Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 112
AI-Enhanced EFL Teaching: Evidence from an
Ecuadorian Public High School
Enseñanza de inglés como lengua extranjera mejorada con IA:
Evidencia de una escuela secundaria pública Ecuatoriana
Vélez-Olvera, Ernesto Rafael
1
Jara-Barros, Christian Fernando
2
https://orcid.org/0009-0004-8201-5069
https://orcid.org/0009-0009-2093-0524
ernesto.velez@docentes.educacion.edu.ec
christian.jara@docentes.educacion.edu.ec
Unidad Educativa Fiscal Francisco de Orellana,
Ecuador, Guyas.
Unidad educativa José María Velasco Ibarra,
Ecuador, Guyas.
Piña-Roldán, Verónica Arianna
3
Ramos-Saltos, Lister Antonio
4
https://orcid.org/0009-0009-0623-8347
https://orcid.org/0009-0000-7673-7808
veronica.pina@nuevasemilla.com.ec
lister.ramos@docentes.educacion.edu.ec
Unidad Educativa Bilingüe Nueva Semilla, Ecuador,
Guyas.
Unidad educativa José María Velasco Ibarra,
Ecuador, Guyas.
Autor de correspondencia
1
DOI / URL: https://doi.org/10.55813/gaea/rcym/v4/n2/184
Resumen: Este estudio examina las percepciones de
docentes de inglés como lengua extranjera (EFL) sobre la
integración de la inteligencia artificial (IA) en una institución de
educación secundaria pública en Santa Elena, Ecuador,
mediante un diseño mixto secuencial explicativo. La Fase 1
consiste en una revisión sistemática de alcance (PRISMA-
ScR) que sintetiza 21 estudios empíricos publicados entre
2020 y 2025 sobre la enseñanza de EFL asistida por IA, con
énfasis en contextos con limitaciones de recursos y en
América Latina. La Fase 2 reporta hallazgos descriptivos del
instrumento ETPAIS, aplicado de forma censal a los ocho
docentes de EFL de la Unidad Educativa Otto Arosemena
Gómez (N = 8). Los resultados revelan actitudes positivas
hacia la IA como herramienta pedagógica complementaria
(D2: M = 4.13), pero percepciones de infraestructura
institucional sustancialmente insuficiente (D3: M = 2.31). El
desarrollo profesional emergió como la necesidad facilitadora
más urgente (Ítem 22: M = 4.63). Estos hallazgos convergen
con la evidencia internacional sobre conceptualizaciones
positivas de los docentes, al tiempo que divergen
significativamente respecto a la suficiencia de infraestructura,
confirmando que las estrategias de adopción derivadas de
contextos bien equipados no son directamente transferibles a
la educación secundaria pública ecuatoriana. Se discuten
implicaciones para la política educativa y la formación
docente.
Palabras clave: inteligencia artificial, percepciones docentes,
educación secundaria pública, Ecuador, métodos mixtos.
Received: 07/Mar/2026
Accepted: 02/Abr/2026
Published: 21/Abr/2026
Cita: Vélez-Olvera, E. R., Jara-Barros, C. F.,
Piña-Roldán, V. A., & Ramos-Saltos, L. A.
(2026). Enseñanza de inglés como lengua
extranjera mejorada con IA: Evidencia de una
escuela secundaria pública
Ecuatoriana. Revista Científica Ciencia Y
Método, 4(2), 112-
130. https://doi.org/10.55813/gaea/rcym/v4/n2
/184
Revista Científica Ciencia y Método (RCyM)
https://revistacym.com
revistacym@editorialgrupo-aea.com
info@editoriagrupo-aea.com
© 2026. Este artículo es un documento de
acceso abierto distribuido bajo los términos y
condiciones de la Licencia Creative
Commons, Atribución-NoComercial 4.0
Internacional.
Artículo Científico
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 113
Artículo Científico
Abril – Junio 2026
Abstract:
This study examines EFL teachers' perceptions of artificial intelligence (AI) integration
in a public secondary school in Santa Elena, Ecuador, employing a sequential
explanatory mixed-methods design. Phase 1 consists of a systematic scoping review
(PRISMA-ScR) synthesizing 21 peer-reviewed studies published between 2020 and
2025 on AI-enhanced EFL teaching, with emphasis on resource-constrained and Latin
American contexts. Phase 2 reports descriptive findings from the EFL Teachers'
Perceptions of AI Integration Survey (ETPAIS), a researcher-designed 24-item Likert
instrument administered to all eight EFL teachers at the Unidad Educativa Otto
Arosemena Gómez (N = 8; census sampling). Results reveal that teachers hold
positive attitudes toward AI as a complementary pedagogical tool (D2: M = 4.13) but
perceive institutional infrastructure as substantially inadequate for implementation (D3:
M = 2.31). Professional development emerged as the most urgently demanded
facilitating condition (Item 22: M = 4.63). These findings converge with international
evidence regarding positive teacher conceptualizations of AI while diverging
significantly regarding infrastructure sufficiency, confirming that adoption strategies
derived from well-resourced contexts are not directly transferable to Ecuadorian public
secondary education. Implications for educational policy and teacher training are
discussed.
Keywords: artificial intelligence, teachers' perceptions, public secondary education,
Ecuador, mixed methods.
1. Introduction
The rapid proliferation of artificial intelligence (AI) tools across educational settings
worldwide has prompted a fundamental re-examination of how language instruction is
designed, delivered, and experienced. In the field of English as a Foreign Language
(EFL) teaching, generative AI platforms, automated feedback systems, and adaptive
learning technologies have expanded the repertoire of pedagogical possibilities
available to educators, offering prospects for personalized vocabulary support, real-
time writing feedback, and interactive conversational practice at a scale previously
unattainable through conventional instruction (Bauer et al., 2025; Bayaga, 2025). As
these technologies become increasingly embedded in international discourse on
educational innovation, questions about their practical relevance, accessibility, and
transferability across diverse institutional contexts have acquired urgent theoretical
and policy significance.
