lin

Transcription

1 Comparative Analysis of English and Chinese Passive Voice in Machine Translation Lin Sun Dissertação/Dissertation Mestrado Internacional em Processamento de Linguagem Natural e Indústrias da Língua International Masters in Natural Language Processing and Human Language Technologies Orientação/Supervised by Professor Jorge Baptista, Universidade do Algarve Professor Izabella Thomas, Universite de Franche Comté 2014

2

3 Comparative Analysis of English and Chinese Passive Voice in Machine Translation Lin Sun Dissertação/Dissertation Mestrado Internacional em Processamento de Linguagem Natural e Indústrias da Língua International Masters in Natural Language Processing and Human Language Technologies Orientação/Supervised by Professor Jorge Baptista, Universidade do Algarve Professor Izabella Thomas, Universite de Franche Comté This project was supported by the European Commission, Education & Training, Eramus Mundus: EMMC , Erasmus Mundus Masters in NLP & HLT 2014 II

4 III

5 Declaração de autoria e autorização de publicação de trabalho académico TÍTULO DA OBRA: Comparative Analysis of English and Chinese Passive Voice in Machine Translation NOME DO AUTOR: LIN SUN Declaro que o presente trabalho é de minha autoria exclusiva, estando os elementos produzidos por terceiros devidamente referenciados. Declaro, também, que o conteúdo não constitui tradução, reorganização de qualquer forma de manipulação de documentos produzidos por terceiros. Declaro, ainda, que a Universidade do Algarve Tem o direito, perpétuo e sem limites geográficos, de arquivar e publicitar este trabalho através de exemplares impressos reproduzidos em papel ou de forma digital, ou por qualquer outro meio conhecido ou que venha a ser inventado, de o divulgar através de repositórios científicos e de admitir a sua copia e distribuição com objetivos educacionais ou de investigação, não comerciais, desde que seja dado crédito ao autor e editor. Faro, 05 de Maio de 2014 IV

6 V

7 Abstract With the development of the human society and the progress of technology, machine translation not only makes it possible for human to translate one source language to the other target languages automatically, but also makes human communication easier than before. It is undeniable that in recent years machine translation has been improved a lot, however, due to the fact that human language is much more complicated and each language has its uniqueness, machine translation still has a long way to go before it can give a qualified output. As a very important linguistic phenomenon, passive voice has been commonly used in different languages around the world, it is also one of the complicated grammatical aspects for a machine to translate. Take the translation of English and Chinese as an example, since these two languages do not belong to the same family, it is normal that they differ in varied grammatical aspects, including passive voice. In this study, the major differences and similarities of the passive voice in these two languages have been pointed out. In field of machine translation, two machine translation system, Google translator and Baidu translator have been tested with three hundred passive voice sentences taken form a English-Chinese parallel corpus, to see their capacity of translating the passive voice from one language to the other. As a result, all the mistakes have been categorized in order to improve the quality of machine translation. Keywords: Passive Voice, English, Chinese, Machine Translation, Google translator, Baidu translator. VI

8 Extended abstract Machine translation, with its undeniable significance, makes it more convenient for people to communicate even they do not speak the same language. It also makes a notable contribution to the progress of globalization. However, even after years of development, it still cannot perfectly translate one language to the other. There are more than 6909 languages spoken in 193 countries all around the world, each of the languages has their own uniqueness, and each country has their own culture. Although languages, sometimes share some similarities, but still, all these factors even make it hard for human to translate one source language to another. With different backgrounds, one sentence may sometimes have several versions of translation done by different persons. It can be imagined how hard it could be for the machine to give a qualified translation. Take the translation between English and Chinese as an example, these two languages belong to two language families. English, comes from the Indo-European language family while Chinese forms one of the branches of the Sino-Tibetan language family. Due to this fact, they are dissimilar in varying grammatical aspects. When a machine is translating these two languages, it can face the problems such as the lexical ambiguity, syntactic ambiguity and semantic ambiguity and so on. In this research, we mainly focused on the study of comparative analysis to the passive voice in English and Chinese. Almost all the languages have passive voice despite of the varied syntactic constructions. English and Chinese are, of course, no exception. However, the passive voices in both languages are differently formed and at the same time share some grammatical similarities. For example, the similarity of the passive voice in these two languages is: Syntactically, the passive voice marked by By/with-phrase in English can be seen as the same like the passive voice marker 被 (bei) in Chinese. While from the perspective of semantic analysis, we can find out that in both English and Chinese, passive voice can be used when it is more important to draw our attention to the person or thing acted upon, and so on. Considered to their differences, for example, in English, passive voice is mainly realized by the copula verb plus past participle, and thus forms such a syntactic structure such as Be + V-ed, but in Chinese, there is no special structure that is clearly different from an active voice. After clarifying all these differences and similarities, it makes easy for us to to detect some major problems faced by machine translation systems while they translating the passive voice sentences. With this purpose, we randomly collected 200 passive sentences from an English-Chinese parallel corpus build by Professor Lu Wei: 100 English passive voice VII

