詞性及語意限制對詞彙歧義解困的影響：中文歧義詞處理的眼動研究 - 政大學術集成

全文

(1)國立政治大學語言學研究所碩士論文. National Chengchi University Graduate Institute of Linguistics Master Thesis. 指導教授：蔡介立 Advisor：Jie-Li Tsai. 立. 政治大. ‧ 國. 學. 詞性及語意限制對詞彙歧義解困的影響：中文歧義詞處理的眼動研究. ‧. n. al. er. io. sit. y. Nat. THE INFLUENCE OF SYNTACTIC CATEGORY AND SEMANTIC CONSTRAINTS ON LEXICAL AMBIGUITY RESOLUTION: AN EYE-MOVEMENT STUDY OF PROCESSING CHINESE HOMOGRAPHS. Ch. engchi. 研究生：陳柏亨. i n U. v. 撰. Student：Po-Heng Chen 中華民國一百零三年七月 July, 2014 0.

(2) THE INFLUENCE OF SYNTACTIC CATEGORY AND SEMANTIC CONSTRAINTS ON LEXICAL AMBIGUITY RESOLUTION: AN EYE-MOVEMENT STUDY OF PROCESSING CHINESE HOMOGRAPHS. 立. 政治大. ‧ 國. 學. BY. ‧. Po-Heng Chen. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. A Thesis Submitted to the. Graduate Institute of Linguistics in Partial Fulfillment of the Requirements for the Degree of Master of Arts. July 2014.

(3) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. Copyright © 2014 Po-Heng Chen All Rights Reserved iii. v.

(4) Acknowledgements. 終於，看似遙遙無期的這一天，意外地到來了。誠摯地感謝我的指導教授蔡介立老師，嚴謹的態度令人折服，是最佳典範。每每與老師討論研究的過程，總是獲益良多。適時的鼓勵，讓我總能堅持下去。感謝我的論文口試委員：李佳穎老師、黃瓊之老師、李佳霖老師。老師們願意抽空閱讀我的論文、細聽我的簡報、並給予我寶貴的建議，對我來說是何等榮幸！此外，也感謝曾與我討論的兩位國外學者：Kara D. Federmeier 及 Alec Marantz。. 政治大回首這三年在政大語言所的日子，豐富而精采。感謝語言所的每位老師從不立同的角度和觀點帶領我認識語言的風貌：徐嘉慧老師的語意學、何萬順老師的語. ‧ 國. 學. 法學、蕭宇超老師的音韻學、黃瓊之老師的兒童語言習得、萬依萍老師的語誤專題、賴惠玲老師的漢語語意學、蔡介立老師的眼動研究方法。. ‧. 感謝在政大遇見的每個人，讓我經驗這一切。感謝萬能的助教，當我在所辦值班的時候，經常分享好吃的點心給我。感謝語言所 100 級的夥伴，忍受我拿著錄音筆蒐集大家的語誤，你們的聲音總能讓我暫時舒緩撰寫論文的壓力。感謝 Henry，在我碩甄口試時讓我借住宿舍還陪考。感謝涵絜學姊，陪我一起分擔只有兩個人選修的語誤課 readings。感謝承翰，你是最貼心的司機。感謝子權，陪我吃飯、聊天、看電影，當你想找人說話的時候，請記得還有我願意傾聽！感謝 EMR Lab 的成員，這裡其實是我第二個家吧！感謝心理系的博班學長姊：熱心助人的家興學長、牛肉麵店小廚娘雅嵐學姊、陪我一起被文獻轟炸的婉雲學姊；感謝同為語言所的學姊們：在 meeting 上總是替我解危的怡璇、不時陪. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 我吃牛肉蓋飯的波乓瑪莉、熬夜陪我準備 PPT 的媛媜；感謝同鄉的宛柔學姊，教我進行實驗與分析資料；感謝善良的美慧學姊，幾次幫我借還師大的圖書；感謝語言所學妹翊倫，當我的 rating VIP；此外，特別感謝語言所同窗孟璋，是你讓我覺得自己並不孤單！感謝我的母校嘉義大學外語系，這裡是我踏上語言學研究的起點。特別要感謝兩位老師：倪碧華老師、龔書萍老師。感謝 Dr. Ni 當年的鼓勵與引薦，搭建了一座橋梁連接我的學士及碩士生涯。感謝書萍老師讓我參與中文篇章隱喻分析和幫忙執行實驗，堅定了我想用實驗方法來探討語言學議題的決心。同時，也感謝助理佩晏和我一起討論隱喻分析。感謝 DFL100 的大學同窗俊翰、政道、德川、阿唐、偉真，你們不時關心我在研究所的生活，讓我備感溫馨。最後，我要感謝我的母親董惠香女士。雖然妳搞不懂我在研究什麼，卻還是 iv.

(5) 默默支持著我，關心我在台北的生活。每當我回家看見您笑容的瞬間，那些讀文獻、跑分析的倦怠感隨即消散，馬上又充滿動力。此時此刻，我想說「我愛妳」。盼自己能時時刻刻記得現在這一點點的成就感！ 2014.07.30. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. v. i n U. v.

(6) Table of Contents. Acknowledgements ................................................................................................. iv List of Tables ........................................................................................................... ix List of Figures ......................................................................................................... xi Chinese Abstract .................................................................................................... xii English Abstract ................................................................................................... xiv Chapter 1 Introduction............................................................................................ 1 1.1 General background ................................................................................ 1 1.2 Research questions .................................................................................. 7. 立. 政治大. ‧ 國. 學. ‧. Chapter 2 Literature Review................................................................................... 7 2.1 Mechanisms of sentence processing......................................................... 8 2.1.1 Syntax-first models ..................................................................... 8 2.1.2 Constraint-satisfaction models .................................................. 12 2.2 Issues of lexical ambiguity resolution .................................................... 14 2.2.1 Meaning dominance .................................................................. 15 2.2.2 Syntactic category..................................................................... 17 2.2.3 Contextual constraints ............................................................... 26 2.2.3.1 Semantic constraint ............................................................. 26 2.2.3.2 Syntactic constraint ............................................................. 30 2.3 Implications of sentence processing models for lexical ambiguity resolution .............................................................................................. 35 2.4 Processing of Chinese ambiguous words ............................................... 37. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Chapter 3 Experiment One: Syntactic category and semantic constraint effects on lexical ambiguity resolution ........................................................... 41 3.1 Method .................................................................................................. 43 3.1.1 Participants ............................................................................... 43 3.1.2 Materials and design ................................................................. 44 3.1.2.1 Norming study 1: Meaning dominance ................................ 48 3.1.2.2 Norming study 2: Meaning relatedness ................................ 50 vi.

(7) 3.1.2.3 Norming study 3: Word predictability and syntactic category bias ..................................................................................... 52 3.1.2.4 Norming study 4: Semantic bias .......................................... 55 3.1.2.5 Norming study 5: Syntactic category judgment of target words ............................................................................................ 55 3.1.2.6 Norming study 6: Sentence plausibility ............................... 56 3.1.3 Apparatus ................................................................................. 56 3.1.4 Procedure.................................................................................. 57 3.2 Data analysis ......................................................................................... 59 3.3 Results .................................................................................................. 62 3.3.1 Target words ............................................................................. 62 3.3.1.1 Duration measures ............................................................... 62 3.3.1.2 Probability measures ........................................................... 64 3.3.2 Post-target words ...................................................................... 67 3.3.2.1 Duration measures ............................................................... 67 3.3.2.2 Probability measures ........................................................... 69 3.4 Discussion ............................................................................................. 73. 立. 政治大. ‧. ‧ 國. 學. y. Nat. sit. n. al. er. io. Chapter 4 Experiment Two: Lexical ambiguity resolution with syntactic category constraint ............................................................................................. 79 4.1 Method .................................................................................................. 81 4.1.1 Participants ............................................................................... 81 4.1.2 Materials and design ................................................................. 82 4.1.3 Norming studies of ambiguous words and sentential contexts ... 84 4.1.3.1 Norming study 1: Meaning dominance ................................ 85 4.1.3.2 Norming study 2: Meaning relatedness ................................ 86 4.1.3.3 Norming study 3: Word predictability and syntactic category bias ..................................................................................... 86 4.1.3.4 Norming study 4: Semantic bias .......................................... 87 4.1.3.5 Norming study 5: Syntactic category judgment of target words ............................................................................................ 89 4.1.3.6 Norming study 6: Disambiguation ....................................... 90 4.1.3.7 Norming study 7: Sentence plausibility ............................... 90 4.1.4 Apparatus ................................................................................. 91 4.1.5 Procedure.................................................................................. 91. Ch. engchi. vii. i n U. v.

