論文

現代九州諸方言における旧上二段動詞の「下二段化」は九州・琉球祖語仮説を支持するか？

現代九州諸方言には，旧上二段動詞の未然・連用形末が母音e を取り旧下二段動詞と統合する，いわゆる「下二段化」が観察される。九州・琉球祖語仮説によるとこの特質は，九州諸方言と琉球諸語がともに経験した音変化の結果であり，この音変化の共有によって九州諸方言と琉球諸語からなる単系統群が定義されるという。しかしながら「下二段化」は音変化ではなく類推変化の可能性が残されている。本稿は，九州諸方言の系統的位置の観点から現代九州諸方言の旧上二段動詞を分析することによって，九州・琉球祖語仮説の妥当性を検討した。その結果，この仮説を支持する証拠が宮崎県中部の方言に認められることを明らかにした。さらに，その他の現代九州諸方言も九州・琉球祖語の子孫とみなしうることを論じた*。
キーワード：上二段活用，九州方言，琉球諸語，日琉祖語，類推

{Non-DP, TP}構造と動詞句内主語における適切なラベル付けと Agreement の適用
――精緻化したPOPの枠組みを用いた場所句倒置の再考――

本稿は，特異な文法構造と情報構造を持つ場所句倒置を遡及的に考察しながら，Chomsky（2013, 2015a）の提唱するPOP の枠組みを精緻化する。動詞句内主語がT とどのように一致するのか，{PP, TP} がどのような一致を行うのか，の2 点に的を絞り，i）Chomsky（2000,2001）の枠組み同様，長距離での一致が当該の枠組みでも維持される，ii）{PP, TP} では，話題素性の継承と一致の結果，ラベル〈Top, Top〉が得られる，と提案する。加えて，場所句が話題，文末主語が焦点となる特異な情報構造に，統語派生がどの程度，影響を与えるかという，統語論と情報構造の相互関係に関しても議論を行う。{PP, TP} での話題の一致及びラベル付けが，統語派生を収束させ，場所句の解釈にも影響を与えるという意味で，その効果が統語論と情報構造の両面で反映されると論じる。

ソテツの葉で具現化するツツバ語の数の概念と Overcounting への改新

ヴァヌアツ共和国の少数言語ツツバ語において，10 台の数はオセアニア祖語を反映した加算法のUndercounting を用いるのに対し，21 以上の数は Overcounting を用いる。Overcounting は700 年代のオルホンの突厥碑文や，マヤ諸語，ゲルマン諸語にも見られる。この一見，難解とも思える数え方は世界でも稀少であり，ツツバ語も含めその生成はこれまで不明であった。
　本論文はツツバ社会において21 以上の数が伝統的な儀礼の場面で用いられることに着眼する。ツツバ社会は豚を財とする。儀礼では財である豚肉の分配・共食が不可欠で，参列者数の把握が必須であった。そのため数という抽象的概念がソテツの葉に具現化され，結果，このソテツの葉を用いた数え上げの手法が Overcounting への改新をもたらしたと結論づける。
　さらに文字の無いツツバ社会では，文字の代わりにソテツの葉が人数・個数を保存する記録簿としての役割を果たしてきたことを発表する*。
キーワード： ヴァヌアツ，オセアニア祖語，数詞，Overcounting，ソテツの葉を用いた数の記録

クプサビニィ語の名詞のディフォルトの長形と不定の短形――共時的用法と通時的発展――

クプサビニィ語（南ナイル，ウガンダ）のほとんどの名詞は，単数形と複数形のそれぞれに，接尾辞の有無により，長形と短形の2 形式がある。本研究は，名詞の長形と短形が談話でどのように使われるかをDryer（2014）の「指示階層」を使って分析する。長形は定の領域に使われるだけでなく，ディフォルトとして，不定の領域の多くにおいても短形よりも頻繁に用いられ，短形を使わなければならない状況は指示が起こり得ないような不定の場合に限られる。長形の意味は，かつて定性であったのが，名詞句の指示対象（のタイプ）について聞き手と知識を共有していると話者が持っている想定に基づくものへと一般化され，使用の文脈が広がったという仮説を立て，この変化について意味と語用の点から説明を試みる。長形と短形の選択のその他の要因（語彙，構文，文脈，話者の世代等）も記述する。また，一般に言語に見られる傾向に反し，より複雑な長形が短形よりも頻繁に起こることも指摘する。

