Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

DOI: https://doi.org/10.46296/yc.v10i19.0947

TUTORÍA CON APOYO DE IA: PRÁCTICA DE CONVERSACIÓN CON

STIMULER EN LA FORMACIÓN DE FUTUROS PROFESORES DE INGLÉS.

AI-SUPPORTED MENTORING: SPEAKING PRACTICE WITH STIMULER IN

FUTURE ENGLISH TEACHER TRAINING

1

2

Chávez-Zambrano Verónica Vanessa ; Vera-Toala Karla Jamileth

1

Docente de la Carrera de Pedagogía los Idiomas Nacionales y Extranjeros de la Universidad

Laica Eloy Alfaro De Manabí. Manta, Ecuador. Correo : veronica.chavez@uleam.edu.ec .

ORCID ID: https://orcid.org/0000-0003-3958-5053

2

Estudiante de la Carrera de Pedagogía los Idiomas Nacionales y Extranjeros de la

Universidad Laica Eloy Alfaro De Manabí. Manta, Ecuador.

Correo: e135016878@live.uleam.edu.ec. ORCID ID: https://orcid.org/0009-0005-3825-1344

Resumen

La habilidad oral es un requisito esencial en la formación inicial de profesores de inglés como

lengua extranjera, ya que los futuros docentes deben ser capaces de comunicarse eficazmente,

controlar la interacción en el aula y proporcionar modelos de lenguaje oral adecuado. Este

estudio empleó un diseño de métodos mixtos, descriptivo, transversal, de campo y no

experimental para evaluar la competencia oral de los estudiantes del programa educativo

“

Pedagogía de Lenguas Nacionales y Extranjeras: Inglés” de la Universidad Laica Eloy Alfaro de

Manabí e investigar las percepciones e ideas de los docentes sobre los factores que determinan

el éxito en este campo. Se matricularon 197 estudiantes en el programa, mientras que la muestra

analítica consistió en 26 evaluaciones diagnósticas completas de expresión oral, seleccionadas

mediante muestreo por conveniencia. Once estudiantes fueron evaluados según los descriptores

A1-A2 y quince, según los descriptores B1. Cuatro profesores de inglés participaron en

entrevistas semiestructuradas, que se analizaron temáticamente. Se utilizaron las escalas de

expresión oral de Cambridge para evaluar la gramática y el vocabulario, la pronunciación, la

comunicación interactiva y el logro global, así como la gestión del discurso en el nivel B1 de los

estudiantes. En cuanto a la pronunciación, fue el área más fuerte en ambos tipos de estudiantes

(

A1-A2: M = 3,27, DE = 1,01; B1: M = 2,53, DE = 0,64); los valores medios más bajos se

obtuvieron en la evaluación de gramática y vocabulario (M = 2,09, DE = 1,45; M = 1,60, DE =

,63, respectivamente). Las percepciones de los docentes coinciden con el perfil diagnóstico en

0

cuanto a la limitada recuperación de vocabulario, la ansiedad, la interacción espontánea

insuficiente, la baja exposición al inglés y las dificultades de comprensión auditiva. Los hallazgos

respaldan las recomendaciones de una interacción guiada más frecuente, práctica integrada de

escucha y expresión oral, colaboración entre pares y retroalimentación sistemática dentro del

Programa de Lengua Inglesa.

Palabras claves: competencia oral, profesores de inglés en formación, evaluación diagnóstica,

métodos mixtos, interacción oral, inteligencia artificial.

Abstract

The speaking skill is an essential requirement in pre-service EFL teacher education as future

instructors must be able to communicate effectively, control classroom interaction and provide

models of proper spoken language. This study involved a mixed-methods, descriptive, cross-

sectional, field-based, and non-experimental design to assess the speaking proficiency of

students of the “Pedagogy of National and Foreign Languages: English” educational program at

Universidad Laica Eloy Alfaro de Manabí and to investigate teachers’ perceptions and ideas

regarding the factors determining success in this field. 197 students were enrolled in the program,

while the analytical sample consisted of 26 complete diagnostic speaking assessments, which

were selected on the basis of convenience sampling. 11 students were assessed according to

Información del manuscrito:

Fecha de recepción: 08 de mayo de 2026.

Fecha de aceptación: 10 de julio de 2026.

Fecha de publicación: 11 de agosto de 2026.

488

Chávez-Zambrano et al. (2026)

A1-A2 descriptors and 15 students, according to B1 descriptors. Four English language teachers

participated in semi-structured interviews, which were analyzed thematically. Cambridge

speaking scales were used to evaluate grammar and vocabulary, pronunciation, interactive

communication, and global achievement, as well as discourse management at the level of B1

students. As to pronunciation, it was the strongest area in both types of students (A1-A2: M =

3

.27, SD = 1.01; B1: M = 2.53, SD = 0.64); the lowest mean values were acquired in the

assessment of grammar and vocabulary (M = 2.09, SD = 1.45; M = 1.60, SD = 0.63, respectively).

Teacher perceptions coincide with the diagnostic profile regarding limited vocabulary retrieval,

anxiety, insufficient spontaneous interaction, low exposure to English, and listening-

comprehension difficulties. The findings support recommendations for more frequent guided

interaction, integrated listening-speaking practice, peer collaboration, and systematic feedback

within the English Language Program.

Keywords: speaking proficiency, pre-service English teachers, diagnostic assessment, mixed

methods, oral interaction, artificial intelligence.

