Vol. 13/ Núm. 3 2026 pág. 3096
https://doi.org/10.69639/arandu.v13i3.2599
The artificial intelligence tool ChatGPT and its impact on the
development of the integrated writing skill in students at the
International Learning' Institute
La herramienta de inteligencia artificial ChatGPT y su impacto en el desarrollo de la
habilidad de escritura integrada en los estudiantes del Instituto International Learning
Paul Sebastián Montenegro Rosero
sebastian.montenegro@upec.edu.ec
https://orcid.org/0009-0006-3203-4717
Universidad Politécnica Estatal del Carchi
Ecuador – Tulcán
Elena Valeria Flores Borja
elenaflores@upec.edu.ec
https://orcid.org/0009-0005-4800-8938
Universidad Politécnica Estatal del Carchi
Ecuador - Tulcán
Artículo recibido: 10 agosto 2026- Aceptado para publicación:16 septiembre 2026
Conflictos de intereses: Ninguno que declarar.
ABSTRACT
The present research analyzed the impact of the pedagogical use of ChatGPT on the development
of English written production in students at the International Learning Institute, located in the city
of Tulcán, Ecuador. The study arose from the difficulties identified in grammatical precision,
spelling, syntactic organization, coherence, cohesion, and lexical variety of the texts produced by
the students. The study was conducted with the entire population, consisting of 45 students aged
between 14 and 22 years, with English proficiency levels ranging from A1 to B2. The research
was developed using a quantitative approach, with a descriptive scope and a single-group
exploratory design with a pretest and posttest. A questionnaire and a written production evaluation
rubric were used for data collection. During the intervention, ChatGPT was used as a
complementary tool to provide immediate feedback, identify grammatical and spelling errors,
suggest improvements in textual structure, and promote vocabulary expansion. The results
obtained through the diagnostic survey evidenced difficulties in sentence structuring, the use of
varied vocabulary, punctuation, and grammatical accuracy, as well as differences in the frequency
with which students used ChatGPT to support their writing. It is concluded that the planned,
ethical, and supervised integration of ChatGPT can constitute a complementary strategy to
strengthen English written production, if its application is accompanied by teacher guidance and
the student's critical participation.
Keywords: chatGPT, english written production, artificial intelligence, language learning,
linguistic feedback
Vol. 13/ Núm. 3 2026 pág. 3097
RESUMEN
La presente investigación analizó la incidencia del uso pedagógico de ChatGPT en el desarrollo
de la producción escrita en inglés de los estudiantes del Instituto International Learning, ubicado
en la ciudad de Tulcán, Ecuador. El estudio surgió ante las dificultades identificadas en la
precisión gramatical, la ortografía, la organización sintáctica, la coherencia, la cohesión y la
variedad léxica de los textos producidos por los estudiantes. Se trabajó con la totalidad de la
población, conformada por 45 estudiantes con edades comprendidas entre 14 y 22 años y niveles
de dominio del inglés de A1 a B2. La investigación se desarrolló mediante un enfoque
cuantitativo, con alcance descriptivo y un diseño exploratorio de un solo grupo con pretest y
postest. Para la recolección de información se emplearon un cuestionario y una rúbrica de
evaluación de la producción escrita. Durante la intervención, ChatGPT se utiliza como
complementario para proporcionar retroalimentación inmediata, identificar errores gramaticales
y ortográficos, sugerir mejoras en la estructura textual y favorecer la ampliación del vocabulario.
Los resultados obtenidos mediante la encuesta diagnóstica evidenciaron dificultades en la
estructuración de oraciones, el uso de vocabulario variado, la puntuación y la corrección
gramatical, así como diferencias en la frecuencia con la que los estudiantes utilizaban ChatGPT
para apoyar su escritura. Se concluye que la integración planificada, ética y supervisada de
ChatGPT puede constituir una estrategia complementaria para fortalecer la producción escrita en
inglés, siempre que su aplicación esté acompañada por la orientación docente y la participación
crítica del estudiante.
Palabras clave: chatGPT, producción escrita en inglés, inteligencia artificial, aprendizaje
de lenguas, retroalimentación lingüística
Todo el contenido de la Revista Científica Internacional Arandu UTIC publicado en este sitio está disponible bajo
licencia Creative Commons Atribution 4.0 International.
RESUMEN
La presente investigación analizó la incidencia del uso pedagógico de ChatGPT en el desarrollo
de la producción escrita en inglés de los estudiantes del Instituto International Learning, ubicado
en la ciudad de Tulcán, Ecuador. El estudio surgió ante las dificultades identificadas en la
precisión gramatical, la ortografía, la organización sintáctica, la coherencia, la cohesión y la
variedad léxica de los textos producidos por los estudiantes. Se trabajó con la totalidad de la
población, conformada por 45 estudiantes con edades comprendidas entre 14 y 22 años y niveles
de dominio del inglés de A1 a B2. La investigación se desarrolló mediante un enfoque
cuantitativo, con alcance descriptivo y un diseño exploratorio de un solo grupo con pretest y
postest. Para la recolección de información se emplearon un cuestionario y una rúbrica de
evaluación de la producción escrita. Durante la intervención, ChatGPT se utiliza como
complementario para proporcionar retroalimentación inmediata, identificar errores gramaticales
y ortográficos, sugerir mejoras en la estructura textual y favorecer la ampliación del vocabulario.
Los resultados obtenidos mediante la encuesta diagnóstica evidenciaron dificultades en la
estructuración de oraciones, el uso de vocabulario variado, la puntuación y la corrección
gramatical, así como diferencias en la frecuencia con la que los estudiantes utilizaban ChatGPT
para apoyar su escritura. Se concluye que la integración planificada, ética y supervisada de
ChatGPT puede constituir una estrategia complementaria para fortalecer la producción escrita en
inglés, siempre que su aplicación esté acompañada por la orientación docente y la participación
crítica del estudiante.
Palabras clave: chatGPT, producción escrita en inglés, inteligencia artificial, aprendizaje
de lenguas, retroalimentación lingüística
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. 13/ Núm. 3 2026 pág. 3098
INTRODUCTION
Writing in English as a foreign language faces recurring barriers, such as the lack of
timely feedback and deficiencies in cohesion and vocabulary. Although generative artificial
intelligence, particularly ChatGPT, emerges as a resource to personalize learning, its unregulated
adoption generates a critical debate: does it act as a pedagogical guide or as a substitute for
cognitive effort? In contexts like the Ecuadorian one, where traditional approaches still
predominate, there is an empirical gap regarding how to structure teacher mediation to guarantee
an ethical and effective use of these tools. The present study addresses this gap by evaluating the
impact of a supervised pedagogical intervention with ChatGPT on 45 students at the
"International Learning" Institute in Tulcán, Ecuador. Unlike passive use approaches, this
research implements a structured model in four phases (planning, drafting, assisted revision, and
editing), where the student must contrast and justify the algorithmic suggestions. The results
provide quantitative evidence that the planned integration of AI significantly improves linguistic
precision and lexical diversity, fostering metacognitive processes and offering a practical
framework for digital literacy in language teaching.
Background of the Problem
Written production in English constitutes a fundamental competence in 21st-century
academic and professional training, as it allows students to articulate ideas, argue rigorously, and
participate in international disciplinary communities. However, those who learn this language as
a second language face recurring difficulties: scarcity of specialized vocabulary, problems with
cohesion and textual structuring, and, above all, the absence of timely feedback that allows them
to identify and correct their errors autonomously. Faced with these limitations, generative
artificial intelligence tools such as ChatGPT have emerged as a possible pedagogical resource to
accompany the writing process. Dou et al. (2025) and Román et al. (2026) highlights that
intelligent environments reduce the barriers associated with the lack of immediate feedback,
particularly among second language learners, and facilitate the creation of personalized learning
experiences that accelerate the development of written competence.
Likewise, Ossa and Willatt (2023) implemented generative AI to provide descriptive
feedback on pedagogical responses elaborated by pre-service teachers, highlighting the tool's
capacity to support formative and large-scale assessment of students' written work. Baldrich and
Domínguez (2024), in their case study on the use of ChatGPT in academic writing, observed
improvements in written tasks regarding coherence, cohesion, and academic language; they
concluded that ChatGPT can serve as a complementary tool for academic work as long as students
already possess fundamental skills in critical thinking, ethics, and argumentation.
The development of written competence in a foreign language requires systematic
guidance, continuous feedback, and interactive learning environments that favor meaningful
INTRODUCTION
Writing in English as a foreign language faces recurring barriers, such as the lack of
timely feedback and deficiencies in cohesion and vocabulary. Although generative artificial
intelligence, particularly ChatGPT, emerges as a resource to personalize learning, its unregulated
adoption generates a critical debate: does it act as a pedagogical guide or as a substitute for
cognitive effort? In contexts like the Ecuadorian one, where traditional approaches still
predominate, there is an empirical gap regarding how to structure teacher mediation to guarantee
an ethical and effective use of these tools. The present study addresses this gap by evaluating the
impact of a supervised pedagogical intervention with ChatGPT on 45 students at the
"International Learning" Institute in Tulcán, Ecuador. Unlike passive use approaches, this
research implements a structured model in four phases (planning, drafting, assisted revision, and
editing), where the student must contrast and justify the algorithmic suggestions. The results
provide quantitative evidence that the planned integration of AI significantly improves linguistic
precision and lexical diversity, fostering metacognitive processes and offering a practical
framework for digital literacy in language teaching.
