In this talk, we present the results of a study on the acquisition of Dutch compound constructions by French-speaking learners in Belgium. Additionally, we compare learners enrolled in Content and Language Integrated Learning (CLIL) programs with learners following traditional second language instruction. Languages are known to vary significantly with respect to their preferences for analytic or synthetic constructions (Schlücker 2019). For instance, Germanic languages tend to use compounds more frequently than Romance languages (Van Goethem 2009; Booij 2010; Schlücker 2019). Van Goethem (2009), for instance, has demonstrated that Dutch has a stronger tendency towards [A+ N] compounding than French (e.g. Du. hoogspanning vs Fr. haute tension ‘high voltage’). Based on the cross-linguistic similarities and differences between the preferences for compound or phrasal constructions in Dutch and French and on the beneficial effect of CLIL on SLA, as already demonstrated in the case of intensifying constructions (Hendrikx 2019), we assume that the CLIL learners’ L2 use of Dutch will integrate more features typical of L1 Dutch use of compounds. We will take into consideration the formal make-up of the compounds, their semantic classification, their frequency and productivity, and possible mistakes in the learner data (for instance, non-target-like phrasal constructions). The data for this study come from a corpus of written productions in the form of fictional e-mails, collected within the context of a research project on CLIL in French-speaking Belgium (cf. Hiligsmann et al. 201, Meunier et al. 2020). From the theoretical point of view, the study takes a Diasystematic Constructionist approach (DCxG) (Höder 2012, 2018; Boas & Höder 2018, 2021), which conceptualizes the linguistic competence of multilingual speakers as an “interlingual network of constructions with different degrees of schematicity†(Höder 2012: 255). Although Höder’s framework has mainly been applied to issues of language contact ...