Recognition Technology on Second

1. Introduction

Language

Speaking Skills of EFL Learners”

Sun, 2023), “Artificial Intelligence in

EFL Speaking: Impact on Enjoyment,

Anxiety, and Willingness to

Pronunciation

and

The effectiveness of speaking skills

determines successful academic

(

participation,

professional

interaction, and English modeling by

teachers. Oral competence goes

beyond general interaction for pre-

service educators and does cover

explanation, question, feedback,

Communicate” (Zhang et al., 2024),

and “Investigating the Impact of

Online Language Exchanges on

Second Language Speaking and

Willingness to Communicate” (Zhou,

classroom

management,

and

mediation of interactions. According

to a meta-studies performed by Faez

et al. (2021), language proficiency

relates to teaching performance but

2

023) point out the interplay of

linguistic competence, affective

states, feedback, and authentic

learning practice.

does

not

fully

explain

the

This phenomenon is relevant to Latin

America and Ecuador as well. In the

paper “An Analysis of the Levels of

English of EFL Teachers in

Ecuador”, Parra-Gavilánez (2022),

stated that teacher proficiency varied

across planning areas of the country.

Narváez-Cantos (2022), in the work

effectiveness of teaching.

Numerous

international

studies

stress the challenges of speaking

development since students have to

coordinate grammar, vocabulary,

pronunciation, discourse, and the

process of interaction. The works like

“The Impact of Automatic Speech

“The Effective English Language

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

Teacher from the Perspective of

Ecuadorian University Students”

concluded that university students

considered pedagogical knowledge,

communicative competence, socio-

affective competence, and language

supported mentoring as a significant

pedagogical matter to be researched

once the achievements of students

are analyzed.

The purpose of this work is to

determine the current English-

speaking proficiency of students

enrolled in the English teacher

education program at Eloy Alfaro

University, to inform evidence-based

pedagogical recommendations. The

research is aimed at answering four

questions: (1) What is the current

level of English-speaking proficiency

of pre-service English teachers

enrolled in the English Education

Program at Eloy Alfaro University of

Manabí? (2) Which components of

speaking proficiency (pronunciation,

grammar, vocabulary, discourse

proficiency

as

interrelated

characteristics. Moving on in time,

Guerrero Rodriguez and Moreira

Baquerizo (2025) pointed out that

limited

instruction

time,

high

emphasis on grammar, and lack of

practice are impediments for

Ecuadorian EFL learning process.

These findings are especially

significant for teacher education. It is

expected that future teachers will use

the English language as the main

means of instruction and will help

students learn how to communicate.

However, the application of general

proficiency definitions provides no

information regarding the type of

difficulties concerning grammar,

pronunciation, and overall speaking

ability of individuals.

management

and

interactive

communication) represent the main

strengths and weaknesses of pre-

service English teachers? (3) How do

English teachers assess students'

speaking

ability

and

factors

influencing it? (4) What pedagogical

strategies could be recommended to

improve oral performance in English

teacher education based on the

findings?

The current research covers this gap

by combining learners’ speaking

assessment with opinions of their

teacher trainers from one Ecuadorian

university. The study does not

involve testing teaching methods or

their efficiency but regards AI-

490

Chávez-Zambrano et al. (2026)

Literature Review

conducted a survey looking into the

students and their understanding of

the effective teachers’ language

proficiency in university EFL context

and concluded that language

proficiency is one of the factors that

define the successful teaching along

with the pedagogical abilities.

•

Previous Research and the

Study Gap

Research into the proficiency of

teachers in terms of language has

established

that

language

knowledge is related to teaching

efficiency, but it can’t be treated as

isolated factor leading to the quality

Guerrero Rodriguez and Moreira

Baquerizo (2025) conducted

a

of

teaching.

In

the

article,

survey of Ecuadorian teachers about

the strategies and difficulties faced

by teachers in the process of

teaching speaking to their students.

As a result of their survey they found

out a number of limitations such as

lack of time, insufficient emphasis on

oral practice, lack of motivation, and

the effectiveness of communicative

activities. On the global level, Zhou

“

Connecting Language Proficiency to

Teaching Ability: A Meta-Analysis,”

Faez and his colleagues (2021)

claimed that the relationship is

positive but varies depending on the

context and measurement used. The

discovery justifies the necessity of

including analytic diagnostics which

define certain aspects of speaking in

comparison to evaluation of the level

of a program or course in terms of the

overall level of spoken proficiency.

(2023) stated that structured online

language exchange has a positive

influence on speaking skills and

willingness to speak, while Sun

The studies that have been carried

out in Ecuador have provided data

that is important for the context, but

they still can be elaborated further.

Thus, Parra-Gavilánez (2022) used

the data coming from standardized

tests to define teacher language

proficiency by national planning

zone, and identified that the zones

did not reach the expected what was

expected proficiency level. He further

(2023) and Zhang et al. (2024)

provided the connection between the

speaking results and technology,

anxiety, pleasure and readiness to

speak.

Considering the above-mentioned

literature, it is possible to identify

three gaps present in the research.

Firstly, there is not enough

491

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

Ecuadorian research concentrating

on the direct language performance

with the spontaneous nature of

communication.

and

interpretation

of

this

The aspects included in the

Cambridge speaking scales suggest

that such view is multidimensional.

Grammar and lexis deal with the

range and mastery of language;

performance by teacher educators.

Secondly, often wide confusions in

speaking proficiency categories

conceals the internal differentiation

of

phoric,

grammatical,

and

pronunciation

shows

how

vocabulary skills, as well as

discourse management, interactive

communication and global result.

Thirdly, there is lots of research

devoted to using technology in

raising speaking proficiency where

the diagnosis is needed.

understandable the person is;

discourse refers to delivering

information; interactive speaking

includes

initiating,

replying,

developing, and negotiating; and

global achievement assesses how

the candidate performs in the task

(Cambridge Assessment English,

2020. Each aspect provides their

own piece of information meaning

that strong pronunciation marks do

not mean that all other components

will be also good.

•

Speaking Proficiency and

Communicative Competence

Speaking is an active and engaging

skill used to find meanings, interact

with others, express ideas, and

change the language depending on

the context of the communication.

