探討閱讀時視覺與語言因素之研究指出眼球移動位置的決定主要受到低階視覺特徵的影響。有些研究也認為此一涉及計算眼跳目標的決策發生在執行眼球移動之前。為了檢視中文詞彙內的視覺複雜度分佈是否影響眼跳目標決定機制 、我們提出了視覺複雜度分佈指標 (visual complexity distribution index, VCD index) 來代表中文雙字詞內的複雜度分佈情形。依據視覺複雜度分佈指標 、本研究挑選出三組不同視覺複雜度分佈的詞彙 (左偏移、右偏移以及無偏移) 、並將這些詞彙箝入於句子中。紀錄中文讀者閱讀這些句子的眼球運動軌跡 、以比較三組實驗情境下的初次觸接凝視時間 (first-pass duration) 、落點位置 (landing position) 以及眼跳機率指標 (probability measures) 差異。使用線性混合模型 (Linear mixed model, LMM) 估計實驗組別效果 、以探討視覺複雜度分佈指標如何影響決定停留時間與眼跳位置的機制。結果發現右偏移組落點位置都落在其他兩組的右側 、而兩組偏移組的凝視時間都較不偏移組短。進一步分析顯示上述結果需在眼跳目標為中文詞彙才可觀察到。這指出由視覺複雜度分佈指標所反映出的中文雙字詞明視度型態 (luminance pattern) 、會在該詞彙被凝視之前影響眼跳位置的決定 、並依落點位置差異而調節了該詞彙被凝視時的處理。 / Previous studies about the visual and linguistic factors that influence the decision about where to move the eyes next in reading suggest a strong influence from low-level features; some studies also assume that this decision involves the computation of a saccade target before the oculomotor program is executed. In order to test whether the distribution of visual components within Chinese words influence the saccade targeting mechanism, we devised a new parameter that reflects the distribution of visual information along 2-character words’ area: the Visual Complexity Distribution (VCD) index. Three groups of words with a marked VCD index (i.e. Left-Bias, Right-Bias and Non-Bias) were identified and embedded in natural sentences; the eye movement of Chinese native speakers was recorded while they read this material in order to contrast first-pass duration, landing position and probability eye movement measures between conditions. The experimental effects were estimated through contrast between conditions using Linear Mixed Models, thus providing evidence about the VCD index’s influence on both, the decision about the when and where to move the eyes next. The analyses on initial fixation position indicate a rightwards shift when sending the eyes towards words with Right-Bias in comparison to the other conditions and shorter fixation durations when biased words are fixated in comparison to the Non-Bias words. Further analyses demonstrated that the results above can only be observed when specifying saccade targets from Chinese words. These results indicate that the luminance patterns within 2-character Chinese words, as reflected by the VCD index, can influence the specification of a saccade target when those words are about to be fixated as well as modulate the fovea load when those words are currently fixated.

