本論文以“中文可以說是一種無時式（tenseless）的語言”（Lin 2006）為啟發，歸納整理中文事件謂語的種類，觀察到單一事件謂語因搭配的時間標記手段（Time-encoding Devices）不同，而可能產生不同的時間錨點（Time Anchorage），進而注意到中文實際上存在依靠單一有界[+telic]事件謂語，即能將時間錨點落在過去的可能性，也就是有界事件謂語存在過去時間默認視角（Default Viewpoint）。除了觀察單一事件謂語外，本研究亦納入了時間副詞、動詞後綴“了1”與篇章語法（Discourse Grammar）為觀察目標，本研究以問卷方式收集資料並進行量化分析，測試並觀察非中文為母語的中文學習者與中文母語人士的回饋有何異同。觀察重點如下：第一，單一有界事件謂語的過去時間默認視角存在與否；第二，單一／多重時間標記手段與單一有界事件謂語共現時，其對受試者理解事件時間產生影響。分析結果顯示，從大部分受試者的回饋中，“有界事件謂語存在過去時間默認視角”此觀念並不是普遍現象。這項結果顯示唯有標示足夠的時間標記手段，才能使聽話者理解事件時間的真正時間錨點。研究者最後以研究結果檢視現今通行的兩套華語文教材，發現中文時間標記教學練習不足，教師在教學現場時應扮演好引導學習者練習的重要角色。時間標記之所以重要，因為在在影響聽話者理解中文事件時間。本研究觀察學習者對時間的理解方式，有助於教學者，進行更有效的教學，協助達到有效的溝通。

傳統的研究將臺灣話的「人」視為人稱代名詞，並著重於它的指示代名詞使用。但出現於言談中的「人」卻多不做此用。因此，本研究旨在探討這種不做指示代名詞用的「人」，及其在言談上的功能。 　　根據Schiffrin提出的言談分析模式，「人」在五個層面--語意結構、行為結構、交談順序結構、訊息狀態、及交談者關係--都有其功能。大體而言，「人」的主要功能是在語意結構上，它在言談間作為一個「逆向承接標記」，它的出現顯示著上下文有著相反或對比的關係。藉由這種承接的作用，能使說話者更明確地表達語意，並使聽話者更精確地掌握語意，因而使得言談更為順暢。 / The present study investigates the behavior of the non-referential lang and the functions it performs in Taiwanese spoken discourse. 　　Based on　Schiffrin's (1987) model of discourse coherence, lang. plays roles on the five planes of discourse. The predominant function of lang is found in the ideational structure, where it explicitly marks contrastive relations between propositions.　In the action structure, it prefaces speech acts like defense, dispute, and refutal when a speaker disagrees the other participant's accusation, assertion, and request.　In the exchange structure, it is employed to claim/fight for or retain the turn.　In the information state, it signals contrasts between the upcoming proposition and speaker knowledge or speaker/hearer meta-knowledge.　In the participation framework, it functions to change the relative relations between interlocutors in a conversation and to indicate a discrepancy between interlocutors in the subjective attitude toward an event. As a whole, lancr serves as a contrastive marker, thus informing a discontinuity in a discourse. 　　The syntactic distribution of lang and its co-occurrence restrictions with Taiwanese sentence-final particles further support our analysis that lang functionsas a marker of contrast.

人智運算(Human Computation)是近年來最熱門的研究領域之一，而適地性服務(Location-based service)也因此衍生出許多研究議題。由於傳統上在搜集資訊時往往會浪費過多的人力資源，所以我們希望使用者藉由玩遊戲的方式背後完成我們想要的事情。在此篇論文中，我們利用GWAP(Games with a Purpose)的概念設計一套行動地理標籤系統，系統採用玩家分享自身附近景點資訊再透過其他玩家來到相同地點做驗證的方式，讓玩家在一邊玩遊戲的過程中搜集相關地理資訊，另一方面達到景點資訊是可信賴的目的。 然而，對系統而言要如何選題給玩家驗證才能提升整體效能，為了解決這些問題，我們在選題策略上提出三種不同的任務分配演算法，再透過一連串的實驗模擬來證明我們所設計的系統確實能夠在選題上達到好的效能，最後我們考慮到現實生活中可能的情形後將此系統實作在智慧型手機上(Android Phone)。 / Human Computation is popular recently and have become one of the hottest research topics, therefore Location-based service also drives a lot of issues. Due to collect information always cause too many manpower- wasted in tradition , so we hoped that the users to complete the things which we want by play game’s way .In this thesis, we based on the concept of ‘Games with a Purpose’(GWAP) to develop a mobile geospatial tagging system, system adopt player share self nearby scenic spots information and then through another players come the same place do the verification, allowing the player to play the game in the side of the process of collecting the relevant geographic information, on the other hand to achieve the purpose of scenic spots information is reliable. However, how to choose topics for player verification can effectively evaluate the system performance, to solve these problems we propose three different selection strategies on the task assignment algorithms, then through a series of simulation experiments designed to prove that our system can indeed achieve good performance on the topics, finally we take into account real life situations and implement in the smart phone (Android Phone).

