On the Status of Stems in Morphological Theory

1 Introduction To a first approximation, languages show two different types of morphological alternation, with 'alternation' construed in a broad sense. One type is concatenative, involving what looks like the affixation of one piece to another; e.g. kick/kick-ed. Another type of alternation involves morphophonological alternations of the type seen in sing/sang. One of the primary tasks of morphological theory is to provide an analysis of such alternations, and to situate them with respect to other parts of the grammar, especially syntax and phonology. Here we examine alternations that have been used in arguments that grammar must contain the 'stem' as a privileged object. We argue that that the move to stems is both unmotivated and problematic, points that we illustrate in a number of case studies, including a discussion of the verbal morphology of Classical Latin. (forthcoming)) advances a piece-based view of word formation, in which the syntax/morphology interface is as transparent as possible. Distributed Morphology posits that there are two types of primitive elements in the grammar that serve as the terminals of the syntactic derivation, and, accordingly, as the primitives of word formation. These two types of terminals correspond to the standard distinction between functional and lexical categories (for more details on the view adopted here see Embick and Halle (forthcoming)): * Aspects of our analysis of the Latin conjugation were presented at Going Romance 2003, and we would like to thank both the organizers of the conference and the conference participants. For comments on a draft version of the material presented here we are indebted to