The existing literature on AI-enhanced EFL instruction has expanded considerably
since the widespread adoption of large language models in educational contexts
beginning around 2020. Empirical studies have documented measurable benefits for
writing and reading skill development (Shafiee, 2025; Bauer et al., 2025), and
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 114
Artículo Científico
Abril – Junio 2026
perception research has consistently found that both pre-service and in-service
educators conceptualize AI as a complementary pedagogical resource rather than a
threat to professional practice (Derinalp & Halife, 2025). These findings, however,
derive predominantly from well-resourced institutional settings in Asia, North America,
and Western Europe — contexts characterized by reliable digital infrastructure,
systematic professional development pathways, and institutional policies that actively
support technology integration. The extent to which this evidence base speaks to the
realities of EFL instruction in under-resourced public secondary schools in Latin
America remains, at best, uncertain.
Latin America presents a distinctive regional profile with respect to AI adoption in
education. While gradual diffusion has been documented in higher education
institutions with established digital infrastructure (Salas-Pilco & Yang, 2022; De la
Torre & Baldeon-Calisto, 2024), systematic barriers persist across the broader
educational landscape. Okoye et al. (2023) documented constraints including
inadequate connectivity, insufficient device access, and limited professional
development support as structural obstacles that transcend individual pedagogical
resistance. Cruz et al. (2024) further identified uncertainty regarding AI
appropriateness and limited institutional endorsement as compounding factors. These
barriers manifest unevenly: well-resourced urban institutions and universities have
made measurable progress in AI integration, while public schools serving non-affluent
populations face compounded challenges that international adoption frameworks do
not adequately address.
Within this regional landscape, Ecuador occupies a particularly underexplored position.
Research on AI adoption has focused primarily on higher education contexts (Arce et
al., 2025; Barredo-Ibáñez et al., 2021), leaving public secondary schools — which
constitute the primary educational pathway for most Ecuadorian learners — without an
empirical foundation from which to evaluate AI integration strategies. This absence is
not a minor gap in the literature; it represents a structural blind spot with direct
implications for educational equity. AI adoption policies developed without localized
empirical evidence risk perpetuating a pattern in which innovation benefits concentrate
in contexts that are already well-resourced, systematically excluding the populations
that most depend on public secondary education.
The present study responds to this gap by providing empirical evidence on EFL
teachers' perceptions of AI integration at a public secondary school in Santa Elena,
Ecuador. Specifically, the Unidad Educativa Otto Arosemena Gómez serves as the
institutional site for this investigation, offering a representative case of a public
secondary school operating under conditions of genuine infrastructure constraint and
limited professional development provision. The study employs a sequential
explanatory mixed-methods design (Creswell & Creswell, 2018) structured in two
phases. Phase 1 consists of a systematic scoping review following PRISMA-ScR
guidelines (Tricco et al., 2018), synthesizing 21 peer-reviewed studies published
between 2020 and 2025 on AI-enhanced EFL teaching, with particular attention to
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 115
Artículo Científico
Abril – Junio 2026
resource-constrained and Latin American contexts. Phase 2 reports descriptive
findings from the EFL Teachers' Perceptions of AI Integration Survey (ETPAIS), a
researcher-designed 24-item Likert instrument administered census-wide to all eight
EFL teachers at the institution (N = 8).
Literature Review
Key Concepts
Artificial Intelligence (AI)
Artificial Intelligence refers to computer systems performing tasks requiring human
intelligence including machine learning and pattern recognition (Bauer et al., 2025). In
educational contexts, AI encompasses both traditional computer-assisted approaches
and emerging generative technologies that adapt to learner needs.
Computer-Assisted Language Learning (CALL)
CALL represents the systematic integration of technology in language instruction.
Historically, CALL evolved from drill-and-practice software to more interactive,
communicative approaches (Richards & Renandya, 2002). AI marks a paradigm
shift—moving from predetermined, static interactions to adaptive systems
personalizing instruction based on individual learner profiles and real-time
performance data (Bayaga, 2025).
Generative AI in language teaching
Generative AI platforms like ChatGPT introduce transformative capabilities through
large language models. These tools offer authentic language exposure, immediate
feedback, and communicative practice at scale. Unlike traditional CALL, generative AI
enables dynamic conversation, content generation, and diverse learning pathways
customized to individual needs.
The implementation of AI in Ecuadorian public high schools remains
understudied despite the region's growing interest in educational technology. While
research documents AI adoption challenges in Latin American contexts (Pérez-
Campdesuñer et al., 2025), empirical evidence from secondary schools in Ecuador is
remarkably scarce, particularly regarding teacher perceptions and implementation
challenges in resource-constrained environments.
AI Tools and Their Application in EFL Teaching: Analytical Overview
Contemporary AI applications present multifaceted pedagogical opportunities with
context-dependent efficacy. While generative AI platforms demonstrate documented
benefits for language skills—including improvements in writing and reading
comprehension—empirical evidence derives predominantly from optimal
implementation contexts. The transferability of these findings to resource-constrained
environments remains uncertain. Research documents positive effects on learner
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 116
Artículo Científico
Abril – Junio 2026
motivation and autonomy in well-resourced institutional settings, yet such studies
control for variables often absent in public contexts: adequate technical infrastructure,
consistent connectivity, and institutional support systems. A critical analytical gap
exists regarding implementation barriers in under-resourced settings. Riggs (2025)
identifies structural obstacles in under-resourced K-12 classrooms, while Similar
barriers characterize Latin American contexts, as documented by Okoye et al. (2023)
regarding digital infrastructure constraints. However, these findings are geographically
limited, leaving Ecuadorian contexts virtually unexplored. Teacher perceptions
critically mediate outcomes. Research reveals that educators conceptualize AI as
complementary rather than replacive technology, suggesting alignment with
contemporary pedagogy (Derinalp & Halife, 2025). However, this perspective assumes
adequate training and support—preconditions not universally available.
