In this poster I will first present the overall framework, objectives and methodology of a new interdisciplinary project on Content and Language Integrated Learning (CLIL) in French speaking Belgium. The project, which aims to assess CLIL at the interface of linguistic, cognitive and educational perspectives, includes four distinct but intertwined PhD research lines. My own research line aims to gain insight in the acquisition of Dutch and English phraseological units by French-speaking CLIL and non-CLIL secondary school pupils. Much literature has been devoted to the prefabricated nature of authentic language use and various researchers agree to say that language is to a large extent formulaic (Wray 2002, Ellis 2008, Erman & Warren 2000), i.e. made of ‘chunks’. Typical examples of phrasemes or phrases include referential phrasemes (such as collocations, compounds, etc.), textual phrasemes (complex prepositions, linking adverbials, etc.) and communicative phrasemes (e.g. speech act formulae, proverbs, etc.), see Granger & Paquot (2008). While phraseological units are said to be acquired mostly unconsciously by native speakers, their acquisition by non-native speakers is less indirect, straightforward or equally complete. Since CLIL-programs provide more target-like and input-rich environments than non-CLIL programs and can therefore be considered closer to L1 acquisition because of their inherent usage-based approach, we can expect pupils in immersion to produce a more phraseological (and target-like) language output than that of students in traditional settings. This will be tested carrying out a longitudinal study covering the 5th and 6th grades of different secondary schools in French-speaking Belgium providing CLIL or non-CLIL education in Dutch and/or English. The analysis will be based on truly longitudinal corpus data (transcribed spoken interactions and written productions), complemented by experimental data. Data collection has now started in Dutch and English CLIL programs and non-CLIL ...