Evaluación de las soluciones existentes de IA generativa en la detección de errores del uso de la lengua
Revista Internacional de Lenguas Extranjeras / International Journal of Foreign Languages · 22.07.2026 · (26) · 181-199
DOI: 10.17345/rile26.4449 ↗Kurz gefasst
Fachartikel in der Zeitschrift Revista Internacional de Lenguas Extranjeras / International Journal of Foreign Languages, online erschienen am 22.07.2026. Verfasst von Rafael Fernández Castillejos und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
This paper explores in depth the capabilities of three generative artificial intelligence tools—ChatGPT, Gemini, and DeepSeek—to solve syntactic problems and provide grammatical reasoning that justifies their solutions. The research is framed within the PALABRIA-CM-UC3M project, carried out by a team of engineers and linguists. The main objective is to evaluate the overall performance and responses to doubts, errors, and linguistic questions of several widely used free generative AI tools among university students. The methodology is based on using as prompts 123 questions proposed by Ignacio Bosque in Repaso de sintaxis tradicional: ejercicios de autocomprobación (1994/2004). The accuracy rate was low in all three cases—ChatGPT (11%), Gemini (20%), and DeepSeek (22%). Although the responses tend to be highly explanatory, consistently including reasoned justifications with concrete examples that illustrate the claims, the tools’ handling of grammatical terminology is weak and includes errors that are not immediately obvious, as they are concealed within lengthy and circular arguments that repeat the same idea. In this regard, these tools are not suitable for solving linguistic problems for students at intermediate or advanced levels.
Zitieren
APA
Castillejos, R. F., & Hernández, A. S. (2026). Evaluación de las soluciones existentes de IA generativa en la detección de errores del uso de la lengua. Revista Internacional de Lenguas Extranjeras / International Journal of Foreign Languages, 181-199. https://doi.org/10.17345/rile26.4449
Vancouver
Castillejos RF, Hernández AS. Evaluación de las soluciones existentes de IA generativa en la detección de errores del uso de la lengua. Revista Internacional de Lenguas Extranjeras / International Journal of Foreign Languages. 2026:181-199. doi:10.17345/rile26.4449
BibTeX
@article{Castillejos2026,
author = {Castillejos, Rafael Fernández and Hernández, Ariana Suárez},
title = {Evaluación de las soluciones existentes de IA generativa en la detección de errores del uso de la lengua},
journal = {Revista Internacional de Lenguas Extranjeras / International Journal of Foreign Languages},
year = {2026},
number = {26},
pages = {181--199},
doi = {10.17345/rile26.4449}
}
EUVIMED