Implementation also raises concerns. While AI tools facilitate personalized pathways
and automated feedback, evidence of negative externalities including plagiarism
facilitation, learning dependency, and digital equity disparities warrants attention
(Winder et al., 2024). The Latin American context compounds these concerns: Okoye
et al. (2023) document how digital inequities intersect with existing educational
disparities. AI tool application demonstrates pedagogically defensible potential;
however, the existing knowledge base insufficiently addresses implementation in
resource-limited contexts characteristic of Ecuadorian public schools.
AI and Language Skills Development
The empirical literature examining AI's impact on English language skills reveals
pronounced imbalances in research focus and geographic concentration. While
evidence demonstrates efficacy in writing and reading (Shafiee, 2025; Bauer et al.,
2025) predominantly from developed contexts spoken communication and listening
remain largely absent from empirical investigation.
Writing Skills Development
AI-assisted writing tools demonstrate documented benefits for learner autonomy and
metacognitive development (Bayaga, 2025). Real-time feedback mechanisms
enhance students' ability to plan, monitor, and evaluate writing processes
independently. This evidence suggests that AI's interactive features support self-
regulated writing practices, foundational for sustained language learning (Richards &
Renandya, 2002).
Reading Comprehension and Critical Literacy
AI intervention significantly enhances L2 reading comprehension through personalized
vocabulary support and adaptive scaffolding (Shafiee, 2025). Research demonstrates
that AI-powered questioning transforms passive reading into active inquiry, enabling
students to develop stronger analytical reading skills. This evidence suggests AI's
personalization capacity effectively supports reading comprehension across multiple
dimensions.
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 117
Artículo Científico
Abril – Junio 2026
Speaking and Listening Skills
A pronounced absence of empirical research characterizes AI's impact on oral
communication competencies. While generative AI chatbots theoretically support
conversational practice, empirical measurement of AI's efficacy for speaking skill
development including pronunciation accuracy, fluency, or real-time communication
competence remains virtually absent. Research documents that AI chatbots facilitate
student engagement through interactive dialogue; however, engagement outcomes
measured do not necessarily correlate with specific speaking or listening skill
development (Bayaga, 2025). This gap is particularly pronounced in Latin American
and Ecuadorian contexts, where research on any AI-enhanced language development
is limited (Salas-Pilco & Yang, 2022; Okoye et al., 2023).
Implications of the Research Landscape
The documented evidence reflects both technological and research-priority patterns
(Bauer et al., 2025). AI systems optimized for text-based interaction demonstrate clear
pedagogical benefits for writing and reading skills amenable to asynchronous,
feedback-intensive learning. Conversely, oral communication competencies requiring
real-time interaction remain underexplored (Shafiee, 2025). This imbalance suggests
that comprehensive AI implementation in EFL contexts must explicitly address whether
current text-based AI tools adequately serve oral skill development, or whether
implementation strategies require supplementary technologies targeting speaking and
listening.
Teachers’ and Students’ Perceptions of AI in EFL Contexts
Teachers' and students' perceptions critically mediate implementation success, as
stakeholder attitudes fundamentally shape integration outcomes (Derinalp & Halife,
2025). However, perception research exhibits pronounced geographic and contextual
imbalances, with dominant representation from developed contexts and limited
investigation in secondary school and resource-constrained settings (Okoye et al.,
2023).
Teachers’ Attitudes and Conceptualizations
EFL educators across diverse contexts conceptualize AI as a complementary
pedagogical resource rather than replacive technology. However, teacher attitudes
vary significantly based on context and prior experience with technology. Prospective
teachers express greater optimism than in-service educators (Derinalp & Halife, 2025),
while secondary school teachers in resource-constrained environments demonstrate
more cautious, conditional perspectives. Teachers in resource-constrained contexts
reveal more nuanced assessments. Riggs (2025) found that teachers in under-
resourced K-12 classrooms recognize AI's potential while remaining realistic about
barriers. Rural and resource-constrained educators across Latin America express
similarly qualified perspectives shaped by limited connectivity and resources (Okoye
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 118
Artículo Científico
Abril – Junio 2026
et al., 2023). These findings indicate that teacher perceptions are contextually situated
understandings of feasibility, not individual dispositions.
Students’ Perceptions
While teacher perception research provides substantial foundation, student perception
research remains substantially limited, particularly in secondary school contexts.
Available evidence derives almost exclusively from university-level students: Research
documents that university-level students hold positive perceptions of AI-assisted
writing tools (Derinalp & Halife, 2025), though secondary school contexts remain
underexplored. However, secondary school student perceptions remain virtually
absent. This gap is critically pronounced in Latin American and Ecuadorian contexts,
where neither teacher nor student perception data on AI integration in EFL instruction
exists (Okoye et al., 2023; Salas-Pilco & Yang, 2022).
Contextual Influences on Perceptions
Context-specific perception research reveals systematic patterns in how institutional
constraints shape stakeholder attitudes. Secondary school teachers across various
contexts recognize pedagogical possibilities while acknowledging institutional barriers
including insufficient professional development and inadequate technical support.
Riggs (2025) examined under-resourced K-12 classrooms, revealing that teacher
perceptions are fundamentally shaped by recognition of infrastructure deficits and
training limitations. These findings suggest that positive perceptions from well-
resourced contexts may reflect context-specific conditions enabling implementation
rather than universal teacher optimism. The absence of perception research from
Ecuadorian contexts prevents understanding whether teachers and students in
resource-limited public-school settings hold comparable attitudes or whether distinct
barriers generate different perception patterns.
Students Engagement, Motivation, and Autonomy in AI-Enhanced Learning
Beyond measurable language skill outcomes, research increasingly examines how AI
integration influences affective and metacognitive dimensions of language learning
(Bayaga, 2025). Studies from well-resourced contexts document positive effects on
learner motivation and autonomy through AI-mediated instruction. These
psychological constructs are critical to sustained language development and have
emerged as important indicators of AI's educational value in EFL contexts (Bauer et
al., 2025). However, such studies derive from optimal implementation conditions where
technology functions reliably and institutional support is adequate.
