Tipo de artículo: Artículo original

Abstract

The growing use of artificial intelligence among English language students raised the need to understand how they viewed the tool as support for speaking practice. The study aimed to describe prior use and perceptions of ChatGPT among students enrolled in the Pedagogy of National and Foreign Languages program. In December 2025, a quantitative, nonexperimental, cross-sectional, and descriptive study was conducted with 30 fourth-semester students at Universidad Estatal del Sur de Manabí. A structured digital questionnaire consisting of seven closed-ended items examined prior use, acceptance, perceived usefulness, and self-assessments related to speaking practice. Twenty-four students (80.0%) reported previous use of ChatGPT; 26 (86.7%) expressed a favorable view of the tool, indicated that they would recommend it, and perceived the feedback it can provide as clear and useful. In addition, 20 participants (66.7%) believed it could support greater fluency and fewer pauses, 24 (80.0%) perceived support for better organization of ideas, and all 30 stated that they could maintain basic conversations. The findings reflect favorable perceptions rather than demonstrated effects or an objective assessment of speaking competence. ChatGPT may be considered a complementary resource when its use is guided by communicative purposes and teacher mediation.

Keywords: ChatGPT; generative artificial intelligence; speaking practice; student perceptions; English as a foreign language

Fecha de recibido: 02/05/2023

Fecha de aceptado: 10/07/2026

Fecha de publicado: 24/07/2026

Introducción

Speaking ability is an essential part of language teacher education. Students preparing to teach English need to communicate effectively and, at the same time, understand how oral interaction can be supported in the classroom. Speaking involves more than producing grammatically correct sentences. It also requires speakers to organize ideas, respond to an interlocutor, maintain interaction, and adjust their message when communication difficulties arise. Communicative competence includes grammatical, sociolinguistic, and strategic dimensions (Canale & Swain, 1980), while the Common European Framework of Reference for Languages describes spoken performance through features such as range, accuracy, fluency, interaction, coherence, and phonological control (Council of Europe, 2020).

In English as a foreign language context, students do not always have enough opportunities to practice speaking regularly. Limited class time, restricted individual participation, and concern about making mistakes in front of others can reduce their willingness to speak (Fathi et al., 2024). These difficulties are particularly relevant in language teacher education because students are developing their own communicative abilities while also becoming familiar with resources they may later use in their professional practice.

Conversational artificial intelligence has created new possibilities for extending speaking practice beyond conventional classroom interaction. Tools such as ChatGPT can maintain a dialogue, respond immediately, and allow learners to practice repeatedly without the pressure of speaking before an entire class. However, access to immediate interaction does not necessarily guarantee accurate feedback or meaningful language development. Its educational value depends on the way the tool is used, the purpose of the activity, and the guidance provided to students.

Research on chatbots for language learning has identified several advantages, including accessibility, immediacy, personalization, and increased opportunities for interaction. At the same time, limitations related to technical performance, cognitive load, novelty effects, and feedback quality have also been reported (Huang et al., 2022). The available evidence is diverse because chatbots differ in their objectives, design, interaction modality, and degree of specialization. Reviews have shown substantial variation in the populations, technologies, tasks, and outcomes examined (Jeon et al., 2023; Koç & Savaş, 2025). These differences make it difficult to treat all conversational systems as if they offered the same educational experience.

Studies involving structured chatbot activities have nevertheless reported promising results. Favorable outcomes for speaking performance and willingness to communicate were found when an English-learning chatbot was incorporated into organized speaking activities (Fathi et al., 2024). Positive speaking outcomes were also reported among school learners who interacted individually or in pairs with a generative chatbot (Tai & Chen, 2024). In both studies, the activities followed a defined instructional design, and speaking was assessed through specific instruments. Their findings suggest that conversational technology can support language practice when it is connected to clear learning objectives.

