Paper short abstract:

In this paper, the development of Proto-Japonic *r in Sakishima will be discussed. Two symmetrical rules will be posited: flap deletion rule, whereby *r got deleted if followed by *i or *u, and flap assimilation rule, whereby *r preceded by *i or *u assimilated with an adjacent consonant.

Paper long abstract:

The aim of this paper is to discuss the development of Proto-Japonic (PJ) *r in Sakishima languages by focusing on the following rules: the flap deletion rule (FDR) and the flap assimilation rule (FAR).

FDR takes place in settings corresponding to Japanese (JP) __ri and, although with less evidence and apparently more complicated conditioning, Japanese __ru. It has been tentatively postulated to reflect PJ *r + close vowel sequences. *r has been found to be deleted in such settings in Miyakoan (MI) and Yonaguni/Dunan (DN): Japanese (JP) hidari = MI psdaz 'left', DN ndai; Old JP agari = MI agaz, DN aŋai 'east'. Since the deletion or lenition of *r before /i/ is also evidenced in Shuri (fidʑai 'left') and in Kagoshima (hidaji), it would appear like an innovation that took place already at a Proto-Kyushu-Ryukyuan stage. The evidence from most Yaeyama regiolects, however, which usually retain *r in this setting, argues to the contrary: Ishigaki (IS) pɨdarɨ, Kuro (KU) pizarɨ, Hateruma (HT) pinari 'left'. The retention of *r in Yaeyama therefore implies that its deletion or lenition elsewhere was an independent innovation.

FAR involves settings corresponding to Japanese Cir__ and Cur__, which makes them symmetrical with the FDR settings. With FAR, the close vowel immediately preceding *r gets deleted, and *r appears to be assimilated with an adjacent consonant. Variants of this rule can be observed in a number of environments in all examined Sakishima ethnolects: JP shirami = MI ssam, IS ssaŋ, DN Ɂtsaːŋ 'louse'; JP abura = MI avva, IS aba, DN anda 'grease, fat'; JP kura- = MI ffa-, IS ffa-, DN dwa- 'dark', while it is not consistently evidenced in North Ryukyuan, and not at all in Kagoshima. Therefore, it is believed that the activation of FAR in some shape can be posited for the Proto-Sakishima stage, although its specific conditioning and development may differ in particular daughter languages.

Both *r-related rules are expected to deepen one's understanding of the history of southern Japonic regiolects. Furthermore, such deletion or assimilation-prone characteristics of Japonic /r/ could be examined in different settings and for different language groups.