This study proposes a novel method of computing language choice and use, by using social media, instead of questionnaires, and especially Facebook, to gather real-time data on individuals daily language use(s), and the how and to whom. The social media data contained more precise information on codemixing cases and enabled to construct a language contact situation typology. Methodological issues, especially regarding privacy and problems linked to language data crowd sourcing via social media, are debated.