integrated in the teaching learning process of schools to improve English oral
fluency levels.
Keywords: English-speaking fluency, Grok, B1 learners, communicative language
teaching, metacognitive strategies
RESUMEN
Muchos estudiantes del nivel B1 de inglés del Colegio “Hispanoamericano “de Guayaquil-
Ecuador tienen dificultades hablando en inglés con fluidez debido al miedo de cometer errores de
pronunciación y por la falta de práctica fuera de la escuela. Esta investigación fue llevada a cabo
con 50 aprendices B1 de inglés como lengua extranjera que asisten a clases los sábados. La cual,
está basada en un diseño de investigación-acción apoyado con el enfoque de Enseñanza
Comunicativa de Lenguas. Además, Grok, una aplicación de Inteligencia Artificial (IA) fue usada
como la herramienta principal para mejorar la fluidez aplicando el Modelo de Análisis, Diseño,
Desarrollo, Implementación y Evaluación. Asimismo, los datos fueron recolectados a través del
uso de encuesta, análisis de chats generados con la entrada de voz de estudiantes y de la
aplicación. Igualmente, los resultados mostraron que los estudiantes se sintieron más cómodos y
motivados cuando practicaban hablar con la aplicación, hicieron menos pausas con la
herramienta, y sus respuestas fueron más largas y más coherentes. La encuesta indicó que la
mejora en la media de confianza y motivación aumentó de 3.2 a 4.2. Finalmente, las conclusiones
e implicaciones revelaron que el uso de la aplicación en un aula mejoró la fluidez, redujo la
ansiedad, y optimizó el uso de estrategias metacognitivas. Por lo tanto, esta investigación refleja
que el uso de herramientas de Inteligencia Artificial como Grok puede ser integrado al proceso
de enseñanza aprendizaje de las escuelas para mejorar los niveles de fluidez oral en inglés.
Palabras clave: fluidez oral en inglés, Grok, estudiantes B1, enseñanza comunicativa de
lenguas, estrategias metacognitivas
Todo el contenido de la Revista Científica Internacional Arandu UTIC publicado en este sitio está disponible bajo
licencia Creative Commons Atribution 4.0 International.
like Grok can be integrated in the teaching learning process of schools to improve English oral
fluency levels.
Keywords: English-speaking fluency, Grok, B1 learners, communicative language
teaching, metacognitive strategies
RESUMEN
Muchos estudiantes del nivel B1 de inglés del Colegio “Hispanoamericano “de Guayaquil-
Ecuador tienen dificultades hablando en inglés con fluidez debido al miedo de cometer errores de
pronunciación y por la falta de práctica fuera de la escuela. Esta investigación fue llevada a cabo
con 50 aprendices B1 de inglés como lengua extranjera que asisten a clases los sábados. La cual,
está basada en un diseño de investigación-acción apoyado con el enfoque de Enseñanza
Comunicativa de Lenguas. Además, Grok, una aplicación de Inteligencia Artificial (IA) fue usada
como la herramienta principal para mejorar la fluidez aplicando el Modelo de Análisis, Diseño,
Desarrollo, Implementación y Evaluación. Asimismo, los datos fueron recolectados a través del
uso de encuesta, análisis de chats generados con la entrada de voz de estudiantes y de la
aplicación. Igualmente, los resultados mostraron que los estudiantes se sintieron más cómodos y
motivados cuando practicaban hablar con la aplicación, hicieron menos pausas con la
herramienta, y sus respuestas fueron más largas y más coherentes. La encuesta indicó que la
mejora en la media de confianza y motivación aumentó de 3.2 a 4.2. Finalmente, las conclusiones
e implicaciones revelaron que el uso de la aplicación en un aula mejoró la fluidez, redujo la
ansiedad, y optimizó el uso de estrategias metacognitivas. Por lo tanto, esta investigación refleja
que el uso de herramientas de Inteligencia Artificial como Grok puede ser integrado al proceso
de enseñanza aprendizaje de las escuelas para mejorar los niveles de fluidez oral en inglés.
Palabras clave: fluidez oral en inglés, Grok, estudiantes B1, enseñanza comunicativa de
lenguas, estrategias metacognitivas
Todo el contenido de la Revista Científica Internacional Arandu UTIC publicado en este sitio está disponible bajo
licencia Creative Commons Atribution 4.0 International.
Vol. 12/ Núm. 4 2025 pág. 1769
INTRODUCTION
Fluent oral English is not just a skill; it is a ticket to the world. Yet countless EFL learners
have difficulties when trying to speak confidently. Reaching a B1 level is the bridge between
hesitation and global opportunity. Also, fluent English speaking opens global doors, but many
EFL learners have a difficult time speaking fluently. If people wish to succeed in many areas in
the world, they would need to be proficient, at least, in the B1 English-speaking level.
Examples of renowned authors are Krashen and Nation who have written about this matter.
Krashen (1982), pointed out the importance of comprehensible input in language acquisition and
stated that pronunciation and fluency come when learners practice meaningful tasks, instead of
only being corrected. Also, Nation (2001) stressed the importance of vocabulary in fluency,
signaling out that pronunciation and proficiency in speaking are made stronger with a large
number of known words.
Research on specific AI apps, like Grok, is still limited. Liu et al. (2025) explained that AI
tools, like ChatGPT, can help students learn English because they give feedback that helps them
improve speaking, writing, and thinking skills. Wangsa et al. (2024) indicated that even though
Grok is a relatively new AI tool, it is notable for its real-time interaction, and its features promise
to help improve English fluency, such features resemble a real person with sarcasm and an
excellent sense of humor, making the learning process very enjoyable. Thus, Grok seems to be a
promising AI tool for helping EFL learners improve their English-speaking fluency.
It was observed a problem in students of third-year BGU of the “Hispanoamericano” High
School of Guayaquil, Ecuador. The problem was that most of them could do all the macro skills
in the B1 English level, but they could not speak, and the reasons why they could not interact
orally were similar because they were nervous, insecure and felt embarrassed about
mispronouncing words in front of their classmates and teachers. These problems made learners
perform in English less proficiently in class and lose confidence in their speaking ability.
As Vygotsky (1934/1986) stated that speaking is what puts together thoughts and real
communication, that without a dialogue, cognition stays in the brain and does not come out. Also,
Dennis (2024) showed how AI conversational tools could give learners personal feedback to
improve English fluency. Likewise, Shikun et al. (2024) explained that AI chatbots help students
practice English speaking in a friendly manner, making learning more enjoyable without feeling
nervous or judged. In this study, the independent variable was the use of Grok and the dependent
variable was speaking fluency.
Thus, the main objective of this study is to demonstrate how this app, which is an AI-
powered tool, can help with its real-time conversational modes and abilities to improve English-
speaking fluency among third-year BGU B1 English level learners. This research aims to give a
possible solution to the mentioned problem above by providing a safe, interactive environment.
INTRODUCTION
Fluent oral English is not just a skill; it is a ticket to the world. Yet countless EFL learners
have difficulties when trying to speak confidently. Reaching a B1 level is the bridge between
hesitation and global opportunity. Also, fluent English speaking opens global doors, but many
EFL learners have a difficult time speaking fluently. If people wish to succeed in many areas in
the world, they would need to be proficient, at least, in the B1 English-speaking level.
Examples of renowned authors are Krashen and Nation who have written about this matter.
Krashen (1982), pointed out the importance of comprehensible input in language acquisition and
stated that pronunciation and fluency come when learners practice meaningful tasks, instead of
only being corrected. Also, Nation (2001) stressed the importance of vocabulary in fluency,
signaling out that pronunciation and proficiency in speaking are made stronger with a large
number of known words.
Research on specific AI apps, like Grok, is still limited. Liu et al. (2025) explained that AI
tools, like ChatGPT, can help students learn English because they give feedback that helps them
improve speaking, writing, and thinking skills. Wangsa et al. (2024) indicated that even though
Grok is a relatively new AI tool, it is notable for its real-time interaction, and its features promise
to help improve English fluency, such features resemble a real person with sarcasm and an
excellent sense of humor, making the learning process very enjoyable. Thus, Grok seems to be a
promising AI tool for helping EFL learners improve their English-speaking fluency.
It was observed a problem in students of third-year BGU of the “Hispanoamericano” High
School of Guayaquil, Ecuador. The problem was that most of them could do all the macro skills
in the B1 English level, but they could not speak, and the reasons why they could not interact
orally were similar because they were nervous, insecure and felt embarrassed about
mispronouncing words in front of their classmates and teachers. These problems made learners
perform in English less proficiently in class and lose confidence in their speaking ability.
As Vygotsky (1934/1986) stated that speaking is what puts together thoughts and real
communication, that without a dialogue, cognition stays in the brain and does not come out. Also,
Dennis (2024) showed how AI conversational tools could give learners personal feedback to
improve English fluency. Likewise, Shikun et al. (2024) explained that AI chatbots help students
practice English speaking in a friendly manner, making learning more enjoyable without feeling
nervous or judged. In this study, the independent variable was the use of Grok and the dependent
variable was speaking fluency.
Thus, the main objective of this study is to demonstrate how this app, which is an AI-
powered tool, can help with its real-time conversational modes and abilities to improve English-
speaking fluency among third-year BGU B1 English level learners. This research aims to give a
possible solution to the mentioned problem above by providing a safe, interactive environment.
Vol. 12/ Núm. 4 2025 pág. 1770
Consequently, to evaluate if Grok app can address this issue, the study used a mixed-
methods approach: a five-question Likert-scale, pre/post-tests to 50 third-year BGU students, who
were selected intentionally. Surprisingly, they were officially B1 on paper, but mute in oral
practice. The interventions consisted of six weekly interventions, each 25 minutes: two 10-minute
Grok voice sessions with a five-minute break in between, applying Communicative Language
Teaching (CLT), role-plays, and metacognitive self-checks to reflect on progress. Sessions were
scheduled weekly to prevent cognitive overload.