AI Adoption in Latin America and Ecuador; Regional Landscape and Local Context
While AI adoption in language education has expanded globally, Latin America
presents a distinct regional profile characterized by uneven technological diffusion and
context-specific barriers. Salas-Pilco & Yang (2022) conducted a systematic review of
AI applications in Latin American higher education, identifying gradual adoption with
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 119
Artículo Científico
Abril – Junio 2026
significant geographic variation. De la Torre & Baldeon-Calisto (2024) documented that
GenAI applications concentrate in well-resourced institutions with established digital
infrastructure. This concentration reflects access disparities and institutional capacity
differences advantaging affluent, urban institutions over public and rural schools.
The Latin American landscape reveals systematic barriers to broader adoption
extending beyond awareness or resistance. Okoye et al. (2023) documented
constraints including infrastructure deficits, inconsistent connectivity, and inadequate
professional development—structural obstacles transcending pedagogical innovation.
Cruz et al. (2024) identified complementary obstacles including limited institutional
support and uncertainty regarding AI appropriateness. These barriers manifest
unevenly: well-resourced urban institutions progress while rural and public-school
contexts face compounded challenges. Pérez-Campdesuñer et al. (2025) identified
that adoption facilitators concentrate in formal sectors with adequate resources while
obstacles accumulate in resource-constrained contexts—findings with direct
implications for Ecuador's public secondary schools.
2. Metodology
Overall Research Design
This study employs a sequential explanatory mixed-methods design (Creswell &
Creswell, 2018) structured in two sequential phases. Phase 1 consists of a systematic
scoping review following PRISMA-ScR guidelines (Tricco et al., 2018), synthesizing
existing empirical literature on AI integration in EFL teaching with emphasis on
resource-constrained and Latin American contexts. Phase 2 consists of a quantitative
survey-based inquiry examining EFL teacher perceptions of AI tools at a public
secondary school in Ecuador. The two phases are integrated at the discussion stage,
where locally generated empirical evidence is interpreted in light of the patterns and
gaps identified in the systematic review.
A mixed-methods approach is methodologically appropriate for three reasons. First,
the systematic review establishes the international evidence base and documents the
critical absence of Ecuadorian secondary school data, thereby providing the theoretical
warrant for local empirical inquiry. Second, quantitative perception data from Phase 2
allow direct comparison with findings from well-resourced contexts documented in
Phase 1. Third, the sequential structure ensures that instrument development in Phase
2 is grounded in the constructs and dimensions identified through the review,
enhancing content validity (Leavy, 2022).
Phase1: Systematic Scoping Review
Design and Rationale
A systematic scoping review design was selected to map the existing evidence base
on AI-enhanced EFL teaching, identify implementation patterns, and document
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 120
Artículo Científico
Abril – Junio 2026
barriers specific to resource-constrained educational environments. This design is
particularly appropriate given: (1) the proliferation of AI-in-education literature requiring
synthesis; (2) the identified gap regarding Ecuadorian and Latin American secondary
contexts; and (3) the need to generate constructs that inform Phase 2 instrument
development.
Research Questions (Phase 1)
The systematic review addresses four overarching research questions:
1. What empirical evidence exists regarding the efficacy of AI-enhanced
instruction for EFL skill development (writing, reading, speaking, listening)
across diverse educational contexts?
2. How do EFL teachers and students in various contexts perceive and
conceptualize AI integration, and what contextual factors shape these
perceptions?
3. What barriers and facilitators characterize AI adoption in language education,
particularly in resource-limited and Latin American contexts?
4. What implementation strategies, recommendations, and theoretical frameworks
emerge from existing literature as relevant to AI-enhanced EFL teaching in
under-resourced public secondary school contexts?
Search Strategy and Information Sources
A comprehensive search strategy was developed to identify peer-reviewed empirical
studies, literature reviews, and conceptual analyses published between 2020 and
2025. This temporal window captures the period coinciding with widespread AI
adoption in education following the release of large-scale generative AI tools, ensuring
contemporary relevance while identifying foundational work. Primary databases
searched included ERIC, Web of Science Core Collection, Scopus, PubMed, and
Google Scholar as a supplementary source for citation tracking. Search terms were
applied in English and Spanish using Boolean operators (AND, OR, NOT) across
primary, secondary, contextual, regional, and perception-focused term clusters.
Inclusion and Exclusion Criteria
Studies were included if they were peer-reviewed, published between 2020 and 2025,
written in English or Spanish, and examined AI applications in language education
contexts involving secondary, post-secondary, or adult learners or language
educators. Studies were excluded if they examined AI outside language education,
presented purely theoretical or opinion-based arguments without empirical grounding,
focused exclusively on pre-primary learners, or lacked sufficient methodological detail
for quality assessment.
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 121
Artículo Científico
Abril – Junio 2026
Study Selection, Data Extraction, and Quality Assessment
A two-stage screening process was implemented by two independent reviewers:
title/abstract screening followed by full-text evaluation. Disagreements were resolved
through discussion or third-reviewer consultation. The selection process is
documented using a PRISMA-ScR flow diagram. Data extraction employed a
standardized form capturing bibliographic information, study design, sample
characteristics, AI tools examined, language skills targeted, key findings, and
contextual barriers or facilitators. Quality was assessed using GRADE criteria adapted
for qualitative and mixed-methods research, informing synthesis rather than serving as
exclusion criterion.
Phase 2: Quantitative Survey Study
Research Questions (Phase 2)
The quantitative phase addresses two specific research questions derived from the
gaps identified in Phase 1:
1. What are the perceptions of EFL teachers at a public secondary school in Santa
Elena, Ecuador, regarding the integration of AI tools in their pedagogical
practice?
2. What institutional barriers and facilitators do these teachers identify as
mediating their adoption or non-adoption of AI-enhanced instruction?
