Knowledge-based intelligent error feedback in a Spanish ICALL system

Abstract

This paper describes the Spanish ICALL system ESPADA which helps language learners to improve their syntactical knowledge. The most important parts of ESPADA for the learner are a Demonstration Module and an Analysis Module. The Demonstration Module provides animated presentation of selected grammatical information. The Analysis Module is able to parse ill-formed sentences and to give adequate feedback on 28 different error types from different levels of language use (syntax, semantics, agreement). It contains a robust chart-based island parser which uses a combination of mal-rules and constraint relaxation to ensure that learner input can be analysed and appropriate error feedback can be generated.