Reading collocations in an L2: do collocation processing benefits extend to non-adjacent collocations?

Various studies have consistently shown that collocations are processed faster than matched control phrases, both in L1 and in L2. Most of these studies focused on adjacent collocations (e.g., provide information). However, research in corpus linguistics normally uses a span to identify collocations (e.g., ± 4 words), and these non-adjacent collocations (e.g., provide some of the information) occur very frequently in language. Nevertheless, how they are processed is less established. A recent study on reading non-adjacent collocations seems to suggest similar processing advantages as for adjacent collocations (Author 2016), but this study was limited to the performance of native speakers. The present study addresses the question of whether advanced non-native speakers also show processing advantages for nonadjacent collocations as native speakers do. Forty advanced non-native speakers of English read collocations in either adjacent or non-adjacent conditions, and their eye-movements were recorded. Mixed-effects analysis of their eye-movements was carried out. The results suggest that non-native speakers read adjacent collocations faster than non-formulaic controls, but this facilitation almost disappears for non-adjacent collocations