论文总字数：57246字

摘 要

词块作为整体存储和提取的语言单位，是语料库语言学研究的重要领域。目前，对于美国主流媒体涉华报道的研究也逐渐成为一个热点问题。

本研究选取2016年纽约时报和华尔街日报涉华报道的两个语料库，库容分别为300018和300349，用Antconc对三词词块进行检索分析，参考Biber等对词块的结构和功能分类标准，对比分析美国主流媒体涉华政治语篇词块的结构和功能特征。

研究显示，在结构上，两家媒体涉华政治语篇词块结构的总体分布略有差异。华尔街日报主要以名词类和介词类词块为主，名词类词块中名词短语 介词短语词块和其他名词短语片段比例相当，动词类词块中其他动词片段比例最高。而纽约时报主要以名词类和动词类词块为主，名词类词块中其他名词词块所占比例略高，动词类词块中其他动词片段比例最高。缺少后置从句修饰以及被动动词片段等分类的词块。纽约时报的关注重点在政治、教育等方面，而华尔街日报侧重于政治、经济等方面，但均格外关注与中国有关的热点问题，如南海问题。

在功能上，指称词块在两个语料库中比例最多，纽约时报指称词块所占比例略低于华尔街日报，而立场词块比例最少，但两个语料库所占比例相当。该研究中，指称词块通常以介词类词块为主，立场词块通常以动词类词块为主。

该研究可以看出本族语者词块使用的特征，语言产出的多样性和地道性以及政治语篇报道一定的客观性。由于对于美国主流媒体涉华政治报道的词块研究较少，一定程度上完善了对该领域的研究。

关键词：美国主流媒体 涉华报道 词块 结构分类 功能分类

Abstract

Lexical bundle, a language unit, which is stored and retrieved as a whole, is an important research in corpus linguistics. Studies of China-related reports in the American mainstream media are becoming a hot issue presently.

This study’s corpora select the China-related reports of New York Times and Wall Street Journal in 2016, whose capacities are respectively 300,018 and 300,349, use the Antconc software to research and analyze the high-frequency three-words, and use the structural and functional taxonomies designed by Biber et al. (1999, 2007), which is aimed at making a contrastive analysis on lexical bundles from the perspective of structure and function.

This study shows that their overall frequency distributions have a slight difference. Wall Street Journal is mainly focused on the NP-based bundles and PP-based bundles, in which the number of NP PP fragment and other NP-phrase fragment accounts for the almost same proportion in the NP-based bundles and other VP-phrase fragment accounts for the largest proportion in the VP-based bundles, while New York Times is mainly focused on the NP-based bundles and VP-based bundles, in which the number of other NP-phrase fragment accounts for the largest proportion in the NP-based fragment and other VP-phrase fragment accounts for the largest proportion in the VP-based bundles. These two corpora are lacking in the clause fragment, passive verb PP fragment and so on. Furthermore, New York Times mainly focuses on the aspect of politics and education, while Wall Street Journal mainly focuses on the aspect of politics and economy and so on. They both concern about some hot issues in China, such as South China Sea.

When it comes to the functional classifications, the number of referential expressions accounts for the largest proportion in the two corpora, with a relatively lower proportion in the New York Times than in the Wall Street Journal. Then, the number of stance expressions accounts for the smallest proportion, with almost the same proportion in the two corpora. These things show the objectivity in political discourses. This study shows that VP-based fragment always embodies the stance expressions and PP-based fragment always embodies the referential expressions.

This study researches the features of lexical bundles, the variety and idiomaticity of language production and the objectivity of political discourse. Due to the fact that studies of features of lexical bundles in American mainstream media are not that enough, this study can improve this kind of research to some degree.

Key words: American mainstream media; China-related reports; lexical bundle; structural classification; functional classification

1. Introduction

1.1 Research Background

It is known to all that there are always different comments on some countries in the international communication of media. Many countries are keen to know about the information of United States because of its leading role in the world stage. China is playing a more and more important role in the stage of international politics and economy with the improving of our comprehensive national strength.