ケチュア語アヤクーチョ方言の示差的目的語標示と情報構造

本稿では，アヤクーチョ方言の-sqa または-na によって作られる体言化従属節の中で起きる示差的目的語標示（differential object marking: DOM）を記述・分析する。アヤクーチョ方言では，体言化従属節内の目的語の標示パターンに-ta と-ø の2 通りが見られる。本稿では，聞き取り調査の結果に基づき，-ta による標示が，対比的焦点および意外性という情報構造上の概念に動機づけられると主張する。このようなアヤクーチョ方言のDOM は，DOM の類型論および情報構造の理論において次の3 点を示唆する。① DOM は多くの言語で報告されてきた有生性，定性・特定性，主題性以外に，対比的焦点や意外性にも動機づけられることがある。②従来情報構造の標示が見られないとされてきた従属節内でも，対比的焦点や意外性といった情報構造上の概念が標示され得る。③情報構造における対比性の有無は，明確に言語形式に反映され得る*。
キーワード：ケチュア語，示差的目的語標示，対比的焦点，意外性，従属節

話題化と左方転位に関する実験研究

本稿は，オンラインの容認性判断課題を通じて，補文内の話題化と左方転位の分布がなぜ制限されているのかを明らかにする。一部の補文内で話題化が容認されないことについて，現在二つの相反する分析が提案されている。干渉分析では，節の左端への演算子の移動を話題化要素が相対的最小性により妨げると分析している（Haegeman 2007, 2010a, 2010b, 2012）。一方，切り詰め分析は，Hooper and Thompson（1973）のC 類とD 類の述語が選択する補文においてTopic Phrase が投射されないことが，話題化と左方転位が容認されない原因であると分析している（Miyagawa 2017）。切り詰め分析の根拠は，移動を伴わないとされる左方転位が節の左端への演算子の移動と干渉しないと考えられることにある。しかし，容認性判断課題の結果，話題化と左方転位の容認度が補文を選択する述語のタイプによって異なることが分かり，話題化と左方転位が生じる投射について再考する必要が生じた。本稿では，干渉と切り詰めの両方が節の左端で作用しているが，どちらによって話題化と左方転位が阻止されるかは，補文を選択する述語のタイプによって異なると論じる。

Articles

Does “shimo-nidanka” in Upper Bigrade Verbs in Modern Kyushu Dialects Support the Proto-Kyushu-Ryukyuan Hypothesis?

In modern Kyushu dialects, so-called “shimo-nidanka” (a change to lower bigrade conjugation) is observed, in which the final vowel of irrealis/conjunctive (mizen/ren’yō) stems of former upper bigrade (kami-nidan) verbs takes /e/ and accordingly merge with former lower bigrade (shimo-nidan) verbs. According to the proto-Kyushu-Ryukyuan hypothesis, this trait is the result of sound changes experienced by both Kyushu dialects and Ryukyuan, and these shared sound changes define the monophyletic group consisting of the aforementioned. However, there is still a possibility that “shimo-nidanka” is a result of analogical changes rather than sound changes. This study examines the validity of the proto-Kyushu-Ryukyuan hypothesis by analyzing the former upper bigrade verbs in modern Kyushu dialects from the perspective of their phylogenetic position. There is evidence to support this hypothesis in the dialects of central Miyazaki Prefecture. This study also argues that other modern Kyushu dialects can also be seen as descendants of the proto-Kyushu-Ryukyuan.

How to Apply Successful Labeling and Agreement to {Non-DP, TP} and the VP-Internal Subject:
Revisiting Locative Inversion in Terms of the Refined POP Framework

This paper aims to elaborate on the POP (Problems of Projection) framework advocated by Chomsky (2013, 2015a) by conducting a reminiscent consideration of Locative Inversion, a construction with the idiosyncratic grammatical array and information structure. Traditionally, the construction has been given a syntactic derivation in which the locative PP undergoes movement to SPEC-T, while the subject DP remains within the VP-domain. This derivation poses two crucial questions for the POP framework: i) how the VP-internal subject undergoes φ-agreement with T and ii) what kind of agreement takes place in {PP, TP}. By refining the POP framework, this paper proposes that i) the POP framework maintains long-distance agreement such as in the Agree framework of Chomsky (2000, 2001) and ii) {PP, TP} receives a label of <Top, Top> as the result of topic-feature agreement via topic-feature inheritance from C by T. In addition, this paper attempts to clarify the interaction between syntax and information structure that is observable in Locative Inversion. Specifically, it demonstrates to what extent the syntactic derivation affects the unambiguous information structure in which the locative PP counts as a topic, while the postverbal subject is presented with focus. This paper argues that this peculiar information structure is associated with the syntactic derivation where {PP, TP} undergoes topic-feature agreement and obtains a label of <Top, Top>. The discussion in this paper suggests that the topic agreement/labeling is reflected in two ways: it helps the syntactic derivation converge and it guarantees that the locative PP functions as the topic of the sentence.*
Key words: Locative Inversion, Problems of Projection (POP), Labeling, Agreement, VP-internal subject

Number Concepts Represented by Cycad Leaves and the Innovation of Overcounting in the Tutuba Language