In summary, this study looks for a possible solution to a problem of many B1-level EFL
students: that is, those who can read, write, and listen in English, but they cannot communicate
orally, due to nervousness about being embarrassed for making pronunciation mistakes in front
of their peers and teachers. Thus, Grok, as an AI-power tool that can give real-time conversational
practice, has the benefits that this app talks like a real person. In turn, learners do not feel nervous
about making mistakes because they know that the app is not a person, who may criticize them if
they distort words, hence, creating a safe and confident environment to improve English-speaking
fluency.
On this respect, Mingyan et al. (2025) explained that AI mobile apps help students practice
English-speaking skills since they give instant feedback and personalized guidance. Likewise,
Quvanch et al. (2024) explained that anxiety makes many students panicky and lose
concentration. Finally, the novelty of this study is about Grok app, specifically, a relatively new
AI-powered application with interactive voice modes that resembles human conversations. Yet,
this study was done only in one group of 50 B1 learners in a specific school, therefore, more
research is needed to see the real impact of this app in different English-speaking levels and
places.
Study Rationale
This study was inspired by Grok, a tool created by xAI, Elon Musk’s company in 2023
which happens to give an interactive and enjoyable way to practice oral skills. As Mohammed et
al. (2025) explained that new AI tools such as ChatGPT, DeepSeek, and Grok make learning
English more creative and interesting, they also said that Grok is becoming a main player because
it gives real-time, engaging support for students’ communication practice.
Thus, many students in Guayaquil do not speak English well, possibly because they might
be shy or may not practice enough outside classes. Similarly, they do not rehearse beyond the
school classes, perhaps because they may not have the financial means to pay for private English
classes. According to INEC (2023) About 62.2% of households have internet, and “72.7% of
people aged 5+ used the internet in the last 12 months. So, Grok app is free, works on phones,
and no fast internet is needed, so this app was a good choice for these students because it is a
promising new AI tool with potential to help acquire the B1 English-speaking level.
Consequently, to evaluate if Grok app can address this issue, the study used a mixed-
methods approach: a five-question Likert-scale, pre/post-tests to 50 third-year BGU students, who
were selected intentionally. Surprisingly, they were officially B1 on paper, but mute in oral
practice. The interventions consisted of six weekly interventions, each 25 minutes: two 10-minute
Grok voice sessions with a five-minute break in between, applying Communicative Language
Teaching (CLT), role-plays, and metacognitive self-checks to reflect on progress. Sessions were
scheduled weekly to prevent cognitive overload.
In summary, this study looks for a possible solution to a problem of many B1-level EFL
students: that is, those who can read, write, and listen in English, but they cannot communicate
orally, due to nervousness about being embarrassed for making pronunciation mistakes in front
of their peers and teachers. Thus, Grok, as an AI-power tool that can give real-time conversational
practice, has the benefits that this app talks like a real person. In turn, learners do not feel nervous
about making mistakes because they know that the app is not a person, who may criticize them if
they distort words, hence, creating a safe and confident environment to improve English-speaking
fluency.
On this respect, Mingyan et al. (2025) explained that AI mobile apps help students practice
English-speaking skills since they give instant feedback and personalized guidance. Likewise,
Quvanch et al. (2024) explained that anxiety makes many students panicky and lose
concentration. Finally, the novelty of this study is about Grok app, specifically, a relatively new
AI-powered application with interactive voice modes that resembles human conversations. Yet,
this study was done only in one group of 50 B1 learners in a specific school, therefore, more
research is needed to see the real impact of this app in different English-speaking levels and
places.
Study Rationale
This study was inspired by Grok, a tool created by xAI, Elon Musk’s company in 2023
which happens to give an interactive and enjoyable way to practice oral skills. As Mohammed et
al. (2025) explained that new AI tools such as ChatGPT, DeepSeek, and Grok make learning
English more creative and interesting, they also said that Grok is becoming a main player because
it gives real-time, engaging support for students’ communication practice.
Thus, many students in Guayaquil do not speak English well, possibly because they might
be shy or may not practice enough outside classes. Similarly, they do not rehearse beyond the
school classes, perhaps because they may not have the financial means to pay for private English
classes. According to INEC (2023) About 62.2% of households have internet, and “72.7% of
people aged 5+ used the internet in the last 12 months. So, Grok app is free, works on phones,
and no fast internet is needed, so this app was a good choice for these students because it is a
promising new AI tool with potential to help acquire the B1 English-speaking level.
Vol. 12/ Núm. 4 2025 pág. 1771
For that reason, the researcher selected 50 students who could read, write, and listen in
English, but could not speak. This moderately sized group allowed to analyze progress using an
automated tool like Grok’s transcription through voice input and teacher observations. No matter
which voice mode was used, this tool knew a lot about many subjects and it could talk about any
topic at any level, so it was a good pedagogical tool to help people speak English more fluently.
The goal was to verify if Grok could help students improve their English speaking, because
many did not speak, perhaps because they get uneasy when they participate orally, and as a result,
they use short sentences, and make pauses saying “um” or “eh”, or they only answer with a “yes”
or a “no” words. So here is where this tool comes in handy and can act as a friend, waiter, or
teacher in voice conversations, so learners can practice real dialogs without feeling nervous.
This study adapts the methodological structure proposed by Aremu et al. (2025), who
applies the ADDIE model, Vygotsky’s ZPD, CLT, and Cognitive Load Theory in an Ecuadorian
EFL context. However, the present research is different because it is done on high school B1
learners, and utilizes Grok, an AI conversational tool.
ADDIE was the model followed to plan carefully the study which consists of Analysis,
Design, Development, Implementation, and Evaluation stages, and it was also included ideas from
experts:
Vygotsky (1978): Students learn better with help from someone more knowledgeable, like
Grok.
Hymes (1972): Speaking activities should be realistic, such as ordering food or asking for
directions, among other communicative functions.
Sweller (1988): Tasks should be simple, manageable to avoid overload and confusion.
Comparative reliability of voice modes: Grok vs. Chatgpt, Copilot, and Gemini
The apps Grok, ChatGPT, Copilot, and Gemini were observed by the researcher for over
six months, and based on the researcher’s personal experience, the app that resembled a human
being the most was Grok; also, it could keep a conversation for an unlimited period of time;
additionally, the app felt as if it could reason like a human. On the other hand, ChatGPT was
almost as good as Grok but it did not keep the conversation flowing; instead, the talk was stopped
by the statement “if there is anything else I can do for you, just let me know.” and finally, Copilot
and Gemini felt like machines; they acted automated. Therefore, the app that resembled the most
a human-like experience was Grok, thus, making the use of the tool more enjoyable.
Theoretical framework
The framework of this research was supported on prominent theories, like CLT, Vygotsky's
ZPD, and Krashen's Input Hypothesis, to explain why Grok app promotes the acquisition of
speaking English fluency and self-confidence.
For that reason, the researcher selected 50 students who could read, write, and listen in
English, but could not speak. This moderately sized group allowed to analyze progress using an
automated tool like Grok’s transcription through voice input and teacher observations. No matter
which voice mode was used, this tool knew a lot about many subjects and it could talk about any
topic at any level, so it was a good pedagogical tool to help people speak English more fluently.
The goal was to verify if Grok could help students improve their English speaking, because
many did not speak, perhaps because they get uneasy when they participate orally, and as a result,
they use short sentences, and make pauses saying “um” or “eh”, or they only answer with a “yes”
or a “no” words. So here is where this tool comes in handy and can act as a friend, waiter, or
teacher in voice conversations, so learners can practice real dialogs without feeling nervous.
This study adapts the methodological structure proposed by Aremu et al. (2025), who
applies the ADDIE model, Vygotsky’s ZPD, CLT, and Cognitive Load Theory in an Ecuadorian
EFL context. However, the present research is different because it is done on high school B1
learners, and utilizes Grok, an AI conversational tool.
ADDIE was the model followed to plan carefully the study which consists of Analysis,
Design, Development, Implementation, and Evaluation stages, and it was also included ideas from
experts:
Vygotsky (1978): Students learn better with help from someone more knowledgeable, like
Grok.
Hymes (1972): Speaking activities should be realistic, such as ordering food or asking for
directions, among other communicative functions.
Sweller (1988): Tasks should be simple, manageable to avoid overload and confusion.
Comparative reliability of voice modes: Grok vs. Chatgpt, Copilot, and Gemini
The apps Grok, ChatGPT, Copilot, and Gemini were observed by the researcher for over
six months, and based on the researcher’s personal experience, the app that resembled a human
being the most was Grok; also, it could keep a conversation for an unlimited period of time;
additionally, the app felt as if it could reason like a human. On the other hand, ChatGPT was
almost as good as Grok but it did not keep the conversation flowing; instead, the talk was stopped
by the statement “if there is anything else I can do for you, just let me know.” and finally, Copilot
and Gemini felt like machines; they acted automated. Therefore, the app that resembled the most
a human-like experience was Grok, thus, making the use of the tool more enjoyable.
Theoretical framework
The framework of this research was supported on prominent theories, like CLT, Vygotsky's
ZPD, and Krashen's Input Hypothesis, to explain why Grok app promotes the acquisition of
speaking English fluency and self-confidence.
Vol. 12/ Núm. 4 2025 pág. 1772
Communicative Language Teaching (CLT)
Richards (2006) explained that CLT is all about using the language to be able to
communicate in real-life situations, not just learning grammar rules. Also, as said by Hymes
(1972), that communicative competence is equal with being capable of knowing when and how
to use the language appropriately based on contexts. Additionally, with this method, students
interact meaningfully in English and learn through oral communication instead of memorizing
grammatical rules. Grok makes this possible through offering real conversation practice.
Sociocultural Theory of Learning
Vygotsky (1978, 1986) proposed in his theory that learning takes place through social
interaction and use of tools and that guidance within the Zone of Proximal Development (ZPD)
gives more productive learning. Because of this, Grok app goes a step further by acting as an
electronic tutor, giving feedback, correction, and illustrations. It allows students to move from
brief answers to more natural conversations. Also, it makes learners more confident and reduces
fear when speaking in English.
