Dephrasing in Kobayashi Japanese: Is it a reality? Yosuke Igarashi Japan Society for the Promotion of Science, National Institute for Japanese Language yosuke.igarashi@kokken.go.jp 1. Introduction In majority of Japanese dialects including Tokyo and Osaka, pitch difference is used to distinguish one word from another. These dialects have pitch specification in the lexicon (lexical tone, henceforth), and are generally classified into the ‘accented’ (yuakusento) dialects. At the same time, not a few of the dialects lack lexical tones. These include for example, Sendai, Utsunomiya, Fukui, and Kumamoto. They are usually called ‘accentless’ (muakusento) dialects [17, 19, 20]. It is an accepted wisdom that the Morokata dialects, including Miyakonojo, Nichinan and Kobayashi (spoken in Miyazaki prefecture) are typologically distinguished from the accentless dialects, though they do not have lexical tones. The most widespread distinction between the accentless dialects and the Morokata might be a feature of ‘pitch pattern fixation’. In the former the pitch pattern of bunsetsu, or prosodic word (ω) presents variability, whereas in the latter it is always fixed. Specifically, a ω in Morokata always shows a pitch rise on the final syllable, exhibiting a ‘high-tailed’ (odaka) pattern. Thus, Morokata has been typologically classified as the ‘one-pattern’ (ikkei) dialects [2]. In other words, this dialectal group can be considered to have a fixed accent system. The present study reports the results of an experiment which investigates prosodic characteristics in Kobayashi Japanese spoken by the younger generation. The first goal is to examine whether the mapping between prosody and syntax (or pragmatics), which most of Japanese dialects are reported to possess, can be observed in this fixed accent dialect. The second goal is to confirm whether Kobayashi can safely be classified as a fixed accent dialect. 2. Prosody of the fixed accent dialects 2.1 Relation between Prosody and syntax The following three syntactic (or pragmatic) items are reported to show a clear correlation with prosody in Japanese dialects. They will be exploited in the experiment to see if they are mapped onto prosody of a fixed accent dialect as well. (1) Branching structure In Tokyo (an accented dialect), a syntactic boundary with a right-branching structure brings about ‘metrical boost’, or pitch range expansion [10]. The sentence with a left-branching structure, on the other hand, does not cause pitch range reset but it is merged into a single prosodic unit. Henceforth, prosodic unit which serves as the domain of pitch range will be called intonational phrase (ip). This mapping has been reported for the accented dialects such as Tokyo [6] Osaka [5], Hirosaki [7], Goshogawara [3], as well as for accentless dialects such as Fukui [11], Kumamoto [12, 8]. The mapping between branching structure and ip in Osaka is in fact controversial. Sugito [18] argues against it, suggesting dialect-specific difference in prosody-syntax mapping. (2) Wh question In Tokyo, a wh word compresses the pitch range for the following words and thus all the words within the scope of the wh word constitute a single ip [11]. This effect is similar to that of focus discussed below. The mapping between wh question and ip is reported for the accentless dialects such as Fukui and Kumamoto as well [11, 12], though the merging of words into an ip is slightly different from Tokyo. Reportedly, prosodic phrasing due to a wh word accompanies deletion of lexical tones in Fukuoka (an accented dialect) [9]. (3) Focus Focus is one of the main factors that influence on intonation in many dialects. In Tokyo, the effects are similar to those of wh word: pitch range is reset at the beginning of the focused word and is compressed in the post-focal words [13]. That is, focus introduces an ip-boundary at the beginning of the focused word and the post-focal words are merged into the same ip. These phenomena are reported for the accented dialects such as Tokyo [6], Osaka [5], Kagoshima [7], Goshogawara [3], as well as for the accentless dialects such as Kumamoto [8]. Sato [16] found the similar effect of focus for Kobayashi (a fixed accent dialect), remarking that the final rise does not occur in the post-focal words. Since it is crucial for typological status of the dialect, her description will be discussed below in a little depth. 2.2 Absence of dephrasing in the fixed accent dialects The well-established view that Morokata has a fixed accent system is challenged by the fact these dialects can show what I refer to as ‘high pitch dislocation’, i.e. a phenomenon where the pitch peak is retracted from the ω-final syllable. Miyakonojo exhibits high pitch dislocation when the ω consists of an auxiliary verb zya [22]. Sato [16] reports for Kobayashi that, when a ω includes a sentence-final particle (shujoshi) such as o, the pitch peak is dislocated to the syllable preceding the particle: muzekatta-o ‘It was cute.’ is pronounced as muzekattáo, instead of muzekattaó (where acute accent indicate high pitch). While what triggers high pitch dislocation should be examined in depth, it is at least clear that even in the fixed accent dialects, pitch pattern is not fixed in a strict sense. The absence of a strictly constant pitch pattern in Morokata might make the difference between these dialects and the accentless dialects ambiguous. In fact, the distinction between the two systems has been found controversial by some researchers [14, 22]. Crucially, even in the accentless dialects pitch pattern is by no means random. As Maekawa [11, 12] demonstrated, accentless dialects such as Kumamoto and Fukui exhibit regularity in pitch patterns of some prosodic unit, albeit ‘degree of freedom’ seems large in the accentless dialects [12,