Background of the Problem
Written production in English constitutes a fundamental competence in 21st-century
academic and professional training, as it allows students to articulate ideas, argue rigorously, and
participate in international disciplinary communities. However, those who learn this language as
a second language face recurring difficulties: scarcity of specialized vocabulary, problems with
cohesion and textual structuring, and, above all, the absence of timely feedback that allows them
to identify and correct their errors autonomously. Faced with these limitations, generative
artificial intelligence tools such as ChatGPT have emerged as a possible pedagogical resource to
accompany the writing process. Dou et al. (2025) and Román et al. (2026) highlights that
intelligent environments reduce the barriers associated with the lack of immediate feedback,
particularly among second language learners, and facilitate the creation of personalized learning
experiences that accelerate the development of written competence.
Likewise, Ossa and Willatt (2023) implemented generative AI to provide descriptive
feedback on pedagogical responses elaborated by pre-service teachers, highlighting the tool's
capacity to support formative and large-scale assessment of students' written work. Baldrich and
Domínguez (2024), in their case study on the use of ChatGPT in academic writing, observed
improvements in written tasks regarding coherence, cohesion, and academic language; they
concluded that ChatGPT can serve as a complementary tool for academic work as long as students
already possess fundamental skills in critical thinking, ethics, and argumentation.
The development of written competence in a foreign language requires systematic
guidance, continuous feedback, and interactive learning environments that favor meaningful
Vol. 13/ Núm. 3 2026 pág. 3099
learning. However, traditional methods of teaching English present considerable limitations: as
Cordero et al. (2025) point out in their study conducted in Ecuador, numerous educational
institutions still depend on approaches focused on grammar and memorization, which restrict
opportunities for communicative practice and hinder the effective learning of writing. Faced with
this lack of pedagogical support, students autonomously resort to artificial intelligence tools.
These findings make it evident that the unregulated use of AI, far from supplying training
deficiencies, can aggravate them, and reinforce the need for a pedagogically mediated integration
that guides the student in the critical and responsible use of these tools.
The incorporation of artificial intelligence applications such as ChatGPT has transformed
the teaching of English through immediate feedback, content personalization, and autonomous
practice. Farrokhnia et al. (2024) argue that ChatGPT functions as a dynamic educational resource
by increasing access to information, facilitating personalized learning, and providing coherent
responses in real time. Adel et al. (2024) point out that ChatGPT can significantly enrich
education by offering dynamic and personalized learning experiences, as well as real-time
feedback, thus boosting student engagement and teaching efficiency when integrated into
responsible ethical frameworks.
Lee and Ko (2026) found that AI-based assistants provide automated feedback on writing,
improving cohesion, accuracy, and fluency in English as a Foreign Language (EFL) teaching
context. Likewise, Zhai and Nezakatgoo (2025) determined that AI facilitates personalized
progress tracking, fostering metacognitive processes that reinforce self-assessment and
independent revision skills. Kohnke et al. (2023) highlight that the pedagogical relevance and
quality of AI suggestions depend largely on effective prompt engineering, which requires users
to iteratively refine their input instructions. Nevertheless, the literature also points out certain
challenges related to the integration of artificial intelligence in educational settings. Yan et al.
(2025) warn that excessive reliance on AI tools can lead to critical problems such as algorithmic
bias, data privacy vulnerability, the weakening of human relationships, and the deterioration of
critical thinking if there is no adequate pedagogical mediation. Baidoo-Anu and Owusu Ansah
(2023) point out that its adoption must be carefully guided to avoid problems such as the
generation of inaccurate information, privacy vulnerabilities, and the amplification of pre-existing
biases in training data.
Specifically in the context of EFL writing, Aliakbar (2024) demonstrates that ChatGPT
improves linguistic precision and lexical variety in students' compositions. However, research
suggests that its pedagogical potential is maximized when used as a mediation tool, rather than as
a total substitute for cognitive effort and writing processes. Liu (2024) documents performance
differences between groups of students assisted by AI and those who received traditional
instruction, generally supporting the positive impact of automated writing evaluation on writing
accuracy.
learning. However, traditional methods of teaching English present considerable limitations: as
Cordero et al. (2025) point out in their study conducted in Ecuador, numerous educational
institutions still depend on approaches focused on grammar and memorization, which restrict
opportunities for communicative practice and hinder the effective learning of writing. Faced with
this lack of pedagogical support, students autonomously resort to artificial intelligence tools.
These findings make it evident that the unregulated use of AI, far from supplying training
deficiencies, can aggravate them, and reinforce the need for a pedagogically mediated integration
that guides the student in the critical and responsible use of these tools.
The incorporation of artificial intelligence applications such as ChatGPT has transformed
the teaching of English through immediate feedback, content personalization, and autonomous
practice. Farrokhnia et al. (2024) argue that ChatGPT functions as a dynamic educational resource
by increasing access to information, facilitating personalized learning, and providing coherent
responses in real time. Adel et al. (2024) point out that ChatGPT can significantly enrich
education by offering dynamic and personalized learning experiences, as well as real-time
feedback, thus boosting student engagement and teaching efficiency when integrated into
responsible ethical frameworks.
Lee and Ko (2026) found that AI-based assistants provide automated feedback on writing,
improving cohesion, accuracy, and fluency in English as a Foreign Language (EFL) teaching
context. Likewise, Zhai and Nezakatgoo (2025) determined that AI facilitates personalized
progress tracking, fostering metacognitive processes that reinforce self-assessment and
independent revision skills. Kohnke et al. (2023) highlight that the pedagogical relevance and
quality of AI suggestions depend largely on effective prompt engineering, which requires users
to iteratively refine their input instructions. Nevertheless, the literature also points out certain
challenges related to the integration of artificial intelligence in educational settings. Yan et al.
(2025) warn that excessive reliance on AI tools can lead to critical problems such as algorithmic
bias, data privacy vulnerability, the weakening of human relationships, and the deterioration of
critical thinking if there is no adequate pedagogical mediation. Baidoo-Anu and Owusu Ansah
(2023) point out that its adoption must be carefully guided to avoid problems such as the
generation of inaccurate information, privacy vulnerabilities, and the amplification of pre-existing
biases in training data.
Specifically in the context of EFL writing, Aliakbar (2024) demonstrates that ChatGPT
improves linguistic precision and lexical variety in students' compositions. However, research
suggests that its pedagogical potential is maximized when used as a mediation tool, rather than as
a total substitute for cognitive effort and writing processes. Liu (2024) documents performance
differences between groups of students assisted by AI and those who received traditional
instruction, generally supporting the positive impact of automated writing evaluation on writing
accuracy.
Vol. 13/ Núm. 3 2026 pág. 3100
One of the challenges facing the teaching of English is written composition; learning
takes place in an educational environment where traditional methods based on repetition and
memorization usually predominate (Párraga Solórzano et al., 2024). Academic writing is a crucial
skill in education worldwide, as it allows students to communicate ideas clearly, coherently, and
effectively in diverse academic disciplines. Despite its importance, many students lack
proficiency in academic writing, which significantly hinders their school performance and limits
their opportunities for academic and professional success.
In the Ecuadorian context, educational institutions continue to present deficiencies in the
teaching and development of writing skills. Students often have difficulties organizing ideas,
maintaining textual coherence, and correctly applying grammar and syntax. Traditional
pedagogical approaches still predominate in classrooms and frequently fail to offer personalized
and effective feedback that fosters writing competence. Furthermore, the implementation of
technological tools, particularly those based on artificial intelligence, remains limited or non-
existent in many classrooms. Faced with this situation, the following research questions arise:
How does the application of the artificial intelligence tool ChatGPT influence the development
of the integrated skill of written production in English? Likewise, the following guiding questions
have been formulated for the study: a) What are the main difficulties faced by students at the
"International Learning" institute in the academic writing process before using artificial
intelligence tools? b) What characteristics does ChatGPT possess that can effectively support
written production? c) How is ChatGPT implemented in the teaching-learning process of
academic English written production?
Research Objectives
In this sense, the research proposes the following general objective: to determine the
impact of the digital tool ChatGPT on the development of the integrated skill of written production
in students at the "International Learning" Institute, located in the city of Tulcán, Carchi province.
To achieve this, the following specific objectives were proposed: a) to identify the main
difficulties in students' written production before the implementation of ChatGPT; b) to apply the
artificial intelligence tool ChatGPT as a pedagogical resource in the teaching-learning process of
written production; and c) to evaluate the changes in the level of students' written production by
comparing the results obtained in the pretest and posttest, in order to formulate pedagogical
guidelines for the effective integration of this tool in the classroom.
Research Hypothesis
According to the research objectives, the following null hypothesis was formulated: H₀:
There is no statistically significant difference between students' writing scores in the pretest and
posttest after the pedagogical use of ChatGPT. H₁: There is a statistically significant difference
between students' writing scores in the pretest and posttest after the pedagogical use of ChatGPT.
One of the challenges facing the teaching of English is written composition; learning
takes place in an educational environment where traditional methods based on repetition and
memorization usually predominate (Párraga Solórzano et al., 2024). Academic writing is a crucial
skill in education worldwide, as it allows students to communicate ideas clearly, coherently, and
effectively in diverse academic disciplines. Despite its importance, many students lack
proficiency in academic writing, which significantly hinders their school performance and limits
their opportunities for academic and professional success.
In the Ecuadorian context, educational institutions continue to present deficiencies in the
teaching and development of writing skills. Students often have difficulties organizing ideas,
maintaining textual coherence, and correctly applying grammar and syntax. Traditional
pedagogical approaches still predominate in classrooms and frequently fail to offer personalized
and effective feedback that fosters writing competence. Furthermore, the implementation of
technological tools, particularly those based on artificial intelligence, remains limited or non-
existent in many classrooms. Faced with this situation, the following research questions arise:
How does the application of the artificial intelligence tool ChatGPT influence the development
of the integrated skill of written production in English? Likewise, the following guiding questions
have been formulated for the study: a) What are the main difficulties faced by students at the
"International Learning" institute in the academic writing process before using artificial
intelligence tools? b) What characteristics does ChatGPT possess that can effectively support
written production? c) How is ChatGPT implemented in the teaching-learning process of
academic English written production?
