To account for vowel alternations in forms such as divine-divinity, Chomsky and Halle propose the Vowel Shift Rule and other rules. This study experimentally assesses the psychological validity and generality of these rules by testing the productivity of vowel alternation. Subjects were required, in a meaningful sentence context, to produce a novel derived form by selecting one of two suffixes and affixing it to a base word (for example, maze + ic/ity, concrete + ify/ic). Items were presented aurally and, in some conditions, orthographically as well. Results were consistent: 90 percent of all responses showed no vowel change. Less than four percent exhibited the Chomsky and Halle predicted vowel changes. The validity of the Vowel Shift Rule is, therefore, highly dubious. (Author/RB)