從語料庫看台灣華語身體部位詞詞彙的詞彙結構、語意表現及詞彙化 - 政大學術集成

全文

(1)National Chengchi University Graduate Institute of Linguistics Master Thesis. :. Advisors : Dr. Huei-ling Lai Dr. Jyi-Shane Liu. 立. 政治大. ‧ 國. 學. Word Component Structures, Meaning Distribution and Lexicalization:. ‧. Linguistic Usages Containing Body-Part Terms liǎn / miàn,. n. al. er. io. sit. y. Nat. yǎn / mù and zuǐ / kǒu in Taiwan Mandarin. Ch. engchi. i n U. : Student: Hsiao-ling Hsu. June, 2017. v.

(2) Word Component Structures, Meaning Distribution and Lexicalization: Linguistic Usages Containing Body-Part Terms liǎn / miàn, yǎn / mù and zuǐ / kǒu in Taiwan Mandarin. BY. 學. ‧ 國. 立. 政治大 Hsiao-ling Hsu. ‧ er. io. sit. y. Nat. n. a lA Thesis Submitted to the i v n Ch U i e n g cofhLinguistics Graduate Institute in Partial Fulfillment of the Requirements for the Degree of Master of Arts. June 2017.

(3) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v.

(4) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Copyright © 2017 Hsiao-ling Hsu All Rights Reserved iii.

(5) Acknowledgements. :. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. :. 、. Ch. engchi. i n U. v. : 、. iv.

(6) 103. ) ). 立. 。. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. v.

(7) TABLE OF CONTENTS. ACKNOWLEDGEMENTS. iv. TABLE OF CONTENTS. vi. LIST OF TABLES. viii. LIST OF FIGURES. x. CHINESE ABSTRACT. xi. ENGLISH ABSTRACT. xii. CHAPTER I: INTRODUCTION. 1. 政治大. 1.1 Motivation and Purpose. 1. 1.2 Research Questions. 3. 立. 1.3 Organization of the Thesis. 4. ‧ 國. 學. CHAPTER II: LITERATURE REVIEW. ‧. 2.1 ‘Word’ in Chinese 2.1.1 Defining the ‘word’ in Chinese. 5 5 7. 2.1.3 Lexicalization. 9. io. sit. y. Nat. 2.1.2 Describing word components: Form class description. n. al. er. 2.2 Metaphor Theory. i n U. v. 2.3 Metaphors and Metonymies with Body-Part Terms. Ch 2.3.1 Liǎn/miàn and face 2.3.2 Yǎn/mù and eye. engchi. 2.3.3 Zuǐ/ kǒu and mouth 2.4 Remarks CHAPTER III: METHODOLOGY. 5. 11 14 15 19 22 26 27. 3.1 Corpus-based and Corpus-driven Approaches. 27. 3.2 Corpora. 27. 3.3 Coding Schema. 28. 3.3.1 Word segmentation. 28. 3.3.2 Filtering. 31. 3.3.3 Form class identity, meanings and lexicalization. 32 vi.

(8) CHAPTER IV: RESULTS & DISCUSSION. 42. 4.1 Word Component Structures. 43. 4.2 Meaning Distribution. 46. 4.3 Lexicalization Types. 65. 4.4 Overall Discussion. 80. CHAPTER V: CONCLUSION. 89. 5.1 Summary. 89. 5.2 Contributions of the Thesis. 90. 5.3 Directions for Future Studies. 92. REFERENCES. 立. 政治大. ‧. ‧ 國. 學. n. al. er. io. sit. y. Nat. 94. Ch. engchi. i n U. v. vii.

(9) LIST OF TABLES Table 2.1. Categories of lexicalization (Packard 2000: 222). 11. Table 2.2. Metonymies of liǎn/miàn and face summarized from Zhao’s. 17. work (2010: 44-47) Table 2.3. Metaphors of miàn and face summarized from Zhao’s work. 18. (2010: 47- 48) Table 2.4. Meaning transference of liǎn/miàn and face summarized from. 18. previous studies (Wen & Wu 2007; Yu 2008; Zhao 2010; Xie 2011) Table 2.5. Metonymies of yǎn/mù and eye summarized from Zhao’s work (2010: 51- 55). Table 2.6. 立. (2010: 55-56). ‧ 國. Metonymies of zuǐ/ kǒu and mouth summarized from Zhao’s work (2010: 71-74). Table 2.8. Metaphors of kǒu and mouth summarized from Zhao’s work. 21 24 25. ‧. (2010: 74-75). Word segmentation criteria proposed by the ROC Computational. y. Nat. Table 3.1. 學. Table 2.7. 政治大. Metaphors of yǎn/mù and eye summarized from Zhao’s work. 20. 28. sit. Linguistics Society (Chen et al., 1996, p.170). 42. Table 4.2. The distribution of word component structures in disyllabic. v i n C hthe six body-part U words containing e n g c h i terms The distribution of literal, metonymic and metaphorical. 44. n. Table 4.3. al. er. Number of types and frequency of the six body-part terms. io. Table 4.1. 47. meanings in the six body-part terms Table 4.4. The distribution of metonymic meanings among the six body-. 48. part terms Table 4.5. The distribution of metaphorical meanings in the six body-part. 54. terms Table 4.6. A comparison of metonymic meanings in liǎn and miàn in. 59. Zhao’s study (2010) and in this study Table 4.7. A comparison of metonymic meanings in yǎn and mù in Zhao’s. 60. study (2010) and in this study. viii.

(10) Table 4.8. A comparison of the metonymic meanings in zuǐ and kǒu in. 61. Zhao’s study (2010) and in this study Table 4.9. A comparison of metaphorical meanings in liǎn and miàn in. 63. Zhao’s study (2010) and in this study Table 4.10. A comparison of metaphorical meanings in yǎn and mù in Zhao’s 64 study (2010) and in this study. Table 4.11. A comparison of metaphorical meanings in zuǐ and kǒu in Zhao’s 65 study (2010) and in this study. Table 4.12. The distribution of the word component structures and their. 67. corresponding lexicalization types in disyllabic words containing liǎn and miàn across metonymic meanings. 學. Table 4.14. 政治大 corresponding lexicalization types in disyllabic words containing 立 yǎn and mù across metonymic meanings The distribution of word component structures and their. ‧ 國. Table 4.13. The distribution of word component structures and their. 70. 72. corresponding lexicalization types in disyllabic words containing. Table 4.15. ‧. zuǐ and kǒu across metonymic meanings. The distribution of word component structures and their. 75. Nat. sit. y. corresponding lexicalization types in disyllabic words containing. er. io. miàn across metaphorical meanings Table 4.16. The distribution of word component structures and their. 77. Table 4.17. The distribution of word component structures and their. 79. al. n. v words containing i corresponding lexicalization types in disyllabic n Ch U yǎn and mù across e metaphorical n g c h imeanings corresponding lexicalization types in disyllabic words containing zuǐ and kǒu across metaphorical meanings. Table 4.18. Number of frequency of the lexicalization types in the six body-. 82. part terms Table 4.19. The distribution of lexicalization types among the six body-part. 82. terms Table 4.20. The distribution of lexicalization types in the six body-part terms 83 with metonymic meanings. Table 4.21. The distribution of lexicalization types in the six body-part terms 83 with metaphorical meanings. ix.

(11) LIST OF FIGURES Figure 3.1. The output of the raw data in the EXCEL templete. 31. Figure 3.2. Form class identity and meanings of liǎn identified in the. 34. Chinese Dictionary Figure 3.3. Form class identity and meanings of liǎn identified in the. 34. Chinese Wordnet Figure 3.4. Form class identity and meanings of miàn identified in the. 35. Chinese Dictionary Figure 3.5. Form class identity and meanings of miàn identified in the. 36. Chinese Wordnet Figure 3.6. Form class identity and meanings of mù identified in the Chinese 36 Dictionary. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. x.

(12) -. 政治大. 立. (.. ‧. ‧ 國. 學. [NN]N. io. [NN]N. n. al. Ch. metaphorical lexicalization. y. [VN]V. engchi. sit. [NN]N. er. Nat. [NN]N ;. i n U. v. ). xi.

