Abstract
Submorphemic units occupy a dubious position between the phoneme and the morpheme in traditional (structuralist-generative) analyses. The present study places submorphemic units within their proper historical context, and concludes that submorphemic units often come about by lexical association (a term adapted from the traditional term “secondary association”). Furthermore, lexical association sometimes causes submorphemic units to develop into morphemes. It is argued, however, that the function and diachronic development of submorphemic units can be but roughly accommodated within the framework of traditional categories. An approach is therefore proposed that is based on prototype theory, which provides a powerful theoretical underpinning for lexical association, and on a dynamic model of the lexicon that will accommodate varying degrees of morphological segmentation among morphologically complex words. It is demonstrated that there is no advantage in forcing a segmentation of all morphologically complex words into discrete morphemes, and it is thus concluded that it is equally unnecessary to require such segmentation of words that contain submorphemic units. The conclusions reached here lend support to the view that the transition from phonology to grammar is a gradient one.*