為了呼應將人際層面融入寫作教學的趨勢，本論文將透過分析人稱代名詞（一種明顯的人際關係標記)在議論文中的使用功能來探討不同程度的台灣大學生如何使用人稱代名詞建構作者與讀者之間的關係。本研究的第一部份著重在七十六篇文章的文本分析。首先，這些文章按照評分結果將其分成高、中、低三組，然後分析人稱代名詞最常出現的搭配語言形式，並歸納出不同人稱代名詞的篇章功能。第二部份則是分析學生問卷及訪談學生，藉以作進一步的闡述。問卷的目的在找出學生對議論文寫作的看法，而訪談學生則是想找出使用不同人稱代名詞的原因。本研究發現不同程度的三組學生在人稱代名詞的整體使用數量、種類、及頻率分配上都有不同，程度高的一組明顯少於中間程度及較低組。同時，結果也顯示這些學生會搭配不同的語言形式（例如動詞、助動詞、加強標記等）來行使不同的篇章功能，而且不同程度的學生在功能運用上也會有所差異。整體而言，低組同學呈現較多的自我投射，中間組同學比較注重與讀者和其他外人的關係，而高組同學在呈現觀點時較為客觀。在選擇人稱代名詞時，學生會從自己本身、讀者、文章寫作等三方面的相互關係作出考量，決定採取主觀或客觀的觀點、表達權威或謙卑的態度、顯示親近或疏離的關係、使用直接或間接的策略。大致上來說，這些學生使用較多的人性訴求來凝聚跟讀者之間的關係，同時也強化自己論點的力道。這樣的策略充分反映出台灣文化中的人道主義和集體主義。本研究發現學生在議論文寫作中會以功能和人際關係為導向來選擇和使用人稱代名詞。 / In response to the call for the incorporation of interpersonal dimension into the writing pedagogy, this study provides a functional analysis of personal pronouns—an explicit interpersonal marker—used in argumentative texts by Taiwanese college students. The purpose is to see how students of different proficiency levels construct the writer-reader relationship through personal pronouns during the composition. The first part of the study centers on the analysis of 76 learner essays. They are first rated and sorted into three groups of different quality—High, Mid, and Low. Later, the linguistic forms associated with personal pronouns are examined, and the discourse functions personal pronouns fulfill in contexts are also identified. The results of the text analysis are further supplemented by the post-writing questionnaires and the oral interviews on students to obtain more in-depth discovery and interpretation. While the questionnaire aims to reveal how the students perceive argumentative writing, the interview intends to find out the reasons for their choices of personal pronouns. The results have shown that the use of personal pronouns in the three groups differs in quantity, type and distribution. The High group writers use significantly fewer pronouns than the other two. Moreover, the students use personal pronouns with salient accompanying linguistic forms (e.g. verbs, modals, emphatic markers) to perform various discourse functions, and students of different levels also vary in maneuvering the functions. Overall, the Low group writers tend to be more self-involved, and the Mid group writers are more likely to include in-group and out-group members in discourse. The High group writers, however, present their arguments more objectively. In selecting personal pronouns, the students usually take account of the interrelationship among the writer, the reader and the text, on whose basis the alternatives between subjectivity and objectivity, authority and modesty, intimacy and detachment, or directness and indirectness are weighed. In general, the students use more personal appeals to achieve mutual solidarity with the reader and to intensify their convictions as well, which reflects humaneness and collectivism that have been highly valued in Taiwanese culture. The study has found that the students' strategic choices of personal pronouns in argumentative writing are usually functionally and interpersonally-oriented.