(8) 4.1.6 Data analysis............................................................................. 91 4.2 Results .................................................................................................. 92 4.2.1 Target words ............................................................................. 92 4.2.1.1 Duration measures ............................................................... 92 4.2.1.2 Probability measures ........................................................... 95 4.2.2 Post-target words ...................................................................... 98 4.2.2.1 Duration measures ............................................................... 98 4.2.2.2 Probability measures ......................................................... 102 4.2.3 Disambiguating words ............................................................ 105 4.2.3.1 Duration measures ............................................................. 105 4.2.3.2 Probability measures ......................................................... 108 4.3 Discussion ............................................................................................111. 立. 政治大. ‧ 國. 學. ‧. Chapter 5 General Discussion ..............................................................................115 5.1 Syntax-first or Constraint-satisfaction .................................................. 117 5.2 Syntactic category constraint ............................................................... 119 5.3 Semantic constraint ............................................................................. 122 5.4 Inherent processing difficulty .............................................................. 124 5.5 Future research .................................................................................... 127. sit. y. Nat. n. al. er. io. References…………………………………………………………………………. 130 Appendixes ........................................................................................................... 140 A. Materials of Experiment 1 ................................................................... 140 B. Examples of the questionnaire in Norming study 1: Meaning dominance…. .................................................................................... ..145 C. Examples of the questionnaire in Norming study 2: Meaning relatedness……………………………………………………………... 146 D. Examples of the questionnaire in Norming study 3 to 6: Word predictability, syntactic category bias, semantic bias, syntactic category judgment of target words, and sentence plausibility ............................. 147 E. Materials of Experiment 2 ................................................................... 148. Ch. engchi. viii. i n U. v.

(9) List of Tables. Table 1. The summarized findings for homographic verbs from Pickering and Frisson (2001) ........................................................................................................ 21 Table 2. The summarized results of Experiment 1 from Folk and Morris (2003) ...... 33 Table 3. Examples of word materials and means (and standard deviations) of word properties for each condition in Experiment 1 ............................................ 45 Table 4. Examples of targets and sentences for each condition in Experiment 1 ....... 47 Table 5. Means (and standard deviations) of words and sentences in all norming studies for Experiment 1 ............................................................................ 48 Table 6. Experiment 1: Means and standard errors of the duration measures for all conditions on target words ......................................................................... 63 Table 7. Experiment 1: Statistic results of all effects for the duration measures on target words ............................................................................................... 64 Table 8. Experiment 1: Means and standard errors of the probability measures for all conditions on target words ......................................................................... 65 Table 9. Experiment 1: Statistic results of all effects for the probability measures on target words ............................................................................................... 66 Table 10. Experiment 1: Means and standard errors of the duration measures for all conditions on post-target words .................................................................. 68 Table 11. Experiment 1: Statistic results of all effects for the duration measures on post-target words........................................................................................ 69 Table 12. Experiment 1: Means and standard errors of the probability measures for all conditions on post-target words .................................................................. 72 Table 13. Experiment 1: Statistic results of all effects for the probability measures on post-target words........................................................................................ 73 Table 14. Means (and standard deviations) of word properties for each condition in Experiment 2 ............................................................................................. 83 Table 15. Examples of targets and sentences for each condition in Experiment 2 ..... 84 Table 16. Means (and standard deviations) of words and sentences in all norming studies for Experiment 2 ............................................................................ 85 Table 17. Experiment 2: Means and standard errors of the duration measures for all conditions on target words ......................................................................... 94. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. ix. i n U. v.

(10) Table 18. Experiment 2: Statistic results of all effects for the duration measures on target words ............................................................................................... 95 Table 19. Experiment 2: Means and standard errors of the probability measures for all conditions on target words ......................................................................... 97 Table 20. Experiment 2: Statistic results of all effects for the probability measures on target words ............................................................................................... 98 Table 21. Experiment 2: Means and standard errors of the duration measures for all conditions on post-target words ................................................................ 100 Table 22. Experiment 2: Statistic results of all effects for the duration measures on post-target words...................................................................................... 102 Table 23. Experiment 2: Means and standard errors of the probability measures for all conditions on post-target words ................................................................ 104 Table 24. Experiment 2: Statistic results of all effects for the probability measures on post-target words...................................................................................... 105 Table 25. Experiment 2: Means and standard errors of the duration measures for all conditions on disambiguating words ........................................................ 107 Table 26. Experiment 2: Statistic results of all effects for the duration measures on disambiguating words .............................................................................. 108 Table 27. Experiment 2: Means and standard errors of the probability measures for all conditions on disambiguating words ........................................................ 110 Table 28. Experiment 2: Statistic results of all effects for the probability measures on disambiguating words ...............................................................................111 Table 29. A comparison of means of first-pass duration measures between Experiment 1 and 2 ..................................................................................................... 123. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. x. i n U. v.

(11) List of Figures. Figure 1. A simplified diagram of Friederici’s auditory sentence processing model (adapted from Friederici (2002), Figure 1) ................................................. 11 Figure 2. A diagram of experimental procedure in Experiment 1 and 2 .................... 58. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. xi. i n U. v.

(12) 國. 立. 政. 治. 大. 學. 研. 究. 所. 碩. 士. 論. 文. 提. 要. 研究所別：語言學研究所. 論文名稱：詞性及語意限制對詞彙歧義解困的影響：中文歧義詞處理的眼. 動研究. 立. 指導教授：蔡介立. 政治大. ‧ 國. 學. 研究生：陳柏亨. sit. y. ‧. Nat. 論文提要內容：(共一冊，39901 字，分 5 章 18 節). er. io. 兩種語句處理模型曾被提出以解釋閱讀句子時語法及語意資訊的互動。句法. al. n. v i n 優先模型(syntax-first models)認為詞性判斷必定先於語意分析，而制約滿足模型 Ch engchi U (constraint-satisfaction models)則認為不同的資訊在語句理解的過程中會同時被處理。本研究檢驗兩種語句處理模型能否解釋語句中的詞彙歧義解困(lexical ambiguity resolution)。許多眼動研究曾發現詞彙歧義詞的次要語義偏向效應(subordinate bias effect)，顯示語意偏向次要語義的語境能加速激發歧義詞的次要語義並且產生語義競爭。然而，語境的語法在詞彙歧義解困中扮演的角色並不清楚。因而，不同. xii.

(13) 語義分屬不同詞性的詞性歧義詞便提供我們一個媒介以檢驗詞彙歧義解困中詞性及語意限制的互動。本研究的目的有二：(一) 檢驗詞性限制能否決定中文詞性歧義詞的語義解困；(二) 檢驗中文歧義詞語義的詞性是否會影響次要語義偏向效應。實驗一我. 政治大. 們將四種不同類型的中文非均勢同形異義詞置於語法及語意皆為次要語義偏向. 立. 的句子裡；實驗二則將四種不同類型的中文非均勢同形異義詞置於語法為次要語. ‧ 國. 學. 義偏向但語意中立的句子裡。受試者閱讀句子時的眼動表現會即時被記錄。. ‧. 實驗一的結果發現：(一) 四類型歧義詞的次要語義偏向效應只反映在目標. Nat. io. sit. y. 詞後區域的二次閱讀指標上；(二) 就效果量而言，NV 歧義詞的次要語義偏向. er. 效應在目標詞及目標詞後兩個區域都比 VN 歧義詞來得大。實驗二的結果則發現：. al. n. v i n Ch (一) VN 歧義詞的次要語義偏向效應從目標詞區的首次閱讀指標就出現，並且 engchi U 持續至目標詞及目標詞後兩個區域的二次閱讀指標；(二) 另外三類型歧義詞的次要語義偏向效應直到所有分析區域的二次閱讀指標才反映出來；(三) NV 歧. 義詞的次要語義偏向效應比 VN 歧義詞出現得更晚也更不明顯。整體而言，本研究的結果顯示詞性限制並不是影響中文詞性歧義詞語義解困的唯一因素。此結果支持制約滿足模型，並反對句法優先模型的預測。. xiii.