Research Objectives
In this sense, the research proposes the following general objective: to determine the
impact of the digital tool ChatGPT on the development of the integrated skill of written production
in students at the "International Learning" Institute, located in the city of Tulcán, Carchi province.
To achieve this, the following specific objectives were proposed: a) to identify the main
difficulties in students' written production before the implementation of ChatGPT; b) to apply the
artificial intelligence tool ChatGPT as a pedagogical resource in the teaching-learning process of
written production; and c) to evaluate the changes in the level of students' written production by
comparing the results obtained in the pretest and posttest, in order to formulate pedagogical
guidelines for the effective integration of this tool in the classroom.
Research Hypothesis
According to the research objectives, the following null hypothesis was formulated: H₀:
There is no statistically significant difference between students' writing scores in the pretest and
posttest after the pedagogical use of ChatGPT. H₁: There is a statistically significant difference
between students' writing scores in the pretest and posttest after the pedagogical use of ChatGPT.
Vol. 13/ Núm. 3 2026 pág. 3101
Justification of the Research
Considering the reviewed evidence, ChatGPT offers significant opportunities to
overcome the limitations of traditional English teaching methods that still predominate in the
classrooms of the Ecuadorian educational system. Educational inclusion, in this sense, is not
reduced to access, but demands the implementation of pedagogical strategies that eliminate
barriers and guarantee quality training, a principle supported by the Constitution of the Republic
of Ecuador and the Organic Law of Intercultural Education (National Assembly of Ecuador,
2008). However, despite the regulatory framework that promotes innovation and educational
equity, research on the use of generative artificial intelligence tools in the development of English
writing remains virtually unexplored in local contexts such as "International Learning". This
absence of empirical studies that characterize the practices, perceptions, and needs of students in
said locality constitutes a gap that limits informed pedagogical decision-making and, therefore,
justifies the realization of the present research.
At the local level, a possible contextual limitation at the "International Learning" institute
would lie in the accentuation of the difficulties described previously. It is likely that teachers do
not employ technology tools based on artificial intelligence in the teaching of writing. If so,
students might be producing texts with deficiencies in structure, clarity, and communicative
purpose, which would limit both their academic development and their capacity for personal
expression. If this scenario persists without pedagogical interventions aimed at modifying it, there
would be a risk that students continue to face difficulties in written production, show low
confidence when expressing themselves in writing, and develop an excessive dependence on
grammatical rules to the detriment of comprehensive communicative competence.
Significance of the Research
This research is justified from a theoretical perspective by contributing to the
understanding of how generative artificial intelligence, particularly ChatGPT, can transform the
teaching of English writing beyond approaches focused on grammar and memorization. Its social
relevance lies in the direct benefit for the students and teachers of the "International Learning"
Institute, who access more functional and motivating learning experiences. This purpose is
articulated with Sustainable Development Goal 4 of the United Nations 2030 Agenda, aimed at
guaranteeing inclusive and quality education through innovative methods (UNESCO, 2024), and
with the National Development Plan 2025-2029, which promotes technological innovation and
digital transformation (National Planning Secretariat, 2025).
MATERIALS AND METHODS
Research Paradigm and Approach
This study is framed within the positivist paradigm, as it adopts a quantitative approach
based on the collection and analysis of numerical data to identify trends and measure the scope
Justification of the Research
Considering the reviewed evidence, ChatGPT offers significant opportunities to
overcome the limitations of traditional English teaching methods that still predominate in the
classrooms of the Ecuadorian educational system. Educational inclusion, in this sense, is not
reduced to access, but demands the implementation of pedagogical strategies that eliminate
barriers and guarantee quality training, a principle supported by the Constitution of the Republic
of Ecuador and the Organic Law of Intercultural Education (National Assembly of Ecuador,
2008). However, despite the regulatory framework that promotes innovation and educational
equity, research on the use of generative artificial intelligence tools in the development of English
writing remains virtually unexplored in local contexts such as "International Learning". This
absence of empirical studies that characterize the practices, perceptions, and needs of students in
said locality constitutes a gap that limits informed pedagogical decision-making and, therefore,
justifies the realization of the present research.
At the local level, a possible contextual limitation at the "International Learning" institute
would lie in the accentuation of the difficulties described previously. It is likely that teachers do
not employ technology tools based on artificial intelligence in the teaching of writing. If so,
students might be producing texts with deficiencies in structure, clarity, and communicative
purpose, which would limit both their academic development and their capacity for personal
expression. If this scenario persists without pedagogical interventions aimed at modifying it, there
would be a risk that students continue to face difficulties in written production, show low
confidence when expressing themselves in writing, and develop an excessive dependence on
grammatical rules to the detriment of comprehensive communicative competence.
Significance of the Research
This research is justified from a theoretical perspective by contributing to the
understanding of how generative artificial intelligence, particularly ChatGPT, can transform the
teaching of English writing beyond approaches focused on grammar and memorization. Its social
relevance lies in the direct benefit for the students and teachers of the "International Learning"
Institute, who access more functional and motivating learning experiences. This purpose is
articulated with Sustainable Development Goal 4 of the United Nations 2030 Agenda, aimed at
guaranteeing inclusive and quality education through innovative methods (UNESCO, 2024), and
with the National Development Plan 2025-2029, which promotes technological innovation and
digital transformation (National Planning Secretariat, 2025).
MATERIALS AND METHODS
Research Paradigm and Approach
This study is framed within the positivist paradigm, as it adopts a quantitative approach
based on the collection and analysis of numerical data to identify trends and measure the scope
Vol. 13/ Núm. 3 2026 pág. 3102
of the impact of an intervention, in this case, the use of the artificial intelligence tool ChatGPT in
the development of English written production (Dávila, 2022).
This study adopts a quantitative approach. According to Vizcaíno et al. (2023), it is based
on the collection and analysis of numerical data to identify trends and measure the scope of the
impact of an intervention. The reasoning is deductive, starting from the theory on the use of AI
in writing to test its effect through the statistical measurement of pretest and post-test results.
Medina et al. (2023) mention that the use of data collection instruments is determined based on
the characteristics of the required information and the goals of the research.
Research Design
Specifically, a pre-experimental single-group design with pre- and post-test evaluation
was used. According to Mills and Jordan (2022), the pretest-post-test structure is valuable in
educational and behavioral research because it allows describing and quantifying the change
within the group between both measurement moments. However, since it is a single-group design
without a comparison group, this structure does not allow isolating the effects of the intervention
against threats to internal validity such as the natural maturation of the participants, history, the
effect of the test itself, or regression to the mean. Consequently, these factors constitute inherent
limitations of the single-group pre-experimental design that are not controlled by the simple fact
of applying a pretest and a post-test but would require the inclusion of a control group or quasi-
experimental designs to be adequately addressed. Despite these limitations, this type of design is
adequate when the objective is to explore and describe the magnitude of the change within a group
after an intervention, without having comparison groups, obtaining evidence of improvement or
intra-group progress, although not causal evidence. Since a control group was not included, the
results are strictly interpreted as changes associated with the intervention and not as causal effects
attributable exclusively to it.
Intervention
The intervention had an effective duration of 12 weeks of instruction, comprising from
January 12 to April 10, 2026, and consisted of 36 sessions of 60 minutes each. ChatGPT (GPT-
4o model, OpenAI) was used as a supervised support tool. Throughout the 12 weeks, students
were guided through the four phases of the writing process: planning, drafting, revising, and
editing.
of the impact of an intervention, in this case, the use of the artificial intelligence tool ChatGPT in
the development of English written production (Dávila, 2022).
This study adopts a quantitative approach. According to Vizcaíno et al. (2023), it is based
on the collection and analysis of numerical data to identify trends and measure the scope of the
impact of an intervention. The reasoning is deductive, starting from the theory on the use of AI
in writing to test its effect through the statistical measurement of pretest and post-test results.
Medina et al. (2023) mention that the use of data collection instruments is determined based on
the characteristics of the required information and the goals of the research.
Research Design
Specifically, a pre-experimental single-group design with pre- and post-test evaluation
was used. According to Mills and Jordan (2022), the pretest-post-test structure is valuable in
educational and behavioral research because it allows describing and quantifying the change
within the group between both measurement moments. However, since it is a single-group design
without a comparison group, this structure does not allow isolating the effects of the intervention
against threats to internal validity such as the natural maturation of the participants, history, the
effect of the test itself, or regression to the mean. Consequently, these factors constitute inherent
limitations of the single-group pre-experimental design that are not controlled by the simple fact
of applying a pretest and a post-test but would require the inclusion of a control group or quasi-
experimental designs to be adequately addressed. Despite these limitations, this type of design is
adequate when the objective is to explore and describe the magnitude of the change within a group
after an intervention, without having comparison groups, obtaining evidence of improvement or
intra-group progress, although not causal evidence. Since a control group was not included, the
results are strictly interpreted as changes associated with the intervention and not as causal effects
attributable exclusively to it.
Intervention
The intervention had an effective duration of 12 weeks of instruction, comprising from
January 12 to April 10, 2026, and consisted of 36 sessions of 60 minutes each. ChatGPT (GPT-
4o model, OpenAI) was used as a supervised support tool. Throughout the 12 weeks, students
were guided through the four phases of the writing process: planning, drafting, revising, and
editing.