本研究使用Schiffrin (1987) 提出的言談結構（discourse structure）為分析架構，旨在觀察漢語兒童在同儕互動中使用口語中常出現的言談標記「好」和「對」的情形，藉以檢驗其如何反映漢語兒童的溝通技巧以及同儕互動的特性。研究語料來自六位五歲的漢語兒童兩兩之間互動的對話，共237分鐘。 研究發現五歲漢語兒童能掌握言談標記「好」和「對」在三個言談結構的功能，且他們使用此兩個言談標記的不同功能時有所異同。首先，「好」和「對」主要使用於交談順序結構（exchange structure）作為表同意的標記。雖然在成人對話中「好」和「對」皆可用於交談順序結構及語意結構（ideational structure），作為表知曉（acknowledgement）的反饋應答標記和話題轉換（topic transition）標記，但研究結果發現漢語兒童在交談順序結構中只使用「好」作為表知曉的反饋應答標記；在語意結構只用「對」作為話題轉換標記。研究結果顯示五歲漢語兒童在與同儕互動時，已能夠使用反饋標記「好」表現參與對方話題的意願，而他們不使用「對」則可能與「對」作為反饋應答標記時的用法和應答詞（backchannels）相似有關。根據先前研究指出應答詞屬於兒童較晚才習得的溝通技巧（Hess & Johnston, 1988）。此外，五歲漢語兒童亦展現出使用話題轉換標記「對」的能力，顯示他們已知道如何使用言談標記幫助建構言談連貫性（discourse coherence）；而他們選擇使用「對」而非「好」來轉換話題，則可能與「好」的此項功能所隱含的發話者權威性有關（Chen & Liu, 2009），若使用這類帶有發話者權威的言談標記，則可能損害其與同儕間的關係。本研究因而推論漢語兒童在同儕互動中言談標記「好」和「對」各種功能的使用，不僅僅顯示出他們的溝通技巧，同時也反映了同儕互動的特性。 / The thesis aims to investigate Mandarin-speaking children’s use of two frequently appearing discourse markers, hao ‘okay’ and dui ‘right’, when interacting with peers in order to examine how their use of these markers may reflect their communicative skills and the characteristics of peer relation. The data included 237 minutes of 5-year-old Mandarin children’s conversations with friends while playing. Schiffrin’s (1987) model of discourse structures, which includes the exchange structure, the action structure, the ideational structures, the participation frameworks, and the information state, was used for the analysis. The results showed that Mandarin-speaking children used hao in the information state, the exchange structure and the action structure while dui in the information state, the exchange structure and the ideational structure. The functions of hao and dui in the present data demonstrated several similarities and differences. Both markers were used by the children in the exchange structure to show the speaker’s agreement. However, only hao functioned as an acknowledgement marker to indicate the receipt of information in the exchange structure while only dui marking topic transitions in the ideational structure, even though both markers can serve these two functions in adult conversation. Mandarin-speaking children’s use of hao and dui to express agreements, which indicates their collaborative stances, may help them establish alliances with each other (Wang et al., 2010). Moreover, Mandarin-speaking children at age five demonstrated their ability to use hao as an acknowledgement marker to show their intention to participate in their peer’s current talk. In addition, that only hao but not dui served as an acknowledgement marker may result from the similarity between the acknowledging function of dui and that of backchannels, which has been considered among the last acquired communicative skills (Hess & Johnston, 1988). Furthermore, Mandarin 5-year-old children had the ability to use dui as a topic transition marker to establish discourse coherence. Meanwhile, that dui, instead of hao, was chosen by the children as a transition marker may reflect the relatively equal relations between peers, since hao is usually used by a speaker with higher status to control the topics in adult conversation (Chen & Liu, 2009). It is concluded that Mandarin children’s use of the two markers not only demonstrates their communicative skills but also reflects the particular nature of peer interaction.