(14) Abstract. Two primary sentence processing models have been proposed to account for the interaction between syntactic and semantic information in reading sentences: Syntax-first models assume that syntactic-category assignment must precede semantic analysis, while constraint-satisfaction models propose that information from different. 政治大. sources is processed and weighed at the same time during sentence comprehension.. 立. The present study examined whether these sentence processing models, which assume. ‧ 國. 學. different contribution of syntactic category and semantic context, can explain the. ‧. resolution of lexical ambiguity in sentences.. Nat. io. sit. y. Several eye movement studies have demonstrated the subordinate bias effect. er. (SBE) for lexical-semantic ambiguous words (i.e., NN/VV homographs), indicating. al. n. v i n C h context can boost that a subordinate-biased semantic e n g c h i U the activation of the subordinate meaning of ambiguous words and causes meaning competition (Duffy, Morris, & Rayner, 1988). However, the role of syntactic context in lexical ambiguity resolution is less clear. Syntactic category ambiguous words (i.e., SCA words; VN/NV homographs), whose alternative meanings differ in syntactic category (e.g., watch in English), serve as a means of examining the interaction between syntactic category and semantic constraints during lexical ambiguity resolution. xiv.

(15) The purpose of the present study was twofold: (a) to examine whether the syntactic category constraint can determine the semantic resolution of Chinese SCA words, and (b) to investigate whether syntactic category of alternative meanings of Chinese homographs can influence the SBE during lexical ambiguity resolution. Four types of Chinese biased homographs (NN, VV, VN, and NV) were embedded into syntactically and semantically subordinate-biased sentences (Experiment 1) and into. 政治大. syntactically subordinate-biased but semantically neutral sentences (Experiment 2).. 立. Participants’ eye movements were recorded as they read each sentence.. ‧ 國. 學. In Experiment 1, the results showed: (1) The SBE for the four types of. ‧. homographs was significant only in the second-pass reading on the post-target words.. Nat. io. sit. y. (2) Numerically, the NV homographs revealed a larger effect size of SBE than VN. er. homographs on both target and post-target words. In Experiment 2, the results showed:. al. n. v i n C hfrom the first-passU reading on the target words and (1) The SBE for VN appeared engchi lasted to the second-pass reading on the target and post-target words. (2) The SBE for. the other types of homographs did not occur until the second-pass reading in all analyzed regions. (3) The SBE for NV occurred much later and less obviously than that for VN. In general, our findings support the constraint-satisfaction models and reject the prediction of the syntax-first models, suggesting that the syntactic category constraint is not the only factor influencing the semantic resolution of SCA words. xv.

(16) Chapter 1 Introduction 1.1 General background Human comprehend countless auditory and written sentences in daily life. To successfully understand the meaning of sentences, people need to integrate plenty of information (e.g., syntactic, semantic, pragmatic information, etc.) provided in the sentences. Thus, it arouses the interests of psycholinguists and neuropsychologists in. 治政 the way information from different sources is integrated 大 in human’s mind and brain. 立 ‧ 國. 學. More specifically, how syntactic and semantic information interact during sentence comprehension is one of the hot issues.. ‧. Two distinct classes of sentence processing models have been proposed to account. sit. y. Nat. for the interaction between syntactic and semantic information—syntax-first. er. io. n. a l and constraint-satisfaction i vmodels. The syntax-first n Ch U engchi (garden-path) models assume that syntactic and semantic information is processed (garden-path). models. serially (Frazier & Rayner, 1987; Friederici, 2002; Friederici, Hahne, & Mecklinger, 1996; Rayner, Carlson, & Frazier, 1983). Syntactic-category assignment must precede semantic analysis. In contrast, the constraint-satisfaction models propose that semantic and syntactic information are processed in parallel (Macdonald, 1993; Macdonald, Pearlmutter, & Seidenberg, 1994; Trueswell & Tanenhaus, 1994; Trueswell, Tanenhaus, & Garnsey, 1994; Trueswell, Tanenhaus, & Kello, 1993). 1.

(17) 2. Different sources of information are processed and weighed at the same time during sentence comprehension. Previous neurolinguistic research on Indo-European languages (e.g., German and French) mostly supported the syntax-first models (Friederici, 2002; Friederici, Gunter, Hahne, & Mauth, 2004; Friederici et al., 1996; Friederici, Steinhauer, & Frisch, 1999; Hahne & Friederici, 2002; Isel, Hahne, Maess, & Friederici, 2007). However, the functional primacy of syntactic category. 政治大. assignment was not obviously found in Chinese studies (Li, 1998; Wang, Mo, Xiang,. 立. Xu, & Chen, 2013; Yu & Zhang, 2008; Zhang et al., 2013; Zhang, Yu, & Boland,. ‧ 國. 學. 2010). The discrepancy might result from the unequal transparency of syntactic. ‧. information manifested in the Indo-European languages and in Chinese (Zhang et al.,. Nat. sit. semantic. information. n. al. contextual. comprehension.. Ch. engchi. during. er. and. io. syntactic. y. 2010). Thus, it is of theoretical significance to investigate the interaction between. i n U. v. Chinese. sentence. Ambiguity resolution can provide an opportunity to examine these sentence processing models. Most literature is established by investigating how people resolve syntactic ambiguity (e.g., The horse raced past the barn fell.) in sentences. Nevertheless, whether these sentence processing models can explain the resolution of lexical ambiguity (e.g., bank in English) in sentences should be examined. Lexical ambiguity is a robust phenomenon of language, in which multiple.

(18) 3. meanings can be expressed through a single linguistic form (either visual or acoustic). Homographic words can be roughly subdivided into two types, homonymy and polysemy (Lyons, 1977). Homonymy contains multiple meanings which are etymologically or semantically unrelated, such as bank (river bank/financial bank), while polysemy involves multiple senses which are etymologically or semantically related, such as hook (fish hook/cloth hook). Recent psycholinguistic and. 政治大. neurolinguistic studies have demonstrated that homonymy and polysemy are. 立. psychological distinct and being processed in different ways (Beretta, Fiorentino, &. ‧ 國. 學. Poeppel, 2005; Frazier & Rayner, 1990; Klepousniotou, 2002; Klepousniotou &. ‧. Baum, 2007; Klepousniotou, Pike, Steinhauer, & Gracco, 2012; Klepousniotou,. Nat. io. sit. y. Titone, & Romero, 2008; Pickering & Frisson, 2001; Rodd, Gaskell, &. n. al. different representations,. er. Marslen-Wilson, 2002). Multiple meanings of homonymy are exclusive and stored as. v i n C h multiple sensesU of polysemy while engchi. are not mutually. exclusive and may share a core representation. To avoid lumping all things together, the present study primarily focuses on homonymy, also called homophonic homographs. Over the past three decades lexical ambiguity resolution has been one of the hot issues in psycholinguistics and neurolinguistics. Researchers have been interested in whether one or multiple meanings are activated when an ambiguous word is.