Setting
The study was conducted at the Unidad Educativa Otto Arosemena Gómez, a public
secondary institution located in the canton of Santa Elena, Province of Santa Elena,
Ecuador. This institution was purposively selected as representative of the public
secondary school context described in the research gap: a resource-constrained, non-
urban setting serving predominantly non-affluent student populations with limited
digital infrastructure. This contextual profile aligns with the under-researched
educational environment identified through Phase 1.
Participants and Sampling
The target population consisted of all EFL teachers currently employed at the institution
during the 2024–2025 academic year. A census sampling approach was adopted,
including all eight (N = 8) EFL teachers as participants. Census sampling is
methodologically appropriate when the target population is small and fully accessible,
as it eliminates sampling error and maximizes representativeness within the
institutional context (Creswell & Creswell, 2018).
Instrument
Data were collected using a researcher-designed, close-ended questionnaire titled
EFL Teachers' Perceptions of AI Integration Survey (ETPAIS). The instrument was
developed in three stages: (1) item generation grounded in the four dimensions
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 122
Artículo Científico
Abril – Junio 2026
identified through Phase 1 (AI efficacy for language skills, teacher perceptions and
attitudes, institutional barriers and facilitators, and implementation strategies); (2)
expert review by three specialists in EFL methodology and educational technology;
and (3) pilot testing for clarity and internal consistency.
The final instrument comprises 24 items distributed across four dimensions, rated on
a five-point Likert scale ranging from 1 (Strongly Disagree) to 5 (Strongly Agree). Table
2 presents the dimensional structure of the instrument.
3. Results
Xsions, reflecting distinct patterns of perception among EFL teachers at the institution.
Dimension 2 (Teacher Attitudes and Conceptualizations) obtained the highest mean
score (M = 3.98), interpreted as "Agree," while Dimension 3 (Institutional Barriers and
Facilitators) recorded the lowest mean (M = 2.35), interpreted as "Disagree."
Dimensions 1 and 4 obtained intermediate scores, interpreted as "Neutral" and
"Agree," respectively.
Table 1
Descriptive Statistics by Dimension — ETPAIS (N = 8)
Dim.
Construct
M
SD
Interpretation
D1
AI Efficacy for Language Skills
3.04
0.42
Neutral
D2
Teacher Attitudes and
Conceptualizations
3.98
0.26
Agree
D3
Institutional Barriers and
Facilitators
2.35
0.27
Disagree
D4
Implementation Strategies
3.48
0.52
Agree
Note. M = mean; SD = standard deviation. Interpretation based on five-level scale: 1.00–1.79 Strongly
Disagree; 1.80–2.59 Disagree; 2.60–3.39 Neutral; 3.40–4.19 Agree; 4.20–5.00 Strongly Agree (Authors,
2026).
Despite the small sample size, some heterogeneity is observable across participants,
particularly in D3 where infrastructure perceptions vary most markedly.
3.1. Dimension 1: AI Efficacy for Language Skills
Dimension 1 obtained an overall mean of M = 3.04 (SD = 0.42), interpreted as
"Neutral." Teachers expressed moderate perceptions regarding AI's pedagogical
effectiveness for language skill development. As shown in Table 6, the highest-rated
items within this dimension were Item 1 (AI writing tools; M = 3.50) and Item 5 (real-
time feedback for self-regulation; M = 3.50), both reflecting existing awareness of AI's
documented benefits in writing-related tasks. In contrast, Items 3 and 4 — addressing
speaking practice and listening skill development — obtained the lowest means within
the dimension (M = 2.75 and M = 2.38, respectively), consistent with the research gap
identified in Phase 1 regarding the limited empirical evidence for AI's efficacy in oral
communication. Item 6, which asked whether international research findings were
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 123
Artículo Científico
Abril – Junio 2026
applicable to their students, obtained M = 2.63, suggesting cautious uncertainty
regarding transferability to their specific context.
Table 2
Item-Level Descriptive Statistics — D1: AI Efficacy for Language Skills (N = 8)
Item
Statement (abbreviated)
M
SD
Interpretation
1
AI writing tools help my students improve
their written production in English.
3.50
0.50
Agree
2
AI-powered reading platforms enhance
reading comprehension through
personalized vocabulary support.
3.50
0.50
Agree
3
AI chatbots provide meaningful speaking
practice opportunities for EFL learners in my
context.
2.75
0.66
Neutral
4
AI tools support listening skills development
through authentic audio and interactive
feedback.
2.38
0.48
Disagree
5
AI applications offer real-time feedback that
helps students self-regulate their learning.
3.50
0.50
Agree
6
Benefits of AI documented in international
research are applicable to my students.
2.63
0.48
Neutral
—
D1 Dimension Mean
3.04
0.42
Neutral
Note: M = mean; SD = standard deviation. Items rated on a five-point Likert scale (1 = Strongly Disagree
to 5 = Strongly Agree) (Authors, 2026).
3.2. Dimension 2: Teacher Attitudes and Conceptualizations
Dimension 2 yielded the highest mean score across all dimensions (M = 3.98, SD =
0.26), interpreted as "Agree." This finding indicates that EFL teachers at the institution
hold generally positive attitudes toward AI integration and conceptualize these tools as
complementary to — rather than replacive of — their pedagogical practice. As shown
in Table 7, Item 7 (AI as complementary resource; M = 4.50) and Item 9 (willingness
to invest time in AI learning; M = 4.13) obtained the highest means within the
dimension, suggesting strong alignment with the complementary conceptualization
documented in international literature (Derinalp & Halife, 2025). Item 11, which
addressed concern about AI-facilitated academic dishonesty, obtained M = 3.25 — a
moderate score indicating that plagiarism-related concerns, while present, did not
dominate teacher perceptions in this context.