Vol. 13/ Núm. 3 2026 pág. 3103
Table 1
Phases of the writing process
Phase Activity Teacher's role ChatGPT's role
Planning (10
min)
Brainstorming,
organizational outline, and
topic selection
Guides idea
generation and
validates the outline
Generates guiding
questions and
suggests possible
thematic approaches
Drafting (20
min)
Production of the first draft
by the student
Supervises individual
work and addresses
any questions.
Does not intervene;
the student writes
autonomously
Assisted
revision (20
min)
The student enters their draft
into ChatGPT using the
structured prompt (see
below) and analyzes the
suggestions
Verifies that the
student contrasts
each suggestion with
their prior knowledge
and justifies their
decisions
Provides feedback in
five dimensions
(mechanics, structure
and syntax,
coherence, cohesion,
and lexicon)
Editing and
closing (10
min)
The student produces the
final version incorporating
the accepted improvements
and reflects on the changes
Collect the final
version and the
student's decision log
Does not intervene
Scope of Research
The scope of this research is quantitative and pre-experimental, with a single-group
design and pre- and post-test evaluation, aimed at exploring and describing the magnitude of intra-
group change in English written production after a pedagogical intervention with the artificial
intelligence tool ChatGPT (GPT-4o model). The study is delimited to a total population of 45
students. Since a control group was not included, the scope of the conclusions is limited to
evidencing the progress or improvement associated with the intervention through the analysis of
intra-subject differences (post-test − pretest) using descriptive statistics and the non-parametric
Wilcoxon signed-rank test. Consequently, the results are strictly interpreted as changes associated
with the intervention and not as causal effects attributable exclusively to it, recognizing as
inherent limitations the impossibility of isolating threats to internal validity such as natural
maturation, history, the effect of the test itself, or regression to the mean.
Population and Sample
La The accessible population for this study consisted of 45 students enrolled at the
"International Learning" English Institute, located in the city of Tulcán, Carchi province, Ecuador.
The participants were aged between 14 and 22 years and had English proficiency levels ranging
Table 1
Phases of the writing process
Phase Activity Teacher's role ChatGPT's role
Planning (10
min)
Brainstorming,
organizational outline, and
topic selection
Guides idea
generation and
validates the outline
Generates guiding
questions and
suggests possible
thematic approaches
Drafting (20
min)
Production of the first draft
by the student
Supervises individual
work and addresses
any questions.
Does not intervene;
the student writes
autonomously
Assisted
revision (20
min)
The student enters their draft
into ChatGPT using the
structured prompt (see
below) and analyzes the
suggestions
Verifies that the
student contrasts
each suggestion with
their prior knowledge
and justifies their
decisions
Provides feedback in
five dimensions
(mechanics, structure
and syntax,
coherence, cohesion,
and lexicon)
Editing and
closing (10
min)
The student produces the
final version incorporating
the accepted improvements
and reflects on the changes
Collect the final
version and the
student's decision log
Does not intervene
Scope of Research
The scope of this research is quantitative and pre-experimental, with a single-group
design and pre- and post-test evaluation, aimed at exploring and describing the magnitude of intra-
group change in English written production after a pedagogical intervention with the artificial
intelligence tool ChatGPT (GPT-4o model). The study is delimited to a total population of 45
students. Since a control group was not included, the scope of the conclusions is limited to
evidencing the progress or improvement associated with the intervention through the analysis of
intra-subject differences (post-test − pretest) using descriptive statistics and the non-parametric
Wilcoxon signed-rank test. Consequently, the results are strictly interpreted as changes associated
with the intervention and not as causal effects attributable exclusively to it, recognizing as
inherent limitations the impossibility of isolating threats to internal validity such as natural
maturation, history, the effect of the test itself, or regression to the mean.
Population and Sample
La The accessible population for this study consisted of 45 students enrolled at the
"International Learning" English Institute, located in the city of Tulcán, Carchi province, Ecuador.
The participants were aged between 14 and 22 years and had English proficiency levels ranging
Vol. 13/ Núm. 3 2026 pág. 3104
from A1 to B2, according to the Common European Framework of Reference for Languages
(CEFR), a classification determined through the institutional placement test applied at the time of
enrollment. The inclusion criteria established that students had to be enrolled in the course under
study during the intervention period, have a proficiency level between A1 and B2 according to
the CEFR, and participate voluntarily. On the other hand, the exclusion criteria contemplated
irregular attendance (less than 80% of the sessions) and dropping out of the program before its
completion.
Given the reduced and finite nature of the target population (N < 50), probabilistic or
random sampling methods were not used; instead, a total population non-probabilistic sampling
approach was used (Cohen et al., 2018). In practice, all 45 students who met the inclusion criteria
completed the process without incurring the exclusion criteria, maintaining an attendance equal
to or greater than 80% of the sessions. Consequently, no dropout rate, loss of participants, or
exclusions due to absenteeism were recorded, so the analyzed sample coincided exactly with the
entire accessible population (n = 45) in both the pretest and posttest and in all subsequent
statistical analyses. It should be noted that, to mitigate the possible effect of the heterogeneity of
levels (A1–B2) as a confounding variable, the score obtained in the pretest was considered as an
individual baseline, so that the analysis of the change focused on the intra-subject difference (post-
test − pretest) and not exclusively on the absolute score of the posttest.
Table 2
Population
Population= 45 students
Sample Frequency Percentage
A1 9 20%
A2 13 29%
B1 15 33%
B2 8 185
Total 45 100%
Source: Data obtained by the academic institution. “International Learning”
Data Collection
Data collection with the 45 students was structured in three phases. The diagnostic phase
included the application of a 15-item questionnaire on AI use and a written production pretest.
During the intervention phase, students carried out writing activities assisted by ChatGPT under
teacher mediation. Finally, in the evaluation phase, a post-test graded with the same analytical
rubric as the pretest (5-20 points) was applied to measure progress. Prior to informed consent, all
data were anonymized and safeguarded for exclusively academic purposes.
from A1 to B2, according to the Common European Framework of Reference for Languages
(CEFR), a classification determined through the institutional placement test applied at the time of
enrollment. The inclusion criteria established that students had to be enrolled in the course under
study during the intervention period, have a proficiency level between A1 and B2 according to
the CEFR, and participate voluntarily. On the other hand, the exclusion criteria contemplated
irregular attendance (less than 80% of the sessions) and dropping out of the program before its
completion.
Given the reduced and finite nature of the target population (N < 50), probabilistic or
random sampling methods were not used; instead, a total population non-probabilistic sampling
approach was used (Cohen et al., 2018). In practice, all 45 students who met the inclusion criteria
completed the process without incurring the exclusion criteria, maintaining an attendance equal
to or greater than 80% of the sessions. Consequently, no dropout rate, loss of participants, or
exclusions due to absenteeism were recorded, so the analyzed sample coincided exactly with the
entire accessible population (n = 45) in both the pretest and posttest and in all subsequent
statistical analyses. It should be noted that, to mitigate the possible effect of the heterogeneity of
levels (A1–B2) as a confounding variable, the score obtained in the pretest was considered as an
individual baseline, so that the analysis of the change focused on the intra-subject difference (post-
test − pretest) and not exclusively on the absolute score of the posttest.
Table 2
Population
Population= 45 students
Sample Frequency Percentage
A1 9 20%
A2 13 29%
B1 15 33%
B2 8 185
Total 45 100%
Source: Data obtained by the academic institution. “International Learning”
Data Collection
Data collection with the 45 students was structured in three phases. The diagnostic phase
included the application of a 15-item questionnaire on AI use and a written production pretest.
During the intervention phase, students carried out writing activities assisted by ChatGPT under
teacher mediation. Finally, in the evaluation phase, a post-test graded with the same analytical
rubric as the pretest (5-20 points) was applied to measure progress. Prior to informed consent, all
data were anonymized and safeguarded for exclusively academic purposes.
Vol. 13/ Núm. 3 2026 pág. 3105
Validity and Reliability of the Data Collection Instrument
The instruments were subjected to validation through expert judgment. According to
Venegas et al. (2025), ensuring validity implies controlling extraneous factors to establish internal
validity and confirm that the instrument measures the intended construct, thus ensuring the
accuracy of the conclusions about students' writing skills. Three specialists in English teaching,
educational research, and instrument design evaluated the pertinence, clarity, coherence, and
correspondence of each item with the variables and dimensions of the study. After the review, the
experts issued their conformity, approving the definitive structure of the questionnaire and the
analytical rubric without major observations, which certified the content validity of both
instruments for their application in the selected sample.
Table 3
Reliability statistics
Dimensión Preguntas (Ítems) Valor Alfa de
Cronbach
1. Teacher's use of AI 1, 3 and 10 0.794
2. Student's use of AI
(Correction/Consultation)
2, 4, 5, 6, 7, 8 and 9 0.879
3. Written production
skills
11, 12, 13 and 14 0.768
Note: Value calculated using Excel.
According to the scaling criteria of George and Mallery (2016), all values exceed the
minimum threshold of acceptability (>0.70). This guarantees that the instrument has adequate
internal consistency and is statistically reliable to measure the study variables in the evaluated
population. Statistical processing was carried out excluding the nominal and single-item variables
(item 15), thus ensuring compliance with the covariance assumptions required by the Cronbach
model.
RESULTS AND DISCUSSION
Results related to the General Objective: determining the impact of the digital tool
ChatGPT on the development of the integrated written production skill
The comprehensive analysis of the data confirms that the pedagogical integration of
ChatGPT had a positive, statistically significant, and large-magnitude impact on the students'
writing competence. Quantitatively, the overall mean of grades increased from 12.96 to 16.51,
completely eradicating the "Insufficient" performance level (from 20% to 0%) and exponentially
increasing the "Outstanding" level (from 2.2% to 37.8%). The non-parametric Wilcoxon test (p <
0.001) and a large effect size (r = .87) validate that this improvement was not a product of chance,
but a direct and robust consequence of the intervention.
