This thesis combines corpus-based and contrastive approaches, seeking to provide a systematic explanation of passive construction in two typologically distinct languages, Mandarin Chinese and Bahasa Indonesia. Previous studies of bei construction and di- passive have mentioned and discussed some verbs which can or cannot be passivized; however, they have not yet provided a systematic categorization of those passivized verbs. Therefore, this thesis aims to reveal the semantic restrictions that affect the compatibility of verbs to appear in passive form. This corpus-based contrastive study also suggests similarities and differences of passives in both languages in terms of structure, meaning, and function. Analysis in this study is based on two corpora: (1) the internet posting of news articles in Bahasa Indonesia from Kompas Cyber Media between January 7-13, 2008; (2) the internet posting of news articles in Mandarin Chinese from 中國時報 ‘China Times’ between January 7-13, 2008. News articles were taken from the same or consecutive days from two main topics: international news and entertainment news. The findings in this study show that transitivity is the most prominent semantic feature of the verbal stems in order to be eligible for passivization in both languages. Bei passives are more often to co-occur with transitive verbs while di- passives can co-occur with both intransitive and transitive verbs. Regarding the relation between structure and meaning, passivization process in Mandarin Chinese only adds a causative-resultative reading and does not change the lexical meaning of the verbal stem. On the contrary, the passivization process in Bahasa Indonesia involves a complex derivation process which alters the transitivity and lexical meaning of the base verb.

本研究以1993年到2017年3月間的193篇FOMC Minutes作為研究素材，先採監督式學習方法，利用潛在語意分析（latent semantic analysis，LSA）萃取出升息、降息及不變樣本的潛在語意，再以線性判別分析(Linear Discriminant Analysis, LDA)進行分類；此外，本研究亦透過非監督式學習方法中的探索性資料分析(Exploratory Data Analysis, EDA)，試圖從FOMC Minutes中找尋相關變數。研究結果發現，LSA可大致區分出升息、降息及不變樣本的特徵，而EDA能找出不同時期或不同類別下的重要單詞，呈現文本的結構變化，亦能進行文本分群。 / In this study, 193 FOMC Minutes from 1993 to March 2017 were used as research materials. The latent semantic analysis (LSA) in supervised learning methods was used to extract the potential semantics of interest rate increased, decreased, and unchanged samples, and then linear discriminant analysis (LDA) was used for classification. In addition, this study attempts to find relevant variables from FOMC Minutes through exploratory data analysis (EDA) in unsupervised learning methods. The results show that LSA can distinguish the characteristics of interest rate increased, decreased, and unchanged samples. EDA can find relevant words in different periods or different categories, show changes in the text structure, and can also classify the texts.

財務報表中的主觀性詞彙往往影響市場中的參與者對於報導公司價值和獲利能力衡量的決策判斷。因此，公司的管理階層往往有高度的動機小心謹慎的選擇用詞以隱藏負面的消息而宣揚正面的消息。然而使用人工方式從文字量極大的財務報表挖掘有用的資訊往往不可行，因此本研究採用人工智慧方法驗證美國財務報表中的主觀性多字詞 (subjective MWEs) 和公司的財務狀況是否具有關聯性。多字詞模型往往比傳統的單字詞模型更能掌握句子中的語意情境，因此本研究應用條件隨機域模型 (conditional random field) 辨識多字詞形式的意見樣式。另外，本研究的實證結果發現一些跡象可以印證一般人對於財務報表的文字揭露往往與真實的財務數字存在有落差的印象；更發現在負向的盈餘變化情況下，公司管理階層通常輕描淡寫當下的短拙卻堅定地承諾璀璨的未來。 / Subjective assertions in financial statements influence the judgments of market participants when they assess the value and profitability of the reporting corporations. Hence, the managements of corporations may attempt to conceal the negative and to accentuate the positive with "prudent" wording. To excavate this accounting phenomenon hidden behind financial statements, we designed an artificial intelligence based strategy to investigate the linkage between financial status measured by annual earnings and subjective multi-word expressions (MWEs). We applied the conditional random field (CRF) models to identify opinion patterns in the form of MWEs, and our approach outperformed previous work employing unigram models. Moreover, our novel algorithms take the lead to discover the evidences that support the common belief that there are inconsistencies between the implications of the written statements and the reality indicated by the figures in the financial statements. Unexpected negative earnings are often accompanied by ambiguous and mild statements and sometimes by promises of glorious future.

