Can we consider "topic" a GRAMMATICAL category? If we cannot, what advantage can there be to calling the "discourse topic" simply "theme" from a PRAGMATIC perspective? 1 Can we bridge this terminological gap by making a distinction between "major" and "minor" topic, as Vasconcellos suggests in this volume? If we assume that there is a "direct" relation between thought and what the message of an utterance is (see for example the position of Givon), what psycholinguistic status (or reality) do topics in discourse have?
Theoretical linguists are preoccupied with the role of terms like "topic" or "theme." Their thinking and the solutions they propose for their description often decide for decades whether a linguistic subject or term is socially acceptable among linguists. 2 The reader may remember a case similar to that of "topic": the "dictum" of Bloomfield and other structuralists in the 1930s regarding the lexicon', they maintained that a "theory of the lexicon" was a subject to be studied only within psychology and not linguistics. While the "lexicon" and its theoretical status have become one of the central concerns of linguistics today, the notion of "topic" is still very much disputed. Although no doubt most linguists would agree that the phenomenon of "topic" provides the needed input energy to speech and communication, it receives surprisingly little attention in linguistic theory.
From the point of view of "descriptive" or "empirical" linguistics, theoretical reasoning might be easy, but once we come down to earth, to "real" data, we have to decide whether the "sentence," the "utterance," the "text," or the "verbal interaction" is the basic UNIT of the definition of "topic." There is no article in this special issue that does not address this question in a historical or systematic way. Givon underpins his definitions with psycholinguistic notions stimulated by current research into "artificial intelligence." Cadiot identifies "topics" in the