This work is essentially based on grammaticalization theory - a branch of linguistics which has gained prominence since the 1980s. It focuses on the interaction between diachrony and synchrony, langue and parole or, for that matter, competence and performance, I-language and ?-language. It does not see these levels as distinct linguistic domains, as much structurally oriented work does. It is important for the present purposes that such an interactionist view entails that performance effects may over time cause new grammatical code relations. Hence the importance of statistical empirical research, which led the author to adopt a predominantly corpus-based approach.