(19) 4. processed. A lot of studies have shown that meaning dominance (i.e., frequency of use) may influence the activation level of alternative meanings of a homograph. The dominant meaning (i.e., high-frequency meaning) is easier to reach a high activation level than the subordinate meaning (i.e., low-frequency meaning) (Burgess & Simpson, 1988; Hogaboam & Perfetti, 1975; Simpson, 1981; Simpson & Burgess, 1985). When the alternative meanings are equally frequent, they reach a high. 政治大. activation level simultaneously and compete with each other (Duffy, Morris, &. 立. 1982; S. C. Sereno, Pacht, & Rayner, 1992).. 學. ‧ 國. Rayner, 1988; Rayner & Duffy, 1986; Seidenberg, Tanenhaus, Leiman, & Bienkowski,. ‧. However, the meaning dominance is not the only factor which can influence. Nat. io. sit. y. lexical ambiguity resolution. An increasing number of studies have been concerned. al. er. with the role of context, especially semantic context, in resolving ambiguous words.. n. v i n There are two views: one isC selective view, and the other is exhaustive access h e naccess gchi U. view. Selective access view assumes that only the context-appropriate meaning is activated (Glucksberg, Kreuz, & Rho, 1986; Schvaneveldt, Meyer, & Becker, 1976; Simpson, 1981; Simpson & Krueger, 1991; Tabossi, 1988; Tabossi, Colombo, & Job, 1987; Tabossi & Zardon, 1993), while exhaustive access view proposes that multiple meanings are activated initially regardless of contextual bias (Onifer & Swinney, 1981; Seidenberg et al., 1982; Swaab, Brown, & Hagoort, 2003; Swinney, 1979; Tanenhaus.

(20) 5. & Donnenwerth-Nolan, 1984; Tanenhaus, Leiman, & Seidenberg, 1979). Recent eye movement studies have consistently found a subordinate bias effect (SBE) for biased homographs (i.e., the frequency of alternative meanings is not equal) (Duffy, Kambe, & Rayner, 2001; Duffy et al., 1988; Pacht & Rayner, 1993; Rayner & Duffy, 1986; Rayner & Frazier, 1989; Rayner, Pacht, & Duffy, 1994; 盧怡璇, 2012). For example, Duffy et al. (1988) observed longer fixation durations on biased homographs than on. 政治大. frequency-matched unambiguous words when the preceding context was biased. 立. toward the subordinate meaning of the homographs. In order to explain the SBE, the. ‧ 國. 學. reordered access model (Duffy et al., 2001; Duffy et al., 1988) was proposed to. ‧. highlight the early impact of semantic context on the activation of multiple meanings.. Nat. io. sit. y. The subordinate-biased context boosts the activation of the subordinate meaning and. al. er. leads to the competition between the alternative meanings.. n. v i n C h and context,Usyntactic category is another factor In addition to meaning dominance engchi. influencing the processing of ambiguous words. Pickering and Frisson (2001) conducted an eye-movement study to investigate the SBE of biased homographic verbs and found there was no SBE in the target region. Thus, they suggested the resolution of biased homographic verbs is not as fast as that of biased homographic nouns. This study implies the necessity of distinguishing different types of homographs based on syntactic category in investigating lexical ambiguity resolution..

(21) 6. Nevertheless, there is a lack of experimental data for examining the influence of syntactic context on lexical ambiguity resolution, since most literature on lexical ambiguity resolution has mainly focused on those ambiguous words whose component meanings share a single syntactic category, (i.e., lexical-semantic ambiguity), such as bank. In fact, there are some ambiguous words whose component meanings cross different syntactic categories, called syntactic category ambiguous. 政治大. words (SCA words), such as watch (a clock/to look). Investigation of SCA words can. 立. shed light on functional primacy of syntactic constraint from the preceding context.. ‧ 國. 學. For example, Folk and Morris (2003) observed an absence of the SBE for SCA words. ‧. in sentences which are semantically- and syntactically-biased toward the subordinate. Nat. io. sit. y. meaning. The observation was considered the evidence that syntactic category. al. v i n C hthe analysis of syntactic in which e n g c h i U information is assumed to. n syntax-first models,. er. constraint can mediate the semantic ambiguity resolution. This result can fit into the. precede the analysis of semantic information. However, an alternative explanation of the absence of the SBE for SCA words is that the inherent difficulty of verb meaning limits the speed of activation and delays the meaning competition. Therefore, the aim of the present study is to systematically investigate the influence of syntactic category and semantic constraints on the resolution of Chinese lexical-semantic and syntactic category ambiguous words. Our ultimate goal is to see whether the syntax-first.

(22) 7. models or the constraint-satisfaction models can explain lexical ambiguity resolution in sentences.. 1.2 Research questions In the present study, we conducted two eye movement experiments using the normal reading task to investigate the influence of syntactic category and semantic. 政治大. constraints on Chinese lexical ambiguity resolution. In Experiment 1, four types of. 立. disyllabic homographs (NN, VV, VN and NV) 1 were embedded into sentences. ‧ 國. 學. involving syntactic category and semantic biases toward the subordinate meaning,. ‧. while in Experiment 2 only the syntactic category constraints preceded the. Nat. io. sit. y. homographs. Each homograph was assigned a frequency-matched unambiguous word. n. al. er. as control, which can fit into the same sentence frame. Specific research questions are listed as follows:. Ch. engchi. i n U. v. (1) Would syntactic category constraints determine the semantic resolution of Chinese SCA words? The syntax-first (garden-path) models predict the absence of the SBE, while the constraint-satisfaction models predict the presence of the SBE. (2) Would syntactic category of alternative meanings of Chinese homographs influence the SBE during lexical ambiguity resolution? 1. NN, VV, VN and NV are used as marks for the four types of homographs. The first and the second letter represent the syntactic category of the dominant and the subordinate meaning, respectively. For example, VN represents the homograph whose dominant meaning is a verb and subordinate meaning is a noun..

(23) Chapter 2 Literature Review. 2.1 Mechanisms of sentence processing The primary task of comprehending a sentence is to integrate plenty of information from different sources (e.g., syntactic structure, word semantics,. 政治大. real-world experience, etc.). Some researchers have been debating how syntactic and. 立. semantic information interact during sentence comprehension. With different. ‧ 國. 學. assumptions, two primary classes of models were proposed: syntax-first models and. ‧. constraint-satisfaction models. The following sections introduce the two models and. Nat. n. al. er. io. sit. y. review some empirical evidence.. 2.1.1 Syntax-first models. Ch. engchi. i n U. v. The most influential model of syntax-first accounts is the garden-path model (Frazier, 1979, 1987, 1989; Frazier & Rayner, 1982; Rayner et al., 1983), which was originally proposed to explain the resolution of syntactic ambiguity. In this model, sentence comprehension is assumed to engage two serial processing stages. At the first stage, sentence comprehension relies on construction of the simplest sentence structure, which is based on the syntactic category information. More importantly, the 8.

(24) 9. mechanism of syntactic processing is modular, impervious to non-syntactic contextual information (e.g., lexical-semantic, pragmatic information). Therefore, the language processor considers only available structural information to generate a single candidate structure. At the second stage, non-syntactic information is taken into consideration. If the initial candidate structure turns out to be inconsistent with non-syntactic information that is processed during the second stage, reanalysis occurs. 政治大. and leads to processing difficulty.. 立. Rayner et al. (1983) conducted an eye-tracking experiment and supported the. ‧ 國. 學. idea that semantic and pragmatic information cannot influence the initial syntactic. ‧. analysis during sentence comprehension. In their experiment, they manipulated the. Nat. io. sit. y. relative likelihood of possible real world events (i.e., the plausibility of sentences) and. n. al. er. the appearance of relative pronoun, constructing four types of sentences: (1) reduced. Ch. engchi. implausible: The dealer sold the car wasn’t. iv n sure U that. it was safe.; (2) reduced. plausible: The teenager sold the car wasn’t sure that it was safe.; (3) unreduced plausible: The teenager who was sold the car wasn’t sure that it was safe.; and (4) active implausible: The teenager sold the car but wasn’t sure that it was safe.. In terms of the total reading time (per character) on whole sentence, they found no differences between the reduced implausible and the reduced plausible sentences, while both types of reduced sentences were read longer than were the unreduced.

