With a few important exceptions, such as Berko (1958) and Derwing and Baker (1977), linguistic morphology has always been a descriptive science, content with investigating actually occurring words and extracting patterns from these. Consequently, it has suffered from a difficulty common to all descriptive endeavours, the fear of obscuring individual cases in abstracting to general principles. In turn, as a result of this deficiency, morphological theory has tended to expend its energy not on general patterns, but rather on single words: more thought has been devoted to the one English word sang than to most other topics. In an earlier paper, Aronoff and Schvaneveldt (1978) showed how an experimental technique borrowed from cognitive psychology, the Lexical Decision Task, could be used to investigate morphological patterns without focusing primary attention on single actually occurring words. The essence of the technique is to ask native speakers of a language (English in this case) whether certain stimuli are words of their language or not.