Abstract
We discuss ways in which to interpret “understanding” and “translating” from a machine's point of view, and we examine some of the reasons why such tasks are so difficult (namely, ambiguity, pragmatics, stylistic resources, world views and the interpretation of metamessages). Then we discuss the successes so far, and our readiness for solutions which may depart from traditional approaches in view of trying to get the whole picture regarding meaning (e.g., by the use of less studied clues such as body language or intonation). We advocate focussing on deduction and on parametrization, and we also promote the use of cognitive sciences results, in particular linguistic theory, while examining the challenges that this difficult integration will entail.
Get full access to this article
View all access options for this article.