According to Goh and Burns (2012),

speaking development involves the

working together of language

competence, discourse, fluency,

pronunciation, and communication.

Thus, oral language skills are more

than being able to produce

syntactically correct sentences; they

Pronunciation is still significant due

to

the

way

it

influences

comprehension, listening effort, and

success in conversation. Metruk

(2024) said in his review that mobile

pronunciation teaching is a newly

emerging area of teaching that has

new opportunities for practice and

feedback. Though intelligibly means

different things and it should be

also

comprise

intelligibility,

effectiveness, responsiveness, and

combined

with

knowledge

of

the given individual’s ability to cope

492

Chávez-Zambrano et al. (2026)

vocabulary, grammar, structure, and

responsiveness to be fully revealed.

combination enjoyment,

willingness to communicate, and

anxiety is significant in AI

of

•

Factors

Influencing

EFL

environments. According to El

Shazly’s (2021) evidence, anxiety

could be lowered through AI-

supported experiences.

Speaking

The performance of speaking in a

language relies on various elements

such

as

linguistic,

cognitive,

The opportunities for interaction

contribute to the development of

speaking. In the study conducted by

Zhou (2023), it was determined that

learners involved in language

affective, and contextual aspects.

Among the elements in the linguistic

dimension, we have vocabulary

limitations like the inability to form

correct grammatical structures or

sentences that would contain

inaccurate information and those

exchange

had

meaningful

conversations, which allowed them

to

practice

speaking.

Ericsson

The

and

pertaining

to

phonetics.

The

researchers,

cognitive aspects come into play in

situations when speakers have to

perform operations such as the

processing of information and

language retrieval; they have to

organize meaning while monitoring

its correctness with very little

preparation time.

Johansson (2023), claim that

technology enables an endless

supply of online practice sessions.

Therefore, even though interaction

appears to play a vital role,

technology does not provide the

ultimate solution to the problems of

speaking production.

The influence of affective elements

appears to be essential when it

comes to spontaneous speech

events and assessments of speaking

proficiency. The findings of Sun

•

Speaking

Assessment

in

Teacher Education

Diagnostic evaluation of learning

indicates the current profile of

learners before specific instructional

programs are designed. In speaking,

analytical scales are important

because they distinguish between

dimensions that may undergo

(2023) showed that the role of

speech recognition has positive

implications for the development of

speech skills; on the contrary, Zhang

et al. (2024) suggested that the

493

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

uneven

development.

The

• AI-Supported Mentoring as a

descriptors given by Cambridge set

out the observable criteria for the

different levels and present the

outcome of evaluation in the format

of signs rather than in the form of a

total score (Cambridge Assessment

English, 2020a, 2020b). This

information is important for the field

of teacher education, as it can help

make decisions related to oral

practices, feedback, peer-mentoring

activities, and language-support

activities sequence.

Pedagogical Implication

There is an emergence of Artificial

intelligence in the field of language

instruction. Various applications of AI

in education can be found in

conversational

agents,

speech

recognition, adaptive practice, and

other tools which ease the learning

process. According to Alshumaimeri

and Alshememry (2024), a number of

ways of implementing AI in foreign

language teaching has been gaining

popularity. Du and Daniel (2024) also

supported the idea that chatbots

could help students to have more

speaking opportunities if tasks were

properly designed.

Assessment also needs to take into

account the conditions under which

performance has taken place. For

example, one task can be impacted

by familiarity with the topic, time

pressure, support from interlocutors,

listening skills, and anxiety. For this

reason, quantitative scores can be

Scholars connect the use of AI tools

in speaking practice with the

improvement of speaking skills,

pronunciation

feedback,

self-

interpreted

through

qualitative

regulation, and anxiety reduction.

Qiao and Zhao (2023) presented a

study where the authors connected

AI-based language training with the

development of speaking abilities

and self-regulation skills. Ding and

Yusof (2025) conducted a mixed-

methods study that showed how

chatbot technology facilitates the

development of speaking skills and

diminishes anxiety levels. Finally,

evidence to facilitate the contextual

understanding of the above scores.

Mixed-method integration allows

direct scores to be compared with

educators’ observations to find the

areas of similarity and differences

instead of regarding both sources as

sufficient independently (Creswell &

Creswell, 2018).

494

Chávez-Zambrano et al. (2026)

Mingyan et al. (2025) found that the

use of an AI-powered mobile

emphasize that informal teaching

and learning through AI depends on

different factors, such as motivation

and context. Thus, the author of the

present work would propose using

AI-supported mentoring as an

addition to human feedback and not

as a method that can replace the

work done by teachers.

application

improves

students'

speaking abilities, although this

improvement was not equal in all

aspects of speaking. The research

results should be further investigated

but cannot be utilized in the settings

without this type of intervention.

AI-assisted

mentoring

requires

human evaluation as well. Ayanwale

et al. (2024) consider the AI literacy

of future educators to be crucial,

while Peña-Acuña and Corga

Fernandes Durão (2024) describe

both possibilities brought about by AI

technologies and the necessity to

2

. Methodology

Research Design

The design for this study was a

mixed-methods, descriptive, cross-

sectional, non-experimental, and

field-based design. The quantitative

part of the study focused on how

students performed on speaking

criteria, and the qualitative part of the

investigate

AI-assisted

English

language learning more closely.

Köbis and Mehner (2021) draw

attention to ethical issues in the use

of AI for mentoring. Alghamdy (2023)

also stresses the importance of

considering pedagogical and ethical

aspects of using AI in EFL contexts.

study

looked

into

teacher-

researchers’ perceptions of various

linguistic, emotional, contextual, and

pedagogical factors that affect oral

performance. The study was non-

experimental in nature because of

not manipulating any variables, the

nonexistence of a control group, and

the nonuse of post-test and AI

intervention (Creswell & Creswell,

One of the most significant is teacher

scaffolding, as automated feedback

may be imperfect and hard to read.

