論文

出雲仁多方言における母音をめぐる音変化

島根県出雲地域で話される出雲仁多方言における母音をめぐる音変化を，古代語との比較にもとづいて，相対年代とともに推定した。仁多方言では，中舌母音化*u, *i > ɨ と母音の低下*u > o, *i > e に加え，r の隠在化と呼ばれる変化が起こった。現代仁多方言の形式のほとんどが，古代語の形式を祖形とし，その祖形が上述の変化を経て成立したものと考えられる。一方，古代語との音対応からは例外的と思われるkusoo「薬」，sɨrosɨ「印」，soso「裾」という3 形式は，先行研究の成果に照らすと，祖語の*o を保持した形式である蓋然性が高いことが明らかになる。このことは，服部（1978–79 [2018]）が中央方言で起こったとした狭母音化*o > u という変化を，仁多方言が経験していないことを示唆する。他の本土諸方言においても，中央方言で狭母音化によって失われた祖語*o や*e が保持されている可能性があり，それらについて比較言語学的観点から再検討していくことがこの分野の今後の課題である^†。
キーワード：出雲仁多方言，狭母音化，母音の低下，比較言語学，日本祖語

L1 日本語の英語学習者によるtough 構文の産出におけるインプットと母語の影響

日英語のtough 構文は本質的にかなり異なる。本研究は，L1 日本語の学習者（中高生）による英語のtough 構文の産出，そしてL2 インプットとL1がそれに与える影響を調べた。JEFILL コーパスと英語の教科書を分析した結果，中高生の英語のtough 構文の産出は限定的であったが，使用頻度が高い述語が教科書分析の結果と合致したことから，インプットの質が影響を与えている可能性が示唆された。一方，インプットは誤文を説明できず，正文に関してもインプット量が十分ではなかった。また，産出されたtough 構文の特徴は正文・誤文共にL1 と合致したことから，L1 転移の可能性が示唆された。さらに，誤文で見つかった残留代名詞に関して追実験でその存在の検証を行った結果，学習者の中間言語がそれを許容していると考えられ，学習者の産出するtough構文も英語母語話者とは本質的に異っている可能性が示唆された*。
キーワード： L1日本語の英語学習者，tough構文，産出，母語の影響，インプット

『文海』の「偽平声」から見る西夏語音韻学の複層性
――西夏文字の字音推定の限界の所在について――

本論文は，西夏語の韻書『文海』の，第三字で調類を指定する特殊な反切について考察する。『文海』巻一（平声巻）掲載字のうち，反切に第三字「」“上”を伴うものは本来上声で読むはずの「偽平声」字の一部である。「偽平声」が生まれた理由は，異なる調類の間で韻目が合併したためと考えられる。換言すれば，「偽平声」字は上声の「欠番韻目」所属字であり，具体的にはR13，R32，R39，R41，R50，R83，R88，R102 の上声韻が「欠番韻目」と考えられる。但し，反切の繫聯状況と「反切下字が調類の指示機能を担う」という前提からは，数多くの字が「偽平声」である可能性が得られてしまう。これに対し本論文は，『文海』内部で反切の理論的基礎と分韻の学理との間に不一致がある可能性，ならびに，この不一致が西夏語音韻学自身の複層性に由来しており，『文海』内部に西夏語音韻学のより古い学理の分析結果が痕跡的に残存している可能性を指摘する*。
キーワード：西夏語，西夏語音韻学，漢語音韻学（中国音韻学），反切，韻書

契丹語の音調

本稿では，10–12 世紀に中国東北方で話された，モンゴル諸語と系統関係を有する契丹語の語音調がどのように実現したかを，当時当地で使用された遼代漢語との対音資料を用いて実証的に明らかにする。予備的作業として，先行研究の発見にかかる，契丹小字文献中の漢語語彙に見られる特殊表記を批判的に検討して定量的に再分析し，そこから遼代漢語の声調体系に関する重大な帰結が導けることを論じる。その帰結を前提として，遼代漢字文献中の契丹語音訳語彙に使用される音訳漢字を音写語内の位置別に定量的に分析し，位置による顕著な声調の選好が存在することを明らかにする。さらにこの選好を遼代漢語の時間的変異性に着目して現代モンゴル諸語の音調も参考に解釈することで，契丹語が語の始端境界と終端境界をL 音調とH 音調とでそれぞれ標示する言語であったことを明らかにする*。
キーワード：契丹小字，契丹語，遼代漢語，声調，境界音調

