6 THE LANGUAGE-SPECIFICITY OF CONCEPTUAL STRUCTURE Path, Fictive Motion, and Time Relations Jürgen Bohnemeyer This chapter addresses the encoding of spatial semantics at Conceptual Structure (CS) in the framework proposed by Jackendoff (1983, 1987, 1996, 2002). The central question con- cerns the aspects of the representation of space at CS that are universal and therefore presum- ably innate. Jackendoff envisions CS as a language-inde- pendent faculty of cognition that generates non- iconic conceptual representations of an algebraic internal structure (a recursive predicate-argu- ment calculus that is syntactically different from both language and predicate logic). Reasoning and any transfer of information between different peripheral systems is divided between CS and another module of higher cog- nition, Spatial Structure (SpS). 1 SpS encodes geometric properties in an “image-schematic” fashion. SpS representations are primarily the product of high-end visual processing, but receive input in other modalities as well, and are themselves a-modal. Jackendoff assumes that language primarily interfaces with CS. Linguistic meaning is a mapping between the syntactic and phonological representations of utterances and some corresponding CS repre- sentations. Lexical meaning components that involve shape, “manner of motion” (Talmy, 2000b), and certain other spatial properties are fully interpreted at SpS (perhaps via some sort of placeholders at CS); but all aspects of syntactic structure map exclusively into CS. The exact division of labor between CS and SpS remains very much an open question within this framework. My concern here is specifically with the representation of Motion events in language and cognition. Jackendoff (1983, 1990) has advanced a number of arguments to the effect that CS encodes notions of Translational Motion (T-Motion) and Path, based on English data. I argue in the following on the basis of evidence from Yucatec Maya that these arguments do not apply universally, and that Yucatec Motion event descriptions do not involve a semantics based on T-Motion and Path (henceforth, a “Path semantics”), but merely a State-Change semantics. In the account proposed here, cogni- tive representations of Motion are comparable between English and Yucatec at the level of SpS, but not at CS. T-Motion involves a homomorphic mappi- ng from the time course of the Motion event into the Path traversed (e.g., Krifka, 1998; Zwarts, 2005), as depicted schematically in Figure 6.1. T-Motion must be encoded on some level of cognition––but to what extent is it encoded in language? It has often been assumed that linguistically, Motion is repre- sented as a special case of State-Change–– Change of Location (e. g., Miller & Johnson- Laird, 1976; Dowty, 1979). 2 Location-Change 111 9780195311129_0111-0137_Malt_MALT_Ch06 20/11/2009 21:22 Page:111 OUP s UNCORRECTED PROOF