文字的意義是怎樣的儲存於我們的記憶？又是透過怎樣的方式來了解文字中的意義呢？本研究藉由語義促發效果的探討，一方面檢視語義與連結促發作用對字彙辨識的影響，另一方面提供中文的辨識歷程及記憶表徵方面的訊息。 本研究在三個相關的實驗中，藉著操弄了各種獨變項來回答上述的問題，第一個獨變項是促發項與目標項的關係；例如，促發項與目標項配對不但會形成（1）雙字詞，且彼此具有相似的意義（如，酣—睡），（2）雙字詞，但彼此具有不同的意義（如，餅—乾），（3）促發項與目標項之間具有相似的意義成分，但不會形成雙字詞（如，扔—丟），以及（4）促發項與目標項之間沒有意義上的關聯。本研究操弄的第二個獨變項是相關配對佔所有刺激的百分比（50％與25％），第三個獨變項則是促發項與目標項之間的聯想強度（高聯想與低聯想）。 綜觀本三個實驗的結果發現（1）當促發項與目標項具有意義上的關聯時，有加速文字辨識的歷程，（2）在高相關配對百分比的情境中，不論高、低聯想強度，意義關聯的刺激配對有顯著的促發效果，（3）在低相關配對百分比的情境中，刺激配對形成雙字詞，但彼此具有不同的意義，則高、低聯想強度都有顯著的促發效果；至於刺激配對形成雙字詞，且彼此具有相似的意義以及刺激配對具有相似的意義成分，但不會形成雙字詞時，則只有高聯想強度有顯著的促發效果，但低聯想強度沒有顯著的促發效果。 這樣的結果顯示在字彙判斷作業中，當促發項與目標項只有意義相似性的成分時，並無法加速受試者對文字的辨識。至於當促發項與目標項形成雙字詞，但彼此具有不同的意義時，並不同於其他的刺激配對，其促發效果可能是藉由組合線索，觸及字彙層次的連結訊息，因而加快文字的辨識速度。

本研究以事件相關電位方法,探討中文字在進行聽覺詞彙辨識的作業中,分別受字形表徵的同音字密度多寡、以及字音-字形對應一致性的影響。首先 進行中文字的字音-字形一致性語料庫的建立,量化中文字的音形對應一致性程度。透過明確的定義並操弄中文字的音形對應一致性,以及中文字同音 字密度的特性,探討在進行中文字聽覺詞彙辨識歷程中,字形表徵屬性如何對語音辨識作業產生影響的認知歷程。實驗分為三種情境(1)同音密度低 (low HD)、(2)同音密度高/音形對應一致性高(high HD/high P-O)、 (3)同音密度高/音形對應一致性低(high HD/low P-O)。前兩項的比較為高同音字密度下的音形對應一致性效果,而後兩項的比較則為在高音形對應一致性下的同音字密度的效果。實驗一採語意判斷作業,研究結果顯示,在高同音字密度時,高音形對應一致性的字引發較大的 N400;而在高音形對應一致性時,同音字密度效果在 LPC 得到顯著的差異。反應中文字在聽 覺詞彙判斷作業上,字形可自動被激活,進而影響語音的辨識。實驗二採押 韻判斷作業,研究結果同樣發現高音形對應一致字引發較大的 N400,實但同音字密度效果性效果在判斷押韻作業上並未達顯著效果。另外,在高音形對應一致性情形下所得到的押韻效果最大,尤以同音字密度高且音形對應一致性高的情況下,押韻效果出現的時間較早。本研究結果支持雙向交互激發模型(BIAM)的假設:中文口語詞彙辨識的歷程中,也會自動激發字形訊息,而語音、字形、語意之間的對應一致性越好,口語詞彙辨識及其整體處理的效能也越高。

因應近年來數位典藏的趨勢日漸發展，越來越多珍貴中文歷史文本 選擇進行數保存，而保存的同時會面對文本的作者遺失或從缺，進而 影響文本的完整性，而本論文提出了一個適用於中文史料文本作者分 析的方法，主要是透過語言模型的建構，為每一位潛在的作者訓練出 一個專屬的語言模型，而搭配不同的平滑方法能避免掉某一受測文本 單詞出現的機率為零的機率進而造成計算上的錯誤，而本論文主要採 用改良式 Kneser–Ney 平滑方法，該平滑方法因其會同時考慮到 N 詞彙 語言模型的高低頻詞的影響，而使其成為建構語言模型普遍選擇的平 滑方式。 若僅將每一位潛在作者的所有文章進行合併訓練成單一的語言模型 會忽略掉許多特性，所以本篇論文在取得附有價值的歷史文本之外， 又加入後設資料 (Metadata) 進行綜合分析，包括人工標記的主題分類 的統計資訊，使建構出來的語言模型更適配受測文本，增加預測結果 的準確性。和加入額外的自定義的字詞以符合文本專有名詞的用詞習 慣，還會在一般建構語言模型的基礎上，加入長字詞的權重，以確定 字詞長度對預測準確度的關係。最後還會採用遞歸神經網路 (Recursive neural networks) 結合語言模型進行作者預測，與傳統的語言模型分析 作進一步的比較。 / In recent years, the trend of digital collections has been developing day by day, and more and more precious Chinese historical corpora have been selected for preservation. The preservation of the corpora at the same time will face the loss or lack of the authors, thus affecting the integrity of the corpora. A method for analyzing the author of the Chinese historical text is mainly through the construction of the language model, for each potential author to train a specific language model, and with a different smoothing method can be avoided zero probability of words and the error is caused by the calculation. This paper mainly adopts the Interpolated Modified Kneser-Ney smoothing method, which will take into account the influence of higher order and lower order n-grams string frequency. So, Interpolated Modified Kneser-Ney smoothing is become a very popular way to construct a general choice of language