In example (78), local information (within the clause where the target verb exists) is not distinctive enough for sense disambiguation. Then, the next step—Contextual dependence, exploring of extra information across clausal boundaries is necessary.
Likewise, in this module, each sense is assumed to be associated with some specific relevant lexical items through semantic linking. Following this criteria, in (79), the relevant items, liao jie 了解 ‘understand’ and zhu yi li 注意力 ‘attention’, are detected as crucial relevant items for sense 1 ‘seeing’ and sense 2 ‘watching’ shown in the re-exemplified example (79a) and (79b), respectively.
(79)a. 宋 明 理學 家 對 世 道 人 心 的影響， 反 而遠 Song ming li xue jia duei xhi dao ren xin de ien xiang fan er yuan Song Ming scientist to public morals DE influence, on the contrary far
不 如 小 說 戲劇來 得 深 入 普 遍。 因為 理學 家 的 著 作[Theme]
bu ru xiao shou xi ju lai de shen ru pu pian yin wei li xue jia de zhu zuo not as novel dramacome DE deep into widespread. Because scientist DE work
只有 知 識分子[Perceiver] 才 能 看 得 懂， 一 般 民 眾 zhi you zhi shi fen zi cai nan KAN de dong yi ban min zhong only intellectuals just can see DE understand, common people
是 沒 有 辦法 了解 的。 (sense 1 ‘seeing’) shi mei you ban fa liou jie de
is not have way understand DE.
‘In the Song and Ming dynasties, Scientists’ effects on the public morals was far behind the influenceof dramas and novels. Because only the intellectuals could undertand the Scientist’s works, and general people could not undertand them at all.’
b. 許多 家 長 帶 孩子 來 諮 詢：教 科書[Theme]，
xiu duo jia zhang dai hai zi lai zi xiuen jiao ke shu many parents bring children come consult: school book
孩子[Perceiver]總 是 看 不下 去，注意力 無 法集中 hai zi zong shi KAN bu xia qu zhu yi li wu fa ji zhong children always watch not down go attention cannot concertrate 煩躁，愛發脾氣。… (sense 2 ‘watching’)
fan zao ai fa pi qi irritable, easy to be angry
‘Many parents bring their children for a consultation: Children always have trouble reading their school books, cannot concentrate, and are irritable and easily angered’
Via similar semantic features, the linking between liao jie 了解 ‘understand’
and the sense of KAN 看 in (79a) or zhu yi li 注意力 ‘attention’ and the sense of
KAN 看 in (79b) is established. In BOW, a number of English synsets belonging to
sense 1 ‘seeing’ are searched and through translation they are equated to some Chinese lexical items which are relevant to sense 1 ‘seeing’, as in (80) (see the English synsets in appendix I):(80)
No Chinese synsets
1 碰見
2 接收
3 來自經驗的
4 明白,了解
5 辨出
… Sense 1 ‘seeing’
Also, there are also some relevant Chinese lexical items of sense 2 ‘watching’ by exploring English synsets in BOW:
(81)
Next, we have to account for the linking between sense1 and its synsets and the connection between sense 2 and its synsets. The lexical item liao jie 了 解
‘understand’ is an indicative anchor in (79a) to help identify sense 1 ‘seeing’. The reason could be that liao jie 了解 ‘understand’ is similar to ming bai/liao jie 明白/了 解 ‘catch’ within the synsets of sense 1 (consider (80)) for they both denote ‘become aware of something’. Equally, zhu yi li 注意力 ‘attention’ is an indicative lexical item because it is linked to zhu yi/liou xin 注意/留心 ‘attend’ within the synsets of sense 2 (see (81)) and they both denote ‘pay attention (to)’. Consequently, in this module, via contextual information, the sense of KAN 看 in example (79a) is identified as ‘seeing’, and in example (79b) as ‘watching’.