Over the past three decades, a huge number of studies have begun to figure out that language is comprised of numerous fixed and semi-fixed formulaic expressions that embody the characteristics of both grammar and vocabulary. Altenberg (1998: 102), according to his study of London-Lund Corpus, found that “more than 80% of natural language by native English speakers is composed of formulaic expressions of various types”.

It is generally agreed that language is formulaic in nature, whether it is spoken or written (Ellis, 1996, 2008; Granger amp; Meunier, 2008; Pawley amp; Syder, 1983; Sinclair, 1991, 2004; Wray, 2002). Pawley amp; Syder (1983:191) argue that “ fluent and idiomatic control of a language rests to a great extent on knowledge of a body of ‘science stems’ which are ‘institutionalized’ or ‘lexicalized’”. Granger (1998:145) also maintains that one of the reasons why the use of recurrent word combinations has become very popular in the teaching of English as a foreign language, for example, is “the formulaic nature of many pragmatic rules”.

The study of formulaic language has become an important research field of many linguists and applied linguists. For example, Becker (1975), as the first linguist to specialize in this linguistic phenomenon, held that “it is the prefabricated patterns rather than the individual words that comprise the minimal unit of humanity’s linguistic communication”. Pawley and Syder (1983) pointed out that “to achieve native-like language selection and fluency, second language learners should not only master the grammatical rules that can produce countless sentences, but they should also equip themselves with enormous lexicalized sentence stems”. Wray and Perkins (2000) suggested that “formulaic sequences can save processing effort by giving us ready-made framework and at the same time achieve social interactions”

The researchers have done a large number of studies from different perspectives, and it included chunk (Sinclair 1991), lexical bundle (Biber et al. 1999, 2004; Cortes, 2004), cluster (Scott 2004), and phraseology (Howarth 1998). Considering its specific research focus, each label has its own history and implication. In this study, I will utilize the lexical bundles, first coined by Biber et al. (1999) to refer to a sequence of three or more words. Specifically, it has been observed that advanced and fluent writing is characterized by appropriate and frequent use of formulaic language, which also helps language users maintain identity in a disciplinary community; conversely, the absence of such formulaic language may indicate writers’ inexperience or lack of expertise in an academic context (Bamberg, 1983; Mc Cully, 1985; Wray, 2002)

Many researchers have made considerable efforts to dig out the lexical bundles both theoretically (e.g., Biber et al., 1999; Benson et al., 1997; Chang, 2004) and empirically (Cortes, 2004, 2006; Hyland, 2008a; 2008b; Ma, 2009; Hu, 2011). These studies are carried out from different perspectives, thus providing scientific significance for English teaching and learning.

Among these studies, many researchers pay much attention to academic bundles--a special type of lexical bundles occurring in academic discourses (Biber 1999, 2004; Cortes, 2004, 2006), but those in political discourses are far from enough, especially in China-related reports in American mainstream media. Given the above limitations of previous studies, the present study intends to make a comprehensive study of the features of lexical bundles on China-related reports of New York Times and Wall Street Journal in 2016. The present study is of great significance considering the following aspects:

First of all, previous comparative studies are centered on lexical bundles in the fields like education, applied linguistics, academic writing and so on. As a consequence, the studies in the field of political discourses are quite rare. The present study can compensate such shortage and better figure out the features of lexical bundles.

What’s more, this study’s corpora are chosen from the China-related reports of New York Times and Wall Street Journal in 2016. Studying the features of lexical bundles of American mainstream media can do good for our understandings for political discourse of America, which is beneficial for mutual understandings and reducing barrier disputes. Given the above limitations, this paper is going to study the three-word lexical bundles of New York Times and Wall Street Journal in 2016.