(13) ABSTRACT This study aims at finding the actual usage patterns of the six body-part terms liǎn, miàn, yǎn, mù, zuǐ, and kǒu extracted from the largest, balanced and widely-used Sinica Corpus 4.0. The word component structures, meaning distribution and lexicalization types in the six body-part terms are systematically analyzed and compared. The results show that, in terms of word component structures, the [NN]N structure is the most dominant word component structure across the six body-part terms. As for meaning distribution, it is found that miàn, mù, and kǒu are more frequently used to denote non-literal meanings (either metonymic or metaphorical meanings) than their counterparts – liǎn, yǎn, and zuǐ. In addition, PERSON and EMOTION are. the most prevalent metonymic meanings across the six body part terms.. 政治大 word component structure is [NN] , while when they denote 立. It is also found that when the body-part terms denote PERSON, the most dominant N. EMOTION,. [NN]N and. [VN]V are the most dominant structures. In general, the [NN]N structure shows the. ‧ 國. 學. highest frequency of occurrences in all the six body part terms when they are used either metonymically or metaphorically. As for the lexicalization types, most of the. ‧. disyllabic words containing the six body-part terms are in metaphorical lexicalization when they denote metonymic or metaphorical meanings. The results may provide. y. Nat. sit. some insights for refining procedures of identifing metaphors and metonymies in. n. al. er. io. corpora so as to accelarate quantification and statistical evaluation.. Ch. i n U. v. Keywords: Meaning Extensions, Word Component Structures, Tagging, Metonymies, Metaphors. engchi. xii.

(14) CHAPTER I INTRODUCTION. 1.1 Motivation and Purpose Language provides a window to uncover human cognition. Linguistic structures are constructed and motivated with reference to our cognitive processing (Langacker 1986). If we view the components of words in terms of their form class identity (i.e., syntactic form class identity or “part of speech”), we can tell the systematic knowledge that native speakers possess in order to understand and use the words. 治政 (Packard 2000). Human body-parts are regarded as大 the most primary source domain 立 ‧ 國. 學. which people map to in order to understand the abstract concepts in the target domain, for we are all born with our body-parts and they are the foremost way by which we. ‧. interact with the world. Our experiences of bodily interaction with the world are thus. sit. y. Nat. the main source for references for people to create metonymies and metaphors.. n. al. er. io. Metaphor and metonymy are considered two important mechanisms that people apply. i n U. v. to understand abstract concepts (Lakoff & Johnson 1980). Studying and analyzing. Ch. engchi. metonymies and metaphors in languages helps us to uncover the veil of human mind and human cognition. Indeed, metaphoric and metonymic expressions containing body-part terms are pervasively found cross-linguistically to express concepts such as personality, emotion, and quality. In English, people often use the linguistic expression - lose face to mean ‘lose one’s own dignity or self-respect’, and a similar expression- diūliǎn. ‘lose face’ is also found in Mandarin Chinese (Yu 2008).. Many studies have been done on the analysis of metaphoric and metonymic expressions with body-part based on systematic cognitive networks. Focusing on the. 1.

(15) classification and the operating mechanism of the body-part metaphoric and metonymic expressions, they suggest that metaphoric and metonymic expressions not only have bodily and psychological basis but also convey some cultural-specifics (Tsai 1993; Yu 2002; Hung & Gong 2011). In the extant literature, the studies of body-part terms metaphors are mostly based on linguistic data limited to dictionaries or underdeveloped corpora, leading to their findings being based more on theoretical reasoning rather than an empirically-based treatment (e.g. Tsai 1993; Yu 2008, 2011b; Wen & Wu 2007; Qin 2008; Zhao 2010;. 治政 Xie 2011; Hung & Gong 2011). Language is always大 changing. Particularly, corpora 立 ‧ 國. 學. which contain a multitude of linguistic expressions have shown that language use is not always stable and fixed but dynamic (Cameron & Deignan 2006). Novel usages. ‧. often emerge from various contexts, and continue to change for different. y. and xǐliǎn. have recently. sit. Nat. communicative purposes. For example, dǎliǎn. n. al. er. io. been used metonymically among the younger generations and even by news reporters. i n U. v. or anchors. The two terms are used not only to express their literal meanings ‘to hit. Ch. engchi. one’s face’ or ‘to wash one’s face’ but have also been extended to express ‘to cause someone to feel embarrassed or ashamed’. Moreover, zuǐ. (mouth), which is. usually considered as a noun, has been used sometimes as a verb in the expression tā hěn xǐhuān zuǐ biérén. to express ‘to attack someone verbally’.. Therefore, it is highly expected that generalized patterns can be found based on the interaction between the form class of the word constituents and their senses from usage-based linguistic evidence. Thus, this study incorporates computational linguistics approaches, corpus-based and corpus-driven approaches, and cognitive. 2.

(16) semantic theories with the aim to better capture the dynamicity and the generalized patterns exhibited by language use to find actual patterns of usage in the real-world context of Chinese. By combining these approaches, we analyze and compare the linguistic expressions of three sets of body-part terms in Chinese –face (liǎn (yǎn. / mù. ), and mouth (zuǐ. / kǒu. / miàn. ), eye. ). The three sets of body-part terms are. selected because in each set, they are counterparts to each other and because they are considered the most distinctive body-parts on which people depend to understand and. 治政大 by looking at their faces, interact with the outside world: people recognize others 立 ‧ 國. 學. perceive the outside world through their eyes, and express themselves through their mouths. Therefore, the investigation of the linguistic expressions of these three. ‧. prominent body-parts will be a good starting point to understand human cognition.. sit. y. Nat. n. al. er. io. 1.2 Research Questions. i n U. v. Based on the motivation and purpose mentioned above, two specific research. Ch. engchi. questions this current study aims to answer are as follows: (A) What word component structures and their corresponding metonymic and metaphoric extensions can be detected in terms of the linguistic usages containing body-part terms in corpora? (B) What lexicalization types can be generalized based on the interaction between the word component structures of the words containing body-part terms and their corresponding metonymic and metaphoric extensions?. 3.

(17) 1.3 Organization of the Thesis This thesis consists of five chapters. Chapter I introduces the motivation and purpose, research questions, and organization of this thesis. Previous studies investigating metaphor theories, metaphors and metonymies with body-part terms, the definition of word, form class identity, and lexicalization are presented in Chapter II. Chapter III provides details about conventions of the data and coding schema adopted in this current study. The data analysis and findings based on our coding schema and discussion are illustrated in Chapter IV. Chapter V concludes this thesis and proposes. 治政大 some directions for future research. 立. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 4. i n U. v.

(18) CHAPTER II LITERATURE REVIEW. 2.1. ‘Word’ in Chinese. 2.1.1 Defining the ‘word’ in Chinese We speakers may have a rough concept about what a word is, but we may encounter some difficulities if we are asked to provide the exact definiton of word. Linguists have been trying to define what a word is for decades and they have provided some definitions :. 治政 A word is a unit which is a constituent at 大 the phrase level and above. It is 立 ‧ 國. 學. sometimes identifiable according to such criteria as (a) being the minimal possible unit in a reply having features such as a regular stress pattern, and. ‧. phonological changes conditioned by or blocked at word boundaries (b). sit. y. Nat. being the largest unit resistant to insertion of new constituents within its. n. al. er. io. boundaries, or (c) being the smallest constituent that can be moved within a. i n U. v. sentence without making the sentence ungrammatical. A word is sometimes. Ch. engchi. placed, in a hierarchy of grammatical constituents, above the morpheme level and below the phrase level. (Hartmann and Stork 1972; Pike and Pike 1982; Cruse 1986; Mish 1991) However, these definitions are still either too general or too vague to be applied in this current study. All of our raw material is extracted from Sinica Corpus 4.0, and we follow the definition of a word and the rules of how words are segmented as they are given in the introduction to the Sinica Corpus (please see 3.3.1). It is observed that the notions of. 5.

(19) linguistic words can be further portrayed on some of the following criteria: orthographic word, sociological word, lexical word, semantic word, phonological word, morphological word, syntactic word and psycholinguistic word (Packard, 2000: 7-14). In computational linguistics, “ the primary goal of segmentation is to identify units to access lexical information (i.e., dictionary lookup)” (Huang et al., 1996: 49). It is found that “ the notion of words must integrate the modular knowledge of phonology, morphology, syntax, and semantics. The lexicon as the knowledge-base of all linguistic knowledge is exactly the locus where such an integrated notion of words. 治政大 units (namely, words ci exist” (Huang et al., 1996: 49). Therefore, the segmentation 立 ‧ 國. 學. ) are defined based on the notion of lexical words (Huang et al., 1996). Lexical. words are defined as follows:. ‧. [T]he lexicon is traditionally seen as that component of the grammar that. sit. y. Nat. contains all that is not predictable, and must therefore be stored in a. n. al. er. io. memorized list. To that extent, ‘words’ are those idiosyncratic, arbitrary. i n U. v. pairings of sound and meaning that cannot be generated by rule ‘on line’. Ch. engchi. that we file away in memory for use in the performance in a speech act. (Packard, 2000: 8-9) Nevertheless, Packard (2000) argues that “listedness”, the main criterion applied to define lexical words, fails to capture the exact definition of ‘word’, for it is common to include some items which are not words and “ fails to include many Chinese words created by rules” (Packard, 2000: 9). Packard (2000) has investigated the morphology of Chinese and he prefers to adopt the syntactic definition of word in his analysis:. 6.

