Data recently collected from some Japanese dialects such as Ibuki Island in the Inland Sea (Setonaikai) and neighboring areas in Kagawa prefecture, or Tsubata and Shiramine in Ishikawa prefecture, reveal a specific tonal pattern which bears a mid-tone mora (or a sequence of mid-tone morae) after 2 initial high tone morae in a single tonal phrase, as shown by #V V V …. This paper presents a hypothesis that this particular tonal pattern is derived, diachronically, from the sequence #HLH …, by rightward spread of the first high tone, thereby causing the downstepping of the second high tone, yielding the pattern #HHM….
Based on the assumption that phrases in the above geographical regions that now display #HHM… used to be, at a certain stage of their historical development, the sequence #HLH…, this paper first suggests a proto-accentual system of the mainland Japanese dialects, and subsequently illustrates the way in which some of the present dialects have developed their tonal systems from the common source. Contrary to the general assumption that the proto-accentual system of mainland Japanese dialects is that of the late Heian period annotated in the Ruijumyogisho, this paper attempts to show that the natural phonetic account of the tonal patterns of these dialects requires a common stage prior to that described in the Ruijumyogisho.
The paper further reconstructs a proto-accentual system of Japanese, based on the data collected in the present Ryukyuan dialects as well as the proto-accentual system of mainland Japanese established in the previous section of this paper. It is then suggested that, at least at one point, proto-Japanese had a balanced tonal-melody system consisting of extensive use of combinations of High and Low tones.

抄録全体を表示

Within the tonal classes in ruibetsu-goi (which is referred to here as “Classified Vocabulary”) developed for the diachronic study of Japanese accent, classes 3, 4, and 5 for two-mora nouns exhibit distinctive patterns of split and merger into two distinct types in Ryukyuan. These two types are called “ita-group” and “iki-group” in this paper. The paper attempts to show that these two groups are distinguished by the vowel length of the first syllables in certain Ryukyuan dialects (e.g. Shuri in mainland Okinawa Island), while they are distinguished by different accentual patterns in some other Ryukyuan dialects (e.g. Tokuwase in Tokunoshima). This paper proposes that such distinction of accentual patterns developed as a result of the transference of the phonological distinctions in Ryukyuan dialects from the vowel length of the first syllable to the accent patterns.
This paper argues that the vowels in the first syllable in “iki-group” nouns were once long, but were subsequently shortened in various Ryukyuan dialects, while their accent patterns retained their former distinctions between "ita-group" and “iki-group”. Thus, being a newly developed accent class in Ryukyuan, formed as a result of the above mentioned process of transference of phonological distinctions, the “iki- group” is a marked category for the two-mora nouns of classes 3, 4, and 5. Thus, the unmarked accentual patterns for classes 4 and 5 are not those of the “iki-group” as most of the previous studies have proposed, but actually “ita-group” patterns. The collorary of this assertion is that, contrary to the traditional analyses that assert that the patterns of tonal mergers in proto-Ryukyuan of 2-mora nouns in the Classified Vocabulary is 12/3/4 5, it is actually 1 2 / 3 4 5/3 4 5, with the newly developed "iki-group" represented by 3 4 5.

抄録全体を表示

首里方言の「iːʧi息, uːʃi臼, wuːki桶」に代表されるように, 北琉球（奄美大島から沖縄本島まで）の各地には, 一部の2音節名詞の語頭音節の母音が長くなっている体系がある。この長音節の出現にアクセントが関与していることは, 服部（1932）によってはじめて指摘された。さらに服部（1979）は, その長音節が

（本稿の日琉祖語）の段階から存在していたことを論じた。本稿は, この2音節名詞の語頭に見られる長音節は, （日琉祖語ではなく）北琉球祖語の段階であらたに生じた, という仮説を提示する。本稿では, この長音節発生の原因は北琉球祖語のアクセント体系に求められるとし, これは（1）同じ体系内の単独形が似た他の型との区別のため, そして（2）体系内の同系列の3音節名詞と同じ型を内部に実現させるため, という2つの理由により生じた, という仮説を提示する。