9 sentences, and their corpus translation, 100 Chinese passive voice sentences with their corpus translation. By using these 200 passive voice sentences, we examined two machine translation system, Google translator (a world-wild machine translation system) and Baidu translator (a Chinese machine translation system) to see their capacity in translating the typical passive voice sentences in different syntactic constructions in both English and Chinese. Together with all these sentences, we manually evaluated the translations provided by Google translator and Baidu translator. All the problems and mistakes had been classified and calculated. The major problems that have been categorized are: the completely wrong translation sentences with different kinds of grammatical problems; partly right sentences with an acceptable change of the synonym words, and so on. Based on our research, we hope that our analysis make some contribution to the comparative analysis of the passive voice in both languages, and also could help to improve machine translation concerned with the passive voice in English and Chinese. Keywords: Passive Voice, English, Chinese, Machine Translation, Google translator, Baidu translator. VIII

10 Resumo Com o desenvolvimento da sociedade humana e do progresso da tecnologia, a tradução automática não só torna possível para o ser humano traduzir de um idioma de origem para seja outras línguas-alvo automaticamente, mas também faz com que a comunicação humana mais fácil do que antes. É inegável que, nos últimos anos, a tradução automática foi melhorar. No entanto, devido ao fato de a linguagem humana ser muito mais complicada e cada língua ter a sua singularidade, a tradução automática ainda tem um longo caminho a percorrer antes de qualidade dar uma saída qualificadade. Como um fenómeno lingüístico muito importante sendo, a voz passiva tem sido comumente usada em diferentes línguas ao redor do mundo considerando, é também um dos aspectos gramaticais complicados de traduzir para uma máquina. Pegue a tradução de Inglês para Chinês como um exemplo, uma vez que estas duas línguas não pertencem à mesma família, é normal que elas diferam em vários aspectos gramaticais, incluindo voz passiva. Neste estudo, as principais diferenças e semelhanças entre a voz passiva nestas duas línguas foram apontadas, Dois sistema de tradução automática, o Google Tradutor e Baidu Tradutor, foram testados com três centenas de frases na voz passiva retiradas de um corpus paralelo Inglês-Chinês, para ver a sua capacidade de traduzir a voz passiva de uma língua para a outra. Como resultado, todos os erros encontrados forma classificados, com vista a melhorar a qualidade de tradução automática. Palavras-chave: Voz passive, Inglês, Chinês, Tradução Automática, Tradutor Google, Tradutor Baidu IX

11 Resumo Alargado A tradução automática TA, com a sua inegável importância, fa tradução automática, com a sua inegável importância, facilita a comunicação entre as pessoas, mesmo quando não falam a mesma língua, constituindo uma contribuição notável. constituíndo uma contribuição notável para o progresso da globalização. No entanto, mesmo depois de anos de desenvolvimento, a TA ainda não consegue traduzir perfeitamente de uma língua para a outra. Há mais de 6909 línguas faladas em 193 países em todo o mundo, cada uma com sua própria singularidade, e cada país com a sua própria cultura. Embora as linguas, às vezes compartilham algumas semelhanças, ainda assim, todos esses fatores tornam difícil para o ser humano para traduzir um idioma de origem para outro. Com diferentes origens, uma frase pode, por vezes, ter várias versões de tradução, realizadas por pessoas diferentes, de qualidade.. Pode-se imaginar o quão difícil pode ser para a máquina para dar uma tradução. considere-se a tradução entre Inglês e Chinês como um exemplo: estas duas línguas pertencem a duas famílias linguísticas distintas. O Inglês, vem da família de línguas indo-europeias, enquanto a lingua Mandarin Chinesa forma um dos ramos da família de línguas sino-tibetanas. Devido a este fato, são diferentes em vários aspectos gramaticais. Quando uma máquina tenta traduzir estas duas línguas, pode enfrentar vários problemas, tais como a ambiguidade lexical, sintática e semântica e assim por diante. Nesta pesquisa, focamos principalmente estudo e análise comparativa da voz passiva em Inglês e Chinês. Quase todas as línguas têm voz passiva. Inglês e Chinês, é claro, não são exceção. No entanto, a voz passiva em ambos os idiomas são diferentemente formadas e ao mesmo tempo compartilhar algumas semelhanças gramaticais. Por exemplo, a semelhança da voz passiva nestas duas línguas é :sintaticamente, a voz passiva é marcada pelo complemento preposicional introduzido por "by" (por) ou "with" (com) em Inglês; por seu turno, em Chinês, o mesmo estatuto pode ser atribuído ao marcador de passiva (bei). Do ponto de vista da análise semântica, podemos considerar que, tanto em Inglês como em Chinês, a voz passiva pode ser utilizada quando. Do ponto de vista da análise semântica, que em Inglês e Chinês, a voz passiva pode ser usada quando é mais importante para chamar a nossa atenção para a pessoa ou coisa posta em prática. Considerando as suas diferenças, por exemplo, em Inglês, a voz passiva é realizada principalmente pelo verbo cópulativo e o particípio passado de verbo, e, assim, forma uma estrutura sintácticqa be + v-ed,contado, em Chinês, não há nenhuma estrutura especial que seja claramente diferente de uma voz ativa. Depois de esclarecer todas essas diferenças e semelhanças, torna-se mais fácil para detectar alguns dos principais problemas enfrentados pelos X