Articles

Historical Study on the Vowels in Izumo-Nita Japanese

Based on the comparison with Old Japanese (OJ), in Izumo-Nita Japanese, which is spoken around the southern part of the Izumo area in Shimane prefecture, three major sound changes are reconstructed: (1) the centralization of high vowels *u, *i > ɨ, (2) the lowering of high vowels *u > o, *i > e, and (3) r-deletion before high vowels. In most cases, forms that appear in OJ are reconstructed as proto forms, from which the forms of Nita could have been derived through the three changes above. From the viewpoint of the sound correspondences between Nita and OJ, however, three forms are considered exceptions: kusoo “medicine,” sɨrosɨ “mark,” soso “cuff.” Referring to previous studies, the mid-high vowels “o” in these forms are thought to be the remnants of *o in proto-Japanese, and they suggest that the mid-vowel raising *o > u, which Hattori (1978–79 [2018]) assumed occurred in OJ, would not have occurred in Nita. In other mainland Japanese dialects, there are what appear to be remnants of *o and *e from proto-Japanese that were lost in the central dialects. Re-examining them from a comparative linguistic perspective is one of the future tasks in this field.

The Influence of Input and L1 on Japanese Learners’ Acquisition of English Tough Constructions

Japanese tough constructions are of four types and quite different in nature from English ones although one of the types superficially corresponds to English ones. This study investigates what kind of tough sentences Japanese learners of English at a relatively early stage of acquisition produce and how L2 input and L1 influence the production. Data from the JEFILL corpus and input data from English textbooks at public junior high and high schools in Japan were analyzed. The results revealed that Japanese junior high and high school students produced a certain number of tough constructions using 11 predicates and with errors. The predicates that the learners frequently used in tough sentences were congruent with the input data. In this respect, the quality of the input seemingly influences Japanese learners’ acquisition of English tough constructions. However, the input frequency appeared to be insufficient for explaining the full range of learners’ data. The input cannot explain learners’ errors either. The various linguistic properties observed in learners’ data, including error sentences, are rather consistent with L1. Thus, it may be the case that, aside from input influence, there is both positive and negative transfer from L1. A follow-up experiment with an acceptability judgment task was also conducted to test whether Japanese learners wrongly allow a resumptive pronoun in tough constructions as the learners’ production data suggests. The results show that the Japanese learners (high-school students) allow it just as in their L1. These findings suggest that Japanese learners’ linguistic knowledge of English tough constructions may differ from that of English native speakers, at least at an early stage of acquisition.

Concerning the Chronological Plurality of Tangut Phonology in the Light of “Pseudo-level Tone” in Wenhai (the Sea of Characters):
The Limitation of Inferring the Sound of Tangut Characters

This paper discusses an untypical type of Fanqie in the Tangut rhyme book, “Wenhai,” in which the third character functions as a tonal specifier. We reveal the existence of “pseudolevel tone” characters. Amongst the Tangut characters in the first volume (level tone volume) of Wenhai, those with the tonal specifier, “rising,” are the de facto rising tone characters and should have been placed in the second volume (rising tone volume) whose theoretical inconsistency results from the inter-tonal mergers of the “rhyme groups.” These mergers have brought several “vacant rhyme groups” to the Tangut phonology of the Tangut Empire period, for example, the rising tone of R13, R32, R39, R41, R50, R83, R88, and R102. Tangut scholars have devoted their intelligence to designing an “ideal” system rather than reporting a strict and modern description. Intentionally and sensibly, they avoided the quantitative unbalance between a minimal pair of level and rising tone.
However, not all possible “pseudo-level tone” characters confess their real tonal category. The suspicion that a specific character belongs to the “pseudo-level” is not always selfevident. If we faithfully accept the axiom of traditional Sinitic phonology that the second character in Fanqie always represents the tone, a large number of possible “pseudo-level tone” characters would become subjects of consideration.
This paper attributes this problem to the clash of earlier and later methodologies. It is probable that Wenhai’s Fanqie contains vestiges of the early theory on the phonological frame. We can derive a new hypothesis from the proposition that the theory of Tangut phonology has a chronological plurality. Specifically, the arrangement of Fanqie characters is probably earlier than the establishment and adoption of the Tangut rhymes’ classification system.

Pitch Realization in Khitan

The aim of this article is to reconstruct the word/phrase pitch pattern in Khitan, an extinct language related to the Mongolic languages that was spoken in Manchuria and northern China during the Liao dynasty (907–1125). We do this by using transcriptional materials of Khitan and Liao Chinese, a local variety of Chinese spoken in that region at that time. First, as preliminary work, we critically review the special notation for tones in Khitan Small Script transcriptions of Chinese words, and reanalyze it from a quantitative perspective, thereby revealing a significant aspect regarding the tonal system of Liao Chinese. Building on this finding, we quantitatively analyze the Chinese characters used for transcribing Khitan words according to their position within the word, showing a clear preference for specific tonal categories based on position. Furthermore, taking into consideration the temporal variability within Liao Chinese and the pitch patterns in modern Mongolic languages, we interpret the tonal preference in Chinese transcriptions of Khitan words to mean that Khitan was a phrase language with an L tone at the left edge and an H tone at the right edge of every word or phrase.