(20) A syntactic word is a form that can stand as an independent occupant of a syntactic form class slot, in other words, a syntactically free form, commonly designated in the literature as X0. This is probably the most common current linguistic characterization of the notion ‘word’, and seems to serve as the basis for identifying the ‘orthographic’ and ‘lexical’ words […]. (Packard, 2000: 13) In Chinese, the syntactic word may be termed as cí liǎnxíng. (Packard, 2000). For instance,. face-shape ‘face shape’ consists of two characters /morphemes, and it. 治政大 because of the also consists of two ci which are nouns. “This is evident 立 ‧ 國. contrast, although diūliǎn. 學. generalizability and productivity of the two constituents” (Packard, 2000: 15). In lose-face ‘lose dignity’ also consists of two characters. ‧. and two morphemes, it is one ci because we cannot substitute another verb for diū. y. without losing the idiomatic meaning of the word.. sit. Nat. or another noun for liǎn. n. al. er. io. In this current study, we follow Packard’s view (2000) on words, for the “form. i n U. v. class description” adopted in this study mostly depends on the notion of the syntactic. Ch. engchi. word. Therefore, the syntactic word is the optimal notion for defining words in this study.. 2.1.2 Describing word components: Form class description Many ways have been proposed to characterize and describe word components: relational description, modification structure description, semantic description, syntactic description and form class description (Packard, 2000: 21-33). Among these descriptions, Packard (2000) favors the “form class” method of word component. 7.

(21) analysis, for this approach avoids problems that cannot be solved under other approaches. Form class description is the method used by Lu (1964) in his work Word Formation in Chinese. Under this approach, a word’s component morphemes are described in terms of their syntactic form class identity, or ‘part of speech’, and they are often coded in this way: [M1 M2]w1, where M1 M2 represent the form class identity of a word’s component morphemes and w1 refers to the form class identity of the word (Packard, 2000). In Packard’s study (2000: 64-65), the form class identity of the. 治政大 criteria: the form class of components of a word is determined based on the following 立 ‧ 國. 學. the gestalt word, the role of the constituent or its position within that word, and the form classes of constituents as they are listed in authoritative prescriptive sources,. ‧. e.g., dictionaries.. sit. y. Nat. Packard develops a working hypothesis for the form class approach based on the. n. al. er. io. notion of the syntactic definition of word where “the outer form class identities are. i n U. v. critically related to the inner form class identities ” (Packard, 2000: 34). Packard finds. Ch. engchi. evidence for this hypothesis and proposes that “the identity of morphemes within words is largely word-driven – the form class identity of the word generally determines the form class identities of its constituents” (Packard, 2000: 65). Packard (2000) also suggests that the form and internal structure of words may be related to the lexicalization of words in syntactic phrases. Some advantages are provided by adopting this approach. One is that if the components of a word are viewed in terms of their form class identity, they can be easily and systematically categorized. Also, “it allows us to account for different types. 8.

(22) of systematic knowledge that native speakers possess regarding the composition of words” (Packard, 2000: 32).. 2.1.3 Lexicalization Lexicalization is an important topic which can be explored by studying metaphoric and metonymic expressions and meaning extensions. Brinton and Traugott (2005) propose a definition of lexicalization by combining the definitions of lexicalization of two scholars (Kastovsky 1982; Lipka 2002) as follows:. 治政大 linguistic contexts speakers Lexicalization is the change whereby in certain 立 ‧ 國. 學. use a syntactic construction or word formation as a new contentful form. with formal and semantic properties that are not completely derivable or. ‧. predictable from the constituents of the construction or the word formation. sit. y. Nat. pattern. Over time there may be further loss or internal constituency and the. n. al. er. io. item may become more lexical. (Brinton and Traugott 2005: 96). i n U. v. By investigating lexicalization in languages, “the nature of the component-word. Ch. engchi. relationship” can be explained (Packard 2000). He further proposes that there is a cline of lexicality formed according to the degree of the semantic and grammatical transparency of the internal components of a word. Packard (2000) sets up five categories of lexicalization based on their degree of lexicality, which are termed as conventional lexicalization (the least lexicalized), metaphorical lexicalization, asemantic lexicalization, agrammatical lexicalization and complete lexicalization (the most lexicalized). The five types of lexicalization are introduced as follows with illustrative examples (Packard 2000: 219-222). In. 9.

(23) conventional lexicalization, the word components still preserve their original meanings and grammatical identity, such as chīfàn. eat rice ‘to eat a meal’. Two. subcategories are distinguished in metaphorical lexicalization. The words are categorized under component metaphorical lexicalization if the original meaning of its components is lost but the grammatical identity of its components still can be recognized, such as chìdào. red-band ‘equator’; while if the original meaning of. the word’s components retain and the grammatical identity of its components still can be recognized but the meaning of the word is not compositional but metaphorical, the. 治政大 words are categorized under word metaphorical lexicalization, such as diànyǐng 立 ‧ 國. 學. electric-shadow ‘movie’. If a word cannot be understood based on the compositional meaning of its components, but the grammatical identity of its components is still. ‧. retained, it is categorized under asemantic lexicalization, for example, wènshì. sit. y. Nat. ask-world ‘to be published’.. n. al. er. io. On the other hand, if the components of a word lose their original grammatical. i n U. v. identity but their meaning still obtains, the word is categorized under agrammatial. Ch. lexicalization, such as zhǔbǐ. eprimary-pen n g c h i‘editor-in-chef’. The last category is. complete lexicalization, in which both the original meaning and the grammatical identiy of the word component are lost, such as zuǒyòu. left-right ‘to influence’.. The five types of lexicalization summarized by Packard (2000:222) are presented in Table 2.1.. 10.

(24) Table 2.1 Categories of lexicalization (Packard 2000: 222) Lexicalization. Word component. Grammatical. type. meaning. identity of. Examples. relations conventional. full. present. páchóng crawl-insect ‘reptile’. metaphorical. metaphoriacal. present. diànyǐng electric-shadow ‘movie’. asemantic. opaque. 立. ask-world ‘to be published’. absent. xuéjiù. 學. metaphorical. study-research ‘pedant’. opaque. ‧. complete. wènshì. 政治大. full or. ‧ 國. agrammatical. present. absent. yāgēn. y. Nat. pressure-root. n. al. er. io. sit. ‘complete’. 2.2 Metaphor Theory. Ch. engchi. i n U. v. In Aristotle’s Poetics, where much of the classical view of metaphors was built, metaphors were considered a figure of speech, a linguistic device, and ornaments of langauges (Halliwell 1987). However, the focus on metaphors shifted from words to thoughts with the publication of Lakoff & Johnson’s work Metaphors We Live By in 1980. Many cognitive linguists have proposed that metaphor is not just a figure of speech: “ Metaphor is pervasive in everyday life, not just in language but in thought and action. Our ordinary conceptual system, in terms of which we both think and act, is fundamentally metaphorical in nature” (Lakoff & Johnson 1980: 3). People make. 11.

(25) their conversation more smooth and economic through the use of metaphors; people understand abstract concepts like emotions by using metaphors as well. Metaphor serves as a kind of mechanism that links different but somehow related concepts in two domains to help or accelerate people’s understanding of abstract subjects or concepts (Lakoff & Johnson 1980). It can be considered natural for people to understand the world through metaphors because the relationships between human cognition and human language are not arbitrary but highly systematic (Lakoff & Núñez 2000).. 立. 政治大. The Contemporary Theory of Metaphor suggests that metaphor is a mental. ‧ 國. 學. mechanism that applies people’s common experience to understand other abstract concepts, and it always appears under this formula “an A is a B” (Lakoff & Johnson. ‧. 1980 ). For example, the surface linguistic expressions you need to budget your time,. y. Nat. er. io. sit. you are wasting my time, I lost a lot of time when I got sick, this gadget will save you hours, and do you have much time left are all created through the underlying. al. n. v i n conceptual metaphor TIME ISCMONEY & Johnson 1980: 11). Lakoff and U h e n(Lakoff i h gc. Johnson propose that people understand and describe the abstract concept of time via conceptual mappings between the source domains (MONEY) and the target domain (TIME). Time is considered valuable and limited in modern Western culture. Therefore, based on a similarity in the characteristics of time and money, people map the partial characteristics of money to talk about time (Lakoff & Johnson 1980). In addition to conventional usages, cognitive metaphor theory also deems that both “dead” and novel metaphors in literary works can be analyzed through the existing conceptual mappings (Lakoff 1993; Kövecses 2002).. 12.