Validity and Reliability of the Data Collection Instrument
The instruments were subjected to validation through expert judgment. According to
Venegas et al. (2025), ensuring validity implies controlling extraneous factors to establish internal
validity and confirm that the instrument measures the intended construct, thus ensuring the
accuracy of the conclusions about students' writing skills. Three specialists in English teaching,
educational research, and instrument design evaluated the pertinence, clarity, coherence, and
correspondence of each item with the variables and dimensions of the study. After the review, the
experts issued their conformity, approving the definitive structure of the questionnaire and the
analytical rubric without major observations, which certified the content validity of both
instruments for their application in the selected sample.
Table 3
Reliability statistics
Dimensión Preguntas (Ítems) Valor Alfa de
Cronbach
1. Teacher's use of AI 1, 3 and 10 0.794
2. Student's use of AI
(Correction/Consultation)
2, 4, 5, 6, 7, 8 and 9 0.879
3. Written production
skills
11, 12, 13 and 14 0.768
Note: Value calculated using Excel.
According to the scaling criteria of George and Mallery (2016), all values exceed the
minimum threshold of acceptability (>0.70). This guarantees that the instrument has adequate
internal consistency and is statistically reliable to measure the study variables in the evaluated
population. Statistical processing was carried out excluding the nominal and single-item variables
(item 15), thus ensuring compliance with the covariance assumptions required by the Cronbach
model.
RESULTS AND DISCUSSION
Results related to the General Objective: determining the impact of the digital tool
ChatGPT on the development of the integrated written production skill
The comprehensive analysis of the data confirms that the pedagogical integration of
ChatGPT had a positive, statistically significant, and large-magnitude impact on the students'
writing competence. Quantitatively, the overall mean of grades increased from 12.96 to 16.51,
completely eradicating the "Insufficient" performance level (from 20% to 0%) and exponentially
increasing the "Outstanding" level (from 2.2% to 37.8%). The non-parametric Wilcoxon test (p <
0.001) and a large effect size (r = .87) validate that this improvement was not a product of chance,
but a direct and robust consequence of the intervention.
Vol. 13/ Núm. 3 2026 pág. 3106
Qualitatively, this impact did not derive from the passive or autonomous use of AI, but
from a supervised pedagogical mediation model structured in four phases (planning, drafting,
assisted revision, and editing). By requiring students to contrast and justify AI suggestions,
ChatGPT ceased to function as a mere automatic corrector. This strategy made it possible to
overcome initial deficiencies in grammar, cohesion, and lexicon, fostering autonomy,
metacognition, and critical thinking.
The numerical results obtained are based on the administration of a structured 15-item
questionnaire with a four-level Likert scale, as well as the application of a pre-test and post-test
for writing evaluated through an analytical rubric (score from 5 to 20 points). The items addressed
two main dimensions: on the one hand, the use and perception of artificial intelligence in the
English classroom (questions 1 to 10); and on the other, the students' self-perception regarding
their competence and difficulties in integrated writing skills (questions 11 to 14). Question 15
evaluated the preference for the tool. To interpret the survey results, the numerical values
corresponding to the response options were recorded into two groups to facilitate comparative
analysis. The first group included the options: “Always (A) + Almost Always (AA)”, which
grouped the favorable or higher frequency responses. The second group included the options:
“Sometimes (S) + Never (N)”, which grouped the lower frequency or unfavorable responses.
Results related to Specific Objective 1: Identification of difficulties and perceptions prior to
the intervention
To identify the main difficulties in the students' writing process and their perceptions of
AI tools before the implementation of ChatGPT, the results corresponding to the 15 questions of
the instrument were analyzed. Table 4 presents the consolidated frequency summary.
Table 4
Consolidated summary of survey response frequencies (N = 45)
Nro. Question / Item Focus Favourable
(A + AA)
Unfavorable
(S + N)
1 Teacher uses online AI tools as teaching
material
13 32
2 Students use ChatGPT to correct writing
errors
6 39
3 Teacher uses ChatGPT for vocabulary
exercises
19 26
4 Perception: AI facilitates English learning 34 11
5 Students use ChatGPT to correct grammatical
errors
8 37
6 Students use ChatGPT to correct spelling
errors
3 42
7 Students use ChatGPT to correct punctuation
errors
4 41
8 Students use ChatGPT to correct text
structure
10 35
9 Students use ChatGPT as a reference to
expand vocabulary
22 23
Qualitatively, this impact did not derive from the passive or autonomous use of AI, but
from a supervised pedagogical mediation model structured in four phases (planning, drafting,
assisted revision, and editing). By requiring students to contrast and justify AI suggestions,
ChatGPT ceased to function as a mere automatic corrector. This strategy made it possible to
overcome initial deficiencies in grammar, cohesion, and lexicon, fostering autonomy,
metacognition, and critical thinking.
The numerical results obtained are based on the administration of a structured 15-item
questionnaire with a four-level Likert scale, as well as the application of a pre-test and post-test
for writing evaluated through an analytical rubric (score from 5 to 20 points). The items addressed
two main dimensions: on the one hand, the use and perception of artificial intelligence in the
English classroom (questions 1 to 10); and on the other, the students' self-perception regarding
their competence and difficulties in integrated writing skills (questions 11 to 14). Question 15
evaluated the preference for the tool. To interpret the survey results, the numerical values
corresponding to the response options were recorded into two groups to facilitate comparative
analysis. The first group included the options: “Always (A) + Almost Always (AA)”, which
grouped the favorable or higher frequency responses. The second group included the options:
“Sometimes (S) + Never (N)”, which grouped the lower frequency or unfavorable responses.
Results related to Specific Objective 1: Identification of difficulties and perceptions prior to
the intervention
To identify the main difficulties in the students' writing process and their perceptions of
AI tools before the implementation of ChatGPT, the results corresponding to the 15 questions of
the instrument were analyzed. Table 4 presents the consolidated frequency summary.
Table 4
Consolidated summary of survey response frequencies (N = 45)
Nro. Question / Item Focus Favourable
(A + AA)
Unfavorable
(S + N)
1 Teacher uses online AI tools as teaching
material
13 32
2 Students use ChatGPT to correct writing
errors
6 39
3 Teacher uses ChatGPT for vocabulary
exercises
19 26
4 Perception: AI facilitates English learning 34 11
5 Students use ChatGPT to correct grammatical
errors
8 37
6 Students use ChatGPT to correct spelling
errors
3 42
7 Students use ChatGPT to correct punctuation
errors
4 41
8 Students use ChatGPT to correct text
structure
10 35
9 Students use ChatGPT as a reference to
expand vocabulary
22 23
Vol. 13/ Núm. 3 2026 pág. 3107
10 Teacher uses ChatGPT to provide writing
feedback
10 35
11 Self-assessment: Correct spelling in writing 22 23
12 Self-assessment: Correct punctuation in
writing
29 16
13 Self-assessment: Correct sentence structure 10 35
14 Self-assessment: Use of diverse vocabulary 9 36
15 Preferred AI tool for writing (ChatGPT) 40 5 (Other tools)
Note: The Always and + Almost Always options were grouped in the (Favorable) column, while
Sometimes and Never were combined in the (Unfavorable) column.
The data show that, regarding practical implementation in the classroom, most students
considered the systematic use of AI applications to be limited. In the questions about teacher
integration (Q1, Q3, Q10) and autonomous student use for grammatical and structural correction
(Q2, Q5, Q8), unfavorable responses clearly predominated. For example, 39 students indicated
that they rarely or never use ChatGPT to correct writing errors (Q2), and 35 reported that teachers
do not use it to provide writing feedback (Q10).
However, the perception changes when analyzing the potential usefulness of these tools
and specific superficial corrections. In Q4, a significant majority (34 students) recognized that AI
facilitates English learning. Furthermore, students reported greater use for discrete rule-based
tasks, such as spelling (Q6) and punctuation (Q7), although the overall frequency remained low.
Regarding self-perceived writing difficulties, the greatest weaknesses were identified in sentence
structure (Q8, Q13) and lexical diversity (Q9, Q14), where unfavorable responses exceeded 35
students. Finally, Q15 revealed a clear preference for ChatGPT (40 students) over other tools such
as Writing Aid or Grammarly, which underscores its perceived versatility for writing
development.
Results related to Specific Objectives 2 and 3: Didactic process and application of ChatGPT
Following the initial diagnosis, a 12-week didactic intervention (36 sessions) was
implemented. Students were guided through the four phases of the writing process (planning,
drafting, revision, and editing) using ChatGPT (GPT-4o) as a supervised mediation tool. Strict
protocols were applied to ensure the critical evaluation of AI suggestions and avoid direct
copying.
To evaluate the initial level of integrated competence in writing, a pre-test was
administered. Students wrote a composition of between 150 and 200 words about a previous trip.
The compositions were evaluated using an analytical rubric focused on grammar, structure,
coherence, cohesion, and lexical competence. Table 5 presents the results of the initial test.
10 Teacher uses ChatGPT to provide writing
feedback
10 35
11 Self-assessment: Correct spelling in writing 22 23
12 Self-assessment: Correct punctuation in
writing
29 16
13 Self-assessment: Correct sentence structure 10 35
14 Self-assessment: Use of diverse vocabulary 9 36
15 Preferred AI tool for writing (ChatGPT) 40 5 (Other tools)
Note: The Always and + Almost Always options were grouped in the (Favorable) column, while
Sometimes and Never were combined in the (Unfavorable) column.
The data show that, regarding practical implementation in the classroom, most students
considered the systematic use of AI applications to be limited. In the questions about teacher
integration (Q1, Q3, Q10) and autonomous student use for grammatical and structural correction
(Q2, Q5, Q8), unfavorable responses clearly predominated. For example, 39 students indicated
that they rarely or never use ChatGPT to correct writing errors (Q2), and 35 reported that teachers
do not use it to provide writing feedback (Q10).