No. Chinese synsets
1 察覺
2 凝視
3 (注意)看
4 留心、留意
… Sense 2 ‘watching’
8. Conclusion
In this paper, a preliminary model of disambiguating polysemous words has been presented. Given the principle of economy, it is assumed that not all the senses of a polysemous word have equal weights and require exactly the same procedure for sense identification. Therefore, three steps are called upon in a sequence when needed. The first step focuses on frame-based information regarding participating frame elements and their expressions, and in most cases, senses can be distinguished in this step. However, in a few cases one basic pattern with core frame elements is shared by different senses. In this case, second step—colloconstruction is proposed to search for further informative syntactic adjuncts to help sense disambiguate. In the second step, word senses are distinguished beyond the expression of core arguments and a detailed lexical as well as grammatical association patterns are sought. However, those word senses sharing the same basic pattern(s) and same colloconstruction(s) or sharing the same basic pattern(s) without finding of distinctive colloconstruction(s) are still unsolved. Thus, the third step—contextual dependence is proposed. The final step to help disambiguate polysemous words is by contextual dependency cues. Through semantic properties, the relevant lexical items are investigated to trigger the target sense. By establishing a linkage to BOW, complicated senses of polysemous words can be identified through connecting the relevant lexical items to the synsets in BOW by their shared meanings. In this step, discourse-level factors are utilized with a clear measure of their semantic relations, just as Biq (1988) stated “Any effort to systematically identify and explain the different types of usage has to consider not only sentential entities but also entities which are outside of the sentences/ proposition yet relevant to the discourse.”
This model is to be applied in computation systems. Therefore, automatic disambiguation is a crucial part. In the preliminary stage, this frame-based model
has the frame elements manually tagged. However, automatic tagging of frame elements is not impossible. For instance, first, through generalizing the categorical composition of each frame elements (Part of Speech (or POS) from Sinica Corpus); a set of categorical groups corresponding to the frame element is explored. For example, the frame element self-mover consists of various categories [N], [N][Conj][N] or other categorical sets. The frame elements Area is also composed of a number of categorical compositions, such as [Nc], [Nec] or other categorical sets (see (82)):
(82) Self-mover < * < Area
[N] [Nc][N][Conj][N] [Nec]
. .. ...
These sets of categorical groups also tell the various structures of each frame element.
The automatic tagging program can be designed following the procedures above.
Further, in the second and third modules, there are also two available searching tools, Sinica corpus and Ontology in BOW. These two searching engines help automatic rummage for intra-clausal colloconstructions and inter-clausal relevant synsets in module 2 and 3 respectively. Combining these three parts, automatic sense disambiguation might be realized. This is an issue for further studies.
By redefining polysemy with operational mechanisms, this study has provided a linguistic model with theoretical validity to develop a computational system for disambiguation. Although, this study is by no means exhaustive, it nevertheless bears some significant implications on both theoretical linguistics as well as computational linguistics:
a) The solution integrates syntax, semantics and pragmatics in a step-by-step manner and make linguistic theories more accessible for computational
applications.
b) With a corpus-based multi-module approach, this model can be universally applied in other languages with the three clearly defined steps.
c) Word senses can be systematically detected via the three steps, incorporating existing linguistic theories that are interactive in nature.
d) As comprehensive investigations of Mandarin lexical semantics are under way (Liu 2002, 2004) and a bilingual ontological Wordnet (Sinica BOW) is also available (Huang et al 2004), the proposed model may offer a
workable resolution to develop a computer system dealing with polysemy resolution.
References
Adriaens, Geert, Steven L. Small, Garrison W. Cottrell, and Michael K. Tanenhaus.
1988. Lexical ambiguity resolution: perspectives from psycholinguistics, neuropsychology, and artificial intelligence. San Mateo, Calif.: Morgan Kaufmann Publishers.
Anatol, S., and Stefan Th. Gries. 2003. Collostructions: investigating the interaction of words and constructions. International Journal of Corpus Linguistics 2.
209-43.
Biber, Douglas, Susan Conrad and Randi Reppen 1998. Corpus linguistics:
investigating language structure and use. Cambridge: University Press.
Bybee, Joan L. 1985. Morphology: a study of the relation between meaning and form.
Amsterdam, Philadelphia: J. Benjamins.
Cau, Feng-Fu, Li-Zhong Cai, and Xiu-Ying Liu. 2001. Shen Ti Yu Pi Yu: Yu Yan Yu Ren Zhi De Shou Yao Jie Mian. Taipei: Wen He.