(26) However, “cognitive linguists do not generally seek to provide an account of novel metaphor use in non-literary discourse” (Cameron & Deignan 2006: 672). Moreover, it appears that cognitive linguists have recently put too much emphasis on thought and have ignored linguistic behavior when they study metaphors. Language is always changing, and language is the window for linguists to uncover human thoughts. Therefore, neither thought nor linguistic behavior can be neglected in understanding metaphors.. 政治大. With the rapid development of technology, corpus linguists and computational. 立. linguists have provided an innovative and combined methodology to examine the. ‧ 國. 學. usages of metaphors (Cameron 2003; Deignan 2005; Ritchie 2003, 2004; Semino et al. 2004; Semino 2005). Cameron and Deignan explore the usages of metaphors not in. ‧. literary works but in spoken discourse (Cameron 2003) and in corpora (Deignan. y. Nat. er. io. sit. 2005). They try to link “the conceptual with the linguistic, in theory and in empirical work” (Cameron & Deignan 2006: 672). Cameron & Deignan (2006) explore three. al. n. v i n C ofh language itself inUunderstanding metaphors. The first issues to show the importance engchi issue is related to the form of linguistic metaphors. They argue that if linguistic. metaphors are all about conceptual mappings, why is it the case that their surface expressions cannot be combined freely, but are restricted in a range of forms, such as [be] in the dark, come/bring to light, and shed light on (Cameron & Deignan 2006: 673). The second one concerns the uneven and inconsistent distribution of linguistic metaphors. They find that the singular flame and the plural flames in the corpora are used to express different metaphoric and metonymic meanings. Based on their data, the plural flames is used to express anger and love, while the singular flame refers to. 13.

(27) faith or idealism, and flame is in only used in a few instances to talk about anger (Cameron & Deignan 2006: 673). This uneven and inconsistent distribution cannot be explained based on The Contemporary Theory of Metaphor which regards linguistic metaphors as surface expressions of systematic conceptual mappings. The last issue deals with cross-linguistic differences in linguistic metaphors. According to their observation and that of other studies on metaphors (e.g., Kövecses 2005) they find that some cross-linguistic differences in linguistic metaphors can be explained by cultural differences, while others still remain unsolved. (Cameron & Deignan 2006:. 治政 674). To deal with the first two issues, they propose大 the concept of “metaphoreme”. 立 ‧ 國. 學. They define “metaphoreme” as “a bundle of relatively stable patterns of language use, with some variation, that, for the time being, describes how people are using the. ‧. lexical items” (Cameron & Deignan 2006: 686). For the last issue, they adopt an. sit. y. Nat. emergentist perspective to explain the cross-linguistic differences in linguistic. n. al. er. io. metaphors. They propose that sometimes closely-related languages and cultures may. i n U. v. also generate various metaphoremes because of their different socio-cultural. Ch. engchi. backgrounds or social values, and this may cause the emergence of the different metaphoremes (Cameron & Deignan 2006: 687).. 2.3 Metaphors and Metonymies with Body-Part Terms Metaphors are everywhere in our daily life. By using metaphors, it is easier for people to understand or express abstract concepts because studies have shown that metaphors are basically created from our bodily experiences (Gibbs Jr & Wilson 2002; Gibbs Jr. 14.

(28) et al. 2004). Human beings are made of body parts and they are keys to human existence in the spatial and temporal domains. We depend on our body-parts in order to interact with the outside world. For instance, we perceive through our eyes, speak through our mouths, and recognize others by their faces. Based on our embodied experience, we employ our body-parts to express abstract ideas. The interaction between body and language is always the central theme in cognitive linguistics for the exploration of human minds. Much analysis has been done on metaphoric and metonymic expressions with body-part terms based on views of systematic cognitive. 治政大 of body-part terms may networks. Tsai (1993) suggests that the semantic change 立 ‧ 國. 學. derive from their metaphorical or metonymic extensions.Yu (2002, 2008, 2011b) has dedicated some of his works to metaphors and metonymies containing body-part. ‧. terms in Chinese. He (2002, 2008, 2011b) provides a fine-grained theoretical analysis. sit. y. Nat. for metaphors and metonymies containing body-part terms, such as heart, head, face,. n. al. er. io. and mouth. Zhao (2010) explores the meaning transference of body-part terms in. i n U. v. metaphorcial and metonymic expressions in Chinese and English. In focusing on the. Ch. engchi. classification and the operating mechanisms of the body-part metaphoric and metonymic expressions, they suggest that metaphoric and metonymic expressions not only have a psychological basis, but also convey some culturally-specific points of view (Tsai 1993; Yu 2002; Hung & Gong 2011).. 2.3.1 Liǎn/miàn and face The value of face-to-face interactions between human beings can never be replaced by technical devices no matter how rapidly new technological innovations are. 15.

(29) invented. Most people are familiar with this experience: chatting with friends via any application such as a smartphone or other electronic device is nothing like talking with friends in person where their faces can be clearly seen without any delay. Faces play a salient role when people interact with each other. Faces, not just languages, are often considered the index to human minds. The important status of the face among human body-parts has attracted many scholars to investigate how the word face in English and liǎn and its counterpart miàn in Chinese are used in both languages. It is suggested that there are at least four non-literal meanings found four face and. 治政大 liǎn/miàn. They are used to express “ , 立. PERSON CHARACTER, EMOTION,. or DIGNITIY”. ‧ 國. 學. (Wen & Wu 2007; Zhao 2010; Xie 2011). Wen & Wu (2007) and Xie (2011) investigated the non-literal meanings of face and liǎn/miàn in terms of metaphors.. ‧. They both suggest that the specific features of face and liǎn/miàn can be mapped to. sit. y. Nat. express abstract concepts such as emotions, characters and social interation in order to. n. al. er. io. help people’s comprehesion. However, Zhao (2010) argues that previous studies put. i n U. v. too much emphasis on the analysis of metaphors containing body-part terms, and he. Ch. engchi. suggests that a complete picture of our cognitive network can be grasped only if the metonymic expressions with body-part terms are included in the analysis. He proposes that several metonymies and metaphors can be identified based on the idealized cognitive models (ICMs) which he utilizes in the analysis of his data. They are summarized in Table 2.2 and Table 2.3.. 16.

(30) Table 2.2 Metonymies of liǎn/miàn and face summarized from Zhao’s work (2010: 44-47) Body-Part. Idealized Cognitive Model. Meaning. Example. Category-and-Property ICM. DIGNITY. diūliǎn. Term Liǎn. ‘lose face’ Category-and-Property ICM. CHARACTER. hòuliǎnpí ‘brazen-faces’. Category-and-Property. chóuméikǔliǎn. EMOTION. 政治大. /Causation ICM. 立. long face’. Thing-and-Part ICM. ‧ 國. PERSON. 學. Miàn. ‘pull/wear/put on a xīnmiànkǒng. Category-and-Property ICM. ‧. ‘new face’ bǎoquánmiànzi. DIGNITY. Category-and-Property ICM. n. al. Ch. engchi. Thing-and-Part ICM. CHARACTER. er. io. sit. y. Nat. ‘save face’. i n U. v. QUANTITY. shuāngmiànrén ‘two-faced’ yímiànqízi ‘a flag’. Face. Thing-and-Part ICM. PERSON. familiar faces. Category-and-Property ICM. DIGNITY. gain great face. Category-and-Property ICM. CHARACTER. bold-faced. Category-and-Property ICM. EMOTION. one’s face falls. ACTION. face the problem. /Causation ICM Perception ICM. 17.

(31) Table 2.3 Metaphors of miàn and face summarized from Zhao’s work (2010: 47- 48) Body-Part. Image Schema. Meaning Example. Location schema. SURFACE. Term Miàn. zhuōmiàn ‘the surface of the table’. Center-periphery schema. běimiàn. SIDE. ‘the north side’ Face. Location schema. SURFACE. 政治大. 立. the face of the earth. ‧ 國. 學. To sum up, as shown in Table 2.4, based on the previous studies (Wen & Wu 2007; Yu 2008; Zhao 2010; Xie 2011), the meaning transference of liǎn has. ‧. developed from the functional characteristics of liǎn, while the meaning transference. Nat. sit. y. of miàn and face have developed from the functional characteristics of miàn and face. n. al. er. io. and also from other two characteristics: the qualities and the location of miàn and face.. Ch. engchi. i n U. v. Table 2.4 Meaning transference of liǎn/miàn and face summarized from previous studies (Wen & Wu 2007; Yu 2008; Zhao 2010; Xie 2011) Characteristics. Liǎn. Miàn. Face. Function. DIGNITY;. DIGNITY;. DIGNITY;. CHARACTER;. CHARACTER;. CHARACTER;. EMOTION. EMOTION. EMOTION; ACTION. Quality. PERSON; QUANTITY; (QUANTIFIER). 18. PERSON.