本論文的主要目的在對於分類詞、性別詞以及複數標記在語言當中的地域分佈提出解釋．在前人的研究當中，這三項元素被認為是名詞句中平衡資訊的重要工具(Greenberg, 1990; Aikhenvald, 2000)：分類詞語言主要位於東南亞和南美洲部分地區，而具有性別詞或複數標記的語言大多出現在歐洲、非洲和美洲部分地區．我們提出的論證如下：即便這三樣元素外表上具有歧異，它們會呈現當今所見的地域分佈原因在於它們共有的兩項標記功能：可數性質及語意分類．分類詞同時滿足兩者而性別詞及複數標記分別滿足其一；依照此邏輯，我們預測有分類詞的語言不會同時具有性別詞及複數標記而反之亦然．本文中我們透過句法形式和語意功能的比較提出論證並透過類型學、地理及歷史的角度分析來自世界上最大的二十個語系（印歐，漢藏，亞非，尼日爾-剛果，南島，達羅毗荼，阿爾泰，南亞，壯侗，尼羅－撒哈拉，烏拉，高加索，等語系）的155個語言．架構上，第一章簡單對研究題目進行介紹，第二章呈現前人研究的匯整，第三章包含我們的理論論證以及我們對於分類詞、性別詞及複數標記分與合的解釋．隨後的第四章中，我們提出類型學和地理資訊系統(GIS)的證據；最後在第五張和第六章我們分別點出本研究的限制以及結論． / This thesis aims at providing an explanation for the typological and areal distribution between numeral classifiers, genders (noun classes) and grammatical plural markers. Within previous studies, these three components are considered as different devices to balance information in noun phrases (Greenberg, 1990; Aikhenvald, 2000). Numeral classifier languages are mainly present in South-East Asia and parts of South-America, while languages with genders and grammatical plural markers are generally attested in Europe, Africa and parts of the Americas. We propose that despite their apparent divergence, the three elements display this particular geographical distribution due to their convergent features of count/mass distinction and semantic classification: Numeral classifiers carry both functions, while genders and plural markers separately fulfill one of them. Following this logic, we expect that a language with numeral classifier do not have simultaneously the systems of genders plus plural markers and vice-versa. Theoretical evidence via formal syntactic form and semantic function comparison is proposed and further supported by typological, geographical and historical analysis of 155 languages that are mainly part of the 20 biggest language groups in the world, e.g. Indo-European, Sino-Tibetan, Afro-Asiatic, Niger-Congo, Austronesian, Dravidian, Japonic, Altaic, Austro-Asiatic, Tai-Kadai, Creole, Nilo-Saharan, Uralic, Quechuan, Hmong-Mien, Mayan, North Caucasian, Language isolates among others. Chapter 1 presents a brief introduction of the subject while chapter 2 displays the literature review. Chapter 3 includes our theoretical discussion proposing explaining the convergence and divergence among numeral classifiers, genders and plural markers, followed by typological and geographical evidence via GIS (Geographic Information System) in Chapter 4. Finally Chapter 5 and 6 contain the limitations of our study and its conclusion.

