ダパ語（中国四川省，チベット＝ビルマ語派）には多様な存在文があり，さまざまな意味の違いに応じて使い分けられる。存在文の意味を決定する要素としては，存在動詞語幹の選択，接辞や助動詞の付加，構成素順，項の有生性がある。本論文では，特に，項の有生性がどのように存在文の意味に影響を及ぼすかに着目し，ダパ語の存在文に関する記述的研究をおこなった。
6つあるダパ語の存在動詞語幹のうち，ˉnʌ は，主語もしくは位格NPに有生物を要求し，主語が有生物の場合は存在文に，位格NPが有生物の場合は分配・獲得を表す特異な存在文になる。その他の存在動詞語幹については，一般に，位格NPが有生物の場合に所有文を形成する。ただし，存在動詞語幹 ˋɕɨ については，位格NPが有生で主語が分離可能である場合に，位格NPが主語を身に付けた状態を表すという現象が見られた。
北米北西海岸地域で話されるハイダ語は，自動詞節の主語として現われる人称代名詞が他動詞節の主語と同じ格の場合と目的語と同じ格の場合の二通りがある（但し，1人称単数と複数，2人称単数に限られる）。従って，ハイダ語は，「分裂自動詞性」（Merlan 1985）を有するといえ，活格型言語の典型的な例の1つと見做され得る。分裂自動詞性を有する様々な言語において，それを決定付けるのは動詞の意味特徴であるが，どのような意味特徴が関与するかは言語によって異なる。
本稿では，「動作性」と「制御性」がハイダ語における分裂自動性に関与すると捉える。これらの意味特徴と更に人称代名詞の人称や格によって，ハイダ語の自動詞は4つに分類することができるが，意味特徴が主たる分類基準であるために，自動詞の分類は，厳密になされるものではない。更に，このことは，活格型言語を一般的に特徴付けることの難しさを示すものと考えられる。
インドネシア国においては国家語であるインドネシア語の勢力が強く，各地の民族語は危機に瀕している。本論文ではインドネシア国スラウェシ島のスラウェシ北部州で話されている少数民族の言語であり，絶滅に瀕しているBantik語を例にとり，消滅に瀕した言語の姿を社会言語学的な観点と言語変化の観点から考察する。Bantik語，インドネシア語マナド方言，インドネシア語標準変種の三つの言語変種の関わり合いをまず述べ，次に高齢層と若年層のBantik語の実際の運用について述べる。伝統的なBantik語の形をかなり保っている高齢層のnarrative speechにも多くのマナド方言が含まれる。Bantik語を日常的に使用する高齢層と異なり，若年層は不完全な形でしかBantik語を使用しない。また音韻的な規則やmorpho-syntacticな規則が，高齢層のそれらとは異なってきており，言語磨耗と考えられる現象が観察される。
水海道方言には，単独用法において摩擦音で現れる語頭の子音が，複合語の第二要素の先頭で閉鎖音（または破擦音）で現れる現象がある。この硬化交替は，四つの音韻プロセス（連濁，p→h，持続性中和，子音の無声化）の不透明な相互作用によって生じる現象である。子音の無声化は連濁とp→hに対して逆投与の関係にあり，持続性中和に対して逆奪取の関係にある。硬化の背後にある音韻的不透明性は古典的な最適性理論では適切な分析ができない。この現象は，レベル間の順序付けを取り入れた弱並列主義の最適性理論（Stratal OT）によって分析が可能になる。共感（Sympathy）理論や候補連鎖（Candidate Chain）理論といった強並列主義の最適性理論の変種では適切な分析を行うことができない。
日本語の語彙的複合動詞におけるタイプ別生産性をプロト意味役割の観点から体系的に調査する。ここでの「タイプ」とは，結合される動詞の項を同定する際の異なる同定方法を指す。
各種の語彙的複合動詞の生産性の違いは，他動詞の数的優位性のみならず，動詞の統語的項の同定における複雑性に起因すると主張する。そのような複雑性と生産性の相関関係を最適化理論（OT）の有標性制約を利用することで説明する方向性を模索する。このような制約は，ただ単に最適な個別の複合語の形を決めるだけでなく，最適な個々の複合動詞を作り出す各種の過程そのものの比較をも可能にする。本稿の特徴は，語彙的複合動詞の生産性が文法関係の視点のみから単に数量的に記述されてきたことから離脱し，なぜそのような数量的生産性の相違が見られるのか説明を企てるところにある。
Ura（2007）は，一部の関西方言において「僕はジョンにそのことをできる（て）思う」のような構文が成り立つことを指摘し，「長距離の例外的格付与（Long-Distance ECM: LD-ECM）」として分析した。
その際の省略可能な補文標識は「弱い」フェイズを形成するとして，「補文標識の省略が許されない言語では，LD-ECMは許されない」という一般化を示した。この一般化に対し，本稿では，60人の非関西方言話者について対応構文を調査し，（i）補文標識が省略可能であるが，LD-ECMが許されない場合（弱い反例），および（ii）補文標識が省略不可能であるが，LD-ECMが許される場合（強い反例）という二種類の反例を提示する。特に（ii）の反例は，Uraの一般化にとって大きな問題であり，補文標識の省略とLD-ECMの成立との間に強い相関関係がないことを指摘する。さらに，フェイズ理論に頼らない分析として，補文の対格名詞句が焦点を当てられるために補文CP指定部に非顕在的に移動し，そこで対格の認可を受けるという分析を示す。この分析では，Uraの分析とは異なり，文法操作の局所性や最短距離性といった文法理論の核を成す制約を変更することなく，日本語にみられるLD-ECMの有標性を捉えることが可能になる。
The nDrapa language (Sichuan, China: Tibeto-Burman) has multiple types of existential sentences, which convey various readings.