Chief, Lian-Cheng, Chu-Ren Huang, Keh-Jiann Chen, Mei-Chih Tsai, and Li-Li Chang. 2000. What can near synonyms tell us? Computational Linguistics and Chinese Language Processing 5.1. 47-60
Cruse, Alan. 1986. Lexical semantics. Cambridge: Cambridge University Press.
2004. Meaning in language: an introduction to semantics and pragmatics. New York: Oxford University Press.
Edmonds, Philip and Graeme Hirst. 2002. Near-Synonymy and Lexical Choice.
Computational Linguistics, 28. 2. 105-144
Fauconnier, Gilles. 1994. Mental spaces: aspects of meaning construction in natural language. Cambridge, NY, USA: Cambridge University Press.
Fauconnier, Gilles. 1997. Mappings in thought and language. Cambridge, NY, USA:
Cambridge University Press.
Fauconnier, Gilles, and Mark Turner. 2002. The way we think: conceptual blending and the mind's hidden complexities. New York: Basic Books.
Fellbaum, Christiane. 1998. WordNet: an electronic lexical database. Combridge, Mass.: MIT Press.
Feng, Heng. 2000. Ci Lei Wei Tan. Taipei: Da Qian Press.
Fillmore, Charles J. 1976. Frame semantics and the nature of language. In Annals of the New York Academy of Sciences: Conference on the Origin and Development of Language and Speech, Volume 280. 20-32.
1982. Towards a descriptiove framework for spatial deixis. In R.J. Jarvella and W.
Klein (eds.), Speech, place and action. New York:Wiley.
and B. T. S. Atkins. 1992. Towards a frame-based organization of the lexicon:
The semantics of RISK and its neighbors. In Lehrer, A and E. Kittay (eds.), Frames, fields, and contrast: new essays in semantics and lexical organization.
Hillsdale: Lawrence Erlbaum Associates. 75-102.
Firth, John Rupert. 1957. Papers in linguistics 1934-1951. London: Oxford University Press.
Gale, William A., Kenneth W. Church and David Yarowsky. 1992. One sense per discourse. Proceedings of the 4th DARPA Speech and Natural Language Workshop.
Geeraerts, Dirk. 1993. Vagueness’s puzzles, polysemy’s vagaries. Cognitive Linguistics, 4.3. 223-272.
Goldberg, Adele E. 1996. Conceptual structure, discourse, and language. Stanford, Calif.: CSLI Publications.
Gorfein, David S. 1989. Resolving semantic ambiguity. New York: Springer-Verlag.
Gries, Stefan Th. 2005. Syntactic priming: A corpus-based approach. Journal of Psycholinguistic Research 34. 365-99.
Hopper, Paul J., and Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Language 2.251-99.
Hu, Yu-Shu and Xiao Fan 1995. Dong Ci Yan Jiu. He-nan: He Nan University Press.
Huang, Chu-Ren, Ru-Ying Chang, and Shiang-Bin. Lee. 2004. Sinica BOW (Bilingual Ontological Wordnet): Integration of Bilingual WordNet and SUMO.
Paper presented at 4th International Conference on Language Resources and Evaluation (LREC 2004). Lisbon. Portugal. 26-28 May, 2004.
Keh-Jiann Chen and Li-Li chang. 1996. Segmentation standard for Chinese natural language processing. Proceedings of the 1996 International Conference on Computational Linguistics (COLING 96). August. Corenhagan, Denmark.
Ikegaya, Yuki, Yasuhiro Noguchi, and Satoru Kogure. 2005. Integration of dependency analysis with semantic analysis referring to the context. Proceedings of PACLIC 19, the 19th Asia-Pacific Conference on Language, Information and Computation. Academia Sinica, Taipei.
Jackendoff, Ray S. 1985. Multiple subcategorization and the theta-criterion: the case of climb. Natural Language and Linguistic Theory, 3. 271-295.
Johnson-Laird, P. N., R. M. J. Byrne, and P. Tabossi. 1989. Resoning by model: the case of multiple quantification. Psychological Review 96, 658-673.
Katz, Jerrold J. 1972. Semantic theory. New York: Harper and Row.
Lakoff, George. 1987. Women, fire, and dangerous things: what categories reveal about the mind. Chicago: University of Chicago Press.