1.2 Research Purposes

Due to the above-mentioned limitations in previous studies, the present study utilizes new corpora to investigate how lexical bundles are used in New York Times and Wall Street Journal in the discipline of political discourse. Two purposes are specified as follows:

First of all, this paper intends to extract and identify the high-frequency three-word lexical bundles in American political discourse, which is helpful for China’s national image and figuring out American mainstream media’s attitudes towards China. With the development of China, some western mainstream media began to focus on China. It makes sense to know China's image in foreign countries' mainstream media, and master the characters and rules of China-related reports for improving the system of China's international communication, as well as building a fine national image.

More importantly, this paper intends to find out the structural and functional characteristics of lexical bundles in New York Times and Wall Street Journal, which can fill the gap in the studies of political discourse to some degree.

1.3 Structure of the thesis

This study is composed of five chapters.

Chapter one is the introduction, which includes the research background, research purposes and the structure of this thesis.

Chapter two is the literature review. It provides an in-depth account of the background literature relevant to the context that my study is situated in and, in doing so, to provide an ‘argument’, ‘case’ or justification for the study. It also provides a review of the non-research literature that summarizes and synthesizes background and contextual information. What’s more, it contains a review of theoretical perspectives that underpin my research project. More importantly, it identifies arguments for and against issues and controversies related to this study.

Chapter three is the research design which contains a description and justification of the methodological approach best suited to my research questions and a discussion of ways in which the validity and reliability of my data collection and data analysis.

Chapter four is the results and discussion. It includes a presentation of the results and findings of my study that are relevant to my research questions. And it also includes an explanation of what the findings mean.

Chapter five is the conclusion of the thesis. It is a reminder of the aims such as research questions and key methodological features of my study. More importantly, it gives pedagogical and points out the limitations of the present study and makes suggestions for further research.

2. Literature Review

This chapter gives a detailed explanation of the theoretical framework on which the present study is based, including the definitions, characteristics and classifications of three-word lexical bundles. Then, it reviews some of the most important empirical studies on lexical bundles conducted by previous researchers both home and abroad.

Many studies have employed a lexical bundle framework to describe expressions typical of different registers, focusing on variation across registers, and describing the discourse functions served by different types of lexical bundles.

2.1 Definitions of lexical bundles

The term “ lexical bundles” was originally coined by Biber et al. (1999) who studied LSWE (Longman Spoken and Written English) corpus. Given that lexical bundles are identified empirically, lexical bundles have both theoretical definitions and operational definitions.

2.1.1 Theoretical definitions of lexical bundles

Many other terms, such as “chunks”, “phraseology”, “clusters”, were widely used by other researchers. Despite the fact that these terms, on most occasions, refer to similar lexical phenomenon, slight differences still exist among them.

Biber et al. (1999: 990) referred to lexical bundles as “recurrent expressions regardless of their idiomaticity, and regardless of their structural status, i.e. lexical bundles are simply sequences of word forms that commonly go together in natural discourse”. To make his study more manageable, he further defined it clearer as “a recurring sequence of three or more words”.

From the above definitions, we can clearly see that sequences have to meet several requirements to be qualified as lexical bundles.

Firstly, lexical bundles are “recurrent” or “recurring”, i.e. they should occur more frequently in a particular register than expected by chance. This frequency of recurrence, as the defining feature of lexical bundles, plays a decisive role in their identification.

Secondly, lexical bundles are sequences of three or more words. Two-word sequences are not lexical bundles despite their high frequency in natural discourse, such as in fact, you know and so on. In identifying lexical bundles, we rely on the orthographic word units (Biber et al. 1999: 990). For example, cannot, don't, and self-control are all regarded as single words in the present study.

Finally, only uninterrupted multi-word sequences are treated as potential lexical bundles. Thus, lexical combinations that span a turn boundary or a punctuation mark are excluded. For example in the present study:

In conclusion, I think it’s time to take actions to solve problems.

In this sentence, the word combination in conclusion I is not regarded as the lexical bundle due to the fact that the comma breaks the continuity of the sequence.