Input Hypothesis and Affective Filter
On the other hand, Krashen (1982) proposed that students learn language as they
understand messages just a little higher than their current level, and that anxiety or fear can stop
this process by the affective filter. Also, they can make the attempt to speaking in a relaxed
scenario with the aid of Grok. This is a proof that AI provides constructive criticism without
judging, reducing the affective filter and making communication easier and more natural.
Cognitive Load and Step-by-Step Learning
Sweller (1988) argued that instruction should be planned in a way to avoid cognitive
overload and advance from simple tasks to more difficult tasks progressively, in a scaffolding
manner. Additionally, the ADDIE model used in this study follows this guideline, so it started
with less complex activities and increased progressively to more difficult activities, and students
were given immediate feedback that helped them stay focused and confident during learning.
Artificial Intelligence in English Language Teaching
Shikun et al. (2024) indicated that AI chatbots could help in language development by
giving a low-stress place to practice. In this same line of reasoning, Liu et al. (2025) said that AI
tools motivated learners to engage and have critical thinking when used in EFL contexts. Also,
Dennis (2024) observed that AI speech recognition tools helped students with pronunciation and
fluency. Therefore, artificial intelligence tools like Grok give learners more chances to practice
English, receive instant feedback corrections, and learn at their own pace. Together with good
teaching, AI is a better and promising tool to help improve oral fluency.
Communicative Language Teaching (CLT)
Richards (2006) explained that CLT is all about using the language to be able to
communicate in real-life situations, not just learning grammar rules. Also, as said by Hymes
(1972), that communicative competence is equal with being capable of knowing when and how
to use the language appropriately based on contexts. Additionally, with this method, students
interact meaningfully in English and learn through oral communication instead of memorizing
grammatical rules. Grok makes this possible through offering real conversation practice.
Sociocultural Theory of Learning
Vygotsky (1978, 1986) proposed in his theory that learning takes place through social
interaction and use of tools and that guidance within the Zone of Proximal Development (ZPD)
gives more productive learning. Because of this, Grok app goes a step further by acting as an
electronic tutor, giving feedback, correction, and illustrations. It allows students to move from
brief answers to more natural conversations. Also, it makes learners more confident and reduces
fear when speaking in English.
Input Hypothesis and Affective Filter
On the other hand, Krashen (1982) proposed that students learn language as they
understand messages just a little higher than their current level, and that anxiety or fear can stop
this process by the affective filter. Also, they can make the attempt to speaking in a relaxed
scenario with the aid of Grok. This is a proof that AI provides constructive criticism without
judging, reducing the affective filter and making communication easier and more natural.
Cognitive Load and Step-by-Step Learning
Sweller (1988) argued that instruction should be planned in a way to avoid cognitive
overload and advance from simple tasks to more difficult tasks progressively, in a scaffolding
manner. Additionally, the ADDIE model used in this study follows this guideline, so it started
with less complex activities and increased progressively to more difficult activities, and students
were given immediate feedback that helped them stay focused and confident during learning.
Artificial Intelligence in English Language Teaching
Shikun et al. (2024) indicated that AI chatbots could help in language development by
giving a low-stress place to practice. In this same line of reasoning, Liu et al. (2025) said that AI
tools motivated learners to engage and have critical thinking when used in EFL contexts. Also,
Dennis (2024) observed that AI speech recognition tools helped students with pronunciation and
fluency. Therefore, artificial intelligence tools like Grok give learners more chances to practice
English, receive instant feedback corrections, and learn at their own pace. Together with good
teaching, AI is a better and promising tool to help improve oral fluency.
Vol. 12/ Núm. 4 2025 pág. 1773
METHOD
Participants
The research group consisted of 50 students, all had the same sociocultural background,
food, music, sports, and they were B1 English level, they could communicate simple ideas but
made mistakes and paused often. For example, a student may say “I go shop” instead of “I’m
going to the shop”. They were chosen because the English teacher knew them from class, and
they wanted to improve their speaking skills.
All participants were 18 and older, nevertheless, this research followed ethical rules to
protect learners, their names and their voice generated by Grok chat texts were kept private (their
real names were not used). Additionally, only the researcher saw the results, and the answers from
the Google Forms surveys were stored safely online. These steps helped make sure that the study
was honest, safe, and respectful for all students.
Table 1 below describes the population (which is the sample) before the interventions of
this research.
Table 1
Group Information
Strengths Weaknesses Opportunities Threats
Cohesive cohort of 50
adult learners (18+),
all Spanish-speaking,
easing Grok’s code-
switching (Table 3).
No prior AI tool
experience,
increasing learning
curve (Table 1).
Scalable to other
schools, with 72.7%
internet usage in
Ecuador (INEC,
2023).
Unreliable internet
or devices in low-
resource settings
(INEC, 2023).
High motivation for
university studies
drives engagement
(Table 1).
Traditional,
grammar-based
instruction lacks
CLT/CLIL
integration (Table
1).
Grok’s novelty invites
further EFL research
(Mohammed et al.,
2025).
Competition from
established AI tools
(e.g., ChatGPT)
(Comparative
Reliability).
Familiarity with
mobile apps
(WhatsApp, TikTok)
supports Grok use
(Table 1).
Limited prior
English exposure
may slow progress
(Table 1).
Potential curriculum
adoption due to
reduced anxiety and
improved oral fluency
(Table 7).
Privacy concerns
regarding voice data
storage (Ensuring
Ethical Standards).
Sampling procedure
For carrying out this research, 50 students were chosen on purpose, because their English
teacher stated that they needed help with speaking skill. Also, all of them lived in Guayaquil city
and spoke Spanish at home. Additionally, many did not have fast internet nor good computers at
home, so most of them used only their phones to interact with Grok. Additionally, their ages were
18 and older, however, parents’ consent was asked to align with ethical standards and guidelines
of the institution.
METHOD
Participants
The research group consisted of 50 students, all had the same sociocultural background,
food, music, sports, and they were B1 English level, they could communicate simple ideas but
made mistakes and paused often. For example, a student may say “I go shop” instead of “I’m
going to the shop”. They were chosen because the English teacher knew them from class, and
they wanted to improve their speaking skills.
All participants were 18 and older, nevertheless, this research followed ethical rules to
protect learners, their names and their voice generated by Grok chat texts were kept private (their
real names were not used). Additionally, only the researcher saw the results, and the answers from
the Google Forms surveys were stored safely online. These steps helped make sure that the study
was honest, safe, and respectful for all students.
Table 1 below describes the population (which is the sample) before the interventions of
this research.
Table 1
Group Information
Strengths Weaknesses Opportunities Threats
Cohesive cohort of 50
adult learners (18+),
all Spanish-speaking,
easing Grok’s code-
switching (Table 3).
No prior AI tool
experience,
increasing learning
curve (Table 1).
Scalable to other
schools, with 72.7%
internet usage in
Ecuador (INEC,
2023).
Unreliable internet
or devices in low-
resource settings
(INEC, 2023).
High motivation for
university studies
drives engagement
(Table 1).
Traditional,
grammar-based
instruction lacks
CLT/CLIL
integration (Table
1).
Grok’s novelty invites
further EFL research
(Mohammed et al.,
2025).
Competition from
established AI tools
(e.g., ChatGPT)
(Comparative
Reliability).
Familiarity with
mobile apps
(WhatsApp, TikTok)
supports Grok use
(Table 1).
Limited prior
English exposure
may slow progress
(Table 1).
Potential curriculum
adoption due to
reduced anxiety and
improved oral fluency
(Table 7).
Privacy concerns
regarding voice data
storage (Ensuring
Ethical Standards).
Sampling procedure
For carrying out this research, 50 students were chosen on purpose, because their English
teacher stated that they needed help with speaking skill. Also, all of them lived in Guayaquil city
and spoke Spanish at home. Additionally, many did not have fast internet nor good computers at
home, so most of them used only their phones to interact with Grok. Additionally, their ages were
18 and older, however, parents’ consent was asked to align with ethical standards and guidelines
of the institution.
Vol. 12/ Núm. 4 2025 pág. 1774
Research design
The interventions combined two strategies, metacognitive and spaced learning.
Metacognitive Strategy
A metacognitive strategy is a three-step cycle, which was used for planning, monitoring,
and reflecting, to improve students’ speaking skill. According to Meher et al. (2024),
metacognition helps learners think about their own learning, and use strategies like thinking aloud,
making idea- chart, and self-assessing to improve their performance in the school.
Planning (before speaking): Before each 10-minute Grok session, students wrote one
personal goal on paper, for example: “I will speak in full sentences without saying ‘um’ or ‘eh’.”
Monitoring (while speaking): After five minutes, they stopped for ten seconds to ask
themselves: “Am I reaching my goal? Am I too nervous?” Then, they made small changes and
continued speaking.
Reflection (after speaking): After finishing, they answered three short questions:
How many pauses did I make?
Did I reach my goal?
What can I do better next time?
Table 2 below indicates the metacognitive steps with examples.
Table 2
Metacognitive Stages
Stage When What Students Did Example
Planning Before speaking Set a goal “I will speak in full sentences.”
Monitoring During speaking Checked progress “Am I too nervous?”
Reflection After speaking Reviewed performance “Next time I’ll pause less.”
Spaced Learning Strategy
The spaced learning strategy is about studying or practicing in short parts with breaks in
between, instead of studying in one long, single session without taking a break. As Kamali et al.
(2024) explained that using spaced learning, with reviewed lessons and short breaks in between
helps students remember information better and feel less nervous about learning. In EFL learning,
it helps learners remember English and not feel tired or bored, also, the breaks in between short
study sessions give the brain time to rest and store the new information, so one example of spaced
learning is when one learner studies for ten minutes, rests for five minutes, and then reviews again
for another ten minutes or when they practice English speaking with Grok to speak English in
two ten-minute sessions, with a five-minute break in between, in total, the practice lasted twenty-
five minutes. Also, participants are encouraged to do the same practices at home to make sure
what has been learned stays in their memory.
