||Foreign language learners’ choice of collocations istraditionally considered tobeoneofthemainmarkersof foreignlanguageness (Korosadowitz-Struzynska 1980: 115), hence relevant in achieving a high degree of compe-tence in the target language. In this study we analyse the Croatian Corpus ofEnglish Learner Essays (CELE), which consists of 298 argumentative essayswritten as part of the state school-leaving exam. The corpus, consisting ofover 74k tokens was collected in 10 different counties, and the essays wereproducedin2010and2011.Thelearner’suseof adjective-noun collocationsis compared against both findings from a native speaker corpus (BNC) and acorpus of learner English (ICLE). Instead of viewing learner usage of colloca-tions as deficient, the claims about the overuse and the underuse of statisti-cally significant collocations are made on the basis of joint findings from theBNC and the ICLE. This approach demonstrates how native speaker data canbe used for comparison, without being the norm against which the learner datashould be measured, and, on the other hand, how other learner data can helpdistinguish between general characteristics of learner language and L1-transfer.