2.1.2 Operational definitions of lexical bundles

Biber et al. (1999) made a distinction among the idiom, collocation and lexis-grammar relationship and put forward the concept of “lexical bundle”. According to Biber et al. (2004), collocation refers to the linear relationship between just two words, while lexical bundle refers to the co-occurrence of three more words, which could be called extended collocation with the characteristic of high frequency and wide dispersion in corpus. This definition made lexical bundle quite operational but still could not eliminate many meaningless word combinations (Ma, 2011).

The operational definition is concerned with two criteria by which lexical bundles are yielded from corpora, i.e. frequency cut-off point and the number of texts in which they occur.

Originally, the actual frequency cut-off point is “somewhat arbitrary” (Biber et al. 2004: 376) considering the research purpose and corpus size. Then, Biber et al. (2004) and Biber amp; Barbieri (2007) adopted a higher relatively higher cut-off point of 40 times per million words.

And at the same time, the number of texts in which lexical bundles occurs is also arbitrary to some degree. On most occasions, the more frequent and wider used the word combination is, the more valuable and meaningful the lexical bundle is. Admittedly, it has many defects. Therefore, Biber and his colleagues (1999) set the rule of a minimum standard of an occurrence in at least five different texts.

2.2 Classifications of lexical bundles

2.2.1 Structural classifications of lexical bundles

Biber et al. (1999) identified high-frequency lexical bundles and the structural classifications are displayed in the following tables:

Table 2.1 Biber et al.’s (1999) classification of three-word lexical bundles in native speaker’s academic prose

2.2.2 Functional classifications of lexical bundles

Lexical bundles can also be classified according to the functions they perform in the discourse. According to the taxonomy designed by Biber et al. (2004), which was later revised by Biber et al. (2007), lexical bundles can be classified into three major categories: stance bundles, discourse organizing bundles, and referential bundles. The following table displays the functional taxonomy :

Table 2.2 Biber et al.’s (2007) functional classification of three-word lexical bundles:

2.3 A brief review of empirical studies on lexical bundles

2.3.1 A brief review of overseas studies on lexical bundles

Biber et al. (1999), as the first to study lexical bundles in a large corpus, carried out a comparative study on lexical bundles in conversation and academic prose on the basis of the Corpus of LWSE (Longman Spoken and Written English). The study found that lexical bundles were extremely pervasive in both spoken and written registers. However, conversation made use of far more lexical bundles than academic prose. What's more, the structural distributions of lexical bundles were strikingly different in spoken and written discourses. Lexical bundles in conversation were rather clausal, about 90% including part of a verb phrase. By contrast, those in academic prose were relatively phrasal, over 60% being composed of noun phrases or prepositional phrases.

Biber et al. (2004) extended previous research by investigating four-word lexical bundles in two university academic registers: classroom teaching and textbooks, and comparing them to those found in conversation and academic prose. The study found that classroom teaching used about twice as many different lexical bundles as conversation did and about four times as many as textbooks did. The extremely high density of lexical bundles in classroom teaching existed because this register relied heavily on lexical bundles in both communicative spoken discourse and informational written discourse. This pattern apparently reflected the complex communicative demands of this register.

The two studies by Biber et al. (1999, 2004) showed that lexical bundles are basic and important building blocks of different academic discourses rather than the accidental by-product of corpus frequency analysis. Especially noteworthy, the two studies have made particular contributions to creating a structural and a functional taxonomy for lexical bundles in spoken and written discourses, according to which most subsequent studies were conducted (Cortes 2004, 2006; Hyland 2008a, 2008b；Chen amp; Baker 2010; Adel amp; Erman 2012; Ma,2009; Yao amp;Lei 2011;etc.).