Research design
The interventions combined two strategies, metacognitive and spaced learning.
Metacognitive Strategy
A metacognitive strategy is a three-step cycle, which was used for planning, monitoring,
and reflecting, to improve students’ speaking skill. According to Meher et al. (2024),
metacognition helps learners think about their own learning, and use strategies like thinking aloud,
making idea- chart, and self-assessing to improve their performance in the school.
Planning (before speaking): Before each 10-minute Grok session, students wrote one
personal goal on paper, for example: “I will speak in full sentences without saying ‘um’ or ‘eh’.”
Monitoring (while speaking): After five minutes, they stopped for ten seconds to ask
themselves: “Am I reaching my goal? Am I too nervous?” Then, they made small changes and
continued speaking.
Reflection (after speaking): After finishing, they answered three short questions:
How many pauses did I make?
Did I reach my goal?
What can I do better next time?
Table 2 below indicates the metacognitive steps with examples.
Table 2
Metacognitive Stages
Stage When What Students Did Example
Planning Before speaking Set a goal “I will speak in full sentences.”
Monitoring During speaking Checked progress “Am I too nervous?”
Reflection After speaking Reviewed performance “Next time I’ll pause less.”
Spaced Learning Strategy
The spaced learning strategy is about studying or practicing in short parts with breaks in
between, instead of studying in one long, single session without taking a break. As Kamali et al.
(2024) explained that using spaced learning, with reviewed lessons and short breaks in between
helps students remember information better and feel less nervous about learning. In EFL learning,
it helps learners remember English and not feel tired or bored, also, the breaks in between short
study sessions give the brain time to rest and store the new information, so one example of spaced
learning is when one learner studies for ten minutes, rests for five minutes, and then reviews again
for another ten minutes or when they practice English speaking with Grok to speak English in
two ten-minute sessions, with a five-minute break in between, in total, the practice lasted twenty-
five minutes. Also, participants are encouraged to do the same practices at home to make sure
what has been learned stays in their memory.
Vol. 12/ Núm. 4 2025 pág. 1775
The intervention consisted of six weekly interventions, each 25 minutes: two 10-minute
Grok voice sessions with a five-minute break in between, conducted on Saturdays, applying
Communicative Language Teaching (CLT) role-plays and metacognitive self-checks to reflect on
progress. Then, after each intervention, students were encouraged to practice speaking at home
using Grok until they felt comfortable, as a strategy to overcome nervousness of speaking in front
of people. The researcher used three main ways to measure progress: surveys before and after the
study, a tool to analyze their chat conversations.
The English teacher was asked to apply the Grok activities throughout all the six-week
period with key CLT principles through meaningful tasks and students’ interaction. Richards
(2006) states that CLT encourages real-life communication and meaningful language use.
There were 12 different voice chats or topics chosen by the teacher, two per week, over a
six-week intervention period, each with Grok acting in a different role, such as a waiter, friend,
or travel agent. The app also has different voice modes that can be applied to each role. The
interventions took place in the virtual classroom. The teacher used his phone as a microphone,
then called on learners one by one to ask questions to the app according to the tasks from Table 3
below. Each intervention lasted 10 minutes with a five-minute recess in between. All
participations were saved by the tool in written form, which answered and corrected mistakes, for
example, changing “I want pizza” to “I’d like pizza.” Students also practiced at home, and
afterward tips were sent via WhatsApp, such as “Try using longer sentences.”
Table 3 below indicates the use of CAP through 12 voice chats activities for 50 students
through
interaction with Grok.
Table 3
Plan for role plays using CAP tool
Talk
Number
Topic Grok’s Role Student’s Task What We Checked
through CAP
1 Greetings Friend Say who you are Pauses, words, logic
2 Travel Travel agent Book a trip Pauses, words, logic
3 Food Waiter Order food Pauses, words, logic
4 Family Cousin Talk about your family Pauses, words, logic
5 Shopping Store worker Ask for clothes Pauses, words, logic
6 Weather News
reporter
Talk about weather Pauses, words, logic
7 Home Neighbor Describe your house Pauses, words, logic
8 School Teacher Talk about school Pauses, words, logic
9 Job Interview Boss Answer job questions Pauses, words, logic
The intervention consisted of six weekly interventions, each 25 minutes: two 10-minute
Grok voice sessions with a five-minute break in between, conducted on Saturdays, applying
Communicative Language Teaching (CLT) role-plays and metacognitive self-checks to reflect on
progress. Then, after each intervention, students were encouraged to practice speaking at home
using Grok until they felt comfortable, as a strategy to overcome nervousness of speaking in front
of people. The researcher used three main ways to measure progress: surveys before and after the
study, a tool to analyze their chat conversations.
The English teacher was asked to apply the Grok activities throughout all the six-week
period with key CLT principles through meaningful tasks and students’ interaction. Richards
(2006) states that CLT encourages real-life communication and meaningful language use.
There were 12 different voice chats or topics chosen by the teacher, two per week, over a
six-week intervention period, each with Grok acting in a different role, such as a waiter, friend,
or travel agent. The app also has different voice modes that can be applied to each role. The
interventions took place in the virtual classroom. The teacher used his phone as a microphone,
then called on learners one by one to ask questions to the app according to the tasks from Table 3
below. Each intervention lasted 10 minutes with a five-minute recess in between. All
participations were saved by the tool in written form, which answered and corrected mistakes, for
example, changing “I want pizza” to “I’d like pizza.” Students also practiced at home, and
afterward tips were sent via WhatsApp, such as “Try using longer sentences.”
Table 3 below indicates the use of CAP through 12 voice chats activities for 50 students
through
interaction with Grok.
Table 3
Plan for role plays using CAP tool
Talk
Number
Topic Grok’s Role Student’s Task What We Checked
through CAP
1 Greetings Friend Say who you are Pauses, words, logic
2 Travel Travel agent Book a trip Pauses, words, logic
3 Food Waiter Order food Pauses, words, logic
4 Family Cousin Talk about your family Pauses, words, logic
5 Shopping Store worker Ask for clothes Pauses, words, logic
6 Weather News
reporter
Talk about weather Pauses, words, logic
7 Home Neighbor Describe your house Pauses, words, logic
8 School Teacher Talk about school Pauses, words, logic
9 Job Interview Boss Answer job questions Pauses, words, logic
Vol. 12/ Núm. 4 2025 pág. 1776
Talk
Number
Topic Grok’s Role Student’s Task What We Checked
through CAP
10 Hobbies Friend Share what you like Pauses, words, logic
11 Daily
Routine
Coworker Describe your day Pauses, words, logic
12 Review Counselor Say what you learned Pauses, words, logic
The researcher used the ADDIE model to plan the study that stands for Analysis, Design,
Development, Implementation, and Evaluation stages.
Analysis
The researcher first talked with students using Grok to understand why they did not speak
English well, so learners said they were afraid of making mistakes and did not practice outside
class, then the researcher also checked school records and saw that less than half of them did not
have a computer at home, so the app was a good choice because it could work on phones and did
not need fast internet and it was user friendly.
Design
The researcher created 12 B1 level talks topics, such as food, travel, or family, and, each
talk was short so that students would not get bored, then Grok asked easy questions at first and
not too easy questions later on, to help them learn step by step.
Development
The tasks were elaborated in a way that they were not too difficult for students. For
example, “Do you like pizza?” is easier to answer than “Explain why do you like pizza?” or it is
easier to ask “Do you have a brother and a sister?” than to ask “Do you have any siblings?”
because the word siblings is not too common for most beginners, then the app was used to act as
a partner in the role play activities, at the same time, as the tasks developed, the app also corrected
mistakes by saying “perhaps you meant to say this”, for example, if a student said “they does”,
the app would correct the mistake by saying “perhaps you meant to say ‘they do’”.
Implementation
The implementation was a straightforward process, the teacher would connect his phone to
a speaker, which would serve as a microphone, also the phone would have the app ready in the
conversation mode, so at this point the app could listen and answer to anybody speaking in the
classroom, and so the teacher would point out each student to participate by asking a question to
the app according to the topics from Table 3, then the app would answer what students asked and
also would give feedback.
Talk
Number
Topic Grok’s Role Student’s Task What We Checked
through CAP
10 Hobbies Friend Share what you like Pauses, words, logic
11 Daily
Routine
Coworker Describe your day Pauses, words, logic
12 Review Counselor Say what you learned Pauses, words, logic
The researcher used the ADDIE model to plan the study that stands for Analysis, Design,
Development, Implementation, and Evaluation stages.
Analysis
The researcher first talked with students using Grok to understand why they did not speak
English well, so learners said they were afraid of making mistakes and did not practice outside
class, then the researcher also checked school records and saw that less than half of them did not
have a computer at home, so the app was a good choice because it could work on phones and did
not need fast internet and it was user friendly.
Design
The researcher created 12 B1 level talks topics, such as food, travel, or family, and, each
talk was short so that students would not get bored, then Grok asked easy questions at first and
not too easy questions later on, to help them learn step by step.
Development
The tasks were elaborated in a way that they were not too difficult for students. For
example, “Do you like pizza?” is easier to answer than “Explain why do you like pizza?” or it is
easier to ask “Do you have a brother and a sister?” than to ask “Do you have any siblings?”
because the word siblings is not too common for most beginners, then the app was used to act as
a partner in the role play activities, at the same time, as the tasks developed, the app also corrected
mistakes by saying “perhaps you meant to say this”, for example, if a student said “they does”,
the app would correct the mistake by saying “perhaps you meant to say ‘they do’”.
Implementation
The implementation was a straightforward process, the teacher would connect his phone to
a speaker, which would serve as a microphone, also the phone would have the app ready in the
conversation mode, so at this point the app could listen and answer to anybody speaking in the
classroom, and so the teacher would point out each student to participate by asking a question to
the app according to the topics from Table 3, then the app would answer what students asked and
also would give feedback.