(25) 10. plausible and active implausible sentences. In addition, the reading times on the unreduced plausible sentence were longer than on the active implausible sentences. They further discovered that the longer reading times on both types of reduced sentences resulted from their longer reading times in the disambiguating region (e.g., wasn’t sure that it was safe), which indicated that readers initially constructed the simplest structure (i.e., the active sentence structure) and encountered processing. 政治大. difficulty when realizing the sentence was a reduced relative-clause sentence. And the. 立. lack of increased reading times on the implausible sentences suggested that the. ‧ 國. 學. plausibility did not influence the initial syntactic analysis. In sum, the study of Rayner. ‧. et al. (1983) supported the idea that during sentence comprehension the initial. Nat. io. sit. y. syntactic analysis was independent and impervious to semantic analysis.. al. er. Recently, Friederici et al. (1996) and Friederici (1995, 2002) also proposed a. n. v i n C h processing on the neurocognitive model of sentence e n g c h i U basis of electrophysiological data. In this model, three phases are proposed to specify the time course of syntactic and semantic processes (see Figure 1). During phase 1, syntactic structure is constructed based on syntactic category information. Violation of syntactic category information usually elicits an early left-anterior negativity (ELAN) around 100-300 ms after target word onset (Friederici, Pfeifer, & Hahne, 1993; Hahne & Friederici, 1999; Hahne & Jescheniak, 2001). During phase 2, morphosyntactic and semantic information is.

(26) 11. processed. Words that are semantically inconsistent with the prior context usually elicit a larger negative brain potential (N400) (300-500 ms), peaking around 400 ms (Kutas & Federmeier, 2000; Kutas & Hillyard, 1984; Neville, Nicol, Barss, Forster, & Garrett, 1991). Finally, phase 3 represents integration and reanalysis of various types of information. Both syntactically anomalous and garden-path sentences usually elicit a positive brain potential (P600), peaking around 600 ms (Hagoort, Brown, &. 政治大. Groothusen, 1993; Osterhout & Holcomb, 1992; Osterhout, Holcomb, & Swinney,. 立. 1994). Interestingly, when a word violates both syntactic category and semantic. ‧ 國. 學. constraints in a sentence, only an ELAN but no N400 appears (Friederici et al., 1999).. ‧. This finding supports syntax-first models, in which syntactic analysis is assumed to. Nat. n. al. er. io. sit. y. precede semantic analysis.. Ch. engchi. i n U. v. Figure 1. A simplified diagram of Friederici’s auditory sentence processing model (adapted from Friederici (2002), Figure 1).

(27) 12. 2.1.2 Constraint-satisfaction models An alternative class of models, constraint-satisfaction models, is proposed to challenge syntax-first models in the 1990s (Macdonald, 1993; Macdonald et al., 1994; Macdonald & Seidenberg, 2006; Trueswell & Tanenhaus, 1994; Trueswell et al., 1994; Trueswell et al., 1993). According to constraint-satisfaction models, two stages of sentence comprehension are unnecessary. Sentence processing engages one. 政治大. mechanism, in which all sources of information are available to interact among one. 立. another. Non-syntactic constraints need not wait until a second stage to influence. ‧ 國. 學. ambiguity resolution. One of the important assumptions in constraint-satisfaction. ‧. models is that multiple candidates are activated initially and weighted by probabilistic. Nat. io. sit. y. constraints (e.g., frequency, plausibility, etc.). When the probabilistic constraints are. al. er. strongly consistent with one analysis, processing is easy because no selection is. n. v i n needed. In contrast, when C multiple are equally supported by constraints, h e nanalyses gchi U processing difficulty takes place due to the competition between multiple candidates. Trueswell et al. (1994) conducted an eye movement experiment and suggested. that the lexical-semantic information (e.g., thematic-role) can be used by readers immediately during comprehending relative clauses. They manipulated the animacy of subject noun (animate: defendant vs. inanimate: evidence), the relative clause type (reduced vs. unreduced), and the type of verb in the relative clause (ambiguous.

(28) 13. regular verb: examined vs. unambiguous irregular verb: drawn). Comparing reducedwith unreduced-relative clauses with animate subject nouns (e.g., reduced: The defendant examined by the lawyer turned out to be unreliable.; unreduced: The defendant that was examined by the lawyer turned out to be unreliable.), they found longer first-pass reading times for reduced-relative clauses in the disambiguating region and longer second-pass reading times both in the verb and the disambiguating. 政治大. regions. These results indicated that readers began to encounter processing difficulty. 立. because they mistook the verb as a main verb initially. For inanimate subject nouns,. ‧ 國. 學. neither first-pass nor second-pass reading times showed significant differences. ‧. between reduced- and unreduced-relative clauses in any region. Thus, there seemed. Nat. io. sit. y. no processing difficulty with the inanimate nouns. Nevertheless, Trueswell et al.. al. er. discovered that their inanimate stimuli varied in patient-typicality (i.e., whether the. n. v i n C hwith regard to theUverb following). For example, their noun is a good patient or theme engchi norming data of patient-typicality indicated the textbook－loved pair obtained a low. score, resulting in weak semantic fit, while the evidence－examined pair received a high score, resulting in strong semantic fit. Further analyses of first-pass reading times revealed that the lack of processing difficulty remained only in the case of strong semantic fit. In the case of weak semantic fit, reduced-relative clauses required longer first-pass reading times than did the unreduced both in the verb and the.

(29) 14. disambiguating regions. Therefore, the study of Trueswell et al. (1994) suggested the influence of semantic information in an early stage of sentence comprehension. In sum, the two primary models of sentence processing have been proposed to elucidate how syntactic and semantic information interact during sentence comprehension. Nevertheless, most evidence comes from the findings of syntactic ambiguity resolution. It remains unclear whether these sentence processing models. 政治大. can explain the resolution of lexical ambiguity. In the next section, some crucial. 立. issues and findings of lexical ambiguity resolution are reviewed first, and then some. ‧ 國. 學. implications of distinct sentence processing models for lexical ambiguity resolution. ‧. are put forward.. n. al. Over the past three. er. io. sit. y. Nat 2.2 Issues of lexical ambiguity resolution. v i n C h psycholinguists decades, e n g c h i U and. neurolinguists have been. interested in lexical ambiguity resolution by using various methodologies, such as cross-modal priming, eye-tracking and event-related potentials (ERPs) methods. A primary issue concerns whether multiple meanings of an ambiguous word are activated at the same time. Previous research has demonstrated that meaning dominance, syntactic category and contextual constraints have an impact on semantic resolution of ambiguous words..

(30) 15. 2.2.1 Meaning dominance Meaning dominance was defined as the relative frequency of usage of alternative meanings involved in an ambiguous word. It can be assessed by using the free-association technique, in which subjects write down the first meaning that came to mind when given the ambiguous word in isolation (Gawlickgrendell & Woltz, 1994; Hogaboam & Perfetti, 1975). Based on meaning dominance, two types of homographs. 政治大. can be differentiated: balanced homographs and biased homographs. Balanced. 立. homographs have multiple meanings whose frequencies of usage are equal, while. ‧ 國. 學. biased homographs have one dominant (high-frequency) meaning and other. ‧. subordinate (low-frequency) meanings. For example, “同志” is a balanced Chinese. Nat. io. sit. y. homograph with two equally-frequent meanings (comrade/homosexual), and “儀表”is. n. al. er. a biased Chinese homograph with a dominant meaning (appearance) and a. Ch. engchi. subordinate meaning (instrument panel).. i n U. v. Previous research has demonstrated that meaning dominance would influence the activations of alternative meanings of homographs (Carpenter & Daneman, 1981; Duffy et al., 1988; Rayner & Duffy, 1986). In a neutral context (i.e., the context where no disambiguating information is provided), the alternative meanings of a balanced homograph were activated simultaneously (Rayner & Duffy, 1986; Swinney, 1979), while the dominant meaning of a biased homograph was retrieved prior to the.

