Score equating requires that the forms to be equated are functionally parallel. That is, the two test forms should rank order examinees in a similar fashion. In language proficiency testing situations, this assumption is often put into doubt because of the numerous tests that have been proposed as measures of language proficiency and the heterogeneous nature of the examinee population to whom the tests are given. A methodology is proposed for handling the problem of equating nominally parallel language tests. In particular, two multiple-choice grammar tests -- one whose items are based on academic texts and cover a wide range of grammar topics, and the other whose non-academic vocabulary and contexts focus on a narrower range of grammar topics -- are examined for functional parallelism through a generalizability study. It was found that these surface differences did not cause any significant reordering of examinee ranks and that score equating could take place. Three commonly used score equating methods, namely those based on linear, equipercentile, and latent trait models, were then applied and evaluated for appropriateness to this particular equating situation. The equipercentile method was found to provide the most accurate and stable results. (Author/PN)