Ma and Chen (2025) provide

evidence explaining that combining

AI with teacher support gives better

results than letting learners work with

AI independently. Liu et al. (2024)

2

018).

495

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

The mixed-methods design of the

research is called convergent design

in the interpretation phase; first,

qualitative and quantitative data were

analyzed separately and then

connected through the joint display

showing the principal diagnostic

patterns as well as emergent themes

based on interviews.

their experience teaching English

and their knowledge of students’

speaking performance. Their

interviews constituted the qualitative

data analyzed in this study.

Data Collection Instruments

The instrument of measurement type

was

composed

of

speaking

assessment scales aligned with

Cambridge and appropriate levels.

The rubric for the A1-A2 level

indicates evaluation of grammar and

Population,

Participants,

and

Analytical Sample

The population that was part of the

study was 197 students enrolled in

the Pedagogy of National and

vocabulary,

interactive

pronunciation,

communication, and

Foreign

Languages:

English

achievements across Parts 1 and 2

of Speaking. The B1 rubric indicates

grammar and vocabulary, discourse

program at Universidad Laica Eloy

Alfaro de Manabí, Ecuador. It was

made of the quantitative analytical

sample which included 26 students

with completeness of their diagnostic

records. The sample was selected by

convenience and from the aspect of

the availability and engagement of

management,

pronunciation,

communication, and

interactive

achievements throughout Parts 1 to

4 of Speaking. The ratings were

given for the above-mentioned

aspects from Band 0 to Band 5

depending on the series of

descriptors provided by Cambridge

(Cambridge Assessment English,

2020a, 2020b).

students

in

the

speaking

assessment; they were selected

through non-probability sampling. It

needs to be mentioned that 11

students were evaluated using A1-

A2 scale and 15 students by B1 level.

The research qualitative tool of

choice was

a

semi-structured

Four English language teachers from

the same program were purposively

selected as key informants due to

interview guide with open questions.

Ten major questions of the tool were

about the overall level of speaking

496

Chávez-Zambrano et al. (2026)

ability, spontaneous usage of

English, fulfilment of the set goals,

quantitative

frequencies

information while

and percentages

conditions

for

doing

good

showed how many performers

scored below Band 3. The use of

Band 3 was for descriptive purposes

only as it was characterized as the

performance, different factors such

as linguistic and affective ones,

teaching practices and methods of

teaching, and recommendations for

the program as a whole. The open-

ended nature of the questions

allowed the interviewers to put

forward their comments on some

usual things without being bound to

particular types of answers.

first

middle

descriptor.

The

researcher did not create an overall

score combining A1-A2 and B1

means due to the instruments used

for measurement.

The interview transcriptions were

analyzed using thematic analysis

whereby meaning units were

extracted and coded into themes of

proficiency variability, spontaneous

communication, linguistic barriers,

emotional and performance factors,

teaching methods used, listening-

speaking partnership, and computer-

aided instruction. The themes were

then generalized up to broader

parameters and compared to

quantitative findings from the

quantitative data (Braun & Clarke,

Procedure and Data Analysis

In this study, the participants

performed speaking tasks at their

respective

performances were recorded for

further processing. The two

levels

and

their

evaluators participated in this review

and one of them evaluated the

analytic

performers while the other i.e., the

global score. No statistical

parameters

of

the

coefficients such as inter-rater

reliability were used in the analysis of

the data that is treated as the

limitation (Braun & Clarke, 2022).

2

022).

Ethical

and

Reporting

Considerations

In this research, quantitative analysis

was conducted for every group

separately. The means and standard

Student and teacher names have not

been disclosed in either the

analytical dataset or the article. The

results were provided in an

deviations

provided

general

497

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

aggregated manner and interviews

were summarized thematically. The

report clearly distinguishes among

the institutional population, the 26

3. Results and discussion

Analytical Sample and Mean

Speaking Bands

complete

diagnostic

included

speaking

in the

The quantitative analysis covered 26

total recordings: 11 learners (42.3%)

in the A1-A2 assessment category

and 15 students (57.7%) in the B1

class. The means and standard

deviations for each criterion are

shown in Table 1. The results are

analyzed separately by group

because the scales for the various

levels utilized different descriptors,

and the A1-A2 rubric did not include

discourse management among the

analytical categories so that was

never a part of the analysis.

assessments

quantitative analysis, and the four

English teachers who participated in

the qualitative interviews. The report

also explicitly states that Stimuler

and other AI applications were not

implemented or evaluated, allowing

for the elimination of causal claims

that are unsupported by the research

design.

Table 1. Mean speaking bands by assessed level

Grammar

and

vocabulary

Discourse

Interactive

Pronunciation

management communication achievement

Global

Group

n

A1-A2 11

B1 15

2.09 (1.45) Not assessed 3.27 (1.01)

1.60 (0.63) 2.20 (0.56) 2.53 (0.64)

2.18 (1.89)

2.13 (0.92)

1.18 (0.60)

2.47 (0.64)

intelligible

undertaken

according

to

the

This

A1-A2 Diagnostic Profile

descriptors.

Pronunciation in the A1-A2 group

capacity should not be considered as

sufficient evidence of oral proficiency

as performance at other levels was

considerably weaker.

has

strongest dimension (M = 3.27, SD =

.01), as only two distraught

been

comparatively

the

1

individuals (18.2%) scored under

Band 3, meaning the overwhelming

With a mean score of 2.09 (SD =

1.45), grammar and vocabulary

majority

of

participants

were

498

Chávez-Zambrano et al. (2026)

constitute the second dimension of

students (eight students or 53.3%)

fell below Band 3 in pronunciation,

while nine (60.0%) of them failed to

meet Band 3 in global achievement,

thus indicating that though the

strongest domains in the B1 band

effectiveness;

seven subjects

(63.6%) received lower than Band 3.

