| This study explores clause-linking devices in Japanese, and especially conditional constructions, from several intersecting functional perspectives: theoretical, contrastive, and pedagogical. It covers, in case-study format, a variety of formally distinct bi-clausal constructions, exploring their syntax, semantics, and pragmatics from the general viewpoint of Fillmore and Kay's Construction Grammar. The primary focus is on the TO, TEMO, and English EVEN IF constructions, each as a constructional family; TARA, (R)EBA, NARA, and IF are also treated. Each construction is highly polyfunctional, comprising a richly articulated polysemy structure which can be illuminated by Lakoff's notion of radial categories. These polysemy structures, each capturing a whole family of subconstructions, show considerable functional overlap--not only within Japanese, but also when comparing a given Japanese construction to its approximate English functional analogue (which may differ formally in major ways). Only by careful attention to these differing polysemy structures (encompassing both central cases and their extensions) can the subtle semantic differences among the constructions be adequately understood; no single list of necessary and sufficient conditions, or single abstract constructional meaning, can do justice to the phenomena. The dominant themes throughout are: (i) constructional idiomaticity and construction-specific semantics and pragmatics, (ii) the polyfunctional and polysemous nature of clause-linking constructions, and (iii) crosslinguistic divergences. These three factors intertwine to produce the daunting complexity characteristic of clause-linking mechanisms.;The contrastive aspect of this study is of fundamental importance. The linguistic difference between the polysemy structure of a Japanese clause-linking construction and that of its English analogue masks a deeper implicit cognitive difference. Concepts and associated semantic categories that come extremely naturally to speakers of a given language may not be automatically shared by speakers of other languages; they may indeed be cognitively grounded and natural in native speakers' minds, yet far from universal; they belong to speakers' "subjective orientation to the world of human experience", and not to "neutral coding systems of an objective reality" (Slobin 1991: 23). The languages also show structural mismatches in preferred event representation patterns. Only with great difficulty can speakers be made aware of these "invisible" etic differences. Yet precisely such an awareness of differences in "thinking for speaking" (Slobin) is crucial for both effective language teaching and insightful theoretical analysis. |