Vol. 12/ Núm. 4 2025 pág. 1777
Evaluation
The researcher analyzed surveys and chat data. The same survey was applied at the
beginning of the six-week study and at the end so a comparison of the results of the survey were
made. Also, the voice chats in written form were analyzed using CAP.
The ADDIE model helped the researcher plan the study in a clear and a step by step
organized way, from understanding where students’ English knowledge was at the beginning of
the interventions, to checking their progress, to the final results, to aid the researcher use the app
better to help improve fluency. This model facilitated understanding how students were more
willing to speak and take part in English conversations. And also, the app’s features like
corrections, practice, and easy design helped them keep learning. Finally, the process showed the
tool can be useful and motivating to improve speaking fluency in B1-level learners.
Instruments
The researcher used two main tools to see if students improved their English speaking:
Survey (Before and After)
Students were given a short survey with five statements to understand how they felt about
speaking English using Grok. The statements used a one to five scale, where one is “I totally
disagree” and five is “I totally agree.” The questions were:
I feel good using Grok to practice English.
I want to speak English more with Grok.
I feel confident when I speak English.
I use correct grammar when I talk with Grok.
Talking with Grok helps me use English in real life.
The same survey was sent through Google Forms, and all 50 students answered both before and
after the study.
Conversation Analysis Protocol (CAP)
This tool analyzed how learners spoke in Grok chats. The researcher created the CAP tool based
on voice chats. It was measured three things:
Pauses: How often students say “um” or “eh.”
Turn Length: How many words they use each time they answer.
Logic Level:
Level 1: Just answer the question.
Level 2: Answer and ask a question back.
Level 3: Start new ideas.
Table 4 below describes the voice modes of Grok.
Evaluation
The researcher analyzed surveys and chat data. The same survey was applied at the
beginning of the six-week study and at the end so a comparison of the results of the survey were
made. Also, the voice chats in written form were analyzed using CAP.
The ADDIE model helped the researcher plan the study in a clear and a step by step
organized way, from understanding where students’ English knowledge was at the beginning of
the interventions, to checking their progress, to the final results, to aid the researcher use the app
better to help improve fluency. This model facilitated understanding how students were more
willing to speak and take part in English conversations. And also, the app’s features like
corrections, practice, and easy design helped them keep learning. Finally, the process showed the
tool can be useful and motivating to improve speaking fluency in B1-level learners.
Instruments
The researcher used two main tools to see if students improved their English speaking:
Survey (Before and After)
Students were given a short survey with five statements to understand how they felt about
speaking English using Grok. The statements used a one to five scale, where one is “I totally
disagree” and five is “I totally agree.” The questions were:
I feel good using Grok to practice English.
I want to speak English more with Grok.
I feel confident when I speak English.
I use correct grammar when I talk with Grok.
Talking with Grok helps me use English in real life.
The same survey was sent through Google Forms, and all 50 students answered both before and
after the study.
Conversation Analysis Protocol (CAP)
This tool analyzed how learners spoke in Grok chats. The researcher created the CAP tool based
on voice chats. It was measured three things:
Pauses: How often students say “um” or “eh.”
Turn Length: How many words they use each time they answer.
Logic Level:
Level 1: Just answer the question.
Level 2: Answer and ask a question back.
Level 3: Start new ideas.
Table 4 below describes the voice modes of Grok.
Vol. 12/ Núm. 4 2025 pág. 1778
Table 4
Grok’s voice modes
Voice Mode Description Platform
Availability
Key Strengths
for EFL
Learners
Potential
Drawbacks
Assistant General-purpose
mode with a neutral
American accent;
handles broad
topics.
Computer (6
total),
Mobile (13
total)
Versatile for
everyday
practice; covers
all topics at any
CEFR level.
None notable—
safe and reliable.
Unhinged
Comedian
Upbeat, humorous
tone; excels at jokes
but may use
profanity.
Both Builds fluency
through fun,
casual dialogue;
encourages
expressive
speaking.
Profanity may
not suit
formal/school
settings.
Grok ‘Doc’ Doctor-like person;
provides
health/medical
advice with deep
knowledge.
Both Useful for
specialized
vocabulary (e.g.,
medical terms);
simulates real
consultations.
Advice is
informational,
not a substitute
for professionals.
Argumentative Debates topics,
provides reasoned
arguments, and
encourages user
input.
Both Promotes critical
thinking and
debate skills;
great for
advanced levels
(B2+).
Can feel
confrontational
for beginners.
(Other Modes) Additional modes
(e.g., up to 13 on
mobile) include
variations like
storyteller or expert
roles (not detailed
here).
Mobile-
exclusive for
extras
Expands role-
playing for
diverse scenarios,
improving
adaptability.
Availability may
vary by update.
Table 4
Grok’s voice modes
Voice Mode Description Platform
Availability
Key Strengths
for EFL
Learners
Potential
Drawbacks
Assistant General-purpose
mode with a neutral
American accent;
handles broad
topics.
Computer (6
total),
Mobile (13
total)
Versatile for
everyday
practice; covers
all topics at any
CEFR level.
None notable—
safe and reliable.
Unhinged
Comedian
Upbeat, humorous
tone; excels at jokes
but may use
profanity.
Both Builds fluency
through fun,
casual dialogue;
encourages
expressive
speaking.
Profanity may
not suit
formal/school
settings.
Grok ‘Doc’ Doctor-like person;
provides
health/medical
advice with deep
knowledge.
Both Useful for
specialized
vocabulary (e.g.,
medical terms);
simulates real
consultations.
Advice is
informational,
not a substitute
for professionals.
Argumentative Debates topics,
provides reasoned
arguments, and
encourages user
input.
Both Promotes critical
thinking and
debate skills;
great for
advanced levels
(B2+).
Can feel
confrontational
for beginners.
(Other Modes) Additional modes
(e.g., up to 13 on
mobile) include
variations like
storyteller or expert
roles (not detailed
here).
Mobile-
exclusive for
extras
Expands role-
playing for
diverse scenarios,
improving
adaptability.
Availability may
vary by update.
Vol. 12/ Núm. 4 2025 pág. 1779
Table 5 below describes the voice tones of Grok.
Table 5
Grok’s voice tones
Tone
Category
Options Description Benefits for EFL
Female Two variations (e.g.,
warm/neutral)
Softer, engaging tones
that can adapt to
conversational flow.
Encourages comfort for
shyer learners; aids
pronunciation practice.
Male Two variations (e.g.,
authoritative/casual)
Deeper, varied pitches for
a broader range of
simulated interactions.
Builds confidence in
diverse dialogues;
mimics real-world
variety.
RESULTS
Data Storage Approach
The researcher saved two types of data:
Surveys: Students answered surveys on Google Forms, the results were saved on Google Sheets.
Students answered the same survey at the beginning of the study and at the end.
Voice Chats in written form: Grok saved 600 voice chats in written form (50 students × 12 talks),
these chats were later pasted on a word document. Each chat included:
What the student said
Grok’s correction
Metrics: pauses, words, and logic level
Example:
Student Ana did Talk 3 from Table 3: “I want pizza, uh, please.”
Grok: “Say ‘I would like a pizza, please.’”
Metrics: 1 pause, 4 words, Logic Level 1
To demonstrate individual progress, see Table 6 below: Ana and Juan's voice chats before and
after Grok intervention.
Table 6
Example Chat stored
Student Talk Student’s
Task
Student’s
Sentence
Grok’s
Corrections
Pauses Words Logic
Ana 3 Order food I want pizza,
uh, please
Say, “I would
like pizza,
please.”
1 4 1
Ana 12 Say what
you learned
I learned to
speak better
without ums
None needed 0 8 2
Juan 6 Talk
weather
It rain, um,
bad
Say, “It is
raining.”
1 4 1
Table 5 below describes the voice tones of Grok.
Table 5
Grok’s voice tones
Tone
Category
Options Description Benefits for EFL
Female Two variations (e.g.,
warm/neutral)
Softer, engaging tones
that can adapt to
conversational flow.
Encourages comfort for
shyer learners; aids
pronunciation practice.
Male Two variations (e.g.,
authoritative/casual)
Deeper, varied pitches for
a broader range of
simulated interactions.
Builds confidence in
diverse dialogues;
mimics real-world
variety.
RESULTS
Data Storage Approach
The researcher saved two types of data:
Surveys: Students answered surveys on Google Forms, the results were saved on Google Sheets.
Students answered the same survey at the beginning of the study and at the end.
Voice Chats in written form: Grok saved 600 voice chats in written form (50 students × 12 talks),
these chats were later pasted on a word document. Each chat included:
What the student said
Grok’s correction
Metrics: pauses, words, and logic level
Example:
Student Ana did Talk 3 from Table 3: “I want pizza, uh, please.”
Grok: “Say ‘I would like a pizza, please.’”
Metrics: 1 pause, 4 words, Logic Level 1
To demonstrate individual progress, see Table 6 below: Ana and Juan's voice chats before and
after Grok intervention.
Table 6
Example Chat stored
Student Talk Student’s
Task
Student’s
Sentence
Grok’s
Corrections
Pauses Words Logic
Ana 3 Order food I want pizza,
uh, please
Say, “I would
like pizza,
please.”
1 4 1
Ana 12 Say what
you learned
I learned to
speak better
without ums
None needed 0 8 2
Juan 6 Talk
weather
It rain, um,
bad
Say, “It is
raining.”
1 4 1
Vol. 12/ Núm. 4 2025 pág. 1780
Juan 12 Say what
you learned
I talk more
now, what is
your job?
None needed 0 8 2
Data Analysis
The researcher examined the data in simple ways to see if students improved.
In surveys, the average (mean) was calculated, the median, the mode which is the most
common answer, and the standard deviation to see how much the answers varied. For example,
students’ confidence started with a mean of 3.2 and increased to 4.2 after using Grok.
On the voice chats, which are the recorded voice inputs, pauses like “um” were counted,
number of words per answer, and logic level.
The focus centered on showing changes in students’ speaking, chats were read to find
improvements, for example, “I am going” instead of “I going”. And the survey was analyzed to
see the level of comfort they had in speaking after using the app as shown in Table 7 below.