models. The combination of all the articles of each potential author into a single language model will ignore many of the features, so this paper in addition to the value of the historical corpora, but also to add the metadata to integrate analysis, including the statistical information of the subject matter classification of the artificial mark, so that the constructed language model is more suitable for the measured text, increase the accuracy of the forecast results, add additional custom words to match the language of the proper nouns, in addition. But also on the basis of the general construction language model, the weight of the long word to join, to determine the length of the word on the relationship between the accuracy of prediction. Finally, recursive neural networks language models are also used to predict the authors and to make further comparisons with the traditional language model analysis.

近年智慧型手機與網路的普及，使得社群網站與線上串流音樂蓬勃發展。臉書（Facebook）用戶截至去年止每月總體平均用戶高達18.6億人 ，粉絲專頁成為公司企業特別關注的行銷手段。粉絲專頁上的貼文能夠在短時間內經過點閱、分享傳播至用戶的頁面，達到比起電視廣告更佳的效果，也節省了許多的成本。本研究提供了一套針對臉書粉絲專頁貼文的分群流程，考量到貼文字詞的複雜性，除了抓取了臉書粉絲專頁的貼文外，也抓取了與其相關的KKBOX網頁資訊，整合KKBOX網頁中的資料，對中文斷詞系統（Jieba）的語料庫進行擴充，以提高斷詞的正確性，接著透過雙分群演算法（Minimum Squared Residue Co-Clustering Algorithm）對貼文進行分群，並利用鑑別率（Discrimination Rate）與凝聚率（Agglomerate Rate）配合主成份分析（Principal Component Analysis）所產生的分佈圖來對分群結果進行評估，選出較佳的分群結果進一步去分析，進而找出分類的根據。在結果中，發現本研究的方法能夠有效的區分出不同類型的貼文，甚至能夠依據使用字詞、語法或編排格式的不同來進行分群。 / In recent years, because both smartphones and the Internet have become more popular, social network sites and music streaming services have grown vigorously. The monthly average of Facebook users hit 1.86 billion last years and Facebook Fan Page has become a popular marketing tool. Posts on Facebook can be broadcasted to millions of people in a short period of time by LIKEing and SHAREing pages. Using Facebook Fan Page as a marketing tool is more effective than advertising on television and can definitely reduce the costs. This study presents a process to cluster posts on Facebook Fan Page. Considering the complicated word usage, we grasped information on Facebook Fan Page and related information on the KKBOX website. First, we integrated the information on the website of KKBOX and expanded the text corpus of Jibea to enhance the accuracy of word segmentation. Then, we clustered the posts into several groups through Minimum Squared Residue Co-Clustering Algorithm and used discrimination Rate and Agglomerate Rate to analyze the distribution chart of Principal Component Analysis. After that, we found the suitable classification and could further analyze it. How posts are classified can then be found. As a result, we found that the method of this study can effectively cluster different kinds of posts and even cluster these posts according to its words, syntax and arrangement.