(31) 16. subordinate meaning (Hogaboam & Perfetti, 1975; Simpson, 1981). Swinney (1979) investigated the semantic resolution of balanced homographs by using a cross-modal priming paradigm with a lexical decision task. In the cross-modal priming paradigm, subjects listened to sentences containing lexical ambiguities (e.g., bugs) followed by a visually-presented. target. (e.g.,. contextually-appropriate:. ant;. contextually-inappropriate: spy; unrelated: sew). Their task was to determine whether. 政治大. the visually-presented target formed a word or not. The results demonstrated. 立. facilitated lexical decisions for the visual targets immediately following the end of the. ‧ 國. 學. homographs, no matter when the target was related to the contextually-intended. ‧. meaning or the contextually-unintended meaning of the ambiguity. Also using the. Nat. io. sit. y. lexical decision task, Simpson (1981) examined meaning activation of biased. al. v i n immediatelyC followed two types h e n g by chi U n. which were. er. homographs in Experiment 1 by presenting the homographs in isolation (e.g., bank), of target words (e.g.,. dominant-related: money; subordinate-related: river). In addition to the prime-target related pairs, prime-target unrelated pairs (e.g., prime: calf; dominant-unrelated targets: money; subordinate-unrelated target: river) were used as controls. Compared to the unrelated controls, only when the target word was related to the dominant meaning was the lexical decision latency facilitated. This study supported the ordered access model of lexical ambiguity resolution (Hogaboam & Perfetti, 1975),.

(32) 17. suggesting that biased homographs were resolved on the basis of the relative frequency of the meanings when there was no disambiguating context. The retrieval of the dominant meaning was faster than that of the subordinate meaning.. 2.2.2 Syntactic category Words from different syntactic categories are linguistically, psychologically and. 政治大. neurologically distinct. Some electrophysiological and imaging studies have reported. 立. distinct brain responses (ERPs / activation areas) to nouns and verbs (Federmeier,. ‧ 國. 學. Segal, Lombrozo, & Kutas, 2000; Liu, Hua, & Weekes, 2007; Rösler, Streb, & Haan,. ‧. 2001; Yokoyama et al., 2006). For example, Federmeier et al. (2000) found ERPs. Nat. io. sit. y. elicited by unambiguous nouns were more negative than those elicited by. al. er. unambiguous verbs between 250 and 450 ms. In addition, unambiguous verbs. n. v i n C h context elicitedUa left-lateralized frontal positivity preceded by appropriate syntactic engchi which was not found for unambiguous nouns. Rösler et al. (2001) also discovered that N400 elicited by nouns had a larger amplitude than that elicited by verbs, and that reaction times were shorter to noun targets than to verb targets in their primed lexical decision task. These findings seem to suggest that nouns and verbs are processed differently in the brain. As reviewed by Vigliocco, Vinson, Druks, Barber, and Cappa (2011), it may take.

(33) 18. more efforts to process verbs than to process nouns in sentences. Semantically, typical nouns often refer to objects, while verbs are usually used to express actions where participants need to integrate several events. Syntactically, verbs need to assign thematic roles (e.g., agent, patient, theme, etc.) to other words (often nouns) in sentences. Morphologically, verbs are more complex and have more inflections than nouns, especially in the Indo-European languages (e.g., German, French, etc.).. 政治大. In terms of behavioral differences, a lot of empirical research has reported a. 立. processing advantage of nouns over verbs, i.e., nouns were processed faster than verbs.. ‧ 國. 學. This processing advantage of nouns were repeatedly replicated in different. ‧. experimental tasks, such as lexical decision (Rösler et al., 2001; J. A. Sereno, 1999; J.. Nat. io. sit. y. A. Sereno & Jongman, 1997), noun/verb categorization (J. A. Sereno, 1999), and. al. er. semantic categorization (Tyler, Russell, Fadili, & Moss, 2001). For example, J. A.. n. v i n Sereno and Jongman (1997)Ccontrolled frequency of occurrence and number h e n goverall chi U of letters for unambiguous pure nouns and unambiguous pure verbs in a lexical decision task and found the latencies to nouns were significantly shorter than those to verbs whenever the stimuli were monosyllabic or disyllabic. Their results indicated that processing verbs was more demanding than processing nouns, which they attributed to differences in inflection structure between nouns and verbs. According to them, in English nouns were used with its bare form more frequently than verbs, so.

(34) 19. the responses to nouns were faster than to verbs when the bare form was presented. Moreover, J. A. Sereno (1999) also replicated the processing advantage of nouns both in a noun/verb categorization task and a lexical decision task. In their experiments, nouns and verbs with either high- or low-frequency were presented to either the left or right visual fields of subjects. Both tasks revealed very similar results: the responses to nouns were significantly faster than those to verbs; however, in the noun/verb. 政治大. categorization task, the advantage of nouns vanished when the stimuli had. 立. low-frequency. In addition, the hemispheric difference was shown only for verbs: the. ‧ 國. 學. responses to verbs were significantly faster in the case of right visual field compared. ‧. to the case of left visual field.. Nat. io. sit. y. Other research investigated reasons for the processing difference between nouns. al. er. and verbs. For instance, Kauschke and Stenneken (2008) used two types of German. n. v i n C h objects) and verbs nouns (man-made vs. biological e n g c h i U (transitive vs. intransitive verbs). with their bare form in a visual lexical decision task. In addition to the processing advantage of nouns over verbs, their results also showed that the responses to intransitive verbs were significantly faster than those to transitive verbs, indicating the influence of syntactic factor since transitive and intransitive verbs differ in their argument structure. In a second experiment, they used inflected word forms by adding plural-suffixes and personal-suffixes to nouns and verbs respectively, constructing.

(35) 20. three. subsets. (e.g.,. NELK-EN—RUPFEN:. pinks—(to/we/they). pick;. KLUB-S—SIEG-T: clubs—(he/she) wins; HENGST-E—HÜPF-E: stallions—(I) bounce). Their reaction time data again demonstrated the noun advantage in all subsets, suggesting that the noun-verb discrepancies in processing cannot be attributed to word form and morphological complexity. In addition, from the perspective of language acquisition, the evidence that verbs. 政治大. are acquired later than nouns may reflect the psychological complexity of verbs. 立. (Gentner, 1982, 2006). Evidence from eye movement showed that the main verb in. ‧ 國. 學. simple active sentences received longer fixation durations than other grammatical. ‧. elements (e.g., function word, the subject, the object, etc.), indicating readers paid. Nat. io. sit. y. more visual attention to the main verb in sentences (Rayner, 1977). Therefore, it is. n. al. er. reasonable to consider more processing loads for verbs than for nouns during sentence comprehension.. Ch. engchi. i n U. v. Actually, the larger difficulties of processing verbs compared to nouns also influence the resolution of ambiguous words. Pickering and Frisson (2001) conducted eye-tracking experiments to investigate the processing of verbs, including verbs with multiple meanings (homographs), verbs with multiple senses (polysemy), and unambiguous verbs. The two types of ambiguous verbs were biased in terms of the frequencies of alternative interpretations. They manipulated whether preceding.

(36) 21. context or succeeding context containing disambiguating information and whether dominant or subordinate interpretation was supported by the disambiguating information. Thus, in the case of ambiguous verbs, four conditions were formed: (a) supportive-dominant, (b) supportive-subordinate, (c) neutral-dominant, and (d) neutral-subordinate. For unambiguous verbs, the presence of disambiguating information in preceding context was the only variable (supportive vs. neutral). For. 政治大. the homographic verbs, they found no context and meaning frequency effects in the. 立. measures of initial processing (i.e., first-pass time and first-pass regressions) in the. ‧ 國. 學. verb region; instead, in the region immediately following the verbs, the context effects. ‧. were observed on first-pass time and the meaning frequency effects emerged on both. Nat. al. er. io. sit. y. total time and second-pass time. Table 1 shows their findings for homographic verbs.. n. Table 1 The summarized findings for homographic verbs from Pickering and Frisson (2001) Measures First-pass time (Gaze duration) First-pass regression Second-pass time (Rereading time) Total time. Ch. Region 1 (context). engchi Region 2 (verb). i n U. v. Sentence. Region 3 (intervening). Region 4 (noun). Region 5 (end). Context. Context & Frequencyb. Context. Context Context. Frequencya Context & Frequencya. Note. Context: neutral > supportive; Frequencya: subordinate > dominant; Frequencyb: subordinate < dominant.

