There is a lack of adequate measurement techniques for testing language proficiency. Researchers compose specific tests for a certain task, but these have only limited general applicability. Often multiple-choice, true-false or fill-in-the-blank tests are used, but these rely heavily on written language and are inadequate for those with poor written but good verbal skills. Such tests tend to compartmentalize learning into components and neglect the overall view, and they generally reflect academic language rather than current vernacular. Norm-referenced, criterion-referenced and non-formal tests involving listing of words following a language cue may be affected by the individual's attitude toward being tested. Indirect tests may solve some of these problems by examining language produced in a non-test situation. Factors studied are sentence length, structural complexity, lexical choice and type-token ration. Occurrences of linguistic mazes and culs-de-sac may also be observed. Second-stage indirect measures such as body language may also be useful. It is possible, however, that proficiency may be a mark of social rather than linguistic status. (CHK)