京都大学 / 新制・論文博士 / 博士(文学) / 乙第13459号 / 論文博第657号 / 新制||文||710(附属図書館) / 京都大学大学院文学研究科中国語学中国文学専攻 / (主査)教授 木津 祐子, 准教授 緑川 英樹, 教授 道坂 昭廣, 教授 齋藤 希史 / 学位規則第4条第2項該当 / Doctor of Letters / Kyoto University / DGAM

University of Macau / Faculty of Education

評論是所有人類行為的核心，因為它影響我們行為的關鍵因素。我們都試著從不同型式的評論分析與研究試著從作者字裡行間的文字呈現內容深入推敲及理解，從而要能過濾出能協助決策的有用資訊。在早期的評論研究將評論視為是文本分類問題，直到2000年前後，從分析評論的主觀句子與評論裡形容詞的程度衡量用詞，學者們開始對解構整篇文本的內容，並試著從語言學的角度分析用字遣詞與情感方向之間的關聯。這種從文字語義關聯分析評論的方式，也使文本挖掘技術必需結合自然語言的處理原則，才能更準確地了解評論的內容。隨著許多新興的機器學習演算法與自然語言處理方法不斷地推陳出新，及網路使用行為拓展至電子商務與線上虛擬社群的建立，情感分析研究亦開始不斷地蓬勃發展。 漢文不同於世界其它語言，它擁有許多獨特表徵：無空格區隔、一字一語素、依詞為語言中表達意義的最小獨立單位，也使得在套用源自西方的情感分析原則時更加困難。然而過去的研究者則加以利用這些語言特徵，建立出專屬中文的情感分析原則。我們務實地討論適用於中文情感分析的情境(a)可取得情感分析資源及專家語言智慧，及(b)可取得領域字詞特徵向量定義的兩個前題下，提出適合的中文情感分析策略。在情境(a)中，我們深入討論運用部首資訊至情感分析中的適用性，並且提出一套能精萃出領域評論文本的觀測字詞/部首組的方法。研究中我們萃取出50個部首組，並運用在領域相近的評論裡得到很好的情感分類成效。而在情境(b)中我們提出適合深層類神經網路學習方法的評論字詞的權重過濾原則，不僅能確保評論字詞在學習過程中仍保有能積旋出合適屬性，並且驗證此權重原則在支援向量機的學習方式下亦有相同的優勢。在研究中，我們亦討論此兩種情境下進行情感分析的必要條件與資訊，並為未來更深入的中文情感分析起到墊腳石的作用。 / Opinion is the core of human behaviors, because it directly influences key factor of our behaviors. Despite of personal or organizational decision making processes, we all constantly conduct various kinds of opinion analysis, including explaining and comprehending what users present. At the beginning, opinion studies considered as a text mining problems, and tried to cluster opinions into positive and negative groups. After 2000, researchers intended to decompose sentences from whole opinions by analysing subjective expressing and adjective words presenting within, as well as explained the relationships between semantics and sentiment from linguistics aspect. Therefore, opinion analysis has to incorporate with natural language processing techniques, so we can understand the opinion contents. Nowadays, sentiment analysis grows event booming due to emerging machine learning and natural language processing approaches, as well as the needs of electronic commerce and virtual community on line. Unfortunately, Chinese is quite unlike other language due to non-space separated, one character as one morpheme, and considering words (compositing with several characters) as minimum semantic expression unit. And those language features also bring difficult to adopted sentiment analysis principles from English. Nevertheless, researchers leveraged Chinese language information to propose specific sentiment analysis approaches dedicated to analyze Chinese opinions. In this study, we practically discussed the situations of conducting sentiment analysis: (a) using sentiment analysis resources and experts’ knowledge; and (b) using word feature vector, called word2vec, and deep learning. In (a) scenario, we propose a Chinese radical-based sentiment analysis approach and experiment the applicability. We also proposed a feature extraction method, so we can generate 50 seeds for further analysis. In (b), we compared 4 different feature selection approaches for deep learning, in order to keep accuracy and make sure understandable feature can be generated in neural network. We also tested feature selection approaches in SVM classifier and retrieved similar results. In this study, we also discussed essential constraints and required information in both scenarios, as well as the results of this study can be the foundation of continuing Chinese sentiment analysis studies.