(37) 22. Pickering and Frisson (2001) suggested that the resolution of homographic verbs was delayed in comparison with that of homographic nouns, since the SBE was not observed in the verb region but in the following region. The delayed resolution allowed alternative meanings of homographic verbs to reach a high level of activation. As a result, the meaning frequency effects did not occur immediately in the verb region. The context effects observed in Region 3 indicated a selection process. 政治大. between multiple meanings. The late occurrence of the meaning frequency effects. 立. context than the subordinate meaning.. 學. ‧ 國. suggested that the dominant meanings were easier to be integrated into the sentence. ‧. In addition to ambiguous verbs, another type of homograph involving a verb. Nat. io. sit. y. meaning is syntactic category ambiguous words (SCA words), whose multiple. al. er. meanings cross different syntactic categories (e.g., watch). People may encounter. n. v i n C hwhen there are SCA syntactic category ambiguities e n g c h i U words in the sentences. Previous. literature on the resolution of syntactic category ambiguities yielded different findings. Frazier and Rayner (1987) proposed a delay model: When no disambiguating information is provided prior to a syntactic category ambiguity, the resolution would be delayed until helpful information is encountered, regardless of interpretation preference of the ambiguity itself. The delay model can account for Frazier and Rayner’s results: (a) reading times on the ambiguous word phrase (e.g., desert trains).

(38) 23. were longer when preceded by a disambiguating determiner (e.g., I know that this desert trains young people to be especially tough.) compared to when preceded by a non-disambiguating determiner (e.g., I know that the desert trains young people to be especially tough.), and (b) reading times on the remainder of the sentence, containing disambiguating information, showed an opposite pattern. Assuming analyzing syntactic categories of words takes time, these results indicated that syntactic category. 政治大. assignment occurred immediately when there is disambiguating information but. 立. delayed until the presence of disambiguating information when the preceding context. ‧. ‧ 國. 學. is neutral.. According to Frazier (1989), the language processor handles ambiguities at each. Nat. io. sit. y. level in different ways. For syntactic ambiguities, only the simplest structure is. al. er. considered and constructed. For purely semantic ambiguities, multiple meanings are. n. v i n C hbased on preference available and selection is rapid e n g c h i U or contextual bias. For syntactic. category ambiguities, selection is delayed until the presence of disambiguation, though multiple meanings and syntactic categories are initially available. In short, a delay mechanism was assumed to handle the resolution of syntactic category ambiguity. However, the resolution of syntactic category ambiguity is delayed probably because it takes more time to process a verb meaning. Pickering and Frisson (2001).

(39) 24. have suggested that the resolution of ambiguous verbs is delayed due to the inherent processing difficulty of verb meanings. It is reasonable to suspect that the difficulty of processing verbs may also affect the resolution of syntactic category ambiguity. It seems unnecessary to assume a special delay mechanism for syntactic category ambiguities. The empirical evidence that both ambiguous verbs and syntactic category ambiguities are not resolved rapidly indicates the influence of the inherent processing difficulty of verbs.. 立. 政治大. On the other hand, some studies supported immediate resolution of syntactic. ‧ 國. 學. category ambiguity (Gibson, 2006; Jones, Folk, & Brusnighan, 2012; Macdonald,. ‧. 1993). For example, Jones et al. (2012) conducted eye movement experiments to. Nat. io. sit. y. investigate how biased SCA words are resolved in disambiguating or neutral contexts.. al. er. In Experiment 1, preceding context provided syntactic category information. n. v i n C h or subordinateUmeaning of SCA words (e.g., I was consistent with either the dominant engchi. shocked to see him counter the offer so quickly.). The results showed no ambiguity effects (ambiguity > unambiguous controls) in gaze duration measures whenever the prior context was biased toward the dominant or subordinate meaning, indicating that the syntactically appropriate meaning was initially selected. However, the ambiguity effects indeed emerged in later eye-movement measures, such as regression-out probability from target words and second-pass time on prior disambiguating context,.

(40) 25. only when the subordinate meaning was biased. These processing costs suggested that readers had difficulty in integrating the subordinate meaning due to the activation of the dominant meaning. In Experiment 2, preceding context was neutral and disambiguating information was provided in post-target region (e.g., We watched her duck eat all of the bread.). The results showed ambiguity effects in different measures consistently in the subordinate-biased condition, including gaze duration on target. 政治大. words 2 , regression-out probability from post-target regions, second-pass time on. 立. target words, and second-pass time in post-target regions. The ambiguity effects. ‧ 國. 學. indicated processing costs of the inconsistency between the initially selected. ‧. (dominant) meaning and the context-intended (subordinate) meaning. In addition, in. Nat. io. sit. y. the subordinate-biased post-target region, first-pass times were shorter for the. al. er. ambiguity condition compared to the control condition. The authors claimed that this. n. v i n resulted from the increasedC regression-out h e n g c probability h i U and may indicate that readers tended to recover the context-intended (subordinate) meaning by making regressions. Taken together, their findings suggested that readers immediately select one meaning for SCA words and rejected the delay model proposed by Frazier and Rayner (1987). To sum up, it is questionable whether syntactic category ambiguity is resolved by a unique delay mechanism. The delay model was proposed on the basis of the delayed 2. The increased gaze durations for biased syntactic category ambiguities were not predicted by either the delay model or the immediate processing theories. Jones (2012) suggested that the increased gaze durations may be caused by parafoveal-on-foveal processing or undershot saccades..

(41) 26. resolution observed on balanced SCA words in neutral contexts. However, the investigation on biased SCA words in disambiguating and in neutral contexts did not support the delay model. This indicates that the resolution of syntactic category ambiguity is not always delayed. Instead, like lexical-semantic ambiguity resolution, the probabilistic constraints matter. Previous research has demonstrated the influence of syntactic category of meanings on lexical ambiguity resolution. Presumably, the. 政治大. inherent processing difficulty of verb meanings would delay the semantic resolution. 立. of SCA words. Therefore, it is of importance to distinguish homographic nouns,. ‧ 國. 學. homographic verbs, and SCA words when examining lexical ambiguity resolution.. 2.2.3.1 Semantic constraint. al. er. io. sit. y. ‧. Nat. 2.2.3 Contextual constraints. n. v i n Hitherto the focus hasCbeen the lexical nature of ambiguous words. h emore n g on chi U. However, more often than not people resolve lexical ambiguity in sentences rather than in isolation. Thus, of great significance is whether sentential contexts could influence the resolution of lexical ambiguity. If any, would different types of contexts exert distinct influence on the resolution of ambiguous words? Most literature has attended to the influence of semantic context. For balanced homographs, prior semantic information seemed effective in constraining meaning activation of the.

(42) 27. homographs, resulting in the selective access. For example, Schvaneveldt et al. (1976) asked subjects to perform the lexical decision task on triples of words, in which the first and the third word were semantically either related or unrelated to the second word, which was a homograph. Decreased reaction time was observed on the third word when the first and third words were both related to the same meaning of the homograph (e.g., SAVE-BANK-MONEY). However, the reaction time on the third. 政治大. word was not significantly facilitated when the first and third words were related to. 立. different meanings of the homograph (e.g., RIVER-BANK-MONEY). The results. ‧ 國. 學. suggested that the prior semantic context established by the first word led people to. ‧. selectively access only one meaning of the homographs. For biased homographs,. Nat. io. sit. y. more mixed results have been shown in previous studies. Some research supported the. al. er. selective access view, that is, contextual information has an early impact on the. n. v i n C h(Glucksberg et al.,U1986; Schvaneveldt et al., 1976; activation of lexical meaning engchi Simpson, 1981; Simpson & Krueger, 1991; Tabossi, 1988; Tabossi et al., 1987; Tabossi & Zardon, 1993). Other studies bolstered the exhaustive access view, in which the meaning activation is assumed to be encapsulated in a modular system and contextual information helps select an appropriate meaning at post-lexical stage (Onifer & Swinney, 1981; Seidenberg et al., 1982; Swaab et al., 2003; Swinney, 1979; Tanenhaus & Donnenwerth-Nolan, 1984; Tanenhaus et al., 1979)..

