Towards modelled testing of productive knowledge of collocations

Accurate use of collocations is seen as a measure of a learner’s proficiency in a second/ foreign language. However, research is not conclusive as to how to best test productive knowledge of collocations and, more fundamentally, there is no agreement on how vocabulary size, i.e. word comprehension, influences collocation production within a constrained or unconstrained context. This article elaborates on a battery of tests aimed at examining these issues. The tests were presented to both English as Foreign Language (EFL) and English as Second Language (ESL) students. The findings indicate that constrained and unconstrained contexts may represent different levels of productive knowledge, casting doubt on the definition of productive knowledge collapsing the two contexts – constrained and unconstrained – into one construct. Furthermore, both levels are influenced by the size of a learner’s receptive knowledge, which confirms Gyllstad’s earlier observation on the relationship between receptive knowledge of collocations and vocabulary size, which is extended here to include productive knowledge. On the basis of these results, we argue that a new definition for ‘productive knowledge of collocations’ is in order, where productive knowledge is categorised according to four levels based on the amount of context provided to the learner