The experimental test measures the ability of a subject to reconstruct obscured words by means of sentence analysis, carried out not as a separate academic task, but in the normal procedure of understanding what the sentence says. It is a task built on the assumption that the ability to handle reduced redundancy is a valid measure of linguistic competence. The reduced redundancy in this type of test is a consequence not of deliberate deletions or masking by superimposed noise, but of the reductions, assimilations, and contractions that normally accompany sentence production by native speakers functioning in a relaxed, informal context. The test is easy to administer and to score and is highly objective. It clearly separates nonnative from native speakers and spreads the nonnatives over an unusually wide range of scores. Performance on the test appears to correlate well with other measures of competence. (Author)