12 sistemas de tradução automática, quando procuram traduzir frases na voz passiva. Com este propósito, foram coligidas aleatoriamente 200 frases passivas de um corpus paralelo Inglês - Chinês, construído pelo Prof. Lu Wei: 100 frases passivas em Inglês e sua tradução em Chinês e 100 frases passivas em Chinês e respetiva tradução para Inglês. Ao utilizar essas 200 frases de voz passiva, examinamos dois sistema de tradução automática, o Google Tradutor (um sistema selvagem mundial de tradução automática mundialmente utilizado) e Baidu Tradutor (um sistema de tradução automática chinês ) para ver a sua capacidade de traduzir frases típicas na voz passiva, em diferentes construções sintáticas, em Inglês e em Chinês. As traduções fornecidas por cada um dos sistemas foram manualmente avaliadas, valiadas manualmente as traduções fornecidas pela Google tradutor e Baidu tradutor. Todos os problemas e os erros foram classificados e contabilizados. Os principais problemas que foram categorizados foram: as frases de tradução completamente errada com diferentes tipos de problemas gramaticais; frases em parte corretas com uma mudança aceitável de certas palavras por sinónimos, e assim por diante. Com base na nossa pesquisa, esperamos fazer alguma contribuição da análise comparativa para a voz passiva em ambas as línguas, e também ajudar a melhorar a tradução automática da voz passiva em Inglês e Chinês. Palavras-chave: Voz passive, Inglês, Chinês, Tradução Automática, Tradutor Google, Tradutor Baidu XI

13 Acknowledgment First and foremost, I would like to show my deepest gratitude to my supervisors, Professor Izabella Thomas and Professor Jorge Baptista. Without their enlightening instruction, impressive kindness and patience, I could not have completed my thesis. I shall extend my thanks to the professors in Universite de Franche-comte for their kindness and help. My sincere appreciation also goes to the European Commission, Education &Training, Erasmus Mundus Masters in NLP & HLT, without their great support, I could not finish my study in such a good way. Last, I would like to thank my love to my parents, my friends for their love and help. XII

14 Table of Contents 1. Introduction...10 Research Objectives Research Questions...10 Layout of the Study Chapter 1 Comparative study of the passive voice in English and Chinese Literature review Different viewpoints on the passive voice in English grammar General ideas on the passive voice in Chinese grammar Comparative studies of the passive voice in English and Chinese Brief Introduction to Passive Voice in English and Chinese English passive voice Chinese passive voice Summary Methodology Corpus Machine translation system Evaluation Data Analysis Conclusion Appendix Appendix Appendix XIII

15 Appendix Index of Tables Table 1 Summary of frequency of syntactic and lexical passives in Chinese; source: Andrew Methven (2006) Table 2 Passive Voice Used in Different Tenses; source: Zhang Daozhen (1998) Table 3 Error analysis: Chinese to English Table 4 Error analysis: English to Chinese Index of Figures Figure 1. Expression of the active voice in Formal Grammar; source: Chomsky (1957) Figure 2 Expression of the passive voice in Formal Grammar; siurce Chomsky (1957)...16 Figure 3 Classification of translation errors...35 XIV

16 1. Introduction It is a common knowledge that verbs are often said to be either active or passive in voice, the voice (also called diathesis) can show the relationships among the subject, the object and the verb. In English language for example, if the subject acts, which means the subject of the sentence is the agent or a doer of the action, then the verb takes the form of active voice. To the contrary, if the subject of the sentence is neither a doer nor a be-er, but is the patient or recipient, which means that the subject is acted upon by some other agent; the verb takes the form of passive voice. Active voice is considered easier to be understood than passive voice. For example in English, the passive voice has an original formation of be/get + v -ed, such as is made, are taken, will be moved, of which the auxiliary verb be is called static passive auxiliary and get is called dynamic passive auxiliary. Also, when an agent of the action (that is, the person or entity performing the action) must be described, a prepositional phrase will be introduced. That is, a by phrase will be used so as to take the agent. Almost all the languages have passive voice despite of the varied syntactic constructions. English and Chinese (in this thesis, all the studies and researches are based on the Mandarin Chinese), of course, no exception. However, the passive voices in both languages are differently formed and at the same time share some grammatical similarities. Also, passive voice has been considered as one of the most difficult grammatical points for any machine translation (MT) system to deal with. Due to this fact, we started our research. Research Objectives In order to launch our investigation, we set three objectives. Firstly, we need to compare both the English and Chinese passive voices, so as to shed light on the major differences existing at lexical, syntactic and semantic levels. Secondly, we selected 150 sentences for each language form a bilingual corpus, and take two commonly used machine translation systems to check their capacity to deal with sentences in passive voice. Thirdly, the data has been analysed, in order to find out the major difficulty for machine to translate the passive voice between two languages, and to see how much the machine translation systems can deal with the possible differences exhibited in passive voices in both languages. By doing so, we hope to find some common problems that machine translation systems fail to cope with. One more thing needs to be stressed here is that as passive voice is one of the most complicate grammatical phenomena in both languages we only focus on passive 10

