Evidence from Antiguan, Japanese and lingua franca English data reveal that a lot of the existing definitions and categories are ethnocentric constructs that do not stand a test with intercultural data and need re-definition. Taking into account the newest approaches regarding co-participation (Schegloff, 1996), collaborative construction of talk-in-interaction (Gramkow, Andersen, 1993 and Firth, 1990) and jointly constructed units of speech (Szartrowski, 1993) together with a combination of grammatical and semantic categories that account for possible units of collaboration may be rewarding in the explanation of lingua franca interactions. The unit of wadan, which is located somewhere between the turn and the exchange (in the sense of Discourse Analysis, cf. Coulthard (1985) or Stenström (1994)), may provide a category capable of grasping the heterogeneity encountered in lingua franca conversations.
Interpreting lingua franca conversational data
Care must be taken when assigning individual pragmatic characteristics of lingua franca talk-in-interaction to either cross-cultural interferences, the existence of a "third" culture or to learner language strategies. It has already been demonstrated (cf. Firth, 1990 and 1996, Gramkow, Andersen, 1993 and Meierkord, 1996 and 1998) that in lingua franca conversations speakers establish their own particular conversational style and that its characteristics can in most cases not simply be interpreted as the results of interferences with the individual speakers' mother tongues. This is also the case with the corpus analysed for the present paper.