(32) Location. SURFACE;. SURFACE. SIDE. 2.3.2 Yǎn/mù and eye The eyes are the most important organ in the visual system for human beings to perceive and experience the outside world. Normally, we use our eyes all the time, so we are very familiar with our eyes; namely, we know the shapes of normal eyes, their qualities, and how they function. Then, because of the concrete bodily experiences,. 政治大. when we encounter something abstract or difficult to understand or express, we. 立. natually map our bodily experiences onto those abstract things or ideas in order to. ‧ 國. 學. help us to comprehend such things. In addition, eyes are often considered the window of soul and true feelings. Thus, eyes are usually the source for people to use to. ‧. describe emotions. Qin (2008) studied the conceptual metaphors of eye in English and. y. Nat. io. sit. yǎn/mù in Chinese and proposes that the metaphorical use of eye and yǎn/mù consists. er. of four major experiential domains: knowledge/intellection; emotion/attitude; social. al. n. v i n C h entities; and shape/time. relationship/relationship between In viewing his data, Qin engchi U (2008) systematically examined the metaphorical use of eye and of yǎn/mù and. suggests that there are 14 metaphorical uses that can be found in these two languages. However, some of the uses that Qin (2008) regards as metaphors are considered to be metonymies, but he does not clarify any difference in detail. Zhao (2010) in his dissertation re-examines the metaphoric and metonymic expressions containing eye in English and yǎn/mù in Chinese, and proposes that at least six metonymies and three metaphors can be identified in both languages, though some differences are found in yǎn and mù. They are presented in Table 2.5 and Table 2.6.. 19.

(33) Table 2.5 Metonymies of yǎn/mù and eye summarized from Zhao’s work (2010: 5155) Body-Part. Idealized Cognitive Model. Meaning. Example. PERSON. yǎnxiàn. Term Yǎn. Thing-and-Part ICM. ‘spy’ Production ICM. yǎnjiān. EYESIGHT. ‘have sharp eyes’. 政治大. Category-and-Property ICM. ‘to signal with. Perception ICM. FOCUS. 學. ‧ 國. 立. yǎnsè. ABILITY. one’s eyes’ bùqǐyǎn. ‧. ‘inconspicuous’. Perception ICM. er. io. sit. y. Nat. INTELLECTION. n. al. Category-and-Property. C ICM /CausationhICM engchi. i n U. v. EMOTION. yǎnguāng ‘judgment’ shùnyǎn ‘pleasing to the eye’. Thing-and-Part ICM. QUANTITY. yìyǎnquán ‘a spring’. Mù. Thing-and-Part ICM. PERSON. ěrmù ‘spy’. Production ICM. EYESIGHT. mùguāng ‘eyesight’. 20.

(34) Category-and-Property. mùguāngrúdòu. ABILITY. ICM ‘short-sighted’ Perception ICM. xǐngmù. FOCUS. ‘eye-catching’ Perception ICM. INTELLECTION. mùguāngyuǎndà ‘ambitious’. Category-and-Property. shǎngxīnyuèmù. EMOTION. ICM /Causation. 政治大. ICM. 立. eyes’. PERSON. private eye. Production ICM. EYESIGHT. dull eyes. Perception ICM. FOCUS. keep an eye on. Perception ICM. INTELLECTION. to western eyes. Category-and-Property. EMOTION. stars in one’s. 學. Thing-and-Part ICM. y. Nat. sit. ICM /Causation ICM. OBSERVATION. n. er. io. Perception ICM. al. ‧. ‧ 國. Eye. ‘a feast for the. Ch. engchi U. v ni. eyes eye somebody up. Table 2.6 Metaphors of yǎn/mù and eye summarized from Zhao’s work (2010: 55-56) Body-Part. Image Schema. Meaning. Example. Apperance Schema. TINY HOLE. qiāngyǎn. Term Yǎn. ‘loophole’ Up-Down Schema. NEAR. yǎnxià. (SPACE). ‘at present’ Front-Back Schema. NEAR (TIME). 21. yǎnqián.

(35) ‘at the moment’ Center-Periphery Schema. CENTER. shīyǎn ‘theme of the poem’. Mù. Up-Down Schema. mùxià. NEAR (SPACE). ‘at present’ Front-Back Schema. 立. NEAR (TIME). 政治大. Center-Periphery Schema. ‘at this moment’. CENTER. ‧ 國. 學. Apperance Schema. TINY HOLE. io. n. Ch. engchi. CENTER. i n U. needle one’s eyes. er. Nat. al. the eye of a in front of. NEAR(SPACE). Center-Periphery Schema. ‘outline’. y. Up-Down Schema. gāngmù. sit. ‧. Eye. mùqián. v. the eye of the hurricane. 2.3.3 Zuǐ/ kǒu and mouth We depend on our mouth to live. As one of the most important part of digestive organs, the mouth functions as an opening to let in food or water to satisfy our physiological needs. As one of the speech organs, the mouth works as a kind of audio equipment which can play the music in our mind and to express our needs or ideas. Because of the close relationship between our mouth and everyday life, it is not hard to imagine that we may use this body-part term to express some abstract ideas, such. 22.

(36) as taste or one’s talking styles. As mentioned above, after systematically examining the metonymic and metaphoric expressions with zuǐ /kǒu in Chinese and mouth in English, Zhao (2010) found four metonymic usages of zuǐ and five of kǒu in Chinese, but six metonymic usages of mouth in English, and two metaphoric usages of kǒu and mouth in both languages (as presented in Table 2.8 & Table 2.9). He has also found that zuǐ is harder to be extended into metaphoical meanings. (?) Later in 2011, Yu (2011b) investigated the Chinese cultural understanding of speech through the investigation of body-part terms containing speech organs, such as zuǐ/kǒu (mouth),. 治政大 chain proposed by Radden she2 (tongue) and hou(throat), based on the metonymic 立 ‧ 國. 學. (2004): speech organ → speaking → speech → language. It is suggested that three. metonymies can be derived from the chain: SPEECH ORGAN FOR SPEAKING, SPEECH. ‧. ORGAN FOR SPEECH, and SPEECH ORGAN FOR LANGUAGE. Yu. (2011b) has found that. sit. y. Nat. the first two metonymies are very common and conventionalized in Chinese, while. n. al. er. io. the last metonymy SPEECH ORGAN FOR LANGUAGE is still rare. He also examined the. i n U. v. word construction of the metonymic expressions containing speech organ terms, and he proposed:. Ch. engchi. When the speech organ term takes the first position, it very often combines with a verbal or adjectival constituent, and the compound takes up the subjectpredicate construction (S–P). When the speech organ term occupies the second position, it can be preceded by a verbal or adjectival element, the internal relationship between the two constituents being, respectively, verb- object (V–O) or adjective-noun (A–N). Occasionally, however, two speech organ terms can combine to form a noun-noun (N–N) construction too. (Yu 2011: 120-121). 23.

(37) Table 2.7 Metonymies of zuǐ/ kǒu and mouth summarized from Zhao’s work (2010: 71-74) Body-Part. Idealized Cognitive Model. Meaning. Example. FLAVOR. zuǐchán. Term Zuǐ. Perception ICM. ‘gluttonous’ Production ICM. UTTERANCES. lòuzuǐ ‘to spill the beans’. 政治大. Category-and-Property ICM. 立. CHARACTER. ‘sweet talker’. ‧ 國. EMOTION. 學. Category-and-Property ICM / Causation ICM. FLAVOR. n. al. er. io. sit. y. ‧. Perception ICM. Nat. Kǒu. zuǐtián. Thing-and-Part ICM. Ch. i n U. v. PERSON. engchi. Category-and-Property ICM. mǐnzuǐ ‘purse one's lips’ jìkǒu ‘be on a diet’ rénkǒu ‘population’. QUANTITY. yì kǒu jǐng ‘a well’. Production ICM. UTTERANCES. kǒucái ‘ways of talking’. Category-and-Property ICM. CHARACTER. xīnzhíkǒukuài ‘have a quick tongue’. 24.