17 voice found in predicate. All the other forms of passive voice are excluded, for example, the past participle, and the non-finite forms of verbs and so on. Research Questions To fulfill our objectives, we thus raise three research questions accordingly. a. What are the major differences between English and Chinese when passive voice is employed? To answer this question, we need to check at almost all grammatical levels, namely, lexical, syntactic and semantic levels. b. What kind of performance does a machine translation system have when translating sentences in passive voice between the given languages, especially those sentences which are constructed in quite different structures? c. What are the common problems found in the machine-translated sentences? And what are the potential reasons that cause such problems? Layout of the Study In the introduction part of this thesis, we introduce the objectives of the research, the research questions, and the layout of the study. Chapter One is the comparative study of the passive voice in English and Chinese. In this part we will examine and introduce some relevant studies done by researchers about English and Chinese passive voice form different grammatical features. In order to have a briefly introducing the basic grammatical features of passive voices in both languages in general and then further compare and differentiate some language-specific features in terms of lexical choices, syntactic constructions and semantic contents. In Chapter two, we focus the study on the comparative analysis of the English and Chinese passive voice in machine translation, the first part, literature review will have a brief introduction of some researches about the machine translation, so as the development of the machine translation in China. Also, the researches about the passive voice in machine translation will be pointed out. After knowing the general situation, we will introduce the methodology, that is the corpus we use, the two machine translation system, and the evaluation skill. The last part of this chapter is the analysis of results, to categorize the different types of mistakes and so on. At last, will be the conclusion of the whole thesis, and the possible suggestions. 11

18 Chapter 1 Comparative study of the passive voice in English and Chinese 1.1. Literature review Passive voice has been a hot topic for linguists throughout the world. We can find many literatures very easily concerning this topic. Different linguists lay focus on varied dimensions of the usage of passive voice. To facilitate our research, we purposely choose some of them so that we can take advantage of their theoretical framework or experiences Different viewpoints on the passive voice in English grammar Passive voice, as an important grammatical feature, a great amount of researches had been done within the development of the English grammar. A. Traditional grammar The study of English grammar started from the early seventeenth centuries. In traditional grammar, the scholars divided the English voice into the active voice and the passive voice. In the book Introduction to the Traditional Grammar(Bella Millett, 2011), it mentions that Verb, a word expressing a state or action...the verb has two voices, active (e.g. the dog bit the man) and passive (e.g. the man was bitten by the dog). The passive voice is formed by the verb be and the past participle of the verb. Traditional linguistics considered the study of the passive voice as the learning of its relation to their active counterparts in both meaning and structure. Ouirk (1985), one of the most well-known linguist of the traditional grammar, holds the view that Voice is a grammatical category which makes it possible to view the action of a sentence in either two ways, without change in the facts reported. The English voice is composed of the active voice and the passive voice. From his point of view, the active-passive relation has two grammatical levels: the verb phrase and the clause. In the verb phrase, the difference between the two voices is that the passive voice adds a form of the auxiliary BE followed by the past participle (-ed participle) of the main verb. While in the clause level, to change the active into the passive requires the arrangement of two clause elements and one addition: (a) the active subject becomes the passive AGENT; (b) the active object becomes the passive subject; (c) the preposition by is added to introduce the agent. The prepositional agent phrase (AGENT BY-PHRASE) of passive sentences is an optional sentence element. 12

19 He also summarized that the transformation of the voice can be expressed by the following forms: noun phrase1 + active verb phrase + noun phrase2 noun phrase2 + passive verb phrase + (by noun phrase1) Although the structure of a sentence will change under the voice transformation, its meaning remains the same. (1) The men respect her. (2). She is respected by the men. Sentence (1) is in active voice while sentence (2) is in passive voice transformed by sentence (1), they are differently formed, however still remains the same meaning. In his description, he also pointed out that the purely formal definition of the passive, viz that the clause contains the construction be (or get) + -ed participle, is very broad. Under this circumstance, the following passive gradient has been pointed out by him. 1. Central passives (1). This violin was made by my father. (2). This conclusion is hardly justified by the results. (3). Coal has been replaced by oil. (4). This difficulty can be avoided in several ways. These four sentences show the example of the passive voice type called central passives, or true passives, while the first two sentences have a direct active-passive relation. Take the first sentence as an example, the passive voice sentence can be transformed into active voice as: My father made this violin. Sentence (3) have some unclarity about the nature of the active counterpart, there are two possible active counterparts, depending on the interpretation of the by-phrase. It means that the passive voice have two possible active forms. The sentence can be transformed as: (1) Oil has replaced coal. (2) (People in many countries) has replaced coal by oil. 13

