Structural Case in Finnish Paul Kiparsky Stanford University 1 Introduction 1.1 Morphological case and abstract case The fundamental fact that any theory of case must address is that morphological form and syntactic function do not stand in a one-to-one correspondence, yet are systematically related. 1 Theories of case differ in whether they define case categories at a single structural level of representation, or at two or more levels of representation. For theories of the first type, the mismatches raise a dilemma when morphological form and syntactic function diverge. Which one should the classification be based on? Generally, such single-level approaches determine the case inventory on the basis of morphology using paradigmatic contrast as the basic criterion, and propose rules or constraints that map the resulting cases to grammatical relations. Multi-level case theories deal with the mismatch between morphological form and syntactic function by distinguishing morphological case on the basis of form and abstract case on the basis of function. Approaches that distinguish between abstract case and morphological case in this way typically envisage an interface called “spellout” that determines the relationship between them. In practice, this outlook has served to legitimize a neglect of inflectional morphol- ogy. The neglect is understandable, for syntacticians’ interest in morphological case is naturally less as a system in its own right than as a diagnostic for ab- stract case and grammatical relations. But it is not entirely benign: compare the abundance of explicit proposals about how abstract cases are assigned with the minimal attention paid to how they are morphologically realized. The position taken here is that the theory of case belongs both to mor- phology and to syntax. It must draw on morphological principles to explain the characteristic organization of case paradigms, such as the patterns of syncretism 1 Warm thanks Arto Anttila for always being ready to share his judgments of the Finnish data, to Kristi´an ´ Arnason for suggesting the possibility of the Dep constraints discussed in section 3, and to Ida Toivonen, Diane Nelson, and Satu Manninen for comments on a draft. The responsibility for errors of fact or interpretation is mine. 1