Several hypotheses are presented which might explain the lower performance of Black Americans on paper and pencil tests of verbal ability. These hypotheses include: 1) Black Americans are not test wise; 2) cultural differences affect verbal skills, i.e. reading ability, vocabulary, word usage, and sensitivity to verbal relationships; 3) cultural differences affect reasoning ability; and 4) Black Americans have a negative attitude toward answering SAT questions. A pilot study is described which uses a think-aloud procedure with antonym, analogy, and data sufficiency items. Subjects were nine inner-city teenagers. Some Ss were told the correct answers and/or the meanings of hard words in the text of the questions. The results of the pilot study and their theoretical analysis are discussed with implications for future research in the field. In general, giving definitions appeared to increase the number of correct (ranked 1) or near-correct (ranked 2) answers given, but giving answers and giving practice with test items seemed to have no effect on verbal reasoning. The use of word association techniques and new dependent measures both appear to help understanding of verbal reasoning on these types of items. The think-aloud procedure appears useful in generating hypotheses, but does not seem to be usable itself in controlled experiments. The usefulness of studies of this type is noted, both for insight into reasoning processes and for improving the validity of test items.