Semantic properties of the so-called past tense morpheme in Late Late Middle Japanese

要旨

This article, which provides the very first formal semantic analysis of Late Late Middle Japanese (LLMJ) data, discusses adjectival (non-preterit) interpretations of the so-called past tense morpheme -ta in LLMJ and contemporary New Japanese (cNJ), accounting for their similarities and differences. We explain the historical semantic change of -ta and its predecessors as a gradual change from a resultative to a perfect and then to a preterit, in compliance with Maslov’s (in: Nedjalkov (ed.) Typology of resultative constructions, 1988) hypothesis. Adjectival -ta is restricted to prenominal modifier cases in cNJ, but is found in both prenominal and matrix clause cases in LLMJ. Assuming that a matrix clause must be tensed, we argue that -ta is used as an aspect morpheme -tastate, and that a verbal complex V-tastate could in principle be suffixed by a tense morpheme. This analysis is needed for LLMJ and constitutes a major departure from the previous works dealing with adjectival -ta in cNJ such as Kusumoto (in: Cuervo et al. (eds.) Formal approaches to Japanese Linguistics 3, 2001) and Ogihara (Linguist Philos 27: 557–608, 2004). According to our account, V-tastate-tapast is illicit in both LLMJ and cNJ. The form V-tastate-∅present is licit in LLMJ, whereas cNJ only accepts the tenseless form V-tastate, which reflects the gradual marginalization of this construction over the course of Japanese language history. This accounts for the distributional constraint on adjectival -ta in cNJ. Regarding semantic interpretation, V-tastate asserts in both LLMJ and cNJ that a target state (Parsons, Events in the semantics of English: A study in subatomic semantics, 1990) of V-ing holds now without entailing that V actually occurred in the past. This semantic proposal is couched in an intensional analysis of -tastate.