Pragmatics and dialogue phenomena
Are they essentially connected?
I argue that the definition of ‘pragmatics’ has drifted a great deal since Morris, but that there is much to be said for retaining and emphasizing his original sense of relatedness to an utterer. I discuss the role of a computational modelling of dialogue within that core conception of pragmatics, arguing it should be near the center on that classic definition, although computational models have never been central to linguistic pragmatics. I review the relevance to this strand of pragmatics of the data versus rules debate within computational linguistics over the last twenty years and argue that we must stay with some form of the data-driven approach since the alternative has not been proved productive.