(43) 28. Recent eye movement studies on lexical ambiguity resolution have clearly depicted the interaction between meaning dominance and semantic contextual influence (Duffy et al., 1988; Rayner & Duffy, 1986). For instance, Duffy et al. (1988) manipulated ambiguity types (balanced vs. biased homographs) and disambiguating locations (pre-target vs. post-target disambiguation) in a normal reading experiment. Target words (either balanced or biased homographs) were embedded into a. 政治大. two-clause sentence, in which one clause contained the target word and the other. 立. contained disambiguating information. Each ambiguous word was also paired with a. ‧ 國. 學. frequency- and length- matched unambiguous control word, which can fit smoothly. ‧. into the same sentence as the corresponding homograph. The results showed that gaze. Nat. io. sit. y. durations on the balanced homographs were longer than those on biased homographs. al. er. or on the control words when preceding context contained no disambiguating. n. v i n information. This ambiguityC effect due to the competition of multiple h ewas n gpresumably chi U meanings of the balanced homographs. In addition, when preceding context contained disambiguating information toward the subordinate meaning of the biased homographs, readers fixated longer on the biased homographs than on the balanced homographs or on the control words. This effect, the subordinate biased effect (SBE), demonstrated the interaction between meaning dominance and semantic contextual information (Rayner et al., 1994). Furthermore, there was no significant difference.

(44) 29. between the gaze durations on the balanced homographs and on the control words in the biasing context. The SBE was robust and consistently found in many eye movement studies (Duffy et al., 2001; Duffy et al., 1988; Pacht & Rayner, 1993; Rayner & Duffy, 1986; Rayner & Frazier, 1989; Rayner et al., 1994; 盧怡璇, 2012). Although the longer gaze durations on balanced homographs in a neutral context can be explained by the. 政治大. exhaustive access view, the absence of ambiguity effect for balanced homographs. 立. preceded by disambiguating information was not predicted. In order to explicate the. ‧ 國. 學. SBE, the reordered access model (Duffy et al., 2001; Duffy et al., 1988) was proposed. ‧. to highlight the early influence of semantic context. In neutral contexts, meaning. Nat. io. sit. y. dominance can mediate the resolution of lexical ambiguity. On the other hand, in. n. al. er. biasing contexts, the semantic information can boost the activation level of the. Ch. engchi. context-intended meaning. As a result, for. iv n balanced U homographs,. the intended. meaning has a higher level of activation and become available earlier than the unintended one, resulting in the absence of meaning competition. For biased homographs, context and meaning dominance jointly determine the availability of alternative meanings. In dominant-biased contexts, only the dominant meaning reaches at a high level of activation due to support from both context and meaning frequency. In the subordinate-biased contexts, the access of the subordinate meaning.

(45) 30. is speeded with the help of context, while the dominant meaning still reaches at a high level of activation. Thus, the alternative meanings compete with each other. To sum up, inconsistent with the assumptions of the modular, exhaustive access view, the reordered access model allows the interaction between semantic contexts and meaning dominance to influence initial access of meaning.. 2.2.3.2 Syntactic constraint. 立. 政治大. In addition to semantic context, syntactic context is also important to word. ‧ 國. 學. processing in sentences (Marslen-Wilson & Tyler, 1980; Tyler & Warren, 1987; West. ‧. & Stanovich, 1986; Wright & Garrett, 1984). Syntactic category ambiguity provides. Nat. io. sit. y. an opportunity for investigating the influence of prior syntactic constraints on lexical. al. er. ambiguity resolution. The key issue is whether leading syntactic category constraints. n. v i n C h of syntactic category can mediate semantic resolution e n g c h i U ambiguities. Previous research has shown discrepant results. Seidenberg et al. (1982) conducted a series of cross-modal naming experiments to probe meaning activations of balanced noun-noun and noun-verb ambiguous words at a short (0 msec) or long (200 msec) stimulus onset asynchrony (SOA). In their Experiment 3, noun-verb ambiguous words were preceded by a syntactically-biased but semantically-neutral context and followed by a target word, which was semantically related to one of the alternative meanings, forming two.

(46) 31. conditions (i.e., congruent: They bought a rose－FLOWER; incongruent: They all rose － FLOWER). Control sentences were also constructed by replacing the ambiguous word with an unambiguous word, which was unrelated to the target word (e.g., They bought a shirt; They all stood). The results showed that subjects named the target faster in both congruent and incongruent conditions than in the controls at the short SOA, but that only in the congruent condition was the target named faster at the. 政治大. long SOA. These results supported that both meanings of syntactic category. 立. ambiguous words were initially activated even with the presence of syntactic category. ‧ 國. 學. constraints. Using the same paradigm, Tanenhaus et al. (1979) also obtained similar. ‧. results and supported the exhaustive access for syntactic category ambiguous words.. Nat. io. sit. y. Some recent electrophysiological studies have also directed attention to the. al. er. processing of SCA words in the syntactically well-specified context (Federmeier et al.,. n. v i n 2000; Lee & Federmeier, C 2006, h e2009). n g cLeeh iandUFedermeier (2009) examined the effects of syntactic and semantic context on ambiguity resolution for noun-verb homographs, using the ERP methodology. In Experiment 1, they manipulated two. variables, word types (NV-homographs vs. unambiguous words) and sentential context types (congruent vs. syntactic prose). Congruent sentences provided syntactic category and semantic information (e.g., He said the long, graceful bird was called a swallow.), while syntactic prose sentences provided only syntactic category.

(47) 32. information but incoherent semantic information (e.g., He realized the young, English life was assigned a swallow.). They found that a sustained frontal negativity (200-700 ms) was elicited by the NV-homographs in the syntactic prose sentences. However, the sustained frontal negativity was largely reduced when additional semantic information was available in contexts, that is, in the congruent sentences. In addition, N400 elicited by the NV-homographs was still more negative than by unambiguous. 政治大. words in the congruent condition. Furthermore, in Experiment 2, they found that. 立. NV-homographs elicited larger N400 than did unambiguous words only when context. ‧ 國. 學. was biased toward the subordinate meaning of homographs. To sum up, the sustained. ‧. frontal negativity suggested that syntactic category information alone cannot. Nat. io. sit. y. exclusively select the context-intended meaning of SCA words. Semantic information. al. v i n C h that the activation indicated e n g c h i U of the. n effect on the N400. er. is also important to the semantic resolution of SCA words. In addition, the ambiguity dominant meaning of. NV-homographs is automatic and cannot be prevented by the subordinate-biased contexts. In contrast, some recent eye movement studies buttressed the view that syntactic category information can guide meaning resolution of SCA words (Folk & Morris, 2003; Jones et al., 2012). For example, in Experiment 1, Folk and Morris (2003) embedded biased noun-noun and noun-verb homographs into sentences which were.

(48) 33. both semantically and syntactically biased toward either the dominant or subordinate meaning of the homographs. They observed longer gaze durations and second-pass times on the noun-noun homographs compared to the unambiguous control words, only when preceding context instantiated the subordinate meaning. This ambiguity effect was also shown in the post-target region, i.e., spillover times. However, for noun-verb homographs, no matter when prior context instantiated the dominant or. 政治大. subordinate meaning, the ambiguity effect was only found on spillover times and. 立. second-pass times, but not on gaze durations. Table 2 shows a summary of their. ‧ 國. 學. results in Experiment 1.. ‧ sit. n. al. Contextual bias. Gaze Duration. er. io. Measures. y. Nat. Table 2 The summarized results of Experiment 1 from Folk and Morris (2003). v A n>i UA NN. CSubordinate h e n g c h i U A = UA Dominance. Word type VN A = UA A = UA. Spillover. Subordinate Dominance. A > UA A = UA. A > UA A = UA. Second-pass Time. Subordinate Dominance. A > UA A = UA. A > UA A = UA. Note. NN: noun-noun homograph; VN: noun-verb homograph; A: ambiguous words; UA: unambiguous words. In Experiment 2, Folk and Morris (2003) further examined the influence of syntactic category constraints alone on the meaning resolution of SCA words. Unlike.