(38) Category-and-Property ICM /. EMOTION. mùdèngkǒudāi. Causation ICM ‘with one’s mouth open in surprise’ Mouth. Perception ICM. make one’s. FLAVOR. mouth water Thing-and-Part ICM. a useless. PERSON. mouth Production ICM. UTTERANCES. keep one’s mouth shut. Causation ICM. mouth. Action ICM. CHARACTER. big-mouthed. EMOTION. make a wry. TALK. 學. ‧ 國. 治政 Category-and-Property ICM 大立 Category-and-Property ICM /. mouth off. ‧. Kǒu. y. sit. Example. er. al. Meaning. n. Term. Image Schema. io. Body-Part. Nat. Table 2.8 Metaphors of kǒu and mouth summarized from Zhao’s work (2010: 74-75). Ch. Apperance Schema. engchi. i n U. v. MOUTH-. kǒudài. SHAPED. ‘pocket’ Location Schema. GATEWAY. ménkǒu ‘doorway’. Mouth. Apperance Schema Location Schema. MOUTH-. the mouth of. SHAPED. a cave. GATEWAY. the mouth of a passage. 25.

(39) 2.4 Remarks According to Cameron and Deignan (2006), both thoughts and linguistic behaviors are the important cores in investigating metaphors, and neither of them can be neglected. Previous studies on metaphors and metonymies containing body-part terms have put too much emphasis on the conceptual aspects. Although they do provide a picture of the operating mechanisms of body-part term metaphors and metonymies, the accounts cannot explain recent usages like tā bèi xǐliǎn / dǎliǎn ‘he is humiliated’ or tā hěn xǐhuān zuǐ biérén. /. ‘he often attacks. 治政大 changes over time. Thus, others verbally’, for their data are limited and language 立 ‧ 國. 學. corpora have come to have an important role in linguistics studies because they may reveal all of the usages in the real-world context.. ‧. However, few studies have explored the metaphoric and metonymic expressions. sit. y. Nat. containing body-part terms based on the data extracted from corpora, and the. n. al. er. io. linguistic behaviors (linguistic forms and structures) in these metaphoric and. i n U. v. metonymic expressions have often been neglected. Linguistic form and meaning. Ch. engchi. should not be regarded as two independent units. They are closely interrelated with each other. As mentioned in the previous section, lexicalization is regarded as an important concept in studying metaphoric and metonymic extensions and it is concerned with the interaction between linguistic structure and meaning. Therefore, this current study will use quantitative and qualitative analyses in a bottom-up approach to systematically explore the metaphoric and metonymic expressions containing body-part terms extracted from the corpora.. 26.

(40) CHAPTER III METHODOLOGY. 3.1 Corpus-Based and Corpus-Driven Approaches In this study, corpus-based and corpus-driven approaches are utilized to investigate and explore the linguistic expressions containing body-part terms in the real world context. Tognini-Bonelli, in her work Corpus Linguistics at Work (2001), makes a distinction between a corpus-based approach and a corpus-driven approach. According to her, the term corpus-based refers to “a methodology that avails itself of. 治政大 and descriptions that were the corpus mainly to expound, test or exemplify theories 立 ‧ 國. 學. formulated before large corpora became available to inform language study”. (Tognini-Bonelli 2001: 65), while the term corpus-driven refers to an approach where. ‧. “the commitment of the linguist is to the integrity of the data as a whole, and. sit. y. Nat. descriptions aim to be comprehensive with respect to corpus evidence” (Tognini-. n. al. er. io. Bonelli 2001: 84). However, Halliday (2004) argues that the two approaches should. i n U. v. be like two poles on a continuum, and that linguistic analyses should be located. Ch. engchi. somewhere in between. Therefore, in this current study, we depend much on the corpora, the content of which represents linguistic behaviors in real life, not only to inspect or account for the previous theoretical hypotheses, but also to find generalized patterns based on the data itself.. 3.2. Corpora. Our data are all extracted from the Academic Sinica Balanced Corpus of Modern Chinese 4.0 (simplified as Sinica Corpus 4.0). The Sinica Corpus was developed by. 27.

(41) the Institute of Information Science and the CKIP group at the Academic Sinica in Taiwan and is considered the first and most representative word-based Chinese corpus. Sinica Corpus 4.0 with 10 million words was completed in 2007 and open to the research community online at (http://www.sinica.edu.tw/SinicaCorpus/) in 2013. The texts dated from 1981 to 2007 are collected from different areas (such as newpapers and literary works) on various topics (such as philosophy, science and society).. 3.3 Coding Schema. 政治大. 立. ‧ 國. 學. 3.3.1 Word segmentation. Word segmentation has been a difficult task in studying Chinese as there are no. ‧. delimiters to mark word boundaries in the written language (Chen et al., 1996). The. sit. y. Nat. word is the basic unit to be analyzed in the field of lexical semantics and it is also the. n. al. er. io. main focus in this current study. Therefore, it is important to illustrate how a word is. i n U. v. segmented to ensure the validity of our analysis. The word segmentation of Sinica. Ch. engchi. Corpus follows the segmentation standard proposed by the ROC Computational Linguistics Society (CKIP group) (Chen et al., 1996). The word segmentation criteria are shown in Table 3.1.. Table 3.1 Word segmentation criteria proposed by the ROC Computational Linguistics Society (Chen et al., 1996, p.170) (1) Segmentation Unitdefinition (2) Segmentation Principles. The smallest string of character(s) that has both an independent meaning and a fixed grammatical category (a). A string whose meaning cannot be derived by the. sum of its components should be treated as a segmentation 28.

(42) unit. (b). A string whose structural composition is not. determined by the grammatical requirements of its components, or a string which has a grammatical category other than the one predicted by its structural composition should be treated as a segmentation unit. (3) Segmentation. (a). Guidelines. Bound morphemes should be attached to neighboring. words to form a segmentation unit when possible. (b). A string of characters that has a high frequency in. the language or high co-occurrence frequency among the components should be treated as a segmentation unit when. 政治大 A string separated by overt segmentation markers. possible.. 立(c). (d). 學. ‧ 國. should be segmented. A string with complex internal structures should be. segmented when possible.. ‧. Nat. sit. y. Using computational linguistics approaches, instances containing either one of. n. al. er. io. the six target body-part terms liǎn/miàn, yǎn/mù, or zuǐ/kǒu in Sinica Corpus 4.0 were. Ch. i n U. v. extracted based on the word segmentation rules mentioned above. First, a search is. engchi. done for the target body-terms liǎn/miàn, yǎn/mù, and zuǐ/kǒu in the corpus. Then, the words containing the target body-part terms are extracted according to the CKIP segmentation. Lastly, the raw data are exhibited in excel templates (as shown in Figure 3.1) with the following information: word, part of speech, frequency, and ranking. (i). Word: It is the most crucial item in the raw data for the preliminary analyses. It is our target instance containing the target body-part terms which are segmented according to the CKIP segmentation. The target. 29.

(43) body-part term as a single word is also found in the raw data, for according to CKIP segmentation, it can also stand for a word with certain part of speech. However, in the cases where the target bodypart term appears as a single word (i.e., miàn. appearing alone as a. word), such cases are excluded from our analysis because they cannot be further analyzed. (ii). Part of Speech: It shows the part of speech of the word that contains the target body-part terms in the texts collected by the corpus. 治政 according to CKIP parsing results. A大 two-stage automatic tagging 立 ‧ 國. 學. process is designed to give the possible parts of speech of each word in Sinica Corpus (Chen et al.,1996). The possible part of speech of each. ‧. word is provided by a hybrid method and re-examined by human. sit. y. Nat. taggers in the first stage, which forms a small corpus as training data. n. al. er. io. for the statistical tagging model in the second stage (Chen et al.,1996). For example, fāngmiàn miànlín. Ch. i n U. v. ‘aspect’ is categorized as a noun and. engchi. ‘to face something’ is categorized as a verb. The. information on the part of speech of the words is important for our form class identity analyses. According to Packard (2000), the interal structure of word components may influence the output of the word, namely, the part of speech of the words. (iii). Frequency: It represents the number of the occurrences of the word containing the target body-part terms appears in the texts collected in the corpus. It needs to be noted that fāngmiàn. 30. ‘aspect’,.