Functions

in

interactive

communication resulted in a similar

mean of 2.18, but the standard

deviation was the biggest one (SD =

provided

some

indicators

of

1

.89);

six

students

(54.5%)

performance, there is still substantial

underperformance relative to what is

considered acceptable performance.

demonstrated a level lesser than

Band 3 and three of them belonged

to Band 0. The results suggest that

some of the learners performed

locally in difficult communication

situations, whereas the majority were

not able to respond.

Both grammar and vocabulary had

the lowest performance (M = 1.60;

SD = 0.63) – out of 15 students, only

one-three scored above Band 3, with

no student achieving Band 4 or 5.

The same concern is relevant to

discourse management (M = 2.20;

SD = 0.56) as its mean result

indicates that 11 students were

below Band 3, which means that

there was difficulty with producing

long answers and keeping the

speech organized.

Finally, the overall score was the

weakest among all groups (M = 1.18,

SD

=

0.60), with 90.9% of

respondents being below Band 3.

Different outcomes in pronunciation

performance

achievement

and

overall

being

indicate

understandable on the level of words

or phrases has not always led to the

completion of tasks.

In addition, concerning interactive

communication, its mean is 2.13 (SD

= 0.92); exactly 10 students scored

below Band 3. Therefore, all the B1

results indicate a profile where

pronunciation was relatively strong

versus other aspects like linguistic

knowledge, organizational skills, or

interactive communication but was

B1 Diagnostic Profile

On the B1 rating scale, pronunciation

achieved the highest measure (M =

2.53; SD = 0.64) and global

achievement was in close second

place (M = 2.47; SD = 0.64).

However, a little more than half of the

499

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

not able to offset the failures in using

an appropriate language.

Table 2. Students performing below Band 3

Group

A1-A2

B1

Dimension

n below Band 3

Percentage

63.6%

Grammar and

vocabulary

7

2

Pronunciation

18.2%

Interactive

communication

6

54.5%

Global achievement

10

14

90.9%

Grammar and

vocabulary

93.3%

Discourse

management

B1

11

8

73.3%

53.3%

66.7%

60.0%

Pronunciation

Interactive

communication

10

9

Global achievement

Figure 1. Percentage of students below Band 3 by assessed dimension

Students performing below Band 3

1

6

4

2

0

8

6

4

2

0

14

1

10

9

8

7

6

2

3,60%

54,50%

90,90%

93,30%

73,30%

53,30%

66,70%

60,00%

18,20%

A1-A2

B1

Percentage

B1

n below Band 3

learners fell within the basic and

intermediate level, while a smaller

group managed to communicate

more autonomously. The teachers

made a clear distinction between

Teacher Interview Findings

The student teachers' statements

reflected a variety of proficiency

levels and speakers. The majority of

500

Chávez-Zambrano et al. (2026)

performance on structured tasks and

spontaneous use of the foreign

language in classroom conditions:

given the appropriate context,

students participated more actively

when they knew the topic, the tasks

were straightforward and group work

was used as a way to reduce

pressure.

The reasons which influenced

students' performance were divided

into affective reasons (such as

anxiety and low self-esteem) and

performance reasons (such as poor

listening and speaking skills and lack

of preparation). The teachers laid

emphasis on positive feedback and

made recommendations to use

known contexts for practicing

speaking skills.

The main linguistic elements of

speech were found to be the issues

of vocabulary knowledge and

grammar, as well as the fact that

The classroom activities that had

been

proposed

included

speaking

and

listening

are

communicative tasks, interviews,

presentations, discussions, different

types of kinesthetic activities, pair

work and delayed feedback, etc. The

teachers considered technology to

be a good method of practice with the

help of the recording and recognition.

interconnected. Teachers reported

that students may sometimes know

the right words but are unable to

respond promptly when it comes to

listening to the message of an

interlocutor. This gives an additional

meaning to the quantitative data

acquired.

Table 3. Integration of diagnostic and interview findings

Relationship to diagnostic

results

Theme

Synthesized teacher evidence

Students were divided into two There was a high variance in the

diametrically opposed groups, with usage of grammar and vocabulary

the first group consisting of highly at A1-A2 level (SD=1.45) and

supported speakers and the interactive speaking (SD=1.89)

Heterogeneous

proficiency

second

group

being

less

dependent on teacher assistance

The rate of the use of English Interactive speaking was below

increased with the delivery of band 3 in 54.5% of A1-A2 level

Limited spontaneous interesting

interaction communicative

topics,

tasks,

clear students and 66.7% of B1 level

and students

pair/group work, but did not

become autonomous

501

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

Relationship to diagnostic

results

Theme

Synthesized teacher evidence

The participant was faced with a Grammar and vocabulary scored

number of linguistic problems, the lowest average in B1 and

such as vocabulary retrieval, average in A1-A2; discourse

grammar, sentence construction, competency was poor in B1

coherence, pronunciation, and the

Linguistic constraints

use of meaningful phrases

Anxiety, low self-esteem, fear of Low performance globally and in

making mistakes, pressure from supported communication might

being evaluated, and feeling suggest a case of interference

ashamed hampered students’ under communicative pressures

participation

Affective and

performance factors

Another

barrier

to

giving This finding may be the reason for

Listening-speaking

relationship

appropriate and timely answers low interactive communication and

was the difficulty of understanding higher pronunciation scores

questions spoken by the teachers

The teachers suggested using The profile indicates the necessity

authentic communication, peer of transferring some elements of

Pedagogical and

technological

response

work,

integrated

delayed

skills,

correction, controlled knowledge into properly

recording, established and independent

shadowing, and AI-supported speech production

practice

English, 2020a; 2020b). This fact is

Discussion

in agreement with the ideas of Goh

and Burns (2012), stating that

speaking involves coordination of

various processes rather than

isolated mastery of sounds or

structures.