The literature also shows that students’ feelings about a tool and their actual language performance are related but distinct. Favorable changes in willingness to communicate, speaking anxiety, and self-perceived competence have been identified among learners who interacted with a chatbot, although speaking-test results did not show significant differences among the groups (Wang et al., 2024). This distinction is important because enjoyment, confidence, acceptance, and perceived usefulness may encourage participation, but they do not by themselves demonstrate improvement in fluency, pronunciation, accuracy, or interaction.

Evidence from the broader field of dialogue-based language-learning systems also points to generally favorable but context-dependent results. A moderate overall effect on speaking development has been estimated, with outcomes varying according to system type, interaction modality, and the constraints placed on communication (Hou & Min, 2026). Many of the systems included in that analysis were rule-based or specifically designed for language learning, rather than general-purpose generative tools. Consequently, their findings support the educational potential of conversational systems without establishing that every use of ChatGPT will produce the same results.

Interest in artificial intelligence for English teaching has also grown in Ecuador. Previous studies have examined its use in higher education, the application of ChatGPT to university speaking practice, and the use of specialized platforms such as SmallTalk2Me in secondary education (Espinoza Molina et al., 2024; Heylighen & Vinueza Jarrín, 2025; Jiménez Iñiguez & Criollo Vargas, 2025). These contributions provide valuable regional evidence, but they involve different educational levels, technologies, populations, and assessment procedures. More context-specific information is therefore needed to understand how students in particular teacher education programs, use and evaluate ChatGPT.

The educational incorporation of generative artificial intelligence also requires teacher guidance and critical digital literacy. Students should be aware that a response may sound clear and convincing while still containing linguistic or factual inaccuracies. For this reason, feedback generated by ChatGPT needs to be reviewed in relation to the objectives of the task and appropriate language criteria. A human-centered approach also requires attention to privacy, human agency, and responsibility for educational decisions (Miao & Holmes, 2023).

Most previous research has focused on structured interventions, specialized language-learning systems, or direct measures of speaking performance. Less descriptive evidence is available on how students in Ecuadorian language teacher education programs use and value ChatGPT as a general-purpose resource for speaking practice. Understanding these perceptions is relevant because students’ acceptance of the tool, their assessment of its feedback, and their experiences during practice may influence whether and how they incorporate it into their learning.

To address this need, the present study used a descriptive quantitative survey with fourth-semester students enrolled in the Pedagogy of National and Foreign Languages program at Universidad Estatal del Sur de Manabí. It did not include a ChatGPT-based intervention, speaking practice task, or direct assessment of speaking performance. Instead, it examined students’ previous use of the tool, their acceptance of it, their perceptions of the feedback it can provide, and their views regarding its potential contribution to fluency, organization of ideas, and basic conversational practice.

The study was guided by the following question: How did fourth-semester students describe their use and perceived value of ChatGPT for English speaking practice? Accordingly, the objective was to describe students’ use, acceptance, and perceived usefulness of ChatGPT as a support resource for English speaking practice.

Materiales y métodos

This study followed a quantitative, nonexperimental, cross-sectional, and descriptive design. Data were collected at a single point in time in December 2025 to describe previous use of ChatGPT and students’ perceptions of its value for English speaking practice. No instructional intervention, ChatGPT practice activity, speaking test, or comparison group was included because the study did not seek to determine causal effects or changes in speaking performance.

The study involved 30 fourth-semester students enrolled in the Pedagogy of National and Foreign Languages program at Universidad Estatal del Sur de Manabí, Ecuador. Participants were selected through non-probability purposive sampling because they belonged to the academic cohort considered relevant to the purpose of the study. Participation was voluntary, and the findings were interpreted within this specific educational context.

Data were collected through a structured digital questionnaire administered by the authors. The instrument consisted of seven closed-ended items addressing previous use of ChatGPT, favorable perception of the tool as a support resource for speaking practice, willingness to recommend it, perceived clarity and usefulness of the feedback it can provide, perceived potential to support greater fluency and fewer pauses, perceived potential to support better organization of ideas, and self-reported ability to maintain basic conversations in English.