20 In the former case, the by-phrase has been interpreted as an agent phrase corresponding to the active subject, but in the latter case, the by-phrase has been given an instrumental interpretation. Sentence (4) shows the most common type of passive, which has no expressed agent, and so leaves the subject of the active counterpart undetermined. 2. Semi-passive (5) We are encouraged to go on with the project. (6) Leonard was interested in linguistics. Sentence (5) and (6) represent the semi-passive or the mixed class, in which the past participle has both verbal and adjectival properties. They are verb-like in having active analogues, for example, sentence (6) can be transformed as Linguistics interested Leonard. On the other hand, their adjectival properties include the possibility of: (a). Coordinating the participle with an adjective; (b). Modifying the participle with quite, rather, more, etc; (c). Replacing be by a lexical copular verb such as feel or seem: So the sentence (6) can also be transformed as 'Leonard seemed very interested in and keen on linguistics'. 3. Pseudo-passives (7) The building is already demolished. (8) The modem world is getting ['becoming'] more highly industrialized and mechanized. These two examples can be called pseudo-passives, since it is chiefly only their superficial form of verb + -ed participle that recommends them for consideration as passives. Although Quirk s explanation of the passive voice system and classification has been adopted during the period of his time, with the lapper of time, some questions have been raised by the newly developed grammarians and linguist. Their doubts like if the semi-passives and the pseudo-passive voices are truly passive voice or can all 14

21 the direct object of activities be changed into passives is hard to answer by the traditional grammarians researches. Under this circumstance, the pure formation of the passive voice be + V + past participle accepted by the traditional linguists can no longer satisfy the strict researches. Also, the simple belief of the active and passive pair is different structured but same in meaning can no longer be stable, for example, the active sentence Tom cannot do it expresses the ability of the doer, while the passive sentence It cannot been done by Tom shows more possibility of an action. B. Transformational-Generative Linguistic Approach With these non-ignorable questions, the newly theory has been put forward. The development of the transformational generative grammar widened the research of the passive structure. Chomsky, in the field of the transformational-generative grammar held the point that the grammar should be formalized. His discussion forced on different kinds of transformational phenomena and gave some complex rules that supplied them. In his book Syntactic Structure (1957), the idea of each sentence in a language has two levels of representation a deep structure and a surface structure has been developed. He believed that the deep structure represented the core semantic relations of a sentence and was mapped on the surface structure via transformations. Specialized in passive voice, Chomsky (1957) pointed out that it is not enough to use phrase-structure rules alone to indicate the relationship between active forms and their passive counterparts, so he used some certain passive transformation rules to explain the derivation of passives. In his view, the active is taken as kernel construction, while the passive is derived from the deep structure by a sequence of transformations. In the transformational grammar, several rules can show the relation of the active-passive pair. For example, the rearrangements: (1) NP1 and NP2 in a different order; (2) the insertion of the auxiliary element of be+en; (3) the insertion of the element of by to the left of the permuted NP1. From his point of view, if the active-passive sentences are, (1) Tom cleaned room. (2) The room was cleaned by Tom. Then sentence (1) can be expressed syntactically in his formal grammar as: 15

22 S NP1 VP N V Aux NP2 Tom cleaned past Det N the room Figure 1. Expression of the active voice in Formal Grammar; source: Chomsky (1957) While the passive voice sentence: The room was cleaned by Tom should be expressed as: S NP1 VP D N Aux V PP The room past be en cleaned p NP2 by Tom Figure 2 Expression of the passive voice in Formal Grammar; siurce Chomsky (1957) Sentence (2) is derived sentence from its corresponding active form sentence (2). These two sentences share the same underlying source-identical kernel sentences, and the deep structure of the passive sentence is basically active in organization. Besides, the object of by in passive is degraded form the subject in the corresponding active sentence, or some grammatical morphemes marking the passive construction (by, be 16

23 and ed participle) are meaningless entities with only formal or syntactic functions. While in his later research, he pointed out the limitations of this passive transformational rule, when the active and its passive pair have qualifiers, they are not potentially synonymous. For example, (1) Everyone in the room speaks at least two languages. (2) At least two languages are spoken by everyone in the room. In this example, sentence (1) is not synonymous with sentence (2). The first sentence focus on the fact that all people can speak at least two languages, while the second sentence refers to that all of them speak the same two or more languages, that is to say when one person in the room can speak English and French, the other can speak Chinese and English, for sentence (1) it is available, however it is unacceptable by sentence (2). In 1971, Chomsky put out a new transformational model and called it the Extended Standard Theory. In this new version, he concludes that semantic interpretation cannot wholly be determined by properties of deep structure because many aspects of surface structure also contribute to the semantic interpretation. In his later research called Trace Theory, he asserts that all interpretation takes place at the level of surface structure. This level is enriched by traces left by the application if the movement rules. He explained that a moved NP is taken to leave behind a trace which remains in surface structure unless it is covered up by the insertion of a lexical item into its place in the grammar tree, the trace is interpreted as a bound variable, with all the relevant properties of its controller, the noun phrase to which it is bound. In this way, the analysis of the passive voice by Chomsky has tried to perfect the transformational on the notion that the passive voice is just the transformation of active voice. However, the disadvantages of his analysis have been pointed out, linguistics like Granger (1983) comments that The complexity and richness of the passive structure are not even hinted at in the transformational accounts of the passive. The main reason why transformational grammar offers so few insights into the workings of the English passive is that the transformationalists and most particularly Chomsky have not investigated the passive for its own sake but as a tool to demonstrate the validity of the different transformational models C. Functional analysis In field of functional grammar, the key of transformational grammar believed has been denied, that is the surface and deep structures lost its value to the 17

