The concept of ‘universal’ and the Case of Japanese

The dominating proposal in generative grammar is that fixed positions in a phrasal structure universally encode information relating to semantic combination: Merge is assumed to be the unique, universal means of combination. Apparent departures from this structural encoding, such as languages which code relations by marking arguments with case, or polysynthetic languages that mark predicates with special affixes, are assumed to be superficial: at some deeper level, all semantic combinations in all languages are encoded by fixed positions in a phrasal structure. This is based on the assumption that having a unique form of coding is maximally efficient. Structural properties derive from the temporal ordering of elements in the sensorimotor apparatus (Kayne, 1994). But this is just one of four modes of giving a form to a semantic combination in the sensorimotor apparatus. Therefore, even if we assume that one of these modes of coding is more basic, we must posit a type of general computational process which can access all modes of coding, in order to recode all of the ‘secondary’ modes into the basic mode. So there is no simplification nor increase in efficiency, in the end, since at some level of processing, all modes must be accessible in any event. In fact, a reductionist model may induce a less efficient processing, since it requires the additional recoding of all ‘secondary’ codings. The reductionist approach results in unexplanatory kinds of parameters in the account of language variation. On the other hand, the traditional view (Meillet, 1949, 1950; Keenan, 1978, among many others) that correlates free order of arguments and case marking can be explained on highly principled grounds: the parametric choice between rigid order and case marking follows from inherent properties of the human sensory and motor apparatus which are motivated on grounds independent from language, properties which are logically anterior to linguistic theory. I illustrate the differences between reductionist and nonreductionist approaches by comparing two reductionist analyses of several phenomena tied to Case in Japanese — Boskovic and Takahashi (1998) and Saito and Fukui (1998) — with a nonreductionist analysis based on fundamental aspects of the A–P system.