Conceptual and procedural information for verb tense disambiguation
This chapter discusses the necessary linguistic and pragmatic support for improving statistical machine translation systems with respect to verbal tenses. The English Simple Past can be translated into French through a series of tenses because of its conceptual, procedural and pragmatic meanings. By testing and validating its usages in offline experiments with a linguistic judgment task, a general predictive model for the cross-linguistic variation of a verbal tense is proposed. The implementation of the procedure encoded by the Simple Past for a statistical machine translation system improved its results in terms of coherence and lexical choices. Thus, this chapter shows that such approach to automatic translation of verbal tenses seems promising and worth pursuing.