In the first part of the dissertation, which corresponds to the theoretical one, a series of cognitive theories are analysed; we also deal with the study of learning strategies; we analyse several studies on the interaction in the field of discourse analysis; we also describe different studies related to discourse and the acquisition of a new language and finally classroom discourse is analysed according to different aspects. Once these aspects are dealt with, we offer a bibliographic review on the topic studied in relation to the aspect of feedback. This review begins with works written in the 1960s and ends nowadays. We describe the works attending to aspects such as main objective, method used and results offered. Finally we deal with characteristics such as the importance of age in the learning process, since we are researching two groups of different ages, and we also describe different learning styles and affective factors. The research part is divided into hypotheses, subjects participating in the two contexts, method, results, discussion and conclusions. Some of the hypotheses we offer are the following ones: 1. In the native teacher class, the corrective exchanges will contain more moves. 2. There will be more confirmation in the class with a lower level of English. 3. There will be more correction in the group with a lower level of English. 4. The more experienced teacher will encourage more self-correction. With respect to the description of subjects in both contexts, we explain that fifteen lessons were audio-taped in two different schools, in two different levels and with teachers who had different characteristics. In this part we also explain the social context which is necessary to understand the characteristics of the two classes. In relation to the results obtained, we observed the following: We have observed that the E.S.O. class is much more interactive than the Bachillerato class, which is shown in the fact that exchanges are much longer in the former, even though the teacher in that class was not native. We consider that recast, which is the corrective technique used mostly for phonological errors, is a very adequate technique and, in fact, it is the technique which students accepted most. We could also observe that it was not always the student the one who chose to reject the acceptance of the correction: There were other options as the case of a different student or the teacher not accepting the correction. In this sense, we would advise teachers to reflect upon their attitude in class and they should consider giving students the opportunity to repeat and accept their corrections. As a conclusion, we must assert that our study of the incidence of error and correction leads us to adopt a positive attitude towards students mistakes. According to the communicative teaching methods used today, errors are not considered as lack of learning, they are rather the proof that learning is taking place. This is a fact broadly accepted today and that can be applied to the learning of both a first language and a second language. We learn through a process of trial and error, constructing and testing hypotheses, and continually revising them to the light of direct correction and the new data we receive from it. We learn a language through using it, rather than learning it first and then use it. Errors must be considered as a visible proof of the invisible process of learning. __________________________________________________________________________________________________ RESUMEN En la primera parte de la tesis, que corresponde a la parte teórica de la misma, se analizan las diferentes teorías cognoscitivas, los distintos enfoques en el estudio de las estrategias de aprendizaje, se lleva a cabo un análisis de los estudios sobre interacción en el campo del análisis del discurso, se relacionan los estudios sobre el discurso y la investigación en la adquisición de una lengua y finalmente se analiza el discurso del aula en varios aspectos. Una vez analizados estos aspectos se realiza una revisión bibliográfica del tema que nos interesa en relación con la corrección. Se describen los estudios realizados dando cuenta de los objetivos de las investigaciones, el método empleado y los resultados obtenidos. Finalmente analizamos aspectos como la diferencia de edad en el aprendizaje, así como los distintos estilos de aprendizaje y aspectos afectivos. En la parte de investigación distinguimos hipótesis del trabajo, sujetos actuando en los dos contextos y método de trabajo, resultados y discusión de los mismos, y conclusiones. Explicamos que 15 lecciones fueron grabadas en dos colegios diferentes, en dos niveles también distintos y con dos profesoras con diferentes características. En cuanto a los resultados obtenidos, se observó en general que la clase de E.SO. es mucho más viva e interactiva que la clase de Bachillerato. También se concluye a partir de dichos resultados que la reformulación empleada para corregir los errores en fonología nos parece una técnica muy adecuada y en efecto es aquella que más aceptada es por los estudiantes. Nuestra investigación sobre la incidencia del error y la corrección nos lleva a adoptar una actitud positiva en cuanto a los errores de los estudiantes. De acuerdo con los actuales métodos de enseñanza de tipo comunicativo, el error no debe ser considerado como una falta de aprendizaje sino más bien la prueba de que el aprendizaje está ocurriendo.