Educators are more and more concerned to integrate human and computer capabilities as efficiently as possible. There is an increasing demand in language teaching for evidence on which to base decisions about use of this ever more prevalent tool. One main yardstick for evaluation of CALL is, naturally, the human teacher. Thus there is demand for comparative data. Attempts to provide such data in a way that allows useful generalisations to be made or accurate analyses of the outcomes are still comparatively limited and difficult to produce. Many attempts have been strongly criticised and the very concept of comparative studies of this kind has been attacked. The belief asserted here is that there is still a need for general studies in a variety of contexts. A longitudinal comparative study is reported that tries to take into account some of the major criticisms, while providing support for the idea that these kinds of study still need to continue. The study provides a detailed description of the design and implementation, an evaluation in terms of Second Language Acquisition (SLA) criteria, and pre and post tests to measure outcomes across a range of language learning tasks.