24 functionalists. In functional grammar, the generate sentences in conformity with grammaticality through transformations is unnecessary, the functionalists concerned with the factual content of natural language and its communicative function, language is a tool to meet the need of human communication. The approach of functional grammar was first pointed out by the Linguistic School of Prague, three main ideas have been stressed. (1) The synchronic study of language is fully justified as it can draw on complete and controllable material for investigation but no rigid theoretical barrier is erected to separate diachronic studies. (2) There is an emphasis on the systemic character of language. It is argued that no element of any language can be satisfactorily analyzed or evaluated in isolation and that assessment can be made only when the relationship of an element is established with coexisting elements in the same language system. In other words, elements are held to be in functional contrast or opposition. (3) Language is looked on as functional in another sense, which is as a tool performing a number of essential functions or tasks for community using it. Halliday (1979), followed the Prague linguists and advocates a synthesis of structural and functional approach. In his research, three function of language has been pointed out: ideational function, interpersonal function and textual function. In his comment, the connection of these three functions can be found: If the ideational component is language as reflection (the speaker as observer of reality), and the interpersonal component is language as action (the speaker as intruder in reality), and the textual component is language as relevance (the speaker as relating to the portion of reality that constitutes the speech situation, the context within which meaning are being exchanged). Detailed in English voice, he divided it into two categories: the middle and the non-middle. He described The way the voice system works as follows. A clause with no feature of agency is neither active nor passive but middle. One with agency is non-middle, or effective, in voice. An effective clause is then either active or passive: active if Agent/Subject, passive if Medium/Subject if the clause is effective, since either participant can then become subject there is a choice between active and passive. The reasons for choosing passive are as follows: (1) to get the Medium as Subject, and therefore as unmarked Theme. (2) to make the agent either late news, by putting it last, or implicit, by leaving it out. ( Halliday, 2000) D. The passive voice in lexical-semantics In field of the lexical semantics, Rick Morneau (2007) described that natural languages have many ways of changing the relative importance or topicality of a verb's arguments. Languages can also remove arguments from the argument structure, 18

25 while implying that they still exist, and make the missing arguments either obliquely expressible or not expressible at all. Finally, languages can also incorporate normally oblique arguments, making them part of the argument structure of the verb. The definition of different voices had been defined as. Active - transitive: The subject is slightly more important or topical than the object. Both must be expressed. Passive: The original object becomes the subject and becomes considerably more topical than the original subject. The original subject is no longer part of the verb's argument structure, and does not have to be expressed. However, it is always implied and may be expressed obliquely (in English, typically using the preposition "by"). Middle: The original object is made more topical and becomes the subject. The original subject is deleted from the verb's argument structure and may not be expressed at all even though it is implied. Anti-passive: The subject is made considerably more salient than the object. The original object is no longer part of the verb's argument structure, and does not have to be expressed. However, it is always implied and may be expressed obliquely. Inverse: The arguments of the active verb are simply reversed. The original object becomes the subject, gaining in importance; and the original subject becomes the object, losing importance. Unlike passive, the original subject is not oblique and must appear. To be specific, he also gave the following example(morneau, 2007), John broke the window. = active voice The window was broken. = passive voice, implied agent The window was broken by John. = passive voice, oblique agent The window was broken with a hammer. = passive voice, oblique instrument, implied agent A hammer broke the window. = incorporated instrument, agent cannot be expressed at all (*by John), new structure is something like I+P-d, where I =instrument. The window broke accidentally. = middle voice, implied agent, agent cannot be expressed at all (*by John). The window broke. 19

26 = P-d verb. This is sometimes confused with middle voice. In the system described in this monograph, this verb is a basic verb and the example is in the active voice. No agent is expressed or implied. John broke something. = anti-passive (this is an approximation - English does not have a true morphological anti-passive construction). The agent alone is prominent. The patient loses its prominence but maybe expressed obliquely. However, even when not expressed obliquely, a patient is always implied. The window broke John. (poetic license needed here) or The window, John broke it. = inverse voice (again, these are approximations - English does not have a regular inverse construction). Patient becomes subject, agent becomes object and must appear General ideas on the passive voice in Chinese grammar Although in Chinese, the use of passive voice is far less than in English, but it is still a very important construction. Yip Po-Ching (2004) mentioned that there are three forms of passive voice in Chinese: (a) the notional passive - where no formal passive marker is employed. wenti jiejue le (lit. problem solve le) The problem was/has been solved. (b) the formal passive - where a passive marker like 被 bei is introduced. wenti zhong bei jiejue (lit. problem finally bei.by solve) The problem was finally solved. (c) the lexical passive - where a verb, indicating that the subject or the topic is the 'receiver' of the action, is followed by a nominalized verbal object. wenti dedao le jiejue (lit. problem receive le solution) A solution was found for the problem. wenti dedao jiejue le (lit. problem receive solution le) 20