However, the perception changes when analyzing the potential usefulness of these tools
and specific superficial corrections. In Q4, a significant majority (34 students) recognized that AI
facilitates English learning. Furthermore, students reported greater use for discrete rule-based
tasks, such as spelling (Q6) and punctuation (Q7), although the overall frequency remained low.
Regarding self-perceived writing difficulties, the greatest weaknesses were identified in sentence
structure (Q8, Q13) and lexical diversity (Q9, Q14), where unfavorable responses exceeded 35
students. Finally, Q15 revealed a clear preference for ChatGPT (40 students) over other tools such
as Writing Aid or Grammarly, which underscores its perceived versatility for writing
development.
Results related to Specific Objectives 2 and 3: Didactic process and application of ChatGPT
Following the initial diagnosis, a 12-week didactic intervention (36 sessions) was
implemented. Students were guided through the four phases of the writing process (planning,
drafting, revision, and editing) using ChatGPT (GPT-4o) as a supervised mediation tool. Strict
protocols were applied to ensure the critical evaluation of AI suggestions and avoid direct
copying.
To evaluate the initial level of integrated competence in writing, a pre-test was
administered. Students wrote a composition of between 150 and 200 words about a previous trip.
The compositions were evaluated using an analytical rubric focused on grammar, structure,
coherence, cohesion, and lexical competence. Table 5 presents the results of the initial test.
Vol. 13/ Núm. 3 2026 pág. 3108
Table 5
Initial test results: Writing competence level (N = 45)
Performance Level Score Range Number of Students Percentage
Outstanding 18–20 pts 1 2.2%
Satisfactory 14–17 pts 19 42.2%
Fair 11–13 pts 16 35.6%
Insufficient 5–10 pts 9 20.0%
Total 45 100.0%
Note: The initial test results reveal that most students obtained scores within satisfactory and fair performance levels.
The results of the initial test reveal that 55.6% of students obtained scores within the fair
and insufficient performance levels. These findings indicated that, although students possessed a
basic capacity to express ideas, they experienced considerable difficulties related to grammatical
precision, text organization, and lexical variety, which confirms the survey results. At the end of
the 12-week intervention, a final test was administered under identical conditions, in which
students were asked to write about a family memory. Table 6 presents the results of the final test.
Table 6
Final test results: Writing competence level (N = 45)
Performance Level Score Range Number of Students Percentage
Outstanding 18–20 pts 17 37.8%
Satisfactory 14–17 pts 23 51.1%
Fair 11–13 pts 5 11.1%
Insufficient 5–10 pts 0 0.0%
Total 45 100.0%
Note: The final test results show that no student obtained an Insufficient grade, while the number
of students classified as Fair decreased considerably.
The post-test results demonstrate a substantial improvement. It is worth noting that no
student obtained an Insufficient grade, and the Outstanding category increased exponentially from
1 to 17 students (37.8%). The overall means increased from 12.96 (SD = 3.01) in the pre-test to
16.51 (SD = 2.41) in the post-test.
Statistical analysis of writing competence
Prior to the comparative analysis, a Shapiro-Wilk normality test was applied to the paired
differences (post-test minus pre-test). The analysis yielded a W statistic of 0.928 with a p-value
of 0.008. Since p < 0.05, the normality hypothesis was rejected; therefore, a non-parametric test
was required. The Wilcoxon signed-rank test was applied to compare the pre-test and post-test
scores. The test yielded a Z statistic of 5.84 and a p-value < 0.001. The obtained p-value is
considerably lower than the established significance level (p < .05), which confirms that the
observed improvement is not due to chance. Furthermore, the calculated effect size (r = .87) is
classified as large, indicating that the magnitude of the improvement was substantial and
Table 5
Initial test results: Writing competence level (N = 45)
Performance Level Score Range Number of Students Percentage
Outstanding 18–20 pts 1 2.2%
Satisfactory 14–17 pts 19 42.2%
Fair 11–13 pts 16 35.6%
Insufficient 5–10 pts 9 20.0%
Total 45 100.0%
Note: The initial test results reveal that most students obtained scores within satisfactory and fair performance levels.
The results of the initial test reveal that 55.6% of students obtained scores within the fair
and insufficient performance levels. These findings indicated that, although students possessed a
basic capacity to express ideas, they experienced considerable difficulties related to grammatical
precision, text organization, and lexical variety, which confirms the survey results. At the end of
the 12-week intervention, a final test was administered under identical conditions, in which
students were asked to write about a family memory. Table 6 presents the results of the final test.
Table 6
Final test results: Writing competence level (N = 45)
Performance Level Score Range Number of Students Percentage
Outstanding 18–20 pts 17 37.8%
Satisfactory 14–17 pts 23 51.1%
Fair 11–13 pts 5 11.1%
Insufficient 5–10 pts 0 0.0%
Total 45 100.0%
Note: The final test results show that no student obtained an Insufficient grade, while the number
of students classified as Fair decreased considerably.
The post-test results demonstrate a substantial improvement. It is worth noting that no
student obtained an Insufficient grade, and the Outstanding category increased exponentially from
1 to 17 students (37.8%). The overall means increased from 12.96 (SD = 3.01) in the pre-test to
16.51 (SD = 2.41) in the post-test.
Statistical analysis of writing competence
Prior to the comparative analysis, a Shapiro-Wilk normality test was applied to the paired
differences (post-test minus pre-test). The analysis yielded a W statistic of 0.928 with a p-value
of 0.008. Since p < 0.05, the normality hypothesis was rejected; therefore, a non-parametric test
was required. The Wilcoxon signed-rank test was applied to compare the pre-test and post-test
scores. The test yielded a Z statistic of 5.84 and a p-value < 0.001. The obtained p-value is
considerably lower than the established significance level (p < .05), which confirms that the
observed improvement is not due to chance. Furthermore, the calculated effect size (r = .87) is
classified as large, indicating that the magnitude of the improvement was substantial and
Vol. 13/ Núm. 3 2026 pág. 3109
pedagogically relevant. Consequently, the null hypothesis (H₀) is rejected, and the alternative
hypothesis (H₁) is accepted, confirming a significant difference in writing competence after the
ChatGPT intervention.
Analysis of results
The initial survey results revealed a gap between the availability of AI tools and their
systematic pedagogical integration, since most students reported infrequent use of AI for
structural or grammatical correction. This coincides with Dou et al. (2025) and Farrokhnia et al.
(2024), who point out that, although students have positive attitudes towards AI, its pedagogical
effectiveness largely depends on structured implementation and teacher support. The high
preference for ChatGPT (Q15) and the recognition of its usefulness (Q4) support Adel et al.
(2024), who highlight that generative AI boosts student engagement when integrated into
appropriate frameworks.
The pre-test results confirmed the root causes of the problem: significant difficulties in
sentence structure, lexical diversity, and overall coherence. However, the post-test results and the
Wilcoxon statistical analysis (p < .001, r = .87) demonstrated that the 12-week mediated
intervention radically transformed these results. The elimination of the "Insufficient" category
and the increase in "Outstanding" results provide empirical evidence that ChatGPT, when used as
a mediation tool rather than a substitute for cognitive effort, significantly improves academic
writing.
These findings coincide with those of Carrillo et al. (2023) and Aliakbar (2024), who
found that AI integration improves textual clarity, linguistic precision, and lexical variety in
students' compositions. Furthermore, the substantial effect size supports Zhai and Nezakatgoo
(2025), who determined that AI facilitates personalized progress tracking, fostering metacognitive
processes that strengthen self-assessment. By requiring students to maintain three preliminary
drafts and justify their acceptance or rejection of AI suggestions, the intervention successfully
mitigated the risks of excessive dependence and erosion of critical thinking warned by Yan et al.
(2025) and Baidoo-Anu and Owusu Ansah (2023). Ultimately, the study confirms that automated
writing evaluation through artificial intelligence, when combined with explicit teacher guidance,
acts as a powerful catalyst for the development of comprehensive writing competence in EFL
contexts.
CONCLUSION
Diagnosis of initial difficulties and the usage gap: before the intervention, students at the
"International Learning" Institute presented marked deficiencies in written production,
particularly in sentence structuring, textual coherence, and lexical diversity. Although there were
a favorable perception of artificial intelligence and a clear preference for ChatGPT, its use was
pedagogically relevant. Consequently, the null hypothesis (H₀) is rejected, and the alternative
hypothesis (H₁) is accepted, confirming a significant difference in writing competence after the
ChatGPT intervention.
Analysis of results
The initial survey results revealed a gap between the availability of AI tools and their
systematic pedagogical integration, since most students reported infrequent use of AI for
structural or grammatical correction. This coincides with Dou et al. (2025) and Farrokhnia et al.
(2024), who point out that, although students have positive attitudes towards AI, its pedagogical
effectiveness largely depends on structured implementation and teacher support. The high
preference for ChatGPT (Q15) and the recognition of its usefulness (Q4) support Adel et al.
(2024), who highlight that generative AI boosts student engagement when integrated into
appropriate frameworks.
The pre-test results confirmed the root causes of the problem: significant difficulties in
sentence structure, lexical diversity, and overall coherence. However, the post-test results and the
Wilcoxon statistical analysis (p < .001, r = .87) demonstrated that the 12-week mediated
intervention radically transformed these results. The elimination of the "Insufficient" category
and the increase in "Outstanding" results provide empirical evidence that ChatGPT, when used as
a mediation tool rather than a substitute for cognitive effort, significantly improves academic
writing.
