This paper discussed future directions of experimental linguistics. The following were discussed: (1) how experimental linguistics is positioned in relation to descriptive linguistics, when we take the tradition of experimental phonetics since Rousselot into consideration; (2) what "experiment" is in experimental linguistics; (3)a case study: an experiment whose results suggest that the semantic category plays a certain role in Japanese grammar; (4) the relationship between experiment and theory.

If we assume that a spoken language and its written counterpart are essentially distinct entities, it turns out that the twentieth century linguistics has studied mismatching aspects of them (sounds and written language) and has paid little attention to the remaining aspects (writing and spoken language). In order to solve this incompatible situation, we need to establish the science of writing (graphetics) and that of spoken language. Based on this spirit, this paper argues that the Japan Experimental Linguistics Society should play an active role in establishing experimental graphetics and refer to Ikeda (2008) as an instance of such endeavor.

The oldest accent patterns of the avian names "hototogisu" and "uguisu" are presumed to be [LLLHL] and [LLHL] respectively and are associated with the pitch patterns of their songs. This fact shows objective evidence that these words are the onomatopoeic expressions which imitate their songs. Another fact that the H of their accent patterns correlates with an acute vowel [i] and the L with grave vowels [u] or [o] shows collateral evidence for the onomatopoetic hypothesis.

According to the past literature and recent questionnaire research, native speakers until the late Heian period had an clearer awareness that "hototogisu" and "uguisu" originated from the avian songs than present-day native speakers. The decreased awareness of their etymology seems to reflect the dissimilarity between the pitch patterns of the original songs and the accent patterns that changed from [LLLHL] and [LLHL] to [HHHLL or LHHLL] and [LHLL or HLLL] respectively.

The purpose of this paper is to examine the involvement of the semantic category “animacy” in the grammaticality of Japanese. This study was designed to elicit ERP by using semantically incompatible phrases. Sixty adnominal constructions such as yotta ishi “drunken stone” were visually presented to twenty right-handed undergraduates. The results were as follows:

(1) When twelve of twenty subjects looked at the incompatible phrases which involved the semantic category [-Human], the peak latencies of negative waves appeared around 350 to 450msec. and were thus considered to be an N400 component.

(2) The incompatible phrases which involved the semantic category [-Human] were more numerous than the compatible phrases which involved the semantic category [+Human].

(3)The fact that the semantic category [Human] plays a certain role in Japanese grammar is proven in brain-scientific reality.

(4)The semantic category [Human] is not a simple bi-value category; [+Human] is a superior feature to [-Human] in terms of the semantic restriction.