本論文探討的主題有四：（一）陳述中文文法手語與台灣自然手語各語言層級結構的異同，並探查兩套手語系統結構差異部分的語言溝通效率與語意清晰度；（二）瞭解受試者兩套手語系統的語言能力及其語言使用情形；（三）調取受試者對兩套手語系統的語言刻板印象與手語政策態度；（四）探討現階段手語政策的實施。 本論文包含量化與質化的研究方法，研究對象為年滿十八歲、居住在大台北地區、且以手語為主要使用語言之聽覺障礙者。量化研究包含手語結構評估問卷及手語使用情形與語言態度問卷。受試者須先完成結構評估測驗，才進行手語使用情形與語言態度問卷之填答。手語結構評估問卷針對兩套手語系統之迅速程度、模糊程度、與歧異程度加以測試。語言結構評估項目共有基本詞彙、詞組、時貌、副詞、簡單句型、複句句型、與篇章等七大類共一百九十四項，以影片方式呈現。受試者每觀看完一項評估項目，即立刻根據影片內容回答問卷上之題目。語言使用情形與語言態度問卷則探查受試者手語能力、手語使用、及語言態度。本論文以非機率抽樣的滾雪球抽樣方式進行量化問卷的發放，共回收75份有效樣本，進行無母數統計分析。質化研究以深入訪談方式進行，以立意抽樣方式共訪問六名受訪者，重點在探查量化問卷所發現的結果之原因。 研究結果顯示，在結構評估測驗中，除了詞組與簡單句型之遞繫句之外，自然手語的溝通效率與語意清晰度皆高於文法手語。受試者的兩套手語系統能力均等，且兩套手語使用頻率主要決定於談話對象的語言使用。語言刻板印象與手語政策態度的調查結果顯示，受試者對自然手語有較高的評價。 依據研究結果，本論文建議現階段手語政策應調整其語言規劃方針，將自然手語納入啟聰學校正式教學語言。對未來手語政策規劃之建議為：研訂相關法令、設立專職機構、擴充手語詞彙並編撰手語字典、明訂啟聰學校教師教學溝通政策與鑑定教師手語能力、培育專業手語翻譯人才、設置手語相關節目並提供無溝通障礙環境、獎勵手語研究與推廣工作、增設相關系所。 關鍵字：台灣自然手語、中文文法手語、語言態度、語言政策 / This thesis aims to (1) compares the language structure of Chinese Signed Language (CSL) with Taiwan Sign Language (TSL), and investigates the efficiency, vagueness, and ambiguity of these two language systems; (2) to provide a preliminary evaluation of the implementation of Sign Language Policy in Taiwan through an investigation of the deaf’s proficiency in CSL and TSL, their use of them, and their attitudes toward both these two linguistic systems and the related policy. Both quantitative and qualitative analyses are adopted. Quantitative analysis includes two tests. On “structure evaluation test,” the efficiency, vagueness, and ambiguity of CSL and TSL are investigated. Evaluated items are taped into 194 segments of films, distributed on 7 linguistic levels, including lexical items, phrases, tense, adverbs, simple sentences, complex sentences, and discourse. Subjects are required to answer the questionnaire immediately after each test item is shown. “Sign language use and attitude questionnaire” is to elicit the deaf’s language proficiency, language use, and language attitudes. 75 questionnaires are collected through nonprobability sampling and nonparamentric statistical test with all the subjects being deaf adults who live in Taipei area and use sign language for communication. For qualitative analysis, 6 informants were interviewed through judgemental samplings to interpret the results of the questionnaires. The results of statistic tests indicate that TSL is more efficient, less vague and less ambiguous than CSL. Moreover, the subjects’ proficiency in the two sign language systems are equally good and their frequency of language use are decided by their interlocutors. As to their attitudes toward the two sign languages and the related language policy, all the subjects show support to TSL. Based on the findings from the quantitative and qualitative analysis, suggestions are given as follows. Legislate the law. Establish a government institution, institute lexicology, set the instructional language of the deaf school and evaluate the proficiency of deaf school teachers. Train the interpreters and provide communication unimpediment environment. Investigate sign language research, and plant institution. Key words: Taiwan Sign Language, Chinese Signed Language, language attitude, language policy