The reading of a sentence is determined by the existential verb stem, the suffix and/or auxiliary, the constituent order, and the animacy of the arguments. In this article, I describe existential sentences in nDrapa, paying special attention to the correlation between the animacy of arguments and the readings of the sentences. Among the six existential verb stems, the implications of the stem ˉnʌ, in particular, change depending on the animacy of the arguments. A ˉnʌ-existential sentence requires an animate argument either as the subject or the locative noun phrase (NP). If the subject is animate, the ˉnʌ-existential sentence conveys the reading of narrowly defined existence. On the other hand, if the locative NP is animate, the sentence conveys a specific "distributing/gaining" implication; that is, someone distributes the subject to the locative NP, or the locative NP gains the subject. Other existential verb stems generally form sentences that indicate possession if the locative NP is animate, except that the stem ˋɕɨ implies that the animate locative NP wears the subject if the subject is alienable.
Haida has been claimed to be a language of the active-stative type in terms of linguistic typology, in which the subjects of intransitive clauses are marked in two different ways: as transitive subjects and as transitive objects. This phenomenon is termed "split intransitivity," and it crosslinguistically underlies active type languages.
The motivations for split intransitivity have been pursued mostly in semantic terms such as "lexical aspect," "agency," and "volition," among others. The split intransitivity in Haida is manifested only when the first (singular and plural) and second (singular) personal pronouns occur as intransitive subjects. Furthermore, intransitive verbs in Haida can roughly be classified into four groups based on the cases and persons of pronouns that occur as their subjects. The present study argues that the two semantic features of [agency] and [control] can be postulated to explain this phenomenon. [agency] describes the situation wherein a verb requires a participant as its subject that performs an activity or instigates a situation. [control] is concerned with one's ability to control the activity or situation. This study also points out inconsistencies in case marking on personal pronouns for certain verbs, as well as for speakers, which may be due to the fact that these two features interact with split intransitivity in a fairly complicated manner. Such inconsistencies may be inevitable since the motivation for the split intransitivity is semantically-conditioned, which in turn leads to difficulties in characterizing active type languages in general.
