Search Results

The emerging interdisciplinary field of Intelligent Computer Assisted Language Learning (ICALL) aims to integrate the knowledge from computational linguistics into computer-assisted language learning (CALL). REAP.PT is a project emerging from this new field, aiming to teach Portuguese in an innovative and appealing way, and adapted to each student. In this paper, we present a new improvement of the REAP.PT system, consisting in developing new, automatically generated, syntactic exercises. These exercises deal with the complex phenomenon of pronominalization, that is, the substitution of a syntactic constituent with an adequate pronominal form. Though the transformation may seem simple, it involves complex lexical, syntactical and semantic constraints. The issues on pronominalization in Portuguese make it a particularly difficult aspect of language learning for non-native speakers. On the other hand, even native speakers can often be uncertain about the correct clitic positioning, due to the complexity and interaction of competing factors governing this phenomenon. A new architecture for automatic syntactic exercise generation is proposed. It proved invaluable in easing the development of this complex exercise, and is expected to make a relevant step forward in the development of future syntactic exercises, with the potential of becoming a syntactic exercise generation framework. A pioneer feedback system with detailed and automatically generated explanations for each answer is also presented, improving the learning experience, as stated in user comments. The expert evaluation and crowd-sourced testing positive results demonstrated the validity of the present approach.