Measuring the effects of explicit and non-explicit grammar instruction : casting the net wide

In order to measure the effects of grammar instruction, it is common to elicit performance on particular target forms before and after instruction, to see what changes have occurred, and to draw inferences from these findings. However, in much instructed SLA research it seems that certain test types are preferred and that only a limited range of tests is administered. In this paper I will consider what it is that common types of test measure and discuss whether the results of different types of test can be combined to give a more rounded picture of the state of the learners’ target form knowledge. Armed with this more detailed picture, it is argued that one would be in a better position to say how instruction had affected the knowledge which underlies a wider range of performance, thus making the findings more robust and potentially more directly useful to language teachers.