本論文主要的目的為找出六個身體部位詞-臉、面、眼、目、嘴、口在台灣華語的真實使用情形。本篇所分析的語料皆來自於目前收錄最多筆語料、體裁種類平衡、被廣泛使用的中央研究院現代漢語平衡語料庫４.０版。我們有系統地分析了包含這六個身體部位詞詞彙的內部構造、語意分佈、與詞彙化類型。研究結果發現，在包含這六個身體部位詞詞彙中，[NN]N為最顯著的詞彙內部構造。而在語意分布的部份，相較於臉、眼、嘴，面、目、口較常被用來表達轉喻及隱喻等非字面上的意義。此外，人與情緒為最為普遍的轉喻意，這兩個轉喻用法在六個身體部位詞詞彙中皆可發現。此外，當身體部位詞詞彙表達人這個轉喻意時，其詞彙內部構造多為[NN]N ; 當身體部位詞詞彙表達情緒這個轉喻意時，其詞彙內部構造多為[NN]N與[VN]V。另外，大體上來說，無論身體部位詞詞彙是表達轉喻意或隱喻意，[NN]N為最普遍的內部構造。並且，當身體部位詞詞彙表達轉喻意或隱喻意時，最常出現的詞彙化類型為隱喻詞彙化（metaphorical lexicalization）。本研究結果期望能提供語料庫辨別轉喻及隱喻的方法，並期望轉喻及隱喻的研究可以從長久以來偏向質性的討論分析，擴展到量化的比較與討論。 / This study aims at finding the actual usage patterns of the six body-part terms liǎn, miàn, yǎn, mù, zuǐ, and kǒu extracted from the largest, balanced and widely-used Sinica Corpus 4.0. The word component structures, meaning distribution and lexicalization types in the six body-part terms are systematically analyzed and compared. The results show that, in terms of word component structures, the [NN]N structure is the most dominant word component structure across the six body-part terms. As for meaning distribution, it is found that miàn, mù, and kǒu are more frequently used to denote non-literal meanings (either metonymic or metaphorical meanings) than their counterparts – liǎn, yǎn, and zuǐ. In addition, PERSON and EMOTION are the most prevalent metonymic meanings across the six body part terms. It is also found that when the body-part terms denote PERSON, the most dominant word component structure is [NN]N, while when they denote EMOTION, [NN]N and [VN]V are the most dominant structures. In general, the [NN]N structure shows the highest frequency of occurrences in all the six body part terms when they are used either metonymically or metaphorically. As for the lexicalization types, most of the disyllabic words containing the six body-part terms are in metaphorical lexicalization when they denote metonymic or metaphorical meanings. The results may provide some insights for refining procedures of identifing metaphors and metonymies in corpora so as to accelarate quantification and statistical evaluation.

XML (eXtensible Markup Language)已經快速成為網際網路上資料交換的標準，其架構適合用來描述複雜的資料結構並提供XML文件驗證的機制。要交換XML文件，就必須使用綱要語言來制定XML語彙，因而由此衍生出XML綱要設計之問題。在本篇研究中，將研究UML資料模型與XML綱要之間轉換的議題並發展一個可進行雙向轉換的模型。此轉換模型包含二部份，一是從UML資料模型結構轉換為XML綱要結構，一是從XML綱要結構反向轉換為UML資料模型結構；這二個模型將分別探討XML DTD (Data Type Definition)及 W3C XML Schema二個主題，將這二個子模型分別從語法及語意轉換之層面切入進行一連串結構轉換規則之延伸及建立。在XML綱要結構反向轉換為UML資料模型結構之模型中亦包含二個子模型，分別由XML綱要結構建立轉換之規則。我們建立此轉換模型之目的在於使UML資料模型結構及XML綱要結構間之轉換方法更為明確、簡單及容易使用，並透過系統實作來驗證其可行性。 / The (eXtensible Markup Language) XML is fast becoming the standard data exchange across the Internet. It is suitable to describe complex-structured data. It also provides a validation mechanism to validate vocabulary used in the XML document. In order to exchange XML document, XML vocabulary needs to be defined. As a result spread out the issues of XML schema design from here. In the research, UML data model and XML schema transformation issue have been studied to develop a bi-directional mapping and transformation model. We design the bidirectional mapping and transformation model in two parts. One is from UML data model to XML schema; another is from XML Schema to UML data model. Each part has two explored subject, including XML DTD (Data Type Definition) and W3C XML Schema. We develop a sequence of transformation rules from the viewpoints of syntactic and semantic transformation. These are also two parts in the transformation model of mapping from XML schema to UML data model. We develop transformation rules from the viewpoints of generic XML constructs step by step. We build up the transformation model to make transformation method more generic, genuine, systematic, explicit, simple and easy to use. Besides, we implement a transformation editor to validate the bi-directional transformation model.