All 30 participants answered the perception and self-assessment items regardless of whether they reported previous use of ChatGPT. Each item was analyzed separately according to the number of affirmative or favorable responses, and the responses were not combined into a composite score. The questionnaire was not treated as a standardized measure of speaking proficiency or as evidence of objective changes in oral performance.

The questionnaire was administered digitally in December 2025. No participant was required to use ChatGPT or complete a speaking task as part of the study. Before accessing the survey items, participants read an informed-consent statement included in the form and electronically indicated their agreement to participate. Responses were collected anonymously, and the information was used exclusively for academic and research purposes.

Responses were analyzed using descriptive statistics. Absolute frequencies and percentages of affirmative or favorable responses were calculated for each item using the total number of participants as the denominator (n = 30). No composite score or inferential statistical test was calculated. Because only aggregate item-level results were available, the analysis did not compare responses according to previous ChatGPT use or other participant characteristics. The findings were interpreted as perceptions and self-assessments rather than evidence of causal effects or objectively measured changes in speaking competence.

Resultados y discusión

The results were organized into two areas: previous use and perceptions of ChatGPT as a support resource; and self-assessments related to English speaking practice. Table 1 presents the absolute frequency and percentage of affirmative or favorable responses recorded for each item.

Table 1. Survey Responses on Perceptions of ChatGPT and English-Speaking Practice.

Indicator	n (%)
Previous use of ChatGPT	24 (80.0%)
Favorable perception of ChatGPT as a speaking-practice support resource	26 (86.7%)
Willingness to recommend ChatGPT for speaking practice	26 (86.7%)
ChatGPT-generated feedback perceived as clear and useful	26 (86.7%)
Perceived potential to support greater fluency and fewer pauses	20 (66.7%)
Perceived potential to support better organization of ideas	24 (80.0%)
Self-reported ability to maintain basic conversations in English	30 (100.0%)

Note. Frequencies and percentages represent favorable responses (n = 30). All participants answered the perception items. No ChatGPT intervention or speaking task was conducted; speaking-related items are perceptions or self-assessments, not performance measures.

Previous Use and Perceptions of ChatGPT

Twenty-four participants (80.0%) reported that they had previously used ChatGPT. This result indicates that the tool was already familiar to most students in the participating cohort. However, the questionnaire did not collect information about the frequency, duration, or purposes of that previous use.

Twenty-six participants (86.7%) expressed a favorable perception of ChatGPT as a support resource for speaking practice, indicated that they would recommend it, and perceived the feedback it can provide as clear and useful. Because the study did not include a ChatGPT activity or intervention, these responses should be understood as students’ opinions about the tool rather than evaluations of a standardized practice experience. Similar favorable perceptions have been reported in research highlighting the accessibility, immediacy, and interaction opportunities provided by language-learning chatbots (Huang et al., 2022; Koç & Savaş, 2025).

The favorable appraisal of ChatGPT-generated feedback does not establish its linguistic accuracy or pedagogical appropriateness. Previous research has shown that chatbot feedback can vary in explicitness, accuracy, and relevance to specific learning objectives (Koç & Savaş, 2025). Consequently, teacher guidance remains important when students use generated responses for language learning.

Perceptions and Self-Assessments Related to Speaking Practice

Twenty participants (66.7%) believed that ChatGPT could support greater fluency and fewer pauses, while 24 (80.0%) perceived that it could help them organize their ideas more effectively. These responses describe students’ expectations and perceptions of the tool’s potential contribution to speaking practice; they do not demonstrate that fluency or discourse organization improved.

Direct evaluation of fluency would require comparable oral samples and analysis of features such as speech rate, pause frequency and duration, continuity, and reformulation. None of these elements was assessed in the present study. The findings should therefore be interpreted as perception data rather than verified linguistic change.