(44) yìfāngmiàn. ‘on one hand’, and lìngyīfāngmiàn. ‘on. the other side’ are counted seperately because they are three separate words as identified by CKIP segmentation, that is, they are mutually exclusive. For example, fāngmiàn corpus, while yìfāngmiàn lìngyīfāngmiàn (iv). appears 4401 times in the appears 853 times and. appears 686 times.. Ranking: It refers to the the ranking of the occurrences of the words containing the target body-part terms used in the texts collected in the. 治政大 is ranked as 1, for example, corpus. The word used the most frequently 立 ‧ 國. ‘aspect’ is ranked as 1 because it is used the most. 學. fāngmiàn. frequently in the corpus.. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 3.1 The output of the raw data in the EXCEL template. 3.3.2 Filtering Words containing body-part terms are composed of different numbers of characters in our data. In this study, we only analyze and compare the form class. 31.

(45) identity of disyllabic words (words containing two syllables/ characters) because we follow Packard’s (2000) criteria which only focuses on these words and their lexicalization types.. 3.3.3 Form class identity, meanings and lexicalization Based on our research aims and the information of our raw data, we develop several steps to analyze the forms and meanings of the words containing the target body-part terms. The form class identity of the word components, the meanings of the. 治政 words, and their lexicalization patterns are analyzed大 according to the following steps. 立 ‧ 國. 學. Step One (form class identification):. ‧. The form classes of the target body-part terms are first identified as they are. sit. y. Nat. listed in the authoritative Chinese Dictionary established by the Ministry of. n. al. er. io. Education, R.O.C1 and in the Chinese Wordnet2. For example, as shown in Figure 3.2 and Figure 3.3, liǎn. Ch. i n U. v. is identified as a noun in both the Chinese. engchi. Dictionary and in the Chinese Wordnet, with their literal and non-literal meanings illustrated with examples. According to the Chinese Dictionary and the Chinese Wordnet, liǎn. , yǎn. while miàn. are identified as nouns, and also as verbs when they. and mù. , kǒu. and zuǐ. are identified as nouns,. represent certain meanings. In the Chinese Dictionary (Figure 3.4), miàn is identified as a verb when it refers to ‘to meet someone’ and ‘to face something or. 1 2. http://dict.revised.moe.edu.tw http://cwn.ling.sinica.edu.tw. 32.

(46) someone’ and it is identified as an adverb when it refers to ‘face to face’, while in the Chinese Wordnet (Figure 3.5), miàn is identified as a verb only when it refers to ‘to face something or to confront’ and miàn is not identified as an adverb. In our analysis, miàn is only considered as either a noun or a verb (when it refers to ‘to face something or someone’) for two reasons. One is that the examples for miàn when it is analyzed as a verb in the Chinese Dictionary (e.g., jiànmiàn. ) and as an adverb (e.g., dāngmiàn. more appropriate when miàn in jiànmiàn. ) are not appropriate. It is. and dāngmiàn. is analyzed. 治政 as a noun. The other reason is that the Chinese大 Dictionary only coincide with the 立 ‧ 國. 學. Chinese Wordnet when miàn is identified as a noun and a verb (only when it. refers to ‘to face something or someone’). The Chinese Wordnet is regarded as. ‧. the more reliable reference in our analysis as it is built on the the basis of lexical. sit. y. Nat. semantics and ontology. Therefore, miàn in our analysis will only be considered. n. al. er. io. as either a noun or a verb (only when it refers to ‘to face something or. i n U. v. someone’). In addition, as shown in Figure 3.6, in the Chinese Dictionary, mù is. Ch. engchi. identified as a noun and a verb (when it represents the meaning of ‘to call/name someone’). However, the verb meaning is not included in the Chinese Wordnet, either. Thus, mù is only considered as a noun in our analysis in line with the same reasons mentioned above.. 33.

(47) 立. 政治大. ‧. ‧ 國. Dictionary. 學. Figure 3.2 Form class identity and meanings of liǎn identified in the Chinese. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 3.3 Form class identity and meanings of liǎn identified in the Chinese Wordnet. 34.

(48) 立. 政治大. ‧. ‧ 國. Dictionary. 學. Figure 3.4 Form class identity and meanings of miàn identified in the Chinese. n. er. io. sit. y. Nat. al. Ch. engchi. 35. i n U. v.

(49) 政治大. Figure 3.5 Form class identity and meanings of miàn identified in the Chinese. 立. Wordnet. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 3.6 Form class identity and meanings of mù identified in the Chinese Dictionary. Step Two (structural analysis): Once the form classes of the target body-part terms are identified, the form classes of the other components that co-occur with the target body-part terms in. 36.

(50) disyllabic words are analyzed based on the role and position of the components within that word and the form classes in which they occur as they are listed in the Chinese Dictionary and Chinese Wordnet (Packard 2000:65). For example, a word like diūliǎn. ‘lose face’ is analyzed as a verb []V composed of a verb. and a noun elements ([VN]V); yǎnjìng. ‘glasses’ is analyzed as a noun []N. composed of two noun elements ([NN]N); zhèngmiàn. ‘front’ is analyzed as. a noun []N composed of an adjective and a noun ([AN] N. 治政大 Step Three (meaning identification): 立 ‧ 國. 學. The various meanings of the six target body-part terms are analyzed and categorized based on their linguistic context and their usages in the real-world. ‧. context, with data from the Chinese Dictionary, the Chinese Wordnet and. sit. y. Nat. previous findings as our references. The linguistic context here refers to their. n. al. er. io. linguistic expressions in words and their usages in the real-world context are. i n U. v. explored in the Sinica Corpus 4.0, which includes the context where the words. Ch. engchi. containing the target body-part terms are used. The words containing the target body-part terms that have compositional meanings are analyzed as the body-part terms which denote literal meanings, such as zuǐchún ‘gauze mask’, liǎnjiá. ‘cheek’, miànshuāng. ‘glasses’, and mùkuāng. ‘lips’, kǒuzhào ‘facial cream’, yǎnjìng. ‘orbit’. The words containing the target. body-part terms which have meanings that are not compositional are analyzed as words which denote non-literal meanings. In total, twelve metonymic meanings are identified: PERSON, EMOTION, CHARACTER, DIGNITY, APPEARACNE,. 37.

(51) CONFRONT (V.) , QUANTITY, ABILITY, INTELLECTION, FOCUS, FLAVOR, UTTERANCES. In. and. general, the target body-part term and its neighboring. component in a disyllabic word are used as a compound. And, by combining the meaning of the target body-part term and its neighboring component, the disyllabic word is used to denote a certain metomymic or metaphorical meaning. For instance, in the disyllabic word shǎngliǎn meaning of shǎng. ‘to honor someone’, it is the. that combines the meaning of liǎn. embed the metonymic meaning DIGNITY in liǎn. 政治大. are defined as follows:. . These metonymic meanings. combining the meaning of the target body-part term and its. 學. PERSON:. ‧ 國. l. 立. to denote and to. neighboring component, the disyllabic word is used to denote the ‘pundit’.. combining the meaning of the target body-part term and its. sit. y. EMOTION:. Nat. l. ‧. meaning of a person, e.g., míngzuǐ. n. al. er. io. neighboring component, the disyllabic word is used to denote certain kinds of feelings, e.g., fānliǎn l. CHARACTER. Ch. engchi. i n U. v. ‘suddenly get mad’.. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is used to denote a person’s personal character, e.g., hēiliǎn l. DIGNITY:. ‘appear as the hatchet man’.. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is used to represent a person’s dignity, e.g., diūliǎn l. APPEARACNE:. ‘lose face’.. combining the meaning of the target body-part term and. its neighboring component, the disyllabic word is used to denote a. 38.

(52) person’s act of appearing in public view, e.g., lòuliǎn l. CONFRONT (V.):. ‘show up’.. combining the meaning of the target body-part term and. its neighboring component, the disyllabic word is used to denote the meaning of ‘to face something’, e.g, miànshì l. QUANTITY:. ‘to be published’.. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is used to denote quantity, e.g., yímiàn l. ABILITY:. as a classifier.. combining the meaning of the target body-part term and its. 治政大word is used to denote a person’s neighboring component, the disyllabic 立 ‧ 國. l. INTELLECTION:. ‘sharp eyes’.. 學. eyesight, e.g., yīngyǎn. combining the meaning of the target body-part term and. ‧. its neighboring component, the disyllabic word is used to denote a. sit. y. ‘discerning eye’.. combining the meaning of the target body-part term and its. n. al. er. FOCUS:. io. l. Nat. person’s intelligence, e.g., míngyǎn. i n U. v. neighboring component, the disyllabic word is used to denote someone. Ch. engchi. or something catch one’s eyes, e.g., yàoyǎn l. FLAVOR:. ‘dazzling’.. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is used to denote a person’s taste experience, e.g., tiāozuǐ l. UTTERANCES:. ‘be fussy about food’.. combining the meaning of the target body-part term and. its neighboring component, the disyllabic word is used to denote a person’s use of uttered sounds for auditory communication, e.g., dǐngzuǐ ‘talk back’.. 39.