27 A solution has been found for the problem. In the book Passive and Voice (1988), together with the other 17 original papers dealing with the voice, Mantaro J. Hashimoto talked about the structure and typology of the Chinese passive construction. Despite of the generally agreement among the passive voice is constructed with the passive marker 被 (bei), and he considered the word 被 (bei) has some colloquial counterparts like: 叫 (jiao, let/by), 让 (rang, allow/by) and 给 (gei, give/by) as the dialect in Northern Chinese. Due to these facts, he argued in his study that (a) the modern Chinese bei (sustain/by) construction is basically an embedding structure of the type found in the southern minority languages, (b) while jiao (let/by), rang (allow/by) and gei (give/by) on the one hand and bei (sustain/by) on the other hand behave more or less the same on the surface, there are some clear deep syntactic and typological differences between these two groups of passive markers, and (c) the change from bei (sustain/by) to jiao (let/by)/rang (allow/by) undoubtedly marks the Altaicization of modern Chinese linguistic stuctures [ perhaps gei (give/by) and its correspondents constituting the intermediate stage ] - the typological transition a step forward from the basically Tai (and perhaps some Austroasiatic) construction to the Altaic structure. We can also find the introduction of the Chinese passive voice in the book The Establishment of Modern Chinese Grammar, Yuzhi Shi listed four markers for the passive construction. (a) Rang 让活儿都让他们干完了 (task) (all) (rang) (them) (do) (finish) (past participle) All of the tasks have been completely done by them. (b) Jiao 叫墨水瓶叫弟弟打翻了 (link) (bottle) (jiao) (brother) (strike down) (past participle) The link bottle was struck down by the brother. (c) Gei 给这本书给你弄脏了 (this)(measure word for book)(book)(gei) (you)(make-dirty) (past participle) 21

28 This book was dirtied by you. (d) Bei 被窗户被工人们刷上了绿漆 (window) (bei) (workers) (paint-up) (past participle) (green) (lacquer) The window has been painted green by workers. In these four markers, he pointed out that In Modern Chinese bei has become purely a passive marker at the expense of its original verbal use, but the other three markers still co-exist with their full verbal uses: rang and jiao are synonymous, means ask, call or make (somebody do something), and gei means give. Speaking to their difference, he considered that the passive marker rang, jiao, gei are more often used in the spoken language, but bei is more commonly used in written language. According to Chaofen Sun, the indicator of passive voice in Chinese is commonly placed in front of a doer/agent and it is often a preposition. He pointed out that The most common prepositional, passive marker is 被 bei. Although many Chinese causative markers such as 让 rang, 给 gei, and 叫 jiao can be used interchangeably, 被 bei, unlike other passive markers, a verb meaning to sugger/to receive in old Chinese, has practically lost all of its ability to function as a verb in modern Chinese. He also brought up the common knowledge that in Chinese, the passive voice with those markers always implies a sense of adversity, or misfortune Comparative studies of the passive voice in English and Chinese In the research of Andrew Methven (2006), he first made a brief introduction of the voice and passive. In his study, he quoted some opinions from several scholars, and pointed out two kind of passive voice,(1) long and short passives, (2)notional passives. As in his description, he said that according to Xiao and Emery (2005), the passive voice is often used as a strategy that allows language users to avoid mentioning the agent, and in majority of cases, the subject of the active verb the agent - is not expressed in the passive. These agentless passives are know as short passives, and the less frequent passives that do have an agent are known as long passives (Xiao et al., 2005; Huang, 1999). While for the Notional Passives, he quoted the description done by Kenneth (1993), that is some verbs in the active voice can express passive meaning, for example these clothes wash well, can have the 22

29 meaning these clothes are washed well. This type of sentence is known as a notional passive. In the second part, he compared the passive voice in English and Chinese from several perspectives. The first perspective is the comparison of passive constructions. In this part, he figured that the English passive voice can be marked by a copular verb followed by the past participle of a verb, for example, be, seem become etc. The unmarked passive form in English is be+past participle, but other copular verbs can be used such as become and get. While in Chinese, the passive voice change is marked overtly through the addition of a passive morpheme, or covertly through implicit semantics. Passive Passive % of total Long passive Short passive Type Marker Occurrences Frequency Percent Frequency Percent Syntactic Bei 87.3% % % Passive Wei suo 60.0% % - - Gei 1.5% % % Jiao 0.4% % - - Rang 1.6% % - - Lexical Ai 52.6% 1 3.3% % Passive Shou 60.4% % % Zao 82.0% % % Table 1 Summary of frequency of syntactic and lexical passives in Chinese; source: Andrew Methven (2006) Except the passive marker bei, there are several other passive morphemes in Chinese, for example rang, jiao, gei. And the archaic structure wei Suo. By using a parallel corpus, he provided a table (Table 1) to show the frequency of syntactic and lexical passives in Chinese. 23