These findings coincide with those of Carrillo et al. (2023) and Aliakbar (2024), who
found that AI integration improves textual clarity, linguistic precision, and lexical variety in
students' compositions. Furthermore, the substantial effect size supports Zhai and Nezakatgoo
(2025), who determined that AI facilitates personalized progress tracking, fostering metacognitive
processes that strengthen self-assessment. By requiring students to maintain three preliminary
drafts and justify their acceptance or rejection of AI suggestions, the intervention successfully
mitigated the risks of excessive dependence and erosion of critical thinking warned by Yan et al.
(2025) and Baidoo-Anu and Owusu Ansah (2023). Ultimately, the study confirms that automated
writing evaluation through artificial intelligence, when combined with explicit teacher guidance,
acts as a powerful catalyst for the development of comprehensive writing competence in EFL
contexts.
CONCLUSION
Diagnosis of initial difficulties and the usage gap: before the intervention, students at the
"International Learning" Institute presented marked deficiencies in written production,
particularly in sentence structuring, textual coherence, and lexical diversity. Although there were
a favorable perception of artificial intelligence and a clear preference for ChatGPT, its use was
Vol. 13/ Núm. 3 2026 pág. 3110
limited, non-systematic, and lacked pedagogical guidance to leverage its potential for formative
feedback.
The application of ChatGPT (GPT-4o model) for 12 weeks, structured in the phases of
planning, drafting, assisted revision, and editing, proved to be a highly effective pedagogical
strategy. The teacher's role was decisive in preventing cognitive dependence, since by requiring
students to justify the acceptance or rejection of AI suggestions, critical thinking, autonomy, and
real appropriation of the writing process were fostered, rather than the mere copying of generated
texts.
The comparison between the pre-test and post-test reveals a statistically significant (p <
0.001) and large-magnitude (r = .87) improvement in the integrated written production skill. The
total eradication of the "Insufficient" level and the substantial increase of students in the
"Satisfactory" and "Outstanding" levels confirm that the immediate, personalized, and
multidimensional feedback (grammar, cohesion, coherence, and lexicon) provided by the AI,
under supervision, accelerates the development of writing competence in English as a second
language learners.
The research concludes that the integration of generative AI tools in the English
classroom should not be seen as a threat to learning, but as an opportunity to transform traditional
methods focused on memorization. For this integration to be successful and ethical, it is essential
that educational institutions train teachers in the design of effective prompts, establish responsible
use protocols, and promote a "dialogue of knowledge" where AI complements, but does not
replace, the student's cognitive effort and human guidance.
limited, non-systematic, and lacked pedagogical guidance to leverage its potential for formative
feedback.
The application of ChatGPT (GPT-4o model) for 12 weeks, structured in the phases of
planning, drafting, assisted revision, and editing, proved to be a highly effective pedagogical
strategy. The teacher's role was decisive in preventing cognitive dependence, since by requiring
students to justify the acceptance or rejection of AI suggestions, critical thinking, autonomy, and
real appropriation of the writing process were fostered, rather than the mere copying of generated
texts.
The comparison between the pre-test and post-test reveals a statistically significant (p <
0.001) and large-magnitude (r = .87) improvement in the integrated written production skill. The
total eradication of the "Insufficient" level and the substantial increase of students in the
"Satisfactory" and "Outstanding" levels confirm that the immediate, personalized, and
multidimensional feedback (grammar, cohesion, coherence, and lexicon) provided by the AI,
under supervision, accelerates the development of writing competence in English as a second
language learners.
The research concludes that the integration of generative AI tools in the English
classroom should not be seen as a threat to learning, but as an opportunity to transform traditional
methods focused on memorization. For this integration to be successful and ethical, it is essential
that educational institutions train teachers in the design of effective prompts, establish responsible
use protocols, and promote a "dialogue of knowledge" where AI complements, but does not
replace, the student's cognitive effort and human guidance.
Vol. 13/ Núm. 3 2026 pág. 3111
REFERENCES
Adel, A., Ahsan, A., & Davison, C. (2024). ChatGPT Promises and Challenges in Education:
Computational and Ethical Perspectives. In Education Sciences (Vol. 14, Number 8).
Multidisciplinary Digital Publishing Institute (MDPI).
https://doi.org/10.3390/educsci14080814
Aliakbar, T. (2024). Exploring the Potential of ChatGPT in EFL Language Learning: Learners’
Reflections and Practices. https://doi.org/10.20944/PREPRINTS202412.2218.V1
Asamblea Nacional del Ecuador. (2008). Constitución de la República del Ecuador 2008 (Decreto
Legislativo 0). Decreto Legislativo 0.
https://www.gob.ec/sites/default/files/regulations/2020-
06/CONSTITUCION%202008.pdf
Baidoo-Anu, D., & Owusu Ansah, L. (2023). Education in the Era of Generative Artificial
Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching
and Learning. Journal of AI, 7(1), 52–62. https://doi.org/10.61969/JAI.1337500
Baldrich, K., & Domínguez, J. C. (2024). El uso de ChatGPT en la escritura académica: Un
estudio de caso en educación. Revista de Medios y Educación.
https://recyt.fecyt.es/index.php/pixel/article/view/103527/79832
Carrillo, C. C. E., Herrera, B. V. A., & Cortes, S. J. N. (2023). Inteligencia Artificial para la
escritura académica en investigación. Ciencia Latina Revista Científica Multidisciplinar,
7(4), 4604–4621. https://doi.org/10.37811/CL_RCM.V7I4.7304
Código Orgánico Integral Penal, COIP (2021). https://www.defensa.gob.ec/wp-
content/uploads/downloads/2021/03/COIP_act_feb-2021.pdf
Cohen, L., Manion, L., & Morrison, K. (2018). Research Methods in Education (Routledge, Ed.;
8th ed.). Taylor & Francis Group. www.routledge.com/cw/cohen
Cordero, M. L. J., Pachar, L. E. R., & Granados, R. J. F. (2025, March). Ethics in the application
of artificial intelligence for the creation of academic work by students in the third year of
high school. 593 Digital Publisher CEIT.
https://www.593dp.com/index.php/593_Digital_Publisher/article/view/3150/2554
Dávila, A. (2022). El investigador y el paradigma de investigación. Revista Investigación,
Transcomplejidad y Ciencia, 3(2), 1–16.
https://revistasuba.com/index.php/INVESTIGACIONTRANSCOMPLEJIDADYCI/articl
e/view/210/139
Dou, A., Xu, W., Li, X., Zhang, S., & Zhang, J. (2025). Artificial Intelligence in Language
Learning: Bridging Gaps, Revealing Patterns, and Charting the Future. International
Journal of Distance Education Technologies, 23(1). https://doi.org/10.4018/IJDET.385045
REFERENCES
Adel, A., Ahsan, A., & Davison, C. (2024). ChatGPT Promises and Challenges in Education:
Computational and Ethical Perspectives. In Education Sciences (Vol. 14, Number 8).
Multidisciplinary Digital Publishing Institute (MDPI).
https://doi.org/10.3390/educsci14080814
Aliakbar, T. (2024). Exploring the Potential of ChatGPT in EFL Language Learning: Learners’
Reflections and Practices. https://doi.org/10.20944/PREPRINTS202412.2218.V1
Asamblea Nacional del Ecuador. (2008). Constitución de la República del Ecuador 2008 (Decreto
Legislativo 0). Decreto Legislativo 0.
https://www.gob.ec/sites/default/files/regulations/2020-
06/CONSTITUCION%202008.pdf
Baidoo-Anu, D., & Owusu Ansah, L. (2023). Education in the Era of Generative Artificial
Intelligence (AI): Understanding the Potential Benefits of ChatGPT in Promoting Teaching
and Learning. Journal of AI, 7(1), 52–62. https://doi.org/10.61969/JAI.1337500
Baldrich, K., & Domínguez, J. C. (2024). El uso de ChatGPT en la escritura académica: Un
estudio de caso en educación. Revista de Medios y Educación.
https://recyt.fecyt.es/index.php/pixel/article/view/103527/79832
Carrillo, C. C. E., Herrera, B. V. A., & Cortes, S. J. N. (2023). Inteligencia Artificial para la
escritura académica en investigación. Ciencia Latina Revista Científica Multidisciplinar,
7(4), 4604–4621. https://doi.org/10.37811/CL_RCM.V7I4.7304
Código Orgánico Integral Penal, COIP (2021). https://www.defensa.gob.ec/wp-
content/uploads/downloads/2021/03/COIP_act_feb-2021.pdf
Cohen, L., Manion, L., & Morrison, K. (2018). Research Methods in Education (Routledge, Ed.;
8th ed.). Taylor & Francis Group. www.routledge.com/cw/cohen
Cordero, M. L. J., Pachar, L. E. R., & Granados, R. J. F. (2025, March). Ethics in the application
of artificial intelligence for the creation of academic work by students in the third year of
high school. 593 Digital Publisher CEIT.
https://www.593dp.com/index.php/593_Digital_Publisher/article/view/3150/2554
Dávila, A. (2022). El investigador y el paradigma de investigación. Revista Investigación,
Transcomplejidad y Ciencia, 3(2), 1–16.
https://revistasuba.com/index.php/INVESTIGACIONTRANSCOMPLEJIDADYCI/articl
e/view/210/139
Dou, A., Xu, W., Li, X., Zhang, S., & Zhang, J. (2025). Artificial Intelligence in Language
Learning: Bridging Gaps, Revealing Patterns, and Charting the Future. International
Journal of Distance Education Technologies, 23(1). https://doi.org/10.4018/IJDET.385045
Vol. 13/ Núm. 3 2026 pág. 3112
Farrokhnia, M., Banihashem, S. K., Noroozi, O., & Wals, A. (2024). A SWOT analysis of
ChatGPT: Implications for educational practice and research. Innovations in Education and
Teaching International, 61(3), 460–474. https://doi.org/10.1080/14703297.2023.2195846
George, D., & Mallery, P. (2016). IBM SPSS Statistics 23 Step by Step A Simple Guide and
Reference fourteenth edition (Taylor & Francis, Ed.). www.routledge.com/9780134320250
Kohnke, L., Moorhouse, B. L., & Zou, D. (2023). ChatGPT for Language Teaching and Learning.