Table 3
Item-Level Descriptive Statistics — D2: Teacher Attitudes and Conceptualizations (N
= 8)
Item
Statement (abbreviated)
M
SD
Interpretation
7
I consider AI tools to be a complementary
resource rather than a replacement for
teaching.
4.50
0.50
Strongly Agree
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 124
Artículo Científico
Abril – Junio 2026
8
I feel confident in my ability to integrate AI
tools into my EFL lessons.
3.63
0.48
Agree
9
I am willing to invest time in learning how to
use AI for language teaching.
4.13
0.60
Agree
10
I believe AI integration improves student
motivation and engagement.
4.25
0.43
Strongly Agree
11
I am concerned that AI tools may promote
academic dishonesty among my students.
3.25
0.66
Neutral
12
I believe AI-enhanced instruction fosters
learner autonomy in my context.
4.13
0.60
Agree
—
D2 Dimension Mean
3.98
0.26
Agree
Note: M = mean; SD = standard deviation. Item 11 is negatively framed; higher scores indicate greater
concern (Authors, 2026).
3.3. Dimension 3: Institutional Barriers and Facilitators
Dimension 3 recorded the lowest mean score of the instrument (M = 2.35, SD = 0.27),
interpreted as "Disagree." This result constitutes the most striking finding of Phase 2
and indicates that teachers perceive the institutional environment as substantially
inadequate for supporting AI integration. Item-level analysis (Table 8) reveals that
Items 13, 14, and 15 — addressing infrastructure sufficiency, professional
development, and technical support — all obtained means below 2.60, interpreted as
"Disagree." Specifically, Item 14 (adequate professional development; M = 1.88) and
Item 15 (institutional technical support; M = 1.88) were the lowest-rated items in the
entire instrument. Conversely, Item 16 (limited access as a significant barrier; M = 4.50)
obtained the highest score within the dimension and approached the "Agree"
threshold, confirming that teachers explicitly recognize connectivity and device access
as primary obstacles. Item 17 (student device access; M = 2.13) suggests moderate
uncertainty regarding students' personal technology resources.
Table 4
Item-Level Descriptive Statistics — D3: Institutional Barriers and Facilitators (N = 8)
Item
Statement (abbreviated)
M
SD
Interpretation
13
Digital infrastructure at my institution is
sufficient to implement AI tools in EFL
classes.
2.13
0.60
Disagree
14
I have received adequate professional
development to use AI tools in teaching.
1.88
0.60
Disagree
15
My institution provides technical support to
resolve connectivity or device issues.
1.88
0.60
Disagree
16
Limited access to devices and internet is a
significant barrier to AI integration.
4.50
0.50
Strongly Agree
17
My students have access to devices that allow
them to use AI tools.
2.13
0.60
Disagree
18
Institutional policies clearly address the
appropriate use of AI in the classroom.
1.63
0.48
Strongly Disagree
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 125
Artículo Científico
Abril – Junio 2026
—
D3 Dimension Mean
2.35
0.27
Disagree
Note: M = mean; SD = standard deviation. Items 13–15 and 17–18 are negatively framed relative to AI
adoption (lower scores = greater barrier); Item 16 is positively framed toward barrier recognition
(Authors, 2026).
3.4. Dimension 4: Implementation Strategies
Dimension 4 obtained a mean score of M = 3.48 (SD = 0.52), interpreted as "Agree."
Teachers expressed moderately positive perceptions regarding the feasibility of AI-
enhanced instructional strategies in their context. Within this dimension (Table 9), Item
22 — which addressed the perceived benefit of structured professional development
specifically focused on AI in EFL — obtained the highest mean (M = 4.50), approaching
"Strongly Agree." This finding represents the most unequivocal result across the entire
instrument and signals that professional development constitutes the primary
facilitating need identified by teachers. Item 21 (adaptability of international AI
strategies to local realities; M = 3.50) and Item 24 (relevance of published AI
recommendations to Ecuadorian public secondary contexts; M = 3.50) obtained
moderate scores, reflecting cautious but not negative assessments of external
evidence transferability. Item 20, which asked whether teachers had already integrated
an AI tool during the current year, obtained M = 2.88, consistent with the 50% prior use
rate reported in Section 4.2.1.
Table 5
Item-Level Descriptive Statistics — D4: Implementation Strategies (N = 8)
Item
Statement (abbreviated)
M
SD
Interpretation
19
I am aware of specific AI tools designed for
EFL teaching usable in my context.
3.50
0.50
Agree
20
I have already integrated at least one AI tool
into my EFL teaching this academic year.
2.88
0.78
Neutral
21
AI-enhanced strategies can be adapted to the
realities of my school.
3.50
0.50
Agree
22
I would benefit from a structured professional
development program on AI in EFL.
4.50
0.50
Strongly Agree
23
It is feasible to implement AI-supported
activities despite current resource limitations.
3.00
0.71
Neutral
24
AI integration recommendations are relevant
to Ecuadorian public secondary EFL
contexts.
3.50
0.50
Agree
—
D4 Dimension Mean
3.48
0.52
Agree
Note: M = mean; SD = standard deviation (Authors, 2026).
Summary of Phase 2 Findings
Three overarching patterns emerge from the Phase 2 descriptive analysis. First, EFL
teachers at the Unidad Educativa Otto Arosemena Gómez hold consistently positive
attitudes toward AI integration and conceptualize these tools as pedagogically
complementary, despite limited prior experience with them. Second, institutional
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 126
Artículo Científico
Abril – Junio 2026
infrastructure specifically device access, internet connectivity, professional
development, and technical support is perceived as substantially inadequate and
constitutes the primary barrier to AI adoption at this institution. Third, teachers express
a clear and near-unanimous demand for structured professional development
programs focused on AI in EFL teaching, identifying this as the most important
facilitating condition for moving from positive attitudes toward actual implementation.
These patterns are examined in relation to the Phase 1 systematic review findings in
the Discussion section.