The analysis provides evidence of an

unbalanced speaking profile rather

than a comprehensive level of

speaking.

Pronunciation

was

recognized as the strongest category

of measurement in the two analyzed

Grammar and vocabulary were found

to be the most consistent weakness

of students, particularly in the B1

group, with as many as 93.3% of

learners’ performance being below

Band 3. Teacher trainers also

referred to vocabulary issues,

problems of construction and errors

in grammar as the most frequent

difficulties encountered by students.

groups,

however,

language

and

achievement,

grammar

vocabulary capabilities, interactive

discourse and management of

interactions remained low. This

finding indicates that the multi-

component view of the Cambridge

scale is justified as it distinguishes

between intelligibility and language

use

(Cambridge

Assessment

502

Chávez-Zambrano et al. (2026)

Guerrero Rodriguez and Moreira

Baquerizo (2025) indicate that

speaking teaching process in

Ecuador is mainly restricted by time

limits and absence of sufficient oral

practice in instruction. The current

study brings direct proof of how these

limitations affect the performance in

analytical speaking performance.

below Band 3 in about half of the

cases. The interviews revealed that

students were able to speak in

English only when discussing familiar

topics or following clear instructions,

working in pairs, or receiving other

types of help from teachers. Zhou

(2023) said that communicating

online positively influenced speaking

and willingness to speak because of

real interaction. However, the

authors of the study did not test the

proposed method and, likewise, did

not seek to adopt this method in their

research, even though they agree

that the activities in the study should

include tasks that make students

take turns, ask questions, clarify,

answer without preparing and

communicate more than giving

simple answers.

The relatively better results in

pronunciation should be interpreted

with caution. Although A1-A2

students

demonstrated

a

considerable level of intelligibility,

their overall performance was still

below the required level; in B1 group,

pronunciation was assumed to be the

strongest criterion but at the same

time over a half of the group failed to

reach Band 3 level. The authors of

Metruk (2024) and Abdelhalim and

Alsehibany (2025) agree that

pronunciation is improved through a

The discoveries relating the state of

knowledge to reasons for the failure

of this knowledge to become

spontaneous communication are

explained with such factors as

anxiety and fear of making mistakes,

insecurity, and evaluation pressure

shown by the teachers’ comments.

This explanation is supported by

studies conducted by El Shazly

combination

feedback

of

practice

and

provided

through

technology. However, the profile in

this paper shows that the understood

pronunciation alone does not

guarantee

good

vocabulary,

coherent speech, and the ability to

communicate independently.

(

2021), Sun (2023), Zhang et al.

2024), who note that oral

The interactive

level

of

(

communication in both groups was

503

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

performance is affected by anxiety

and willingness to communicate.

Thus, it may be concluded that the

students,

only

26

complete

diagnostic speaking assessments

were available for analysis, while the

nature of convenience sampling

undermines the possibility of

generalizing the results of the study.

Furthermore, the classification of the

A1-A2 and B1 groups led to an

evaluation based on level-specific

descriptors, which does not allow

direct comparisons across levels.

Besides, no inter-rater reliability

coefficient is available for the data.

anxiety-friendly

environment

is

created through the use of low-stress

speaking activities, peer learning,

clear rules for evaluation, delay in

feedback.

In addition, the importance of

listening comprehension mentioned

by the teacher educators may

provide an interesting insight

regarding the interactive issue.

Learners may know many words and

speak with perfect articulation but be

unable to answer the question they

heard and understood late. The

solution of this problem may be found

in integrated listening-speaking task,

where students have to find

important information and give

answers to questions that require

negotiating meaning. This approach

to teaching corresponds to the whole

approach suggested by Goh and

Burn (2012), allowing teachers to

move from rote learning to real

communication and production of

language.

The cross-sectional design only

reflects

students’

speaking

proficiency at one point in time, thus

not assessing the causes of the

phenomena in question or the

changes in performance over time.

Future studies should include larger

samples of students, conduct

additional interviews, and analyze

inter-rater

evaluators

agreement

among

the

to strengthen

methodological rigor of the study.

4. Conclusions

The diagnostic assessment indicated

that pre-service English language

teachers who were in the records

Limitations and Future Research

The study has several limitations.

Out of the total population of 197

showed oral

heterogenous

proficiency profiles. Pronunciation

504

Chávez-Zambrano et al. (2026)

was found to be the strongest

component of speech output;

however, it was not leading to a

globally high speech performance.

Mentoring with the use of AI

technology can provide more options

for individual practice, feedback, and

self-assessment, however, it has not

been investigated so far owing to the

absence of the AI application in the

study. There was no intervention

applied to the study, therefore, the

results presented here do not claim

that the performance of the students

was improved with the help of

Stimuler or any other tool; thus,

future projects should integrate

Grammar

and

vocabulary

represented the major weaknesses,

especially because of the B1 level;

however, interactive speaking skills

and discourse management skills

also needed attention. The results

show that using separate criteria for

assessment of speaking is of much

more importance than simple

identification of speaking proficiency

level.

mediation,

transparent

ethical

issues,

and

assessment,

evidence showing the results of the

project.

Teacher educators’ perceptions

were similar with the results of the

diagnostic assessment with limited

spontaneity of English use and poor

vocabulary retrieval, difficulties in

grammar and discourse, lack of

experience, issues with listening

comprehension, anxiety, fear of

making mistakes and pressure

during assessment. The coincidence

of direct assessment data and

qualitative information means that

oral development requires linguistic

support as well as frequent

communication practice, integrated

listening and speaking practice, and

constructive feedback in a classroom

where risk is encouraged.

Recommendations

The English language program ought

to encompass speech tasks that will

be performed often but in short

periods.