過去多數英文介系詞語意之研究皆以介系詞為多義詞，並藉由語意網絡去分析其詞義（e.g. Lindner, 1983; Tyler and Evans, 2001），或透過意象圖式理論（image schema theory）來探究介系詞語意，依不同的前景名詞（figure）與背景名詞（ground）組成可產生不同的形貌結構，從而衍生出其特殊語意（e.g. Lakoff, 1987）。然而，有關介系詞片語中詞彙搭配詞組如何影響介系詞使用之研究則較不多見。故本論文著重在分析前景與背景名詞等搭配詞組之語意特徵，探究其對介系詞in和on語意及介系詞片語組成之影響。以英國國家語料庫（British National Corpus）和政治大學外語學習者語料庫（The NCCU Foreign Language Learner Corpus）作分析，比較學習者與以英語為母語者使用in和on介系詞片語在語意上的差異。 本研究共分三部分。第一部分的詞義分析比較學習者與母語人士在使用in與on之詞義之異同，在介系詞in的部分，學習者與母語人士語料相較之下僅有細微的差異，多數語料皆為原型場景（proto-scene）之詞義的使用。介系詞on的詞義使用則稍有不同；學習者較傾向作具體位置的指涉，然母語者則較傾向呈現此介系詞的抽象意涵。第二部分的語意特徵分析則比較介系詞in和on在具體（literal）與抽象（metaphorical）片語組成上，前景與背景語意特徵之異同。研究結果顯示不同的介系詞片語會由特定類型的前景與背景名詞組成，故在組成不同語意之介系詞片語時，特定語意特徵出現的頻率會呈現顯著性地差異。第三部分為學習者語料的錯誤分析，探究學習者誤用in和on介系詞片語之情形。 綜合上述研究結果可得知，組成介系詞片語的名詞詞組會依上下文語意而產生不同之語意偏好（semantic preferences），介系詞的語意除了會受到後接名詞的影響，周圍名詞也可能造成語意上的差異。這些名詞可能會選擇特定的介系詞來使用，亦會影響介系詞在片語中之語意。 本論文深入地探討了介系詞片語詞組之組成，透過大量的語料分析，為英語介系詞複雜的語意層面提出了一個較系統化且完整之解釋，期能將研究發現應用於教材設計上，希望能在學習介系詞使用上有所幫助，也供未來介系詞相關研究作為參考。 / The meanings of English preposition have been explored greatly through vast directions of research. Most studies agree that preposition is a polysemous lexical item whose senses can be construed from the semantic network (e.g. Lindner, 1983; Tyler and Evans, 2001). Its senses has also been examined under the image-schema theory from the cognitive perspective (e.g. Lakoff, 1987), in which the figure (or trajector) and ground (or landmark) interact differently to form various configurations that contribute to sets of distinct senses. Though these studies attempt to provide a comprehensive network for the meanings of prepositions, how the semantic features of the figure and ground might influence the choice of a preposition and the construction of a prepositional phrase is hardly seen. The thesis adopted a native speaker corpus (British National Corpus, BNC) and a learner corpus (The NCCU Foreign Language Learner Corpus, NCCU) to probe into how the semantic features of the figure and ground nouns can help explicate the semantic profiles of the preposition in and on, and to compare learners’ understanding of them with that of the native speakers’. Based on the corpora, the study is composed of three major analyses. The sense analysis compared the sense distributions of in and on, from which similarities and differences may be found between BNC and NCCU. The result of this analysis showed that in the sense categories of in, the distribution did not differ greatly since most of the data were categorized into the proto-scene sense in both BNC and NCCU, and only slight variations could be found in other categories. In the sense categories of on, in senses that refer to location, higher frequency could be found in NCCU, while more metaphorical constructions of on were identified in BNC. In the second analysis, the semantic feature analysis, the distribution of the semantic features are compared between literal and metaphorical constructions of in and on and between BNC and NCCU respectively. For the comparison of in and on, the statistical results showed that different types of nouns could be observed in the data of particular prepositions. Thus, for nouns in prepositions of different meanings, there may be significant differences in the semantic features identified in these nouns This result implied that the nouns surrounding a preposition might have some influences toward the meaning of this preposition. In the third analysis, the errors in the learner data were identified and examined. Based on the results, we proposed that in a prepositional phrase formed by a particular meaning of a preposition, there exist a range of semantic preferences shown in their co-text. The meanings of a preposition are influenced by the lexical words surrounding the preposition, rather than by the word that goes after the preposition (the ground). This thesis extends from the previous studies on the semantics of prepositions and includes the important linguistic elements in forming the prepositional phrases in the analysis. By incorporating a large amount of data, it also provides a systematic analysis and a more comprehensive explanation toward the complex semantics of English prepositions. The findings can be applied to the design of English teaching materials and techniques and may hopefully bring some insights to further preposition-related studies.

過去許多研究著眼於探討英文介系詞的語意，其中許多學者專注在單一介系詞的探討（e.g., Boers, 1996; Lindstromberg, 2010），其它學者則分析由動詞與介系詞組成之動詞片語、由介系詞與名詞組成之介系詞片語的語意（e.g. Larsen-Freeman & Celce-Murcia, 1999; Lindner, 1983; Quirk, Greenbaum, Leech, & Svartik, 1985）。過去這些研究大多在單一介系詞的框架下進行，鮮少有包含雙介系詞的句構的研究。本論文所研究之句構為：一個由動詞及UP/DOWN組成之動詞片語加上一個由IN與名詞組成之介系詞片語，在本句構中包含兩個連用之介系詞，本研究的分析包含雙介系詞的語意、動詞片語以及介系詞片語的語意，另外還包含此句構中所有語意的語意關連。 本研究採納並調整前人對介系詞、動詞片語以及介系詞片語的語意類別，以調整過的語意類別分析句構。研究結果發現在此句構中，雙介系詞大多含有隱喻概念，而大多的動詞片語則用來表達完成的動作語意，介系詞片語則多描繪空間概念或狀態。除此之外，我們發現此句構中的語意間有所關連，另外我們更發現UP和DOWN在本句構中並沒有完全對比的語意。 在本研究中，我們不同於以往研究只專注於一個介系詞或一個片語的分析，而是由三個角度切入探討一個含有雙介系詞的句構，未來期望能將本研究的結果運用在對比學習者對此句構的語言表現，並對介系詞的教學有更多貢獻。