(53) Ten metaphorical meanings are identified: SURFACE, CONDITION, SIDE, NEAR SPACE, NEAR TIME, TINY HOLE, CENTER, ENTRY, MOUTH-SHAPED, and GATEWAY.. These metaphorical meanings are defined as follows: l. SURFACE:. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is projected to denote something’s surface, e.g., dìmiàn l. CONDITION:. ‘ground’.. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is projected to denote. 治政大‘market conditions’. something’s state, e.g., shìmiàn 立. combining the meaning of the target body-part term and its. 學. SIDE:. ‧ 國. l. neighboring component, the disyllabic word is projected to denote ‘the north side’.. combining the meaning of the target body-part term and. sit. y. NEAR SPACE:. Nat. l. ‧. certain direction, e.g., běimiàn. n. al. er. io. its neighboring component, the disyllabic word is projected to denote. i n U. the small distance in space, e.g., yǎnqián l. NEAR TIME:. Ch. engchi. v. ‘at present’.. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is projected to denote the small distance in time, e.g., mùqián l. TINY HOLE:. ‘at this moment’ .. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is projected to denote tiny hole, e.g., wǎngyǎn l. CENTER:. ‘tiny hole of a net’.. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is projected to denote. 40.

(54) something’s center or core, e.g., cíyǎn l. ENTRY:. ‘motif of lyrics’.. combining the meaning of the target body-part term and its. neighboring component, the disyllabic word is projected to denote something’s entries, e.g., shūmù l. MOUTH-SHAPED:. ‘bibliography’.. combining the meaning of the target body-part term. and its neighboring component, the disyllabic word is projected to denote an entity’s mouth-shaped external appearance, e.g., húzuǐ ‘spout’. l. 治政 : combining the meaning of大 the target body-part term and its 立. GATEWAY. ‧ 國. entrance, e.g., jìnkǒu. 學. neighboring component, the disyllabic word is projected to denote an ‘import’.. ‧ sit. y. Nat. Step Four (lexicalization categorization):. n. al. er. io. Packard’s categories of lexicalization (2000) were adopted as our model to. i n U. v. analyze the lexicalization of the words containing the target body-part terms in. Ch. engchi. Chinese. For example, the expression yǎnxiàn. may be coded under two. kinds of lexicalization: conventional lexicalization and metaphorical lexicalization. It is coded as conventional lexicalization when its word component meanings and the grammatical identity of the word components are fully retained, and its meaning is compositional, referring to ‘eyeline’. On the other hand, it is categorized into metaphorical lexicalization if the grammatical identity of the word component meanings are retained, but the meaning is not compositional but metaphorical, which refers to ‘spy’.. 41.

(55) CHAPTER IV RESULTS & DISCUSSION. This study analyzes and compares the linguistic expressions of three sets of body-part terms extracted from the largest, balanced and widely-used Mandarin Chinese corpus, the Sinica Corpus 4.0, and aims to answer the following questions: (A) What word component structures and their corresponding metonymic and metaphoric extensions can be detected in terms of the linguistic usages. 政治大 What lexicalization 立 types can be generalized based on the interaction containing body-part terms in corpora?. (B). ‧ 國. 學. between the word component structures of the words containing body-part terms and their corresponding metonymic and metaphoric extensions?. ‧. The word component structures of the linguistic usages containing body-part terms. y. Nat. io. sit. and their corresponding metonymic and metaphorical extensions are presented and. er. discussed in 4.1 and 4.2. The lexicalization types of the words containing body-part. al. n. v i n C h in 4.3. In addition, terms are displayed and discussed an overall discussion is engchi U. presented in 4.4. Before the main results are presented and discussed, some basic information about the six body-part terms is presented in Table 4.1.. Table 4.1 Number of types and frequency of the six body-part terms Liǎn. Miàn. Yǎn. Mù. Zuǐ. Kǒu. Total. 66. 334. 192. 122. 66. 275. 1,055. (%). 6.26. 31.66. 18.20. 11.56. 6.26. 26.07. 100. Frequency. 1134. 24855. 5848. 18783. 605. 8196. 59,421. (%). 1.91. 41.83. 9.84. 31.61. 1.02. 13.79. 100. Number of types. 42.

(56) Table 4.1 shows the number of types and the frequency of the six body-part terms in this study extracted from Sinica Corpus 4.0. In total, 1055 types of words containing the target body-part terms were analyzed: 66 for liǎn; 334 for miàn; 192 for yǎn; 122 for mù; 66 for zuǐ; and 275 for kǒu. In terms of the number of types, miàn (31.66%) shows the highest percentage in the six body-part terms, and liǎn (6.26%) and zuǐ (6.26%) show the lowest percentage. In terms of frequency of occurence, miàn (41.83%) is used the most frequently in the six body-part terms, and zuǐ (1.02%) is used the least frequently. The six body-part terms can further be paired according to. 治政大 miàn as a pair (which refers to the body-part they refer to, i.e., liǎn and its counterpart 立 ‧ 國. 學. faces); yǎn and mù as a pair (which refers to eyes); and zuǐ and kǒu as a pair (which. refers to mouths). If they are viewed in pairs, liǎn/miàn shows the highest percentage. ‧. of occurrences in types (37.92%) and frequency (43.74%) among the three pairs. As. sit. y. Nat. for yǎn/mù and zuǐ/kǒu, zuǐ/kǒu (32.33%) occur more frequently in types than yǎn/mù. n. al. er. io. (29.76%), while yǎn/mù (41.45%) occur more frequently than zuǐ/kǒu (14.81%).. Ch. engchi. 4.1 Word Component Structures. i n U. v. Word component structures and the meanings of a word are the two crucial factors which determine the type of lexicalization of that word. As mentioned in Chapter III, the word component structures of a word are analyzed based on the role of the components of a word and on their position within that word and based on the form classes as they are listed in the Chinese Dictionary and Chinese Wordnet (Packard 2000:65). Under this approach, the morphemes of the components of a word are described in terms of their syntactic form class identity, or part of speech, and they are. 43.

(57) often coded in this way: [M1 M2]w1, M1 and M2 represent the form class identity of the components of a word and w1 refers to the form class identity of the word (Packard 2000). The distribution of the word component structures in disyllabic words containing the six body-part terms is exhibited in Table 4.2.. Table 4.2 The distribution of word component structures in disyllabic words containing the six body-part terms %. Miàn. %. Yǎn. %. Mù. %. Zuǐ. %. Kǒu. %. 800. 70.55. 16455. 66.20. 4491. 76.80. 17573. 93.56. 443. 73.22. 4870. 59.42. 1. 0.01. 1. 0.17. 3. 0.04. 184. 2.24. [NN]V. 立. [NN]A [NN]D. 2.39. 12. 0.06. 9.79. 485. 1.95. 6. 0.10. 29. 0.15. 3. 0.50. 969. 11.82. 144. 12.70. 1139. 4.58. 577. 9.87. 613. 3.26. 125. 20.66. 1700. 20.74. 2. 0.03. 0. 0. 1. 0.01. 49. 0.84. 31. 0.17. 0. 85. 1.04. 12. 0.06. 56. 0.68. 146. 1.78. 70. 0.85. 1. 0.09. 176. 0.71. 29. 0.12. 127. 0.51. io. [NV]V. 140. al. 4.10. 273. n. [NV]A. 240. [VV]N. 1.45. 2. 0.33. er. [NV]N. 1.79. Nat. [VN]D. 446. ‧. [VN]A. 0.02. 111. ‧ 國. [VN]V. 1. 學. [VN]N. 政治大. y. [NN]N. Liǎn. sit. Structures. Ch. engchi U. v ni. 6. 0.02. 3806. 15.31. 1093. 4.40. 236. 4.04. 71. 0.38. 19. 3.14. 8. 0.10. [AN]V. 125. 0.50. 9. 0.15. 149. 0.79. 7. 1.16. 2. 0.02. [AN]A. 182. 0.73. 14. 0.24. 9. 0.05. [AN]D. 780. 3.14. 26. 0.44. 10. 0.05. 7. 0.12. 67. 0.82. [VV]V [AN]N. 35. 3.09. [NA]N [NA]V. 43. 3.79. 6. 0.02. 50. 0.85. Total. 1134. 100. 24855. 100. 5848. 100. 44. 18783. 100. 5. 0.83. 35. 0.43. 605. 100. 8196. 100.