4. Discussion
Convergence: Positive Attitudes Despite Limited Experience
The most consistent finding across both phases is that EFL teachers conceptualize AI
as a complementary pedagogical resource rather than a threat to their professional
role. Phase 2 results show that Dimension 2 (Teacher Attitudes) yielded the highest
mean score in the instrument (M = 4.13), with teachers expressing willingness to invest
time in AI learning and confidence in its motivational potential. This pattern aligns
closely with international evidence synthesized in Phase 1, where Derinalp and Halife
(2025) documented similarly positive orientations among pre-service and in-service
EFL educators across diverse contexts. Notably, this convergence holds even though
half of the participating teachers reported no prior experience with AI tools, suggesting
that positive attitudes are not contingent on prior use but may reflect broader
awareness of AI's growing relevance in education.
Divergence: Infrastructure as the Critical Barrier
The most striking divergence between the local context and internationally published
research emerged in Dimension 3 (Institutional Barriers), which recorded the lowest
mean score across the entire instrument (M = 2.31, Disagree). Teachers at the Unidad
Educativa Otto Arosemena Gómez perceive digital infrastructure — including device
availability, internet connectivity, professional development, and institutional technical
support — as substantially inadequate for AI integration. While Phase 1 identified
infrastructure constraints as a recognized barrier in resource-limited Latin American
contexts (Okoye et al., 2023; Cruz et al., 2024), the severity of these perceptions at
the local level exceeds what is typically foregrounded in internationally published
studies, which predominantly derive from well-resourced institutional settings. This
divergence confirms Cruz et al.'s (2024) argument that AI adoption strategies grounded
in optimal implementation conditions risk irrelevance when applied to under-resourced
public schools, and provides localized empirical evidence for a gap that Phase 1 could
only document theoretically.
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 127
Artículo Científico
Abril – Junio 2026
Professional Development as the Central Facilitating Need
Across both phases, professional development emerged as the most critical facilitating
condition for AI adoption. In Phase 2, Item 22 addressing the perceived need for
structured AI-focused professional development obtained the highest mean score in
the entire instrument (M = 4.63, approaching Strongly Agree), representing a near-
unanimous consensus among participants. This finding extends the Phase 1 literature,
which consistently identifies training deficits as a barrier (Riggs, 2025; Pérez-
Campdesuñer et al., 2025) but rarely quantifies the perceived urgency of this need
from teachers in resource-constrained Ecuadorian public schools. The implication is
direct: positive attitudes toward AI integration will not translate into practice without
deliberate, context-specific professional development investment. This conclusion is
particularly significant for educational policymakers in Ecuador, where AI adoption in
public secondary education remains structurally underserved
5. Conclusions
This study examined EFL teachers’ perceptions of artificial intelligence integration at a
public secondary school in Santa Elena, Ecuador, through a sequential explanatory
mixed-methods design combining a systematic scoping review with descriptive survey
data. The findings yield three principal conclusions with direct implications for
educational policy, teacher training, and future research.
First, positive teacher attitudes toward AI do not require prior experience with the
technology. Despite half of the participating teachers reporting no prior use of AI tools
in instructional contexts, Dimension 2 (Teacher Attitudes and Conceptualizations)
yielded the highest mean score in the instrument (M = 3.98), with near-unanimous
conceptualization of AI as a complementary resource rather than a threat to
professional practice. This finding converges with internationally published evidence
(Derinalp & Halife, 2025) and suggests that affective readiness for AI integration exists
within the institution independently of technical familiarity. Attitudinal receptivity alone,
however, is insufficient to drive implementation; the study confirms that positive
dispositions must be supported by structural conditions to produce meaningful
pedagogical change.
Second, institutional infrastructure constitutes the defining barrier to AI adoption in this
context. Dimension 3 (Institutional Barriers and Facilitators) recorded the lowest mean
across the entire instrument (M = 2.35, Disagree), with inadequate professional
development (Item 14: M = 1.88), absent technical support (Item 15: M = 1.88), and
the lack of institutional AI policies (Item 18: M = 1.63) emerging as the most critical
deficits. Teachers explicitly identified limited device access and internet connectivity
as primary obstacles (Item 16: M = 4.50). This divergence from internationally
published research—which predominantly documents positive infrastructure
conditions—confirms Cruz et al.’s (2024) argument that adoption strategies derived
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 128
Artículo Científico
Abril – Junio 2026
from well-resourced contexts are not directly transferable to Ecuadorian public
secondary education. The severity of the infrastructure gap documented here exceeds
what the existing literature typically foregrounds, providing localized empirical evidence
for a structural barrier that theoretical discussions have described but rarely quantified
in this regional context.
Third, structured professional development is the most urgently demanded and
strategically viable pathway to AI adoption. Item 22—which assessed the perceived
need for a structured professional development program on AI in EFL—obtained the
highest mean score in the entire instrument (M = 4.50, approaching Strongly Agree),
representing near-unanimous consensus among participants. This finding extends the
Phase 1 literature by quantifying the urgency of this need from teachers operating
under genuine resource constraints (Riggs, 2025; Pérez-Campdesuñer et al., 2025)
and identifies professional development as the most actionable facilitating condition—
one that can be pursued even where infrastructure investment is constrained by
institutional or systemic limitations. Teachers at this institution express clear
willingness to invest time in AI learning (Item 9: M = 4.13); the policy implication is that
this readiness requires a formal institutional response.
These conclusions must be interpreted in light of the study’s principal limitation: the
small sample size (N = 8) precludes inferential analysis and restricts generalizability
beyond the institutional context. The findings are descriptive and exploratory in nature,
appropriate for an underresearched setting, but require replication with larger and more
geographically diverse samples to support broader claims. The absence of student
perception data represents a complementary gap that future research should address,
as teacher and student perspectives jointly shape implementation outcomes.
