Elementary and junior high school students took part in a test involving the probability texture of the language to determine whether spelling ability is related to a student's understanding of possible letter combinations in English words. Two similar groups of 124 fifth grade students and 129 eighth grade students were ranked in spelling ability by performance on the California Achievement Test, with those scoring lower than the 50th percentile classified as poor spellers. A test instrument was constructed of 120 five-letter nonsense words of four degrees of approximation to English, ranging from combinations most remote to English (xkjzt) to those most similar to English (comer). Students were asked to select from the two words the one that looked most like English. The results of the study indicated (1) that good spellers performed better than poor spellers on the word-choice discrimination test except on problems of extreme difficulty, when performance of both dropped to chance levels, (2) that the relationship between spelling achievement and verbal intelligence was only moderate, and (3) that spelling achievement reflected, in part, increased knowledge of the probability texture of English. (MP)