All 30 participants stated that they could maintain basic conversations in English. This response represents a general self-assessment of conversational ability and was not linked to a ChatGPT-based task. It does not demonstrate a particular proficiency level. The Common European Framework of Reference for Languages requires observable evidence of production and interaction, including fluency, coherence, accuracy, range, and phonological control (Council of Europe, 2020).

The distinction between perception and demonstrated performance is consistent with previous research. Favorable changes in willingness to communicate, speaking anxiety, and self-perceived communicative competence have been reported without corresponding significant differences in speaking-test performance (Wang et al., 2024). Such evidence confirms that self-reported outcomes should be interpreted separately from directly assessed speaking ability.

Other studies have reported improvements in speaking performance when chatbots were integrated into structured instructional interventions and evaluated through specific speaking measures (Fathi et al., 2024; Tai & Chen, 2024). Those studies differed from the present descriptive survey because they included guided activities and direct assessments. Their findings therefore illustrate what could be examined in future intervention research but cannot be treated as outcomes of the present study.

Implications for Language Teacher Education

The favorable perceptions observed suggest that students may regard ChatGPT as a complementary resource for English speaking practice. Its potential value lies in offering additional space to rehearse ideas, simulate interaction, and obtain immediate responses beyond regular classroom speaking time.

These findings do not establish that unstructured use of ChatGPT improves speaking competence. Previous research indicates that educational outcomes depend on the clarity of instructions, the design of prompts, the communicative purpose of the activity, and the connection between chatbot interaction and subsequent human communication (Koç & Savaş, 2025; Lin & Mubarok, 2021). In language teacher education, any future use of ChatGPT should therefore involve clearly delimited tasks, teacher review of generated feedback, and opportunities for interaction with peers and teachers.

Its incorporation should also include critical evaluation of generated responses, attention to data protection, and preservation of human responsibility in educational decision-making (Miao & Holmes, 2023). These are pedagogical implications supported by literature, not variables measured in the survey.

Scope and Limitations

The study involved 30 fourth-semester students from one academic program and one university. The results therefore describe the participating cohort and should not be generalized to all students in the program or to other institutions.

The questionnaire consisted of seven closed-ended items, and the analysis was limited to aggregate frequencies and percentages of affirmative or favorable responses. All participants answered the perception items, but the available data did not permit comparisons between students with and without previous ChatGPT use. Relationships among items and differences according to participant characteristics were not examined.

The psychometric properties of the questionnaire were not examined, and no composite score was calculated. The study also did not include open-ended questions, a ChatGPT intervention, speaking tests, comparable oral samples, trained raters, or a comparison group. Consequently, the findings describe perceptions and self-assessments but do not establish causal effects or demonstrate improvements in fluency, pronunciation, accuracy, or communicative competence.

Future research could compare perceptions according to previous experience with ChatGPT and combine survey data with clearly defined chatbot-supported activities, pre- and post-intervention speaking tasks, analytic rubrics, trained evaluators, and larger and more diverse samples.

Conclusions

The study showed that students held generally favorable perceptions of ChatGPT as a support resource for English speaking practice. Most participants reported previous familiarity with the tool, and 26 expressed a favorable view of its potential use, indicated that they would recommend it, and perceived the feedback it can provide as clear and useful. These findings fulfilled the descriptive objective of characterizing prior use, acceptance, and perceived usefulness within the selected university cohort.

Participants also expressed favorable views regarding ChatGPT’s potential to support fluency, reduce pauses, and help organize ideas. In addition, all 30 students reported that they could maintain basic conversations in English. These responses were perceptions and self-assessments collected through a questionnaire; they were not the result of a ChatGPT intervention and do not constitute direct evidence of speaking proficiency or linguistic improvement.

Within these limits, ChatGPT may be considered a complementary resource for speaking practice when its use is organized around clear communicative purposes, critical evaluation of generated responses, and teacher mediation. It should complement rather than replace interaction with teachers and peers.