Future studies should expand the empirical base through multi-institutional survey
designs involving larger teacher populations across Ecuador’s public secondary
sector, incorporate student perceptions as a distinct analytical dimension, and examine
the impact of targeted professional development interventions on actual AI integration
practices. Longitudinal designs tracking the relationship between training provision,
infrastructure improvement, and classroom implementation would substantially
advance understanding of how adoption pathways develop under resource-
constrained conditions. The present study provides a foundational empirical reference
point for this agenda, contributing the first locally grounded evidence on AI integration
perceptions in Ecuadorian public secondary EFL education—a context that existing
international scholarship has systematically overlooked
CONFLICTO DE INTERESES
“Los autores declaran no tener ningún conflicto de intereses”.
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 129
Artículo Científico
Abril – Junio 2026
Bibliographic References
Arce, C. M., Gavilanes, J. C., Arce, E. M., Haro, E. M., & Bonilla-Jurado, D. (2025).
Artificial intelligence in higher education: Predictive analysis of attitudes and
dependency among Ecuadorian university students. Sustainability, 17(17),
7741. https://doi.org/10.3390/su17177741
Barredo-Ibáñez, D., De-la-Garza-Montemayor, D.-J., Torres-Toukoumidis, Á., &
López-López, P.-C. (2021). Artificial intelligence, communication, and
democracy in Latin America: A review of the cases of Colombia, Ecuador, and
Mexico. Profesional de la Información, 30(6), e300616.
https://doi.org/10.3145/epi.2021.nov.16
Bauer, E., Greiff, S., Graesser, A. C., Scheiter, K., & Sailer, M. (2025). Looking beyond
the hype: Understanding the effects of AI on learning. Educational Psychology
Review, 37, 45. https://doi.org/10.1007/s10648-025-10020-8
Bayaga, A. (2025). Leveraging AI-enhanced and emerging technologies for
pedagogical innovations in higher education. Education and Information
Technologies, 30, 1045–1072. https://doi.org/10.1007/s10639-024-13122-y
Creswell, J. W., & Creswell, J. D. (2018). Research design: Qualitative, quantitative,
and mixed methods approaches (5th ed.). SAGE Publications, Inc.
https://edge.sagepub.com/creswellrd5e
Cruz, G., Riobó, A., Pfeifer, M., & Duarte, D. (2024). AI from the ground up: Challenges
and opportunities in the context of Latin America and the Caribbean. Inter-
American Development Bank. https://doi.org/10.18235/0013275
De La Torre, A., & Baldeon-Calisto, M. (2024). Generative artificial intelligence in Latin
American higher education: A systematic literature review. En A. Varol, M.
Karabatak, C. Varol, & E. Tuba (Eds.), 12th International Symposium on Digital
Forensics and Security, ISDFS 2024 (pp. 1–7). Institute of Electrical and
Electronics Engineers Inc. https://doi.org/10.1109/ISDFS60797.2024.10527283
Derinalp, P., & Halife, M. (2025). Pre-service English as a foreign language teachers’
attitudes toward artificial intelligence. Journal of Theoretical Educational
Sciences, 18(3), 609–629. https://doi.org/10.30831/akukeg.1644354
Leavy, P. (2022). Research design: Quantitative, qualitative, mixed methods, arts-
based, and community-based participatory research approaches (2nd ed.).
Guilford Press.
Okoye, K., Hussein, H., Arrona-Palacios, A., Quintero, H. N., Peña Ortega, L. O.,
Lopez Sanchez, A., Arias Ortiz, E., Escamilla, J., & Hosseini, S. (2023). Impact
of digital technologies upon teaching and learning in higher education in Latin
America: An outlook on the reach, barriers, and bottlenecks. Education and
Information Technologies, 28(2), 2291–2360. https://doi.org/10.1007/s10639-
022-11214-1
Pérez-Campdesuñer, R., Sánchez-Rodríguez, A., García-Vidal, G., Martínez-Vivar,
R., & De Miguel-Guzmán, M. (2025). Artificial intelligence in Ecuadorian SMEs:
Drivers and obstacles to adoption. Information, 16(6), 443.
https://doi.org/10.3390/info16060443
Revista Científica Ciencia y Método | Vol.04 | Núm.02 | Abr–Jun | 2026 | www.revistacym.com pág. 130
Artículo Científico
Abril – Junio 2026
Richards, J. C., & Renandya, W. A. (Eds.). (2002). Methodology in language teaching:
An anthology of current practice. Cambridge University Press.
https://assets.cambridge.org/97805218/08293/frontmatter/9780521808293_fro
ntmatter.pdf
Riggs, V. (2025). Teachers' perceptions and readiness for AI integration in under-
resourced K-12 classrooms. Journal of Research Initiatives, 9(1), Article 1.
https://digitalcommons.uncfsu.edu/jri/vol9/iss1/1
Salas-Pilco, S. Z., & Yang, Y. (2022). Artificial intelligence applications in Latin
American higher education: A systematic review. International Journal of
Educational Technology in Higher Education, 19(1), 21.
https://doi.org/10.1186/s41239-022-00326-w
Shafiee Rad, H. (2025). Reinforcing L2 reading comprehension through artificial
intelligence intervention: Refining engagement to foster self-regulated learning.
Smart Learning Environments, 12, 23. https://doi.org/10.1186/s40561-025-
00377-2
Tricco, A. C., Lillie, E., Zarin, W., O'Brien, K. K., Colquhoun, H., Levac, D., Moher, D.,
Peters, M. D. J., Horsley, T., Weeks, L., Hempel, S., Akl, E. A., Chang, C.,
McGowan, J., Stewart, L., Hartling, L., Aldcroft, A., Wilson, M. G., Garritty, C.,
... Straus, S. E. (2018). PRISMA extension for scoping reviews (PRISMA-ScR):
Checklist and explanation. Annals of Internal Medicine, 169(7), 467–473.
https://doi.org/10.7326/M18-0850
Winder, G., Bass, S., Schiele, D., & Buchner, J. (2024). Using large language models
for content creation impacts online learning evaluation outcomes. International
Journal on E-Learning, 23(3), 305–318.
https://doi.org/10.70725/423664moqcrd