本論文著眼於「共時」與「歷時」兩個視角，並從方言、跨語言現象及歷史語料討論「喫」「吃」「乞」的相關議題，包括歷史音韻、方言語法、歷史語法。 根據我們的研究結果顯示，「喫」「吃」兩字來自於「齧」「齕」的部件取代，而「喫」「吃」混用的機制是聲符「乞」「契」的音同所致，最終「喫」「吃」形成異形同義字。另外，也針對表示飲食義的「喫」進行歷史音韻的推測，按照音變的規則「見」系「溪」母的「喫」在現代漢語應讀為顎化音，但在詞彙擴散的效應下部份的聲母，流入中古的「照」系，其他的例子像是「廈」「閘」也循此變化模式。 文獻指出元明以後「喫」「吃」「乞」三者都有當作被動標記的用例，並且相互的混用。透過閩方言的音韻資料顯示，「喫」「吃」「乞」表示被動的語義根源並不相同，三者的混用主要是在歷史音變中形成一組同音字。 閩方言中有一個單用的入聲「乞」兼表給予、使役、索取、被動。而在閩南地區被動用法則為「乞與」或以授予動詞「與」兼用。因此，我們建立了一個假設：閩方言內部的被動標記「乞」可能有不同的語義根源，一個是和授予動詞「與」複合的「乞與」朝向授予>使役>被動發展；一個是「乞」由索求到被動。為了論證這個觀點，本文從英語的“get”作為觀察，並討論了「與格轉換」、「詞義分解」、原型施事受事理論。 從方言語料顯示「乞」可以出現在動詞及與格標記的位置，形成S+乞+DO+乞+IO；以及雙賓結構「乞+IO+DO」。透過與格轉換理論，前者可以推導出後者，另外歷史上也見到「乞與+IO+DO」的雙賓句式，這兩種句式的存在就是為了辨義作用，如同漢語的「借」和「借給」。閩方言裡不使用「乞與+IO+DO」的雙賓句式，因此我們推測當「乞與」形成之初是一組反義並列，隨著「與」的共現和語法化「乞與」複合成使役動詞並列，並朝向被動標記發展。對照「乞與」的發展推測，單用的「乞」其來源就值得探究。 閩方言單用的被動標記「乞」基本上都必須帶有施事者，形成長被動「NP1+乞+NP2+V」，但有少數的用例顯示，仍可以接受不帶施事的短被動「NP1+乞+V」。動詞「乞」的語義本身就帶有下對上的位階關係，因此操控權並非只限於 「乞」的主語。據此，論文的最後一章推測短被動「乞」的形成有兩個重要條件：動詞作格化(被動化)、動作事件的發生是在非自願性(不幸說)。歷史上「乞+N」最早出現，但是因為名詞動詞的模糊性產生「乞+N」>「乞+V」如「乞降」。 而長被動「乞」的產生，透過其他被動式的觀察，其發展應和「NP1+被+V+於+NP2」這個結構有關，透過句法操作、句式趨向最終產生「NP1+乞+NP2+V」。 兩相對照，短被動是在「乞+V」中「乞」重新分析為一個次要動詞(副動詞)；而長被動是在句法操作生成。 關鍵字: 詞彙擴散 被動標記 與格轉換 作格化(被動化) / Abstract This paper is aimed at two perspectives, “synchronic time” and “diachronic time,” and explores issues relating to “喫” “吃,” and “乞” from dialect, cross-linguistic phenol, and historical corpus discussions, including historical phonology, dialectic grammar, and historical grammar. The results show that the two words “喫” and “吃” are replaced by the “齧” and “齕” parts, while the mixed mechanism of “喫” and “吃” is derived from the homophone of “乞 (beg)” and “契”(bond), thus leading to the synonyms “喫” and “吃” Additionally, targeting the word “喫” that means diet, historic-phonology-related speculations were made. According to the sound shift rules, the word “喫” with “見” system and “溪” constituent should be pronounced as a palatalized sound (tilde) in the modern Chinese language. However, under the influence of lexical diffusion, some consonants fall under the “照” system. Other examples such as “廈” and “閘” also follow this pattern. According to literatures, there have been cases of “喫” “吃” and “乞” used as passive markers, which have been alternatively used. According to the phonological information of the Min dialect, it shows that the semantic roots of “喫” “吃” and “乞” are not the same and that the alternative use is mainly to form a set of homophones from the historical sound shifts. In the Min dialect, there is a single checked tone “乞” which means “give, causative, beg, supplicate, and passive.” In the Minnan region, the passive usage is “乞與” or the dative verb to be given “與.” Hence, the hypothesis is the passive marker “beg” in the Min dialect may have different semantic roots: one is the dative verb that compounds with “與” and “乞與” which are headed toward the direction of give>causative>passive; the other is the word “乞” that shifts from demand to passive. In order to demonstrate this point, the English word “get” was observed in this paper. The dative shift, lexical decomposition, and Porto agent Porto patient were explored. The dialectic corpus shows that “乞” can appear in the position of the verb and dative marker, thus forming S+乞+DO+乞+IO and the double object construction 乞+IO+DO. Through the dative shift theory, the latter can be deduced from the former. Additionally, the “乞與+IO+DO” double object construction has also been throughout history. The existence of the two sentence patterns plays the role of “to distinguish,” just like “borrowing” and “lending” in the Chinese language. In the Min dialect, the “乞與+IO+DO” double object construction is not used. Therefore, it is speculated that “乞與” was first formed as a set of opposite (meaning) antonym. With the collocation of “與” and the grammaticalization of “乞與” compounded to form and tie with the causative verb and head toward passive marker development. In conjunction with the development of “乞與” it is speculated that the source of the sole use of “乞” is worth exploring. Basically, the passive marker “乞” used in the Min dialect must have a causal agent, forming long passive construction “NP1+乞+NP2+V.” However, few cases show that short passive construction “NP1+乞+V” without a causative agent is still acceptable. The verb “乞” itself possesses the semantic bottom-to-top relationship. Thus, the control is not restricted to the “乞” subject. Accordingly, the last chapter in this paper covers two speculated important conditions that contribute to the formation of “乞”: ergativization/passivization and occurrences of actions that are involuntary (sad to say). In history, “乞+N” first appeared, but due to the ambiguity of the nouns and verbs, “乞+N” > “乞+V” such as “beg to surrender or got tamed” resulted. On the other hand, the generation of the long passive “乞,” as observed through other passives, is associated with the “NP1+被+V+於+NP2” structure. Through syntactic operations and sentential convergence, “NP1+乞+NP2+V” eventually resulted. In contrast of the two, short passive is the reanalysis of “乞” in “乞+V” to derive at a secondary verb (coverb), while long passive is generated through syntactic operation. Keywords: Lexical diffusion, Passive markers, Dative shift, Ergativization (passivization)

隨著電腦以及網際網路的普及，越來越多各領域的資料被數位化，利用電腦幫助儲存及管理資料。有許多資料在數位化的過程中，採用tree的資料結構來表達以及儲存。也因此，如何查詢這些龐大的資料，就成為重要的課題。在本論文中，我們針對具有rooted、labeled以及ordered或unordered等特性的tree結構資料的索引問題，提出稱之為PCS-trie之全新的主記憶體資料庫(Main Memory Database)索引結構，並提出相關的增刪資料及搜尋演算法，以達成加速處理sub-tree query之目標。此索引方法的基礎，在於將tree database中的tree編碼為可完整代表其結構的PREOD code字串，之後再以我們所提出的PCS-trie加以索引。PCS-trie索引結構支援資料庫的動態增刪，且我們也提供了有效率的新增及刪除資料的演算法。本論文中也提出了多種搜尋演算法，使能夠在PCS-trie中進行exact matching、處理query tree含有don’t care部分之查詢、以及fault tolerant等不同類型的查詢。最後，我們以實驗的方法，配合人工產生的實驗資料，來對PCS-trie索引方法的時間及空間等各方面的效率加以檢驗。 / With the popularization of computer and the Internet, more and more data in various domains are digitized in order to take advantage of the power of computing and storing, and use the Internet to spread these informationz. Many data use tree structure to store them in the process of digitization. For this reason, it is a challenge to deal with these enormous data. In this paper, we present an approach to the search problem for these rooted labeled trees. We show a novel index structure, PREOD code Search trie (PCS-trie), and related algorithms for construction and search of PCS-trie, to speed up the sub-tree query to a tree database. The fundamental of this reaseach is to encode trees in tree database by PREOD code, in which the structure information of a tree can be reserved completely. Then we index these PREOD codes by PCS-trie. PCS-trie supports dynamic insertion and deleteion of PREOD codes. PCS-trie can handle three different types of query requirements: exact sub-tree query, query with don’t cares, and fault-tolerant query. Finally, we have conducted a series of experiments to evaluate the performance of PCS-trie and related algorithms. Experimental results obtained by running our techniques on synthetic data demonstrate the good performance of the proposed approach.