Such

activities

like

conversation clubs, cross-semester

collaboration, role plays, information

gap activities, problem-solving tasks,

debates, interviews, and English-

only activity moments should enforce

engagement, meaning making, and

immediate responses instead of

relying on memorized speeches.

Teachers’ combine

ought

to

speaking and listening through the

505

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

use of listening exercises and

speaking. Analytic criteria for these

types of activities ought to be

interpret

machine-generated

suggestions. Evaluation about the

use of this technology should be

based on keeping a complete

presented

beforehand,

while

feedback should be focused on a

small number of issues in the field of

grammar, vocabulary, discourse,

pronunciation, and interaction.

database

of

students

who

participated in the experiment,

engagement of trained evaluators,

documentation of reliability, checking

using pre-test/post-test evaluation,

and qualitative research.

The program should introduce

supportive norms that would help

diminish fear of errors and negative

peer response towards mistakes.

Low-pressure practice, pairing up

students in different combinations,

gradually increasing difficulty of

tasks, self-evaluation, and peer

feedback should help increase

participation of students while

creating clear requirements for

participation in speaking activities.

References

Abdelhalim, S. M., & Alsehibany, R.

A. (2025). Integrating AI-

powered in EFL

pronunciation instruction:

Effects on accuracy and L2

motivation. Computer

Assisted Language Learning,

–25.

https://doi.org/10.1080/09588

21.2025.2534015

tools

1

2

Alghamdy, R. Z. (2023). Pedagogical

and ethical implications of

artificial intelligence in EFL

Students experiencing permanent

obstacles in speaking should be

supported

according

to

their

context: review study.

English Language Teaching,

6(10), 87–98.

A

individual problems instead of using

the same speaking practice aimed at

all students.

1

https://doi.org/10.5539/elt.v16

n10p87

Any future use of Stimuler,

conversational AI, or pronunciation

applications should be built upon

established learning goals, teacher

education, privacy check, access

process, and instructions for how to

Alshumaimeri, Y. A., & Alshememry,

A. K. (2024). The extent of AI

applications in EFL learning

and

teaching.

IEEE

Transactions on Learning

Technologies, 17, 653–663.

https://doi.org/10.1109/TLT.2

023.3322128

506

Chávez-Zambrano et al. (2026)

Ayanwale, M. A., Adelana, O. P.,

Molefi, R. R., Adeeko, O., &

https://doi.org/10.1057/s4159

9-025-05550-z

Ishola,

Examining

A.

M.

(2024).

artificial

Du, J., & Daniel, B. K. (2024).

Transforming

education:

language

systematic

intelligence literacy among

pre-service teachers for future

classrooms. Computers and

Education Open, 6, 100179.

https://doi.org/10.1016/j.caeo.

A

review of AI-powered chatbots

for English as a foreign

language speaking practice.

Computers and Education:

2024.100179

Artificial

100230.

Intelligence,

6,

Braun, V., & Clarke, V. (2022).

Thematic analysis: A practical

guide. SAGE.

https://doi.org/10.1016/j.caeai

.2024.100230

Cambridge Assessment English.

El Shazly, R. (2021). Effects of

(

2020a). A2 Key for Schools

artificial

English speaking anxiety and

speaking performance:

case study. Expert Systems,

38(3), e12667.

https://doi.org/10.1111/exsy.1

667

intelligence

on

handbook for teachers for

exams from 2020. Cambridge

University

A

Press

&

Assessment.

Cambridge Assessment English.

2020b). B1 Preliminary for

Schools handbook for

2

(

Ericsson, E., & Johansson, S.

(2023). English speaking

practice with conversational

AI: Lower secondary students’

educational experiences over

teachers for exams from 2020.

Cambridge University Press &

Assessment.

Creswell, J. W., & Creswell, J. D.

time.

Education:

Intelligence,

Computers

and

Artificial

100164.

(2018). Research design:

Qualitative, quantitative, and

mixed methods approaches

5,

https://doi.org/10.1016/j.caeai

2023.100164

(5th ed.). SAGE.

.

Ding, D., & Yusof, A. M. B. (2025).

Investigating the role of AI-

powered conversation bots in

enhancing L2 speaking skills

Faez, F., Karas, M., & Uchihara, T.

2021). Connecting language

(

proficiency to teaching ability:

A meta-analysis. Language

Teaching Research, 25(5),

and

reducing

speaking

anxiety: A mixed methods

study. Humanities and Social

7

54–777.

https://doi.org/10.1177/13621

8819868667

Sciences

2,

Communications,

1223.

6

1

507

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

Goh, C. C. M., & Burns, A. (2012).

Teaching speaking: A holistic

teacher scaffolding. Frontiers

in Psychology, 16, 1568239.

https://doi.org/10.3389/fpsyg.

2025.1568239

approach.

Cambridge

University Press.

Guerrero Rodriguez, S. E., & Moreira

Baquerizo, A. S. (2025).

Ecuadorian EFL teachers’

Metruk, R. (2024). Mobile-assisted

language

learning

and

pronunciation instruction: A

systematic literature review.

Education and Information

Technologies, 29, 16255–

16282.

experiences

in

fostering

students’

English-speaking

skills: Insights into strategies

and challenges in public and

private schools. UNESUM-

Ciencias. Revista Científica

Multidisciplinaria, 9(2), 124–

https://doi.org/10.1007/s1063

9-024-12453-0

Mingyan, M., Noordin, N., & Razali,

A. B. (2025). Improving EFL

speaking performance among

undergraduate students with

an AI-powered mobile app in

after-class assignments: An

1

36.

https://doi.org/10.47230/unes

um-ciencias.v9.n2.2025.124-

136

Köbis, L., & Mehner, C. (2021).

Ethical questions raised by AI-

supported mentoring in higher

empirical

Humanities

Sciences

12,

investigation.

and

Social

Communications,

370.

education.