This table indicates the survey results, calculating the mean, median, mode, and standard
deviation before and after the use of Grok.
Table 7
Survey Results
Question Befor
e
Mean
Before
Media
n
Befor
e
Mode
Befor
e SD
After
Mea
n
After
Media
n
After
Mod
e
Afte
r SD
Chang
e
Comfort
with Grok
3.3 3 3 0.8 4.1 4 4 0.9 +0.8
Motivatio
n
3.5 4 4 0.7 4.3 4 5 0.8 +0.8
Confidenc
e
3.1 3 3 0.9 4.3 4 4 0.8 +1.0
Grammar 3.0 3 3 0.8 3.9 4 4 0.7 +0.9
Real-Life
Use
3.4 3 3 0.9 4.0 4 4 0.8 +0.6
Chart 1 below shows the mean scores from Table 7.
Juan 12 Say what
you learned
I talk more
now, what is
your job?
None needed 0 8 2
Data Analysis
The researcher examined the data in simple ways to see if students improved.
In surveys, the average (mean) was calculated, the median, the mode which is the most
common answer, and the standard deviation to see how much the answers varied. For example,
students’ confidence started with a mean of 3.2 and increased to 4.2 after using Grok.
On the voice chats, which are the recorded voice inputs, pauses like “um” were counted,
number of words per answer, and logic level.
The focus centered on showing changes in students’ speaking, chats were read to find
improvements, for example, “I am going” instead of “I going”. And the survey was analyzed to
see the level of comfort they had in speaking after using the app as shown in Table 7 below.
This table indicates the survey results, calculating the mean, median, mode, and standard
deviation before and after the use of Grok.
Table 7
Survey Results
Question Befor
e
Mean
Before
Media
n
Befor
e
Mode
Befor
e SD
After
Mea
n
After
Media
n
After
Mod
e
Afte
r SD
Chang
e
Comfort
with Grok
3.3 3 3 0.8 4.1 4 4 0.9 +0.8
Motivatio
n
3.5 4 4 0.7 4.3 4 5 0.8 +0.8
Confidenc
e
3.1 3 3 0.9 4.3 4 4 0.8 +1.0
Grammar 3.0 3 3 0.8 3.9 4 4 0.7 +0.9
Real-Life
Use
3.4 3 3 0.9 4.0 4 4 0.8 +0.6
Chart 1 below shows the mean scores from Table 7.
Vol. 12/ Núm. 4 2025 pág. 1781
Chart 1
Survey Changes
Chat analysis results
The results from the chats show that there was an overall fluency improvement, during the
first interventions, students used short sentences with mistakes and many pauses like “I go uh
shopping” instead of “I am going shopping” and towards the end of the intervention period, they
used more words per answers with less pauses, and some of them showed logic level 2 according
to the use of CAP analysis as shown in Table 8. Grok helped by correcting mistakes and giving
feedback.
Table 8 below shows the chat results through the use of CAP.
Table 8
Chat Results using CAP
Talk Number Pauses (Mean) Words per Answer (Mean) Logic Level (Mean)
1 1.2 4.2 1.0
3 0.9 5.5 1.1
6 0.6 6.3 1.3
9 0.5 7.0 1.5
12 0.4 7.8 1.8
Chart 2 below shows progress from voice chat converted to text by Grok.
3,3
3,5
3,1
3
3,4
4,1
4,3
4,3
3,9
4
Survey Results showing the Mean
Comfort
Motivation
Confidence
Grammar
Real-life Use
Chart 1
Survey Changes
Chat analysis results
The results from the chats show that there was an overall fluency improvement, during the
first interventions, students used short sentences with mistakes and many pauses like “I go uh
shopping” instead of “I am going shopping” and towards the end of the intervention period, they
used more words per answers with less pauses, and some of them showed logic level 2 according
to the use of CAP analysis as shown in Table 8. Grok helped by correcting mistakes and giving
feedback.
Table 8 below shows the chat results through the use of CAP.
Table 8
Chat Results using CAP
Talk Number Pauses (Mean) Words per Answer (Mean) Logic Level (Mean)
1 1.2 4.2 1.0
3 0.9 5.5 1.1
6 0.6 6.3 1.3
9 0.5 7.0 1.5
12 0.4 7.8 1.8
Chart 2 below shows progress from voice chat converted to text by Grok.
3,3
3,5
3,1
3
3,4
4,1
4,3
4,3
3,9
4
Survey Results showing the Mean
Comfort
Motivation
Confidence
Grammar
Real-life Use
Vol. 12/ Núm. 4 2025 pág. 1782
Chart 2
Chat Progress
Study Observations
The researcher checked each student’s chats carefully as shown in Table 9 below, Maria
had six pauses in Talk one in the first week but only one pause in Talk 12 on the sixth week.
Carlos used three words per answer first week, but eight words by the last week, and, he
even asked Grok an additional question: “What is your job?”, showing Level 2 logic according to
CAP.
Table 9 below indicates the results for 10 students from the 50-student population using
CAP.
Table 9
Group Progress
Chart 3 below shows examples of individual word count per answer, from week one chat
blue to week six chat orange.
1,1 0,9 0,7 0,5 0,4
4,1
5,5 6,2 7 7,9
TALK 1 TALK 3 TALK 6 TALK 9 TALK 12
Progress in Grok Chats
Pauses per minute Words per answer
Student Pauses
Week One
Pauses
Week Six
Words
Week One
Words
Week Six
Logic
Week
One
Logic
Week
Six
Ana 1.2 0.4 4.0 7.5 1.0 1.8
Juan 1.3 0.5 4.5 8.0 1.0 2.0
Maria 1.5 0.3 3.8 7.2 1.0 1.7
Carlos 1.1 0.4 4.2 7.8 1.0 2.0
Sofia 1.0 0.3 4.0 7.0 1.0 1.5
Pedro 1.4 0.5 4.3 7.6 1.0 1.8
Valeria 1.2 0.4 3.9 7.4 1.0 1.6
Diego 1.3 0.3 4.1 7.9 1.0 1.9
Camila 1.1 0.4 4.0 7.3 1.0 1.7
Luis 1.2 0.5 4.2 7.7 1.0 1.8
Chart 2
Chat Progress
Study Observations
The researcher checked each student’s chats carefully as shown in Table 9 below, Maria
had six pauses in Talk one in the first week but only one pause in Talk 12 on the sixth week.
Carlos used three words per answer first week, but eight words by the last week, and, he
even asked Grok an additional question: “What is your job?”, showing Level 2 logic according to
CAP.
Table 9 below indicates the results for 10 students from the 50-student population using
CAP.
Table 9
Group Progress
Chart 3 below shows examples of individual word count per answer, from week one chat
blue to week six chat orange.
1,1 0,9 0,7 0,5 0,4
4,1
5,5 6,2 7 7,9
TALK 1 TALK 3 TALK 6 TALK 9 TALK 12
Progress in Grok Chats
Pauses per minute Words per answer
Student Pauses
Week One
Pauses
Week Six
Words
Week One
Words
Week Six
Logic
Week
One
Logic
Week
Six
Ana 1.2 0.4 4.0 7.5 1.0 1.8
Juan 1.3 0.5 4.5 8.0 1.0 2.0
Maria 1.5 0.3 3.8 7.2 1.0 1.7
Carlos 1.1 0.4 4.2 7.8 1.0 2.0
Sofia 1.0 0.3 4.0 7.0 1.0 1.5
Pedro 1.4 0.5 4.3 7.6 1.0 1.8
Valeria 1.2 0.4 3.9 7.4 1.0 1.6
Diego 1.3 0.3 4.1 7.9 1.0 1.9
Camila 1.1 0.4 4.0 7.3 1.0 1.7
Luis 1.2 0.5 4.2 7.7 1.0 1.8
Vol. 12/ Núm. 4 2025 pág. 1783
Chart 3
Group Progress
Study limitations
The population-sample for this research was only 50 students, so more study needs to be
done on this.
Grok saves the voice input into written form, so the results from home practices could not
be heard, they could only be read.
Ensuring ethical standards
Parents were asked to sign consent forms for students, but they could leave the study at any
time. We used fake names, like Ana or Juan, to protect privacy. All chats were kept on a locked
phone and only the researcher and the English teacher had access to it, and all procedures
complied with the ethical guidelines and regulations of the “Hispanoamericano” High School and
the Universidad Bolivariana del Ecuador to make sure the protection and confidentiality of student
information.
Grok’s Effectiveness
Grok was helpful because it is fast, free, funny, easy to use and learners liked talking to it
because it is not a real person, it is a “robot”, so they did not feel judged or embarrassed of making
pronunciation mistakes as they would in front of real people.
Surveys showed that 80% of students felt comfortable using the tool, and 86% felt more
confident when speaking English.
Chats showed that students used more words and paused less over time.
DISCUSSION
Grok has many features, but the ones that are of interest for this research are the
conversational ones, and it helps to point out that these features are almost identical, either in a
computer or in a mobile phone. The only difference is that in a phone, the app has more “voice
modes (roles)”, but in spite of this, the app is the same. So, Grok’s conversational features are
4 4,5 3,8 4,2 4 4,3 3,9 4,1 4 4,2
7,5 8 7,2 7,8 7 7,6 7,4 7,9 7,3 7,7
ANA JUAN MARÍA CARLOS SOFÍA PEDRO VALERIA DIEGO CAMILA LUIS
Words per Answer by Student
Start words End words
Chart 3
Group Progress
Study limitations
The population-sample for this research was only 50 students, so more study needs to be
done on this.
Grok saves the voice input into written form, so the results from home practices could not
be heard, they could only be read.
Ensuring ethical standards
Parents were asked to sign consent forms for students, but they could leave the study at any
time. We used fake names, like Ana or Juan, to protect privacy. All chats were kept on a locked
phone and only the researcher and the English teacher had access to it, and all procedures
complied with the ethical guidelines and regulations of the “Hispanoamericano” High School and
the Universidad Bolivariana del Ecuador to make sure the protection and confidentiality of student
information.