RELC Journal, 54(2), 537–550. https://doi.org/10.1177/00336882231162868
Lee, J., & Ko, Y. (2026). AI versus humans: Impact of IPA, AI apps, and peer interactions on the
English skills of EFL adolescents. System, 140, 104040.
https://doi.org/10.1016/J.SYSTEM.2026.104040
Liu, W. (2024). A Systematic Review of Automated Writing Evaluation Feedback: Validity,
Effects and Students’ Engagement. Language Teaching Research Quarterly, 45, 86–105.
https://doi.org/10.32038/ltrq.2024.45.05
Medina, R. M., Rojas, L. R., Bustamante, H. W., Loaiza, C. R., Martel, C. C., & Castillo, A. R.
(2023). METODOLOGÍA DE LA INVESTIGACIÓN TÉCNICAS E INSTRUMENTOS DE
INVESTIGACIÓN (Wilson Sucari, Patty Aza, & Antonio Flores, Eds.; Instituto
Universitario de Innovación Ciencia y Tecnología Inudi Perú S.A.C.
https://doi.org/10.35622/inudi.b.080
Mills, G. E., & Jordan, A. W. (2022). Educational Research: Competencies for Analysis and
Applications (13th ed.). https://www.pearson.com/en-us/subject-catalog/p/educational-
research-competencies-for-analysis-and-
applications/P200000001512/9780137534944?srsltid=AfmBOopLZZuTcNTfhvJnuVdkL
Kw-WZK9cJur8LTssmCgdchWk742MTBP
Ossa, C., & Willatt, C. (2023). Providing academic writing feedback assisted by Generative
Artificial Intelligence in initial teacher education contexts. European Journal of Education
and Psychology, 16(2). https://doi.org/10.32457/EJEP.V16I2.2412
Párraga Solórzano, R. J., Acosta Isidor, Z., Ricardo Suárez, J., & Limones Borbor, J. (2024).
Impacto de las estrategias metodológicas en el aprendizaje del idioma inglés en estudiantes
de nivel elemental en comunidades de la provincia de Santa Elena. LATAM Revista
Latinoamericana de Ciencias Sociales y Humanidades, 5(5).
https://doi.org/10.56712/LATAM.V5I5.2766
Román, M. G. M., Ibarra, B. S. N., Jordán, E. R. P., Chiluisa, E. A. P., Ibarra, R. E. N., & Pérez,
J. C. M. (2026). La inteligencia artificial generativa y la formación del juicio crítico en la
adolescencia. Revista Veritas de Difusão Científica, 7(2), 1555-1579.
Secretaría Nacional de Planificación. (2025, August). PLAN NACIONAL DE DESARROLLO
2025-2029.
Farrokhnia, M., Banihashem, S. K., Noroozi, O., & Wals, A. (2024). A SWOT analysis of
ChatGPT: Implications for educational practice and research. Innovations in Education and
Teaching International, 61(3), 460–474. https://doi.org/10.1080/14703297.2023.2195846
George, D., & Mallery, P. (2016). IBM SPSS Statistics 23 Step by Step A Simple Guide and
Reference fourteenth edition (Taylor & Francis, Ed.). www.routledge.com/9780134320250
Kohnke, L., Moorhouse, B. L., & Zou, D. (2023). ChatGPT for Language Teaching and Learning.
RELC Journal, 54(2), 537–550. https://doi.org/10.1177/00336882231162868
Lee, J., & Ko, Y. (2026). AI versus humans: Impact of IPA, AI apps, and peer interactions on the
English skills of EFL adolescents. System, 140, 104040.
https://doi.org/10.1016/J.SYSTEM.2026.104040
Liu, W. (2024). A Systematic Review of Automated Writing Evaluation Feedback: Validity,
Effects and Students’ Engagement. Language Teaching Research Quarterly, 45, 86–105.
https://doi.org/10.32038/ltrq.2024.45.05
Medina, R. M., Rojas, L. R., Bustamante, H. W., Loaiza, C. R., Martel, C. C., & Castillo, A. R.
(2023). METODOLOGÍA DE LA INVESTIGACIÓN TÉCNICAS E INSTRUMENTOS DE
INVESTIGACIÓN (Wilson Sucari, Patty Aza, & Antonio Flores, Eds.; Instituto
Universitario de Innovación Ciencia y Tecnología Inudi Perú S.A.C.
https://doi.org/10.35622/inudi.b.080
Mills, G. E., & Jordan, A. W. (2022). Educational Research: Competencies for Analysis and
Applications (13th ed.). https://www.pearson.com/en-us/subject-catalog/p/educational-
research-competencies-for-analysis-and-
applications/P200000001512/9780137534944?srsltid=AfmBOopLZZuTcNTfhvJnuVdkL
Kw-WZK9cJur8LTssmCgdchWk742MTBP
Ossa, C., & Willatt, C. (2023). Providing academic writing feedback assisted by Generative
Artificial Intelligence in initial teacher education contexts. European Journal of Education
and Psychology, 16(2). https://doi.org/10.32457/EJEP.V16I2.2412
Párraga Solórzano, R. J., Acosta Isidor, Z., Ricardo Suárez, J., & Limones Borbor, J. (2024).
Impacto de las estrategias metodológicas en el aprendizaje del idioma inglés en estudiantes
de nivel elemental en comunidades de la provincia de Santa Elena. LATAM Revista
Latinoamericana de Ciencias Sociales y Humanidades, 5(5).
https://doi.org/10.56712/LATAM.V5I5.2766
Román, M. G. M., Ibarra, B. S. N., Jordán, E. R. P., Chiluisa, E. A. P., Ibarra, R. E. N., & Pérez,
J. C. M. (2026). La inteligencia artificial generativa y la formación del juicio crítico en la
adolescencia. Revista Veritas de Difusão Científica, 7(2), 1555-1579.
Secretaría Nacional de Planificación. (2025, August). PLAN NACIONAL DE DESARROLLO
2025-2029.
Vol. 13/ Núm. 3 2026 pág. 3113
https://intranet.msp.gob.ec/images/Documentos/Descargas/013%20PlanNacionalDeDesar
rollo25-29_EcuadorNoSeDetiene_compressed-1.pdf
UNESCO. (2024). Guía para el uso de IA generativa en educación e investigación. Unesco, 1–
49. https://unesdoc.unesco.org/ark:/48223/pf0000389227.locale=en
Venegas, M. V. L., Esquivel, G. J., Gonzales, B. M. T., & Altamirano, J. Y. (2025). Validez y
confiabilidad en investigaciones en ciencias sociales y políticas. Análisis comparativo en
los enfoques cuantitativo, cualitativo y mixto. Cuestiones Políticas, 43(83), 97–121.
https://doi.org/10.5281/ZENODO.17757052
Vizcaíno, Z. P. I., Cedeño, C. R. J., & Maldonado, P. I. A. (2023). Metodología de la investigación
científica: guía práctica. Ciencia Latina Internacional, 7(4), 9723–9760.
https://ciencialatina.org/index.php/cienciala/article/view/7658/11619
Yan, Y., Liu, H., & Chau, T. (2025). A Systematic Review of AI Ethics in Education: Challenges,
Policy Gaps, and Future Directions. Journal of Global Information Management, 33(1).
https://doi.org/10.4018/JGIM.386381
Zhai, Y., & Nezakatgoo, B. (2025). Evaluating AI-Powered Applications for Enhancing
Undergraduate Students’ Metacognitive Strategies, Self-Determined Motivation, and
Social Learning in English Language Education. Scientific Reports, 15(1), 35199.
https://doi.org/10.1038/S41598-025-19118-Z
https://intranet.msp.gob.ec/images/Documentos/Descargas/013%20PlanNacionalDeDesar
rollo25-29_EcuadorNoSeDetiene_compressed-1.pdf
UNESCO. (2024). Guía para el uso de IA generativa en educación e investigación. Unesco, 1–
49. https://unesdoc.unesco.org/ark:/48223/pf0000389227.locale=en
Venegas, M. V. L., Esquivel, G. J., Gonzales, B. M. T., & Altamirano, J. Y. (2025). Validez y
confiabilidad en investigaciones en ciencias sociales y políticas. Análisis comparativo en
los enfoques cuantitativo, cualitativo y mixto. Cuestiones Políticas, 43(83), 97–121.
https://doi.org/10.5281/ZENODO.17757052
Vizcaíno, Z. P. I., Cedeño, C. R. J., & Maldonado, P. I. A. (2023). Metodología de la investigación
científica: guía práctica. Ciencia Latina Internacional, 7(4), 9723–9760.
https://ciencialatina.org/index.php/cienciala/article/view/7658/11619
Yan, Y., Liu, H., & Chau, T. (2025). A Systematic Review of AI Ethics in Education: Challenges,
Policy Gaps, and Future Directions. Journal of Global Information Management, 33(1).
https://doi.org/10.4018/JGIM.386381
Zhai, Y., & Nezakatgoo, B. (2025). Evaluating AI-Powered Applications for Enhancing
Undergraduate Students’ Metacognitive Strategies, Self-Determined Motivation, and
Social Learning in English Language Education. Scientific Reports, 15(1), 35199.
https://doi.org/10.1038/S41598-025-19118-Z