Lai, Huei-ling. 2004a. Same constructions but different grammaticalization paths of Hakka lau and tung. Proceedings of Fifth International symposium on Taiwanese Languages and Teaching, 195-210. Prudence University, Taichung.
Lehrer, Adrienne, and Eva Feder Kittay. 1992. Frames, fields, and contrasts :new essays in semantic and lexical organization. Hillsdale, N.J.: L. Erlbaum Associates.
Li, W., Q. Lu, and R. Xu. 2005. Similarity based chinese synonym collocation extraction. Computational Linguistics and Chinese Language Processing 1.
123-44
Liu, Mei-chun. 1994. Discourse explanations for the choice of jiu and cai in Mandarin conversation. Chinese Language and Linguistics. 2, 671-709
2002. Mandarin verbal semantics: a corpus-based approach. Taipei, Taiwan:
Crane Publishing.
and Yiching Wu. 2004. A frame-based analysis of polysemy-verbs of Encoding in Mandarin. Paper presented in The 12th Annual Conference of the IACL.
Nankai University. Tianjin, China.
Lien, Chin-fa. 2000. A frame-based account of lexical polysemy in Taiwanese.
Language and Linguistics 1.1. 119-138. Beijing: Shang Wu Yin Shu Guan.
Xiou-ying, Liu. 2004b Taiwan South-Min yi dong dong ci ‘Zou’de duo yi xing ji gai nian jie gou: yu yi ien shen de tu jing. Yu Yan Xue Lun Cong. 29. 334-371.
Lin, Chienjer Charles and Ahrens, Kathleen. 2005. How many meanings does a
word have? Meaning estimation in Chinese and English. In James Menitt and
William S-Y Wang (eds.) Language Acquisition, Change and Emergence:Essays in Evolutionary Linguistics. City University of Hong Kong Press.
Lv, Shu-Xiang. 1992. Zhong Guo Wen Fa Yao Lue. Taiwan: Wei Shi Zhe Press.
Perter, Harder. 1996. Functional Semantics: a theory of meaning, structure and tense in English. New York: Mouton de Gruyter.
Pustejovsky, James. 1995. The generative lexicon. Cambridge, Mass.: MIT Press.
Pustejovsky, James and Branimir Boguraev. 1996. Lexical semantics: the problem of polysemy. New York: Oxford University Press.
Ravin, Yael. 1990. Lexical semantics without thematic roles. Oxford: Oxford University Press.
Robins, Robert Henry. 1979. A short history of linguistics. Bloomington: Indiana University Press.
Rosch, Eleanor. 1977. Human categorization, in N. Warren ed., Advances in cross-cultural psychology, vol. 7. London: Academic Press.
Saeed, John I. 1997. Semantics. Cambridge, Mass.: Blackwell Publishers.
Schutze, Hinrich. 1997. Ambiguity resolution in language learning: computational and cognitive models. Stanford, Calif.: CSLI Publications.
Su, Lily, I-Wen. 2002. What can metaphors tell us about culture? Language And Linguistics 3.3, 589-613.
Swinney, David A. 1979. Lexical access during sentence comprehension: (re) consideration of context effects. Journal of Verbal Learning and Verbal Behavior 18, 645-659
Taylor, John. 1989. Linguistic categorization: prototypes in linguistic theory. Oxford:
Oxford University Press.
Towell, Geoffrey and Ellen M. Voorhees. 1998. Disambiguating highly ambiguous words. Computational Linguistics 24.1, 125-145
Tsai, Jia-Lin., Wen-Lian. Hsu, and Jebg-Woei. Su. 2002. Word sense disambiguation and sense-based NV event frame identifier. Computational Linguistics and Chinese Language Processing 1. 29-46.
Tsai, Mei-Chih. 2005. Word sense disambiguation based on syntactic construction.
Computational Linguistics and Chinese Language Processing 10.4:483-494.
Biq, Yung-O 1988. From focus in proposition to focus in speech situation: cai and jiu in Mandarin Chinese. Journal of Chinese Linguistics 16.1. 72-108.
Wang, Huei. 2002. A study on noun sense disambiguation based on syntagmatic features. Computional Linguistics and Chinese Language Processing 2. 77-88.
