SMILLE for Portuguese: Annotation and Analysis of Grammatical Structures in a Pedagogical Context.

In Second Language Acquisition (SLA), the exposure of learners to authentic material is an important learning step, but the use of raw text may pose problems, because the information that the learner should be focusing on may be overlooked. In this paper, we present SMILLE for Portuguese, a system for detecting pedagogically relevant grammatical structures in raw texts. SMILLE’s rules for recognizing grammatical structures were evaluated in random sentences from three different genres, achieving an overall precision of 84%. The automatic recognition of pedagogically relevant grammatical structures can help teachers and course coordinators to better inform the choice of texts to be used in language courses, while also allowing for the analysis of grammar profiles for SLA. As a case study, we used SMILLE to analyze pedagogical material used in a Portuguese as foreign language course and to observe how the predominance of grammatical content in the texts is related to the described grammatical focus of the language levels.