Future studies should examine whether these favorable perceptions correspond to observable changes in speaking performance by using clearly defined interventions, comparable oral tasks, analytic assessment criteria, trained evaluators, and larger and more diverse participant groups.

References

Canale, M., & Swain, M. (1980). Theoretical bases of communicative approaches to second language teaching and testing. Applied Linguistics, 1(1), 1–47. https://doi.org/10.1093/applin/I.1.1

Council of Europe. (2020). Common European Framework of Reference for Languages: Learning, teaching, assessment—Companion volume. Council of Europe Publishing. https://rm.coe.int/common-european-framework-of-reference-for-languages-learning-teaching/16809ea0d4

Espinoza Molina, N. D., Intriago Alcívar, W. S., Suing Nagua, Á. M., & Velasquez Tonato, J. C. (2024). Inteligencia artificial: Su influencia en las clases de inglés como idioma extranjero en la educación superior. Kronos – The Language Teaching Journal, 5(1), 93–106. https://doi.org/10.29166/kronos.v5i1.6639

Fathi, J., Rahimi, M., & Derakhshan, A. (2024). Improving EFL learners’ speaking skills and willingness to communicate via artificial intelligence-mediated interactions. System, 121, 103254. https://doi.org/10.1016/j.system.2024.103254

Heylighen, H. J. M., & Vinueza Jarrín, W. M. (2025). Estudio de caso: ChatGPT para mejorar la comunicación oral en inglés como lengua extranjera en estudiantes universitarios. Realidad y Reflexión, 1(62), 164–196. https://doi.org/10.5377/ryr.v1i62.21737

Hou, Z., & Min, S. (2026). Dialogue-based computer-assisted language learning systems for second language speaking development: A three-level meta-analysis. ReCALL, 38(1), 40–56. https://doi.org/10.1017/S0958344025100268

Huang, W., Hew, K. F., & Fryer, L. K. (2022). Chatbots for language learning—Are they really useful? A systematic review of chatbot-supported language learning. Journal of Computer Assisted Learning, 38(1), 237–257. https://doi.org/10.1111/jcal.12610

Jeon, J., Lee, S., & Choe, H. (2023). Beyond ChatGPT: A conceptual framework and systematic review of speech-recognition chatbots for language learning. Computers & Education, 206, 104898. https://doi.org/10.1016/j.compedu.2023.104898

Jiménez Iñiguez, C., & Criollo Vargas, M. I. (2025). Mejora de las habilidades de expresión oral en inglés mediante la herramienta basada en inteligencia artificial SmallTalk2Me. Kronos – The Language Teaching Journal, 5(2), 95–108. https://doi.org/10.29166/kronos.v5i2.7668

Koç, F. Ş., & Savaş, P. (2025). The use of artificially intelligent chatbots in English language learning: A systematic meta-synthesis study of articles published between 2010 and 2024. ReCALL, 37(1), 4–21. https://doi.org/10.1017/S0958344024000168

Lin, C.-J., & Mubarok, H. (2021). Learning analytics for investigating the mind map-guided AI chatbot approach in an EFL flipped speaking classroom. Educational Technology & Society, 24(4), 16–35. https://doi.org/10.30191/ETS.202110_24(4).0002

Miao, F., & Holmes, W. (2023). Guidance for generative AI in education and research. UNESCO. https://doi.org/10.54675/EWZM9535

Tai, T.-Y., & Chen, H. H.-J. (2024). Improving elementary EFL speaking skills with generative AI chatbots: Exploring individual and paired interactions. Computers & Education, 220, 105112. https://doi.org/10.1016/j.compedu.2024.105112

Wang, C., Zou, B., Du, Y., & Wang, Z. (2024). The impact of different conversational generative AI chatbots on EFL learners: An analysis of willingness to communicate, foreign language speaking anxiety, and self-perceived communicative competence. System, 127, 103533. https://doi.org/10.1016/j.system.2024.103533