本文透過操弄雙字詞詞首的語意(sense)多寡和左右視野，試圖探討中文雙字詞的語意表徵和左右大腦對於多意詞(polysemy)的處理機制。實驗一顯示的左右腦結果和Pylkkänen等人在2006年的MEG研究相似，也就是左腦的多意詞促進效果，支持多意詞單一表徵的型態；然而，右腦卻呈現多意詞抑制的效果。這樣的現象產生兩者可能解釋：(1) 右腦還是屬於單一語意表徵，但由於右半腦處理語意的特性，導致和左腦得到不同的結果；(2)右腦的結果是來自於右腦屬於語意多重表徵(separate entries)的因素。為了要釐清這些說法，實驗二進一步的改變作業深度，讓受試者做詞類判斷作業，企圖讓受試者進行比較深層的語意處理。實驗二結果顯示，在改變作業深度之後，我們的確得到右腦語意促進效果，所以證明右腦的語意屬於單一表徵，在比較深層作業處理階段，因為左右腦處理語意的特性，使得右腦有機會呈現實驗預期的結果。另外，在動詞、名詞事後分析的結果中，我們也發現動詞、名詞的語意效果在大腦有不同的分布區位。名詞的語意效果分布在大腦中間偏後的位置；動詞則是主要分布在大腦前額一帶 總結以上發現，本研究的發現支持過去學者所提出的多意詞單一表徵的說法；第二、本研究對左右半腦處理語意特性，也符合過去的假設，也就是左腦擅長主要、細微的辨識，右腦則擅長維持次要、普遍語意。第三、本研究額外的發現是，動詞、名詞的語意效果在大腦有不同的分布，意味著不同的詞類在大腦可能有不同的表徵。 / Acknowledgements …………………………………………………………iv Tables…………….……………………………………………………………ix Figures …………………………………………………………………………x Chinese Abstract …………………………………………………………xii English Abstract ………………………………………………………xiii CHAPTER 1. INTRODUCTION ……………………………………………………..……1 1.1 What are senses? Homonymy vs. Polysemy …………………….1 1.2 English words vs. Chinese compounds ………………………….3 1.3 Hemispheric processing of semantic ambiguity ……………4 2. REVIEW OF RELATED PSYCHOLINGUISTIC RESEARCH ………………6 2.1 Neighborhood size effect in English …………………………6 2.2 Neighborhood frequency effect …………………………….……9 2.3 Event-related potentials (ERPs) vs. neighborhood size effect....11 2.3.1 Event-related potentials ………………………………….11 2.3.2 The advantages of electrophysiological techniques …12 2.3.3 Language-related ERP components ……………………….…12 2.3.4 The neighborhood size effect and. ERPs ……………..14 2.4 Neighborhood size effect in Chinese ……………………….16 2.5 Lexical ambiguity in English—homonymy vs. polysemy……… 22 2.5.1 Mixed results of ambiguity effects ………………………23 2.5.2 Polysemy—separate entries or single entry? …………25 2.5.3Some evidence for single entry hypothesis of senses…27 2.6 Lexical ambiguity in Chinese …………………….……………26 2.7 Hemispheric asymmetry in lexicon processing ……………33 3. EXPERIMENT 1 ………………………………………………………………38 3.1 Experiment 1... ..……………………………………….….....39 3.1.1 Participants …………………………………………………………39 3.1.2 Materials ……………………………………………………………39 3.1.3 Procedure ……………………………………………………………40 3.2 EEG recording parameters …………………………………………41 3.3 EEG data analysis procedure …………………………….....42 3.4 Results ……………………………………………………………………43 3.4.1 Behavioral data of sense effect ……………………………43 3.4.2 Behavioral data of lexicality effect ……………………44 3.4.3 Event-related potentials ………………………………….…45 N170 (150- 180 ms) …………………………………………………46 Frontal P200 (220-260 ms) …………………………………….