In Indonesia, while Indonesian, its national language, is full of vitality and immensely influential, many indigenous languages are in danger of extinction.
Bantik, spoken in North Sulawesi Province, is one of such languages and serves to illustrate how languages change in various ways as their speakers diminish in number. The focus of this paper is to describe some of the significant changes the Bantik language is undergoing in response to the sociolinguistic situation in which it finds itself. After presenting the interaction of Bantik, Manado Malay, and Standard Indonesian, all of which are spoken in North Sulawesi Province, I will describe the actual performances of aged and young speakers of Bantik.
In narrative speech, aged speakers, while retaining most traditional aspects of Bantik, use many Manado Malay words. In contrast to the aged speakers, who use Bantik in everyday conversations with family and friends of the same age, young speakers only have a limited knowledge of the language. They also have phonemic and morpho-syntactic rules quite distinct from those of the traditional Bantik, which strongly suggests that Bantik is undergoing attrition.
The Mitsukaido dialect of Japanese has a phenomenon in which a stem-initial fricative turns into a stop (or an affricate) when the stem is the second member of a compound.
This hardening alternation results from opaque interactions among four phonological processes: sequential voicing, debuccalization (p→h), continuancy neutralization and consonant devoicing. Consonant devoicing counterfeeds sequential voicing and p→h, and counterbleeds continuancy neutralization.
Classic Optimality Theory cannot deal with the phonological opacity behind the hardening alternation. Stratal OT, a weak parallelist OT extension which incorporates level ordering, provides a solution for this phenomenon. Strict parallelist extensions such as Sympathy theory and Candidate Chain theory make the wrong prediction for the hardening alternation.
The type-wise productivity of lexical V-V compounds in Japanese is investigated systematically from the perspective of thematic proto-roles. It is shown that the degrees of productivity are sensitive not only to the quantitative advantage of transitive verbs but also to the complexities encountered in the process of argument matching between two verbs to be combined via compounding. The correlation between such complexities and productivity is captured by an optimality-oriented approach that takes advantage of a set of markedness constraints.
The constraints are not simply capable of selecting the most optimal candidate of a single comparison, but useful for comparing different winning candidates from separate and mutually independent comparisons as well. The current approach distinguishes itself from those based on the traditional concept of transitivity of verbs, which merely describe - but are not capable of predicting - the observed patterns of productivity.
Based on the observation on a special type of ECM construction in a certain variety of Kansai Japanese, Ura (2007) argues that dialect allows a true instance of long-distance case assignment whereby an object NP in an embedded clause is assigned an accusative case by the main verb, as in Boku wa John-ni sono koto-o dekiru (te) omou 'I think (that) John [DATIVE] can do that [ACCUSATIVE]'. According to his analysis, such an unorthodox operation is made possible by the deletable complementizer in this dialect, which renders the phase of the embedded finite clause 'weak', thereby exempting it from the Phase Impenetrability Condition (PIC). From this, Ura suggests the generalization that no language allows long-distance ECM (LD-ECM) unless its complementizer is deletable.
This paper investigates the acceptability of the corresponding construction with sixty speakers of non-Kansai Japanese and adduces two types of counterexample to Ura's generalization: (i) LD-ECM is disallowed even though the complementizer is deletable, and (ii) LD-ECM is allowed even though the complementizer is not deletable. These counterexamples indicate that there is no strong correlation between the deletability of a complementizer and the acceptability of LD-ECM. We thus reject Ura's analysis utilizing the 'strength' of a phase and present an alternative analysis in which the accusative NP in an embedded clause, typically focused with heavy stress, undergoes covert movement to the Spec of CP and gets its accusative case licensed there. The proposed analysis can capture the marked nature of the LD-ECM construction without altering the definitions of such general notions as locality and minimality