Artificial

Frontiers

Intelligence,

in

4,

6

24050.

https://doi.org/10.3389/frai.20

1.624050

https://doi.org/10.1057/s4159

9-025-04688-0

2

Narváez-Cantos, D. (2022). The

effective English language

teacher from the perspective

Liu, G. L., Darvin, R., & Ma, C.

2024). Exploring AI-mediated

(

informal digital learning of

English (AI-IDLE): A mixed-

of

Ecuadorian

university

students. Maskana, 13(2), 5–

method

investigation

of

14.

Chinese EFL learners’ AI

adoption and experiences.

Computer Assisted Language

https://doi.org/10.18537/mskn

.13.02.01

Parra-Gavilánez, L. (2022). An

analysis of the levels of

English of EFL teachers in

Ecuador. Medwave, 22(S2),

eUTA153.

Learning,

https://doi.org/10.1080/09588

21.2024.2310288

1–29.

2

Ma, Y., & Chen, M. (2025). The

human touch in AI: Optimizing

language learning through

self-determination theory and

https://doi.org/10.5867/medw

ave.2022.S2.UTA153

508

Chávez-Zambrano et al. (2026)

Peña-Acuña, B., & Corga Fernandes

Durão, R. (2024). Learning

English as a second language

with artificial intelligence for

investigation. Frontiers in

Psychology, 14, 1210187.

https://doi.org/10.3389/fpsyg.

2023.1210187

prospective A

systematic review. Frontiers in

Education, 9, 1490067.

https://doi.org/10.3389/feduc.

024.1490067

teachers:

Zhang, C., Meng, Y., & Ma, X.

(

2024). Artificial intelligence in

EFL speaking: Impact on

enjoyment, anxiety, and

willingness to communicate.

System, 121, 103259.

2

Qiao, H., & Zhao, A. (2023). Artificial

intelligence-based language

https://doi.org/10.1016/j.syste

m.2024.103259

learning:

Illuminating the

impact on speaking skills and

self-regulation in Chinese EFL

Zhou, A. (2023). Investigating the

impact of online language

context.

Frontiers

in

exchanges

language

on

speaking

second

and

Psychology, 14, 1255594.

https://doi.org/10.3389/fpsyg.

willingness to communicate of

Chinese EFL learners: A

2023.1255594

Sun, W. (2023). The impact of

automatic speech recognition

mixed

methods

study.

Frontiers in Psychology, 14,

1177922.

technology

on

second

language pronunciation and

speaking skills of EFL

learners: A mixed methods

https://doi.org/10.3389/fpsyg.

2023.1177922

Appendix A. Diagnostic Speaking Assessment Summary

Table A1. A1-A2 speaking descriptors

Grammar and

vocabulary

Interactive

communication

Band

5

Pronunciation

Good control of simple

forms and a range of

appropriate everyday

vocabulary.

Maintains simple

exchanges with very

little prompting or

support.

Mostly intelligible, with some

control of phonological

features.

Maintains simple

exchanges with some

limited phonological control. difficulty and requires

prompting.

Sufficient control of simple

forms and appropriate

everyday vocabulary.

Mostly intelligible despite

3

1

Considerable difficulty

Very limited phonological

Limited control of a few

maintaining

forms and isolated words control and often

or phrases. unintelligible.

exchanges; requires

additional support.

509

Revista Científica Multidisciplinaria Arbitrada YACHASUN. Volumen 10, Número 19 (Ed. jul – dic. 2026) ISSN: 2697-3456

AI-Supported mentoring: Speaking practice with Stimuler in future English teacher training.

Grammar and

vocabulary

Interactive

communication

Band

0

Pronunciation

Performance below Band

Performance below

Band 1.

Performance below Band 1.

1

.

Table A2. B1 speaking descriptors

Grammar and

vocabulary

Discourse

Interactive

communication

Band

Pronunciation

management

Produces

Initiates,

responds,

develops

interaction, and

negotiates with

very little support.

Good control of

simple forms,

attempts complex

forms, and uses a

Intelligible, with

extended, relevant

generally

speech with

cohesive devices

intonation, stress,

despite some

5

appropriate

range of vocabulary.

and clear sounds.

Mostly intelligible,

hesitation.

Extends beyond

short phrases,

simple forms and an remains mostly

Initiates and

responds

with some control at

appropriately with

very little

Good control of

3

appropriate range

for familiar topics.

relevant, and uses utterance and word

basic cohesive

devices.

level.

prompting.

Produces short

Sufficient control of phrases with

simple forms and a frequent

Mostly intelligible

despite limited

control of

phonological

features.

Maintains simple

exchanges with

difficulty and

1

0

limited appropriate

vocabulary.

hesitation,

repetition, or

digression.

requires support.

Performance below Performance

Band 1. below Band 1.

Performance below Performance

Band 1. below Band 1.

Appendix B. Interview Guide for English Teacher Educators

Purpose: To collect English teachers’ perceptions regarding the current speaking

proficiency of students in the English Language Program.

Instructions: Please answer honestly. Information will be used only for academic

research and kept confidential.

1. How would you rate the overall speaking level of the students you currently

teach in the English Language Program?

2. How often do your students use English spontaneously during class?

3

. Do your students reach the expected speaking level according to the semester

they are in?

510

Chávez-Zambrano et al. (2026)

4. In your opinion, what performance factors do you believe might affect the

students’ speaking skills?

5

6

7

8

9

1

. In your opinion, what linguistic factors do you believe might affect the students’

speaking skills?

. From your view, what affective factors do you believe might affect the students’

speaking skills?

. In your opinion, what are the main methodological factors that affect students’

speaking performance?

. Which teaching strategies do you usually use to help students improve their

speaking skills?

. Do you think technology can support the development of students’ speaking

skills? If so, which tools?

0. Which recommendations would you give to improve speaking skills in the

English Language Program?

511