……47 N400 …………………………………………………………………48 3.5 Discussion ……………………………………………………………51 4. EXPERIMENT 2 ……………………………………………………………57 4.1 Experiment 2 …………………………………………………………58 4.1.1 Participants ………………………………………………………58 4.1.2 Materials …………………………………………………………58 4.1.3 Procedure ……………………………………………………………59 4.2 Results …………………………………………………………………60 4.2.1 Behavioral data ……………………………………………………60 4.2.2 ERP data ……………………………………………………………61 N170 (150-180 ms) ………………………………………………....62 Frontal P200 (220-260 ms) …………………………………………63 N400 (350-500 ms) …………………………………………………63 4.3 Discussion …………………………………………………………………….65 Nouns and verbs ………………………………………………………67 4.4 Re-analyses …………………………………………………………69 4.4.1 Behavioral data ……………………………………………………69 4.4.2 ERP data ……………………………………………………………....71 Nouns …………………………………………………………………71 Verbs …………………………………………………………………74 4.5 Discussion 2 ………………………………………………………77 5. GENERAL DISCUSSION AND CONCLUSIONS ………………………81 5.1 Separate entries or single entry? …………………………81 5.2 Hemispheric processing of polysemy in different depth of tasks ………....82 5.3 Nouns and verbs ………………………………………………………84 5.4 Conclusions …………………………………………………………….85 References ……………………………………………….……………………86 Appendixes ………………………………………………………….…….94 / The current study used the manipulation of visual field and the number of senses of the first character in Chinese disyllabic compounds to investigate the representation of senses and the hemispheric processing of semantic polysemy. The ERP results in experiment 1 revealed crossover patterns in the LH and RH, which resembled the MEG data in Pylkkänen et al.’s study (2006). The sense facilitation in the LH was in favor of the assumption of single entry representation for senses. However, the inhibition in the RH yielded two possible interpretations: (1) the nature of hemispheric processing in dealing with semantic ambiguity; (2) the semantic activation from the separate-entry representation for senses. To clarify these possibilities, the depth of the task was changed. Experiment 2 was designed to push subjects to a deeper level of lexical processing through the word class judgment task. The results revealed the sense facilitation effect in the RH and suggested that in a deeper level, the RH had more possibility to observe the sense facilitation due to different efficiency of cerebral hemispheres in dealing with ambiguity. By chance, planned comparisons of the sense effect in different word classes suggested different distributions of the sense effects for nouns and verbs. For nouns, the sense effects were located in central-to-parietal areas while for verbs, the sense effects mainly were from the frontal area. In sum, the current study was in support of the account of single entry representation for senses, which was consistent with previous findings proposed by Beretta et al. (2005), Pylkkänen et al. (2006), and Rodd et al. (2002). Second, the research demonstrated that cerebral hemispheres played a role in semantic activation in a complementary way in which the LH was engaged in fine and focused semantic coding while the RH was more sophisticated in coarse coding and maintaining alternate meanings (e. g. Beeman & Chiarello, 1998; Burgess and Simpson, 1988). When the depth of tasks was changed, the RH advantage for the processing of semantically related senses was observed. Third, different distributions of the sense effects for nouns and verbs implied the distinct representations for different parts of speech in the brain.