Grok’s Effectiveness
Grok was helpful because it is fast, free, funny, easy to use and learners liked talking to it
because it is not a real person, it is a “robot”, so they did not feel judged or embarrassed of making
pronunciation mistakes as they would in front of real people.
Surveys showed that 80% of students felt comfortable using the tool, and 86% felt more
confident when speaking English.
Chats showed that students used more words and paused less over time.
DISCUSSION
Grok has many features, but the ones that are of interest for this research are the
conversational ones, and it helps to point out that these features are almost identical, either in a
computer or in a mobile phone. The only difference is that in a phone, the app has more “voice
modes (roles)”, but in spite of this, the app is the same. So, Grok’s conversational features are
4 4,5 3,8 4,2 4 4,3 3,9 4,1 4 4,2
7,5 8 7,2 7,8 7 7,6 7,4 7,9 7,3 7,7
ANA JUAN MARÍA CARLOS SOFÍA PEDRO VALERIA DIEGO CAMILA LUIS
Words per Answer by Student
Start words End words
Vol. 12/ Núm. 4 2025 pág. 1784
basically three, namely, the voice modes, the names of the voices that can resemble either a man’s
or a woman’s voice, and the ability to imply and infer meaning.
So, Grok’s roles act like real life impersonations, for example, in a computer, the tool has
six roles, but in a phone, it has thirteen. For instance, one Grok’s role is “Assistant”, this role is
very general in the way it talks, and, most of the time, it has a normal American accent, and it can
answer all types of general topics. Besides this, there is also a role called “Unhinged Comedian”
which has a different tone from the “Assistant”, the “Unhinged Comedian” has an upbeat tone of
voice, and it is excellent being funny and telling jokes, however, it might not be for everybody
because it can talk profanity.
Additionally, another voice mode that is worth mentioning is “Grok ‘Doc’”, this acts as a
doctor, and, amazingly, it can give recommendations about health and medicine, it actually knows
a great deal about the medical field. Another role is the “Argumentative” one, it is interesting in
the sense that it can actually argue about any topic, and gives the basis for its points of view, and
it actually encourages people. Furthermore, regardless of the role used, Grok has an almost
endless amount of knowledge of any field into it, so it can talk about any topic in any level so
makes it a promising tool to improve fluency in the English-speaking skill.
The second Grok’s voice feature is that the user can choose from four different tones of
voice, it has two female, and two male tones, this feature makes it versatile and encouraging.
And the third conversational feature, and perhaps the most important one, is the ability
Grok has to imply and infer meaning, it is almost as if Grok can reason like a human would, this
feature enables it to maintain a human like conversation, it listens and responds, and keeps the
subject going for as long as a person talks to it, and if an EFL learner asks to talk slower, or in
any of the six English levels A1, A2, B1, B2, C1, or C2 according to the Common European
Framework of Reference for the Languages (CEFR), it would.
For the aforementioned reasons: “voice modes”, “tones of voice”, and “inference,
implying, and almost reasoning” are what make it a promising AI tool that can be implemented
in schools to help students progress from one English speaking level to the next by imitating
original and authentic settings that may support oral fluency.
After using Grok, students improved their English speaking in many ways. They spoke
longer sentences and paused less. At the beginning, they produced short sentences like “I go shop”
and paused about 1.2 per answer, and by the last talk, learners used around 7.8 words per answer
and paused only 0.4 times.
Some students started asking questions back, like “Why is it raining?” showing that they
could use Level 2 logic. Also, common mistakes, such as saying “I go” instead of “I’m going” or
“I like pizza” instead of “I would like a pizza,” were corrected immediately by the app and learners
tried again and improved.
basically three, namely, the voice modes, the names of the voices that can resemble either a man’s
or a woman’s voice, and the ability to imply and infer meaning.
So, Grok’s roles act like real life impersonations, for example, in a computer, the tool has
six roles, but in a phone, it has thirteen. For instance, one Grok’s role is “Assistant”, this role is
very general in the way it talks, and, most of the time, it has a normal American accent, and it can
answer all types of general topics. Besides this, there is also a role called “Unhinged Comedian”
which has a different tone from the “Assistant”, the “Unhinged Comedian” has an upbeat tone of
voice, and it is excellent being funny and telling jokes, however, it might not be for everybody
because it can talk profanity.
Additionally, another voice mode that is worth mentioning is “Grok ‘Doc’”, this acts as a
doctor, and, amazingly, it can give recommendations about health and medicine, it actually knows
a great deal about the medical field. Another role is the “Argumentative” one, it is interesting in
the sense that it can actually argue about any topic, and gives the basis for its points of view, and
it actually encourages people. Furthermore, regardless of the role used, Grok has an almost
endless amount of knowledge of any field into it, so it can talk about any topic in any level so
makes it a promising tool to improve fluency in the English-speaking skill.
The second Grok’s voice feature is that the user can choose from four different tones of
voice, it has two female, and two male tones, this feature makes it versatile and encouraging.
And the third conversational feature, and perhaps the most important one, is the ability
Grok has to imply and infer meaning, it is almost as if Grok can reason like a human would, this
feature enables it to maintain a human like conversation, it listens and responds, and keeps the
subject going for as long as a person talks to it, and if an EFL learner asks to talk slower, or in
any of the six English levels A1, A2, B1, B2, C1, or C2 according to the Common European
Framework of Reference for the Languages (CEFR), it would.
For the aforementioned reasons: “voice modes”, “tones of voice”, and “inference,
implying, and almost reasoning” are what make it a promising AI tool that can be implemented
in schools to help students progress from one English speaking level to the next by imitating
original and authentic settings that may support oral fluency.
After using Grok, students improved their English speaking in many ways. They spoke
longer sentences and paused less. At the beginning, they produced short sentences like “I go shop”
and paused about 1.2 per answer, and by the last talk, learners used around 7.8 words per answer
and paused only 0.4 times.
Some students started asking questions back, like “Why is it raining?” showing that they
could use Level 2 logic. Also, common mistakes, such as saying “I go” instead of “I’m going” or
“I like pizza” instead of “I would like a pizza,” were corrected immediately by the app and learners
tried again and improved.
Vol. 12/ Núm. 4 2025 pág. 1785
Survey results also showed improvement. Confidence increased from 3.1 to 4.2, Comfort
from 3.3 to 4.1, and Motivation from 3.5 to 4.3. Additionally, students said they felt less anxious
using the app. For example, Ana said, “Grok is like a friend, I feel good using Grok,” and Juan
added, “I talk more now, no worry.” These results show that it helped them speak more freely,
make fewer mistakes, and gain confidence.
Additionally, overall, the ADDIE plan worked: starting with easy questions, gradually
increasing difficulty, and using role-play and immediate corrections helped students improve,
also, the combination of short, repeated sessions, feedback, and student`s reflections using
metacognitive strategy made learning safe and effective.
According to the general objective of analyzing the impact of Grok on learners’ academic
performance, this study demonstrated that the use of this AI tool can be a possible solution to
improve students’ English speaking fluency, with more participation and confidence. Besides, the
data collected through CAP and surveys indicated that it is a valuable tool for helping improve
the teaching and learning process in English as a foreign language in general Education.
CONCLUSION
In conclusion, this study showed the effectiveness of the proposed solution, which is the
integration of Grok app that significantly improved English-speaking fluency. It also, reduced
communication anxiety, and increased confidence in B1 students of “Hispanoamericano” High
School in Guayaquil-Ecuador. Additionally, scores for confidence improved from 3.1 to 4.2, also,
the average number of pauses per answer decreased from 1.2 to 0.4, and response length increased
from 4.2 to 7.8 words. Based on Communicative Language Teaching (CLT) and metacognitive
strategies, and the ADDIE model, the intervention gave students a safe place to talk without fear
of being judged that helped them grow English language fluency reaching the objectives.
Although the population-sample was only of fifty students, the results showed strong proof that
Grok is a helpful tool that can be used in any classroom to improve English fluency. Therefore,
this study suggests that Grok can potentially be used in classrooms and at home as a
complimentary help to improve English fluency.
Survey results also showed improvement. Confidence increased from 3.1 to 4.2, Comfort
from 3.3 to 4.1, and Motivation from 3.5 to 4.3. Additionally, students said they felt less anxious
using the app. For example, Ana said, “Grok is like a friend, I feel good using Grok,” and Juan
added, “I talk more now, no worry.” These results show that it helped them speak more freely,
make fewer mistakes, and gain confidence.
Additionally, overall, the ADDIE plan worked: starting with easy questions, gradually
increasing difficulty, and using role-play and immediate corrections helped students improve,
also, the combination of short, repeated sessions, feedback, and student`s reflections using
metacognitive strategy made learning safe and effective.
According to the general objective of analyzing the impact of Grok on learners’ academic
performance, this study demonstrated that the use of this AI tool can be a possible solution to
improve students’ English speaking fluency, with more participation and confidence. Besides, the
data collected through CAP and surveys indicated that it is a valuable tool for helping improve
the teaching and learning process in English as a foreign language in general Education.
CONCLUSION
In conclusion, this study showed the effectiveness of the proposed solution, which is the
integration of Grok app that significantly improved English-speaking fluency. It also, reduced
communication anxiety, and increased confidence in B1 students of “Hispanoamericano” High
School in Guayaquil-Ecuador. Additionally, scores for confidence improved from 3.1 to 4.2, also,
the average number of pauses per answer decreased from 1.2 to 0.4, and response length increased
from 4.2 to 7.8 words. Based on Communicative Language Teaching (CLT) and metacognitive
strategies, and the ADDIE model, the intervention gave students a safe place to talk without fear
of being judged that helped them grow English language fluency reaching the objectives.
Although the population-sample was only of fifty students, the results showed strong proof that
Grok is a helpful tool that can be used in any classroom to improve English fluency. Therefore,
this study suggests that Grok can potentially be used in classrooms and at home as a
complimentary help to improve English fluency.