30 In the third part of his comparative analysis, he compared the usage of the passive voice in both language, and he summarized that it has been shown that the usage and implications of the passive in Chinese is very different to that of English. Therefore the manipulation of transitivity to convey intentionality and ideology (Hatim and Mason) in English is unlikely to be reflected using same constructions in Chinese due to genre and semantic restrictions. Therefore having recognised such patterns in an English text, alterations to the transitivity patterns have to be implemented in order to reflect the same relations of power and ideology in the target text. Han Qin (2002) focuses on the differences and similarities of the English and Chinese passive voices. She describes in great detail the features of the passive voice found in both languages. While Long Jianghua (2005) discusses how passive voice is used in news items and how they can be well translated into Chinese. He also focuses on the frequency of passive voices used in news items and concludes that passive voice is used more in news items compared with other writing styles. In her monograph on Chinese syntax, Zhang Yufeng (2003) puts in one chapter the discussion of the passive voice in Chinese language from all aspects: lexical, syntactic, semantic and even pragmatic Brief Introduction to Passive Voice in English and Chinese According to the previous study, we will have a brief introduction of English and Chinese passive voice in this chapter, basically from syntactic structure, lexical choices and semantic content English passive voice A- Lexical Choices In English, passive voice is mainly realized by copula verb plus past participle. Generally speaking, few words are specifically used to express the passive voice such as become, get and with. As Chinese is considered as a morphologically impoverished language and lacks the similar verb constructions, lexical items or in many cases, particles play a very important role in expressing some grammatical meanings, including passive meaning. Unlike English, in Chinese, to express the meaning of passiveness, we often choose certain Chinese characters to help construct the passive 24

31 structure among which the most frequently used is 被 bei. But it is not the only character used in passive voice, there are other characters, namely, 让 rang, 叫 jiao, 给 gei, 受 shou and 为 wei and so on. These particles, semantically conveying more or less the similar meaning can often be used alternatively. In some cases, they cannot be used interchangeably (Daozhen Zhang, 1998). For example, 由于全球变暖, 城市可能被完全地淹没 (due to) (global) (warming) (city) (could) (bei) (completely) (submerge) Due to global warming, cities could be completely submerged. In this sentence, the particle 被 bei cannot be replaced by any of them: 让 rang, 叫 jiao, 给 gei, 受 shou and 为 wei. A. Syntactical constructions of English passive voice As mentioned in the previous sections, the passive voice in English is always considered as the structure with an original formation of be/get + v ed. Meanwhile, some passive voices are marked by by-phrase or with-phrase. That is to say, lexically, there are two English prepositions employed in passive voice. As we know, passives with be as an auxiliary verb are most commonly used, and are always called the Be-passive. So the first type of English passive voice is: a. The passive voice in original formation of be/get + v-ed. The syntactical structure is: NP + Be/Get + V-en (1) The cake has been eaten. (2) The roads have been covered. (3) The captain has been attacked. Another type of English passive voice is: b. The passive voice marked by by-phrase or with-phrase. With its syntactical structure as: NP1 + Be + V-en + By/With + NP2. (1) The cake has been eaten by Tom. 25

32 (2) The roads have been covered with snow. (3) The captain has been attacked by a knife. From these examples we can see that in first type of English passive voice, the sentence has a receiver and an action. While in the second type of passive voice, despite of these features, there is also an agent in the sentence. According to this difference, these two types of passive voice can be used in different conditions. However, since the passive voice expresses the relationships between doers and receivers, which mean that, the subject in passive voice is acted upon as patient. Because of this reason, there is the third type of English passive voice. The non-marked passive voice. (1) The car needs preparing. (2) The box weighs ten kilogram. This two sentences are the examples of the third type of passive voice in English, both the subject in these two sentences had been acted upon as a patient, and all these action had done by someone or something else to these subject.. We can see that this type of passive voice does not look like other passives as they do not possess the same form, i.e. Be + v-ed, or with By/Ger phrases. However, semantically they are passives. Subject Auxiliary Past Participle Tense The cake is/are made Present The cake has/have been made Present perfect The cake was/were made Past The cake had been made Past perfect The cake will be made Future The cake will have been made Future perfect The cake is/are being made Present progressive The cake was/were being made Past progressive Table 2 Passive Voice Used in Different Tenses; source: Zhang Daozhen (1998) 26

33 What is more, passive voice can be used in various tenses due to the different situation. According to this fact, Zhang Daozhen (1998) provided a table (Table 2) to show the detailed information of passive voice used in various tenses. B. Semantic analysis The passive voice is wildly used in English language. to analyze the passive voice from a semantic point of view, we will find that the passive voice conveys more or less the following semantic implications: a. The actor or the do-er is unclear. The house has been pulled down. This sentence only expressed the fact that the house has been pulled down. However, it did not clearly indicate the doer who conducted the act. Therefore, it is clear that when the doer of the action is unclear, the passive voice can be used. b. The actor or the do-er is not important enough to mention. This kind of car is made in Shanghai. The main focus of this sentence is that the producing area of this kind of car is in Shanghai, and it is unnecessary for readers to know who make this car. So we can use the passive voice and leave out the doer. Or in other words, it is considered as unnecessary to point out the doer or it is better not to mention the doer. c. To stress the actor, the do-er or the action. The paper is cut by the knife. Also, we can see very clearly that in this sentence, knife has been stressed, and there is no doubt that the knife has been used to cut the paper. It is perfect to use a passive voice to stress the key point of the sentence. 27 d. To balance the sentence or for any other contextual need.