抄録全体を表示

It is a well-known fact that the following“sound law”(rule of phonetic correspondences) is found between the Kyoto-Tokyo dialect of Japanese and the Shuri dialect, which was the standard language of the former Ryukyuan Kingdom:
Kyoto ki ke ka ko ku; gi ge ga go gu
Shuri t_??_i ki ka ku ku; d_??_i gi ga gu gu
On the other hand, several exceptions are found to this rule of correspondences, but all of them can be explained, so far as the author knows, in terms of comparative method principles.
In this article, some exceptional phonetic correspondences found in Sino-Japanese words and signemes (morphemes) are explained as due to borrowing from Japanese into Ryukyuan. For example,
Chinese characters and«meanings»Japanese old orthography and [the present pronunciation] Shuri forms Chinese pronunciation of the characters
假病«feigned illness»kebyau [kebjo_??_] t_??_ibjo_??_ (chia³-ping⁴)
系圖«pedigree»keidzu [ke_??_zu] t_??_i_??_dzi (his⁴-t_??_u²)
見物«sightseeing»kembutsu [ke_??_zu] t_??_imbutsi (chien⁴-wu⁴)
聖賢«sages»seiken [se_??_ken] _??_i_??_t_??_i_??_ sheng⁴-hsien²
藝能«public entertainments»geinou [ge_??_no_??_] dzi_??_nu_??_ (I⁴-nêng²)
玄関《vestibule》 genkwan [ge_??_ka_??_] d_??_i_??_kwa_??_ (hsüan²-kwan¹)
日限《time-limit》 nichigen [nit_??_ige_??_] nit_??_id_??_i_??_ (jih⁴-hsien⁴)
As contrastively shown in the above table, these Shuri words are evidently loanwords from Japanese, not from Chinese. However, if we adhere to the “sound law”mentioned above, the expected Shuri forms should be [kibjo:, ki:dzi, kimbutsi, _??_i:kiη, giηnuη, giηkwaη, nit_??_igiη]. That is to say that these Sino-Japanese words make exceptional correspondences, which are very difficult to explain.
As the result of his researches to linguistically compare Japanese and Ryukyuan dialects, the author had come to a hypothesis, to the effect that the Shuri dialect has passed at least the following three stages of development:
the A period *ki *ke *ka *ko *ku; *gi *ge *ga…
the B period *ki *kii *ka *ku *ku; *gi *gii *ga…
the C period t_??_i ki ka ku ku; d_??_i gi ga…
(including the present)
If the above Sino-Japanese words would have been borrowed in the period A or C from Japanese, the present forms of Shuri would have had [ki] and [gi] instead of [t_??_i] and [d_??_i]. Accordingly, the only possibility is that they were borrowed in the B period from the western Japanese dialects, where [k] in [ke] must have been palatalized to some extent, as it is so nowadays. The ears of the contemporary speakers of Ryukyuan must have been very keen to hear the feature of palatalization in [k], which discriminated [ki] from [kii], so that the Japanese [ke] sounded to the Ryukyuan ears as [ki] rather than [kii].
Studying the Yu-yin Fan-i 語音翻譯 (1501), the author has revealed that this document represents exactly the B stage of Ryukyuan, and conducting researches into various contemporary and later documents, he comes to the opinion that the B period probably lasted from around 1400 until the late 16th century. He supposes that a number of Japanese high priests who came to stay in Shuri around 1500 and taught the Ryukyuans in Buddhism, Japanese and Chinese classics apparently caused the borrowing of those Sino-Japanese words and signemes (morphemes), which are fairly numerous and firmly established in the present Shuri dialect.

抄録全体を表示

The author introduces Morris Swadesh's method of lexical statistics or glotnchronology to the linguistic circle of Japan, applying it to the comparative study of Old Japanese and modern dialects.
He discusses its significance as a comparative method of related languages, and points out the following two major advancements. While the comparative method so far concerns the phonology and grammar rather than the vocabulary, Swadesh deals with the latter. Swadesh's method is statistical and accordingly more objective than the usual comparative method.
Although it is possible to comment critically on the method at several points, its effectiveness in general should be acknowledged. For instance, there are no“concepts and experiences common to all human groups”(Salish Internal Relationships, 157), because the dispositions of the speakers of different languages are often different even to “the same things”, that is, the “sememes”(as the author calls them) of synonymous words of any two languages are often not identical. Thus, Japanese /me/«eye» denotes the eye which opens and closes rather than the eye-ball, while (Chakhar) Mongol/nüdä/ which is the only word for “eye” denotes the latter rather than the former. Nevertheless we can find words which refer to things and events common to all human groups, and thus we can generally fill Swadesh's questionnaire.

抄録全体を表示