In Tutuba, a minority language of the Republic of Vanuatu, the numbers 11–19 use the undercounting additive method reflecting the Proto-Oceanic language, whereas numbers 21 and higher use overcounting. Overcounting can also be found in Orkhon Turkic inscriptions of the 700s and Mayan and Germanic languages. This way of counting, which at first can appear confusing, is rare worldwide, and its creation in conjunction with the Tutuba language, was unknown until now.
　　This paper focuses on how the Tutuba society used numbers 21 and higher in the context of traditional rituals. In the Tutuba society, pigs were highly valued as an indicator of wealth. At rituals, distribution and communal meals of pork were essential, and having a full accounting of the number of attendees was thus crucial. Accordingly, this paper concludes that at the Tutuba rituals, the abstract concept of numbers was embodied by the cycad leaf, thus, the method of enumerating by using cycad leaves brought about innovation of overcounting.
　　Furthermore, this paper reveals that the Tutuba society, which did not have letters, used cycad leaves to replace letters and play the role of record books that maintained quantities and attendee numbers.

Long Default and Short Indefinite Noun Forms in Kupsapiiny:
Synchronic Usage and Diachronic Development

In Kupsapiiny, a Southern Nilotic language of Uganda, nearly all common nouns in the singular or the plural have two forms depending on the presence vs. absence of a suffix. This study examines how long and short common noun forms (with and without the suffix, respectively) occur in discourse by means of Dryer’s (2014) Reference Hierarchy. It shows that short forms are only infrequently used for indefinites, with their mandatory use restricted to two types of indefinites, semantically nonspecific indefinites involving negation and true-predicate nominals, whereas long forms serve as frequently used, default forms, which occur not only in definite situations, but also – in fact, more commonly than short forms – in most indefinite situations. The present study hypothesizes that long forms have semantically generalized from forms marking definiteness, i.e., the speaker’s assumption about the identifiability or accessibility of the referent to the hearer (Givón 1995, 2001), to those indicating the speaker’s assumption about the accessibility of the referent (type) to the hearer (or sometimes the speaker themself) with the result that their context extended from definite situations to include most indefinite situations, as reported about the diachronic development of definite articles in other languages (e.g., Greenberg 1978). It speculates that the context extension for long forms was brought about by associations between referents in discourse based on the Gricean maxim of relation as well as subjectification. Other types of factors than the speaker’s assumption about the referent (type) accessibility and indefiniteness in the choice between long and short forms (lexical, constructional, contextual, and generational differences as well as special uses of short forms for definites) are also described. This study also points out that the Kupsapiiny noun system as a whole contradicts the grammatical form–frequency correspondence hypothesis (Haspelmath 2021).*
Key words: (in)definiteness, Dryer’s Reference Hierarchy, form–frequency noncorrespondence, grammaticalization, Kupsapiiny (Kalenjin, Southern Nilotic, Uganda)

Differential Object Marking and Information Structure in Ayacucho Quechua

This paper offers a description and analysis of differential object marking (DOM) observed in -sqa or -na subordinate clauses in Ayacucho Quechua. This language allows for two patterns of object marking, accusative -ta and -ø, only within nominalized subordinate clauses. It is argued that this DOM is triggered by two information structural properties: contrastive focus and unexpectedness. The accusative -ta marking is preferred to -ø marking when an object NP is in contrastive focus and/or is unexpected from common grounds. Such DOM in Ayacucho Quechua has three implications for the typology of DOM and information structure. First, DOM can be triggered by contrastive focus or unexpectedness, although these two factors have not been widely reported to be relevant to DOM. Second, it is possible to recognize information structure–based grammatical markings in subordinate clauses, which have been said to be “pragmatically flat” (Bybee 2002). Last, contrastiveness as an information structural property can be formally marked.

Topicalization and Left Dislocation: An Experimental Study

This study aims to elucidate the restrained distributions of embedded topicalization and left dislocation (LD) through an online acceptability judgement task. Currently, there are two opposing approaches regarding their confined distributions. The intervention approach by Haegeman (2007, 2009, 2010a, 2010b, 2012) argues that topicalization gives rise to the intervention effect with operator movement to the left periphery. The truncation approach by Miyagawa (2017), on the other hand, notes that the truncation of a Topic Phrase excludes both topicalization and LD in the complements of Class C and D predicates of Hooper and Thompson (1973). The rationale for the latter approach is that LD, which does not involve movement, would not intervene with operator movement. The results of the acceptability judgement task reveal hitherto unnoticed differences of acceptability among various complements, and we are led to reconsider the exact syntactic projections hosting topicalization and LD. It is argued that both intervention and truncation are operative at the periphery, but their effect on embedded topicalization and LD varies depending on the matrix predicate class.*
Key words: topicalization, left dislocation, cartography, online experiment