The objective of this article is to analyze the didactic functionality of a chatbot to improve the results of the students of the National University of Distance Education (UNED / Spain) in accessing the university in the subject of Spanish Language. For this, a quasi-experimental experiment was designed, and a quantitative methodology was used through pretest and posttest in a control and experimental group in which the effectiveness of two teaching models was compared, one more traditional based on exercises written on paper and another based on interaction with a chatbot. Subsequently, the perception of the experimental group in an academic forum about the educational use of the chatbot was analyzed through text mining with tests of Latent Dirichlet Allocation (LDA), pairwise distance matrix and bigrams. The quantitative results showed that the students in the experimental group substantially improved the results compared to the students with a more traditional methodology (experimental group / mean: 32.1346 / control group / mean: 28.4706). Punctuation correctness has been improved mainly in the usage of comma, colon and periods in different syntactic patterns. Furthermore, the perception of the students in the experimental group showed that they positively value chatbots in their teaching–learning process in three dimensions: greater “support” and companionship in the learning process, as they perceive greater interactivity due to their conversational nature; greater “feedback” and interaction compared to the more traditional methodology and, lastly, they especially value the ease of use and the possibility of interacting and learning anywhere and anytime.

本文的目的是分析聊天机器人的教学功能，以提高国立远程教育大学 (UNED / 西班牙) 的学生在访问西班牙语学科的大学时的成绩。为此，设计了一个准实验性实验，在控制组和实验组中通过前测和后测采用定量方法，比较两种教学模式的有效性，一种更传统的基于纸上练习，另一种基于关于与聊天机器人的互动。随后，通过文本挖掘和潜在狄利克雷分配（LDA）、成对距离矩阵和二元组的测试，分析了实验组在学术论坛上对聊天机器人教育用途的看法。定量结果表明，与采用更传统方法的学生相比，实验组的学生在成绩上有显着提高（实验组/平均值：32.1346/对照组/平均值：28.4706）。标点正确性的改进主要体现在不同句法模式中逗号、冒号和句号的使用上。此外，实验组学生的看法表明，他们在三个维度的教学过程中积极重视聊天机器人：在学习过程中更多的“支持”和陪伴，因为他们的对话性质使他们感知到更多的交互性；与更传统的方法相比，更多的“反馈”和互动，最后，