This article is a reaction to M. Komarek's essay Communication versus system? (1999) and is primarily concerned with the critical analysis of the dichotomic concept of natural language. In particular, the absence of empirical evidence for a language system (langue) is pointed out, which creates serious issues for the entire structuralist approach. That is, if it is impossible to have empirical experience with a language system (langue), it is thus impossible from the position of empirical science to make any sort of claim regarding the relationship between this system and concrete instances of speech. It is thus deemed necessary to reject the langue-parole dichotomy in linguistics. The aim of non-dichotomic linguistics is, then, to create models of the speech behavior of language users, not the reconstruction of a language system (langue). As natural language in actual communication is quite varied, these models will have a merely approximative character.