Wierzbicka, Anna. 1990. Prototypes save: on the uses and abuses of the notion of prototypes in linguistics and related fields’, in S. L. Tsohatzidis ed., Meanings and prototype: studies in linguistic categorization. London: Routledge and Kegan Paul.
Wittgenstenin, L. 1953. Philosophical investigations. Oxford: Basil Blackwell and Mott.
1958. The blue and brown books. Oxford: Basil Blackwell and Mott.
Appendix I
Examples of tagged corpus (for the purpose of economy, not all tagged data are
listed)
others
隨心所欲，一般機場內皆設有租車服務。如果從倫敦清晨出發，[Self-mover]走Ｍ１Ａ１公 路[Path]下午３點左右便可抵達愛丁堡[Goal]。Ｍ表示高速公路，
Self-mover < * < Direction
政 大 東 語 系 洪 士 培 的 男 友 是 日 本 人 ， 她 說 雙 方[Self-mover] 已 決 定 未 來 要 走 的 方 向
Self-mover < Direction < *
1 覺得這些教法真是至高無上的真理，他覺得身心輕安，[Self-mover]又再向前走。沒想到，
Theme < Source < Goal <*
優良表現，特別計畫了兩天一夜的短程旅行。車隊[Theme]從臺北[Source]往宜蘭[]走，經南
還有一段很長的路要走，[Theme]如何在未來的資訊高速公路[Path]上走得平穩安全，端賴 吸收既有的經驗，提早做準備。
輛的車隊。這個車隊在仙跡岩停車，[Theme]要沿著峭壁中間的公路[Path]往前走三公里，
才能看到滑坡的場面。仙跡岩過去三四百公尺，就遇到 theme < Direction < *
「博物館的生命是永久，中華文化[Theme]也一直在往前[Direction]走，」秦院長表示，為
他的鬼魂走出雕像，拖著瘦弱的身體，[CNI/Self-mover]再走一趟宿命的旅程[Area]。幽魅 的鬼魂，口中哼著送葬的哀歌，一語道斷
Self-mover < Area < *
東海岸的東管處,順便到泰源幽谷看看,再來我們[Self-mover]到南橫[Area]走一趟,走了這一 趟回來,我發現我們都已經不怕冷了
有種難以言喻的美，繁複的生命景致讓人目不暇給，如果你[Self-mover]到紅樹林[Area]走 一遭，就能夠同時進行賞樹、賞鳥、賞蟹等貼近自然的活動，到處
帆船，乃荷蘭村內最具象徵性的展示物。小人國；[Self-mover]在小人國內[Area]走一遭，
彷彿自己便是童話中的巨人格列弗。威簾塔（
Area < Self-mover < *
著塔克金溪縱谷與司馬庫斯部落[Area]遙遙相望。我[Self-mover]希望下次有機會去走一 趟。鎮西堡靠近新光，標高超過２０００公尺，此地早晚溫差
中國人常做的運動，[Self-mover]最好就是哪一天一大清早到公園[A]去走一趟。早晨在公園 裡運動的，男女老少都有
Sense 4 ‘leaving’
self-mover < *
師父教過我的，怎地忘了？」瓦耳拉齊喝道：「你[Self-mover]再不走，我要殺你了！」
我ＫＴＶ和同伴們一定會感激不盡的。現在，我[Self-mover]要走了，希望下次再來的時候，
住的是一片安全美麗的大地
尼奧卻說：「你們玩吧，我和秀子[Self-mover]吃完了就走。」「何必太嚴肅呢？看看何妨？」
急著想出來，聽聽屋子裡很久沒有聲音了，料想皇帝[Self-mover]大概已經走了，就一邊兒 把頭伸出來一邊兒問：老頭子走了吧﹖
*＜self-mover
確實，幾年前香港移民走了一批高級職員[Self-mover]，形成真空，底下的人藉此機會升了 上來。
所以今天能決定中國將來之運命者，第一件事便是請走宋子文[Self-mover]，否則政府必然 垮台
Self-mover < * < Area
滾滾古路長，滿目空雲待夕陽，回頭一望家三遠，不知何事[SCNI/elf-mover]走他鄉[Area]。」
請問這是出家人的境界嗎？