Analysis of EFL speech production corpora according to the Speech Learning Model

Abstract

[ES] Este TFG trata de mostrar cómo pronuncia una persona del País Vasco al hablar en Inglés (teniendo esta lengua como lengua extranjera) de acuerdo a un modelo teórico sobre el aprendizaje de segundas lenguas. También trata de mostrar si éste modelo es capaz de explicar todos los fallos de pronunciación de la hablante y, en caso de que no lo haga, sugerir que puede que haya necesidad de desarrollar una teoría enfocada al aprendizaje del habla de lenguas extranjeras.

[EN] Nowadays learning a language other than one's own mother tongue is seen as a must in our society. The reasons for considering this are various: possibilities of moving from one place to another (travelling), immigrating to start a better life in a foreign country, bigger chances to achieve a better career and the like. In this context, it is important to speak the language of the place of destination accurately to be understood by the native population. This means that grammar is not the only important aspect to be acquired in a new language but also its speech patterns and pronunciation. One of the main researchers investigating the phenomenon of pronunciation is J. E. Flege. He has conducted numerous experiments and pieces of research regarding second language (L2) speech acquisition and developed the Speech Learning Model (SLM) (1995) to explain and predict the pronunciation mistakes L2 learners make. A similar phenomenon to L2 learning is foreign language (FL) learning. Nevertheless, it has to be stressed that, although similar, they are not the same event. One of the main differences between these two learning processes is that L2 learners are in a linguistic immersion setting whereas FL students are not; and also that the latter ones receive formal classroom instruction mostly oriented to grammar skills rather than to oral skills. Thus, from this first impression, it can be deduced that L2 and FL learning do not occur in the same circumstances. As a consequence, a need to verify whether an L2 learning oriented theory was sufficient to explain all phenomena occurring in FL learning was detected. Therefore, the aim of this study is on the one hand, to analyse FL speech production and see whether the SLM can explain all pronunciation errors of a FL speaker; and on the other, to shed some light on FL speech teaching methodologies so that it becomes more effective and efficient. The confirmation of the existence of unexplained or unpredicted mistakes might mean that a FL speech learning theory would be needed; and a big amount of predicted mistakes might suggest that the teaching practices should be remodelled. The methodology followed in the analysis of errors was listening to the data with Praat and classifying the mistakes according to the SLM (as new, same or similar sound categories). The results showed that some of the participant's mistakes were not explained by the model (such as orthographically motivated mistakes). Taking into account that, as demonstrated by many studies -such as the one by Gómez Lacabex & García Lecumberri (2010)-, perception and production ii training do help improving those abilities, it would be interesting to incorporate such training sessions into the classrooms. Finally, this dissertation also aims to encourage more research within the field of FL speech learning in order to firstly, understand better this phenomenon and secondly, improve the teaching system.