Vol. 12/ Núm. 4 2025 pág. 1786
REFERENCES
Aremu, I. O., Paredes Espinosa, K. E., Intriago Cañizares, F., & Bonilla Tenesaca, J. R. (2025).
AI-powered podcast interventions for enhancing speaking skills in English language
teaching (ELT) adult A1 students. Arandu Utic, 12(3), 2702–2723.
https://doi.org/10.69639/arandu.v12i3.1509
Creswell, J. W., & Creswell, J. D. (2018). Research design: Qualitative, quantitative, and mixed
methods approaches (5th ed.). SAGE Publications.
Dennis, N. K. (2024). Using AI-powered speech recognition technology to improve English
pronunciation and speaking skills. IAFOR Journal of Education: Technology in
Education, 12(2), 107–126. https://doi.org/10.22492/ije.12.2.05
Hymes, D. (1972). Sociolinguistics: Selected readings (J. B. Pride, Ed.). Penguin.
Instituto Nacional de Estadística y Censos. (2023). Tecnologías de la información y comunicación
(tic) en los hogares. https://www.ecuadorencifras.gob.ec
Kamali, M., Singh, M., Pallathadka, H., Etkov, I., Prasad, K. D., Ismail, S. M., … & Adefila, A.
(2024). The efficacy of the spaced learning teaching method on student engagement and
learning anxiety. Research Square (Preprint). https://doi.org/10.21203/rs.3.rs-
4712072/v1
Krashen, S. D. (1982). Principles and practice in second language acquisition. Pergamon Press.
Liu, J., Sihes, A. J., & Lu, Y. (2025). How do generative artificial intelligence (AI) tools and large
language models (LLMs) influence language learners’ critical thinking in EFL education?
A systematic review. Smart Learning Environments, 12(48), 1–35.
https://doi.org/10.1186/s40561-025-00406-0
Meher, V., Baral, R., & Bhuyan, S. (2024). Assessing the impact of metacognitive interventions
on academic achievement of higher secondary school students in education. Indonesian
Educational Research Journal, 1(3), 144–157. https://doi.org/10.56773/ierj.v1i3.27
Mingyan, M. N. (2025). Improving EFL speaking performance among undergraduate students
with an AI-powered mobile app in after-class assignments: An empirical investigation.
Humanities and Social Sciences Communications, 12(370), 1–15.
https://doi.org/10.1057/s41599-025-04688-0
Mohammed, A. A. Q., Mudhsh, B. A., Bin-Hady, W. R. A., & Al-Tamimi, A. S. (2025).
DeepSeek and Grok in the spotlight after ChatGPT in English education: A review study.
Journal of English Studies in Arabia Felix, 4(1), 13–22.
https://doi.org/10.56540/jesaf.v4i1.114
Nation, I. S. (2001). Learning vocabulary in another language. Cambridge University Press.
REFERENCES
Aremu, I. O., Paredes Espinosa, K. E., Intriago Cañizares, F., & Bonilla Tenesaca, J. R. (2025).
AI-powered podcast interventions for enhancing speaking skills in English language
teaching (ELT) adult A1 students. Arandu Utic, 12(3), 2702–2723.
https://doi.org/10.69639/arandu.v12i3.1509
Creswell, J. W., & Creswell, J. D. (2018). Research design: Qualitative, quantitative, and mixed
methods approaches (5th ed.). SAGE Publications.
Dennis, N. K. (2024). Using AI-powered speech recognition technology to improve English
pronunciation and speaking skills. IAFOR Journal of Education: Technology in
Education, 12(2), 107–126. https://doi.org/10.22492/ije.12.2.05
Hymes, D. (1972). Sociolinguistics: Selected readings (J. B. Pride, Ed.). Penguin.
Instituto Nacional de Estadística y Censos. (2023). Tecnologías de la información y comunicación
(tic) en los hogares. https://www.ecuadorencifras.gob.ec
Kamali, M., Singh, M., Pallathadka, H., Etkov, I., Prasad, K. D., Ismail, S. M., … & Adefila, A.
(2024). The efficacy of the spaced learning teaching method on student engagement and
learning anxiety. Research Square (Preprint). https://doi.org/10.21203/rs.3.rs-
4712072/v1
Krashen, S. D. (1982). Principles and practice in second language acquisition. Pergamon Press.
Liu, J., Sihes, A. J., & Lu, Y. (2025). How do generative artificial intelligence (AI) tools and large
language models (LLMs) influence language learners’ critical thinking in EFL education?
A systematic review. Smart Learning Environments, 12(48), 1–35.
https://doi.org/10.1186/s40561-025-00406-0
Meher, V., Baral, R., & Bhuyan, S. (2024). Assessing the impact of metacognitive interventions
on academic achievement of higher secondary school students in education. Indonesian
Educational Research Journal, 1(3), 144–157. https://doi.org/10.56773/ierj.v1i3.27
Mingyan, M. N. (2025). Improving EFL speaking performance among undergraduate students
with an AI-powered mobile app in after-class assignments: An empirical investigation.
Humanities and Social Sciences Communications, 12(370), 1–15.
https://doi.org/10.1057/s41599-025-04688-0
Mohammed, A. A. Q., Mudhsh, B. A., Bin-Hady, W. R. A., & Al-Tamimi, A. S. (2025).
DeepSeek and Grok in the spotlight after ChatGPT in English education: A review study.
Journal of English Studies in Arabia Felix, 4(1), 13–22.
https://doi.org/10.56540/jesaf.v4i1.114
Nation, I. S. (2001). Learning vocabulary in another language. Cambridge University Press.
Vol. 12/ Núm. 4 2025 pág. 1787
Quvanch, Z. Q. (2024). Analyzing levels, factors, and coping strategies of speaking anxiety
among EFL undergraduates in Afghanistan. Cogent Education, 11(1), Article 2413225.
https://doi.org/10.1080/2331186X.2024.2413225
Richards, J. C. (2006). Communicative language teaching today. Cambridge University Press.
Shikun, S., Grigoryan, G., Huichun, N., & Harutyunyan, H. (2024). AI chatbots: Developing
English language proficiency in the EFL classroom. Arab World English Journal, 15(3),
292–305. https://doi.org/10.24093/awej/ChatGPT.20
Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science,
12(2), 257–285. https://doi.org/10.1207/s15516709cog1202_4
Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes (M.
Cole, V. John-Steiner, S. Scribner, & E. Souberman, Eds.). Harvard University Press.
Vygotsky, L. S. (1986). Thought and language (A. Kozulin, Ed. & Trans.; Original work
published 1934). MIT Press.
Wangsa, K. K. (2024). A systematic review and comprehensive analysis of pioneering AI chatbot
models from education. Future Internet, 16(7), 219. https://doi.org/10.3390/fi16070219
Quvanch, Z. Q. (2024). Analyzing levels, factors, and coping strategies of speaking anxiety
among EFL undergraduates in Afghanistan. Cogent Education, 11(1), Article 2413225.
https://doi.org/10.1080/2331186X.2024.2413225
Richards, J. C. (2006). Communicative language teaching today. Cambridge University Press.
Shikun, S., Grigoryan, G., Huichun, N., & Harutyunyan, H. (2024). AI chatbots: Developing
English language proficiency in the EFL classroom. Arab World English Journal, 15(3),
292–305. https://doi.org/10.24093/awej/ChatGPT.20
Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science,
12(2), 257–285. https://doi.org/10.1207/s15516709cog1202_4
Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes (M.
Cole, V. John-Steiner, S. Scribner, & E. Souberman, Eds.). Harvard University Press.
Vygotsky, L. S. (1986). Thought and language (A. Kozulin, Ed. & Trans.; Original work
published 1934). MIT Press.
Wangsa, K. K. (2024). A systematic review and comprehensive analysis of pioneering AI chatbot
models from education. Future Internet, 16(7), 219. https://doi.org/10.3390/fi16070219
Vol. 12/ Núm. 4 2025 pág. 1788
ANNEXES
APPENDIX A: Solicitud de aprobación presentada a la Universidad Bolivariana del Ecuador para
el trabajo de titulación.
https://drive.google.com/file/d/1NGlWuf7X4gn5ROLNsnbw48cArxHzCirV/view?usp=
sharing
APPENDIX B: (i) Solicitud de autorización enviada al rector de la Unidad Educativa
Hispanoamericano. (ii) Autorización otorgada por el Economista Felipe Bocca, rector.
https://drive.google.com/file/d/18MqBiNfAjdA0g4G0vfi0Tt2BYa9rxgRW/view?usp=s
haring
APPENDIX C: Encuesta Likert pre- y post-intervención.
https://drive.google.com/file/d/1gAFN2HTGtnyEHta1G13eddQv8BgZpZZ_/view?usp=
sharing
APPENDIX D: Protocolo de Análisis de Conversación (CAP)
https://drive.google.com/file/d/1bKlD5ds8jNajuHIIjgR4ZilcMTtGZQZb/view?usp=sha
ring
ANNEXES
APPENDIX A: Solicitud de aprobación presentada a la Universidad Bolivariana del Ecuador para
el trabajo de titulación.
https://drive.google.com/file/d/1NGlWuf7X4gn5ROLNsnbw48cArxHzCirV/view?usp=
sharing
APPENDIX B: (i) Solicitud de autorización enviada al rector de la Unidad Educativa
Hispanoamericano. (ii) Autorización otorgada por el Economista Felipe Bocca, rector.
https://drive.google.com/file/d/18MqBiNfAjdA0g4G0vfi0Tt2BYa9rxgRW/view?usp=s
haring
APPENDIX C: Encuesta Likert pre- y post-intervención.
https://drive.google.com/file/d/1gAFN2HTGtnyEHta1G13eddQv8BgZpZZ_/view?usp=
sharing
APPENDIX D: Protocolo de Análisis de Conversación (CAP)
https://drive.google.com/file/d/1bKlD5ds8jNajuHIIjgR4ZilcMTtGZQZb/view?usp=sha
ring