As can be seen from the previous studies (Cortes 2004, 2006；Hyland 2008b；Chen amp; Baker 2010; Adel amp; Erman 2012), different population groups differed significantly in their use of lexical bundles. However, these studies have not reached complete consensus in some aspects. For example, Hyland (2008b) adopted a functional taxonomy different from other studies, which will to some extent affect the results. Furthermore, the relationship between the number of lexical bundles and writing proficiency indicated by Hyland's (2008b) is just the opposite to that in Chenamp; Baker's (2010) and Adel amp; Erman's (2012).

2.3.2 A brief review of domestic studies on lexical bundles

Ma (2009) conducted an empirical study on the use of three-word lexical bundles in L2 timed essays of English majors. A total of 191 most frequent three-word bundles were identified in the native speakers' corpus and used as target bundles. Results of the study showed that there exists a huge gap between native speakers and Chinese L2 learners of English in the use of lexical bundles. Nearly two-thirds of the target bundles had a lower output rate in the L2 timed essays than that in the native writings and half had a lower output rate than the given frequency cut-off. A closer look into the bundles that produced an output rate lower than the given frequency cut-off showed that bundles with past-tense verbs, NP PP bundles as well as bundles with appositive clause or attributive clause had extremely low output in the L2 timed essays, and therefore should be specially strengthened. In addition, the use of prepositional bundles also needed improvement in L2 timed writing. It is also found that there exists a significantly positive correlation between the output rates of target bundles in the L2 writings and their occurrence rates in the native speakers' essays, suggesting a correlation between input and output. The paper also analyzed the possible causes for the high output rate of target bundles in the L2 writing, including the input frequency effect as well as the distinctive features of lexical bundles, such as simple vocabulary components and strong discourse functions.

Xu (2010) presented an empirical study on the use of three-word and four-word lexical bundles in L2 writings by non-English major masters, with specific focus on the structural analysis. The study aimed to discover the patterns of use of lexical bundles by advanced L2 learners. The research corpus was made up of 327 personal statements. Results showed that the order of the structural categories of lexical bundles in L2 writings in terms of frequency was VP-based bundles, NP-based bundles, PP-based bundles, adjective bundles, clause bundles, conjunctive bundles and adverbial bundle. This order corresponded to that of the native speakers' expository writing identified in Ma's (2009), showing that advanced English learners were similar to native speakers in terms of their orders of bundle types in writing. This study also revealed a shocking lack of prevalence and variety of the lexical bundles used by Chinese postgraduates. It was finally inferred in the study that writing style and context, identity of the writers, and distinctive features of lexical bundles may be possible factors affecting the use of lexical bundle in L2 writing.

Domestic studies on lexical bundles yielded some similar findings. Firstly, Chinese English learners' competence of lexical bundles is far from satisfactory in general. There exists a huge gap between Chinese learners and native speakers of English in the use of lexical bundles. Secondly, Chinese learners, be it undergraduates or postgraduates, tended to over-rely on a considerable number of lexical bundles to develop their writing, which was recognized as a problematic feature of L2 writing. Thirdly, Chinese learners showed some features of native spoken English in the use of lexical bundles. Finally, Chinese learners' inadequate or erroneous use of certain bundles was greatly influenced by the negative transfer of mother language.

2.4 Unsettled Issues

Indeed, many linguists have spared no efforts to study the features of lexical bundles both theoretically and empirically and most overseas studies have proved that lexical bundles are important building blocks of academic writing, which provides scientific significance for English teaching and learning.

However, these studies mainly pay attention to academic discourses and focus on the comparison of lexical bundles between native speakers and L2 learners, so those in political discourses are far from enough, especially in native country’s political discourses. Therefore, the structural and functional characteristics of lexical bundles in American mainstream media are still not that clear.

What’s more, studies of features of lexical bundles of China-related reports in American mainstream media are also quite rare.

Given the above limitations of previous studies, the present study intends to make a comprehensive study of the features of lexical bundles on China-related reports of New York Times and Wall Street Journal.

3. Research Methodology

As have been mentioned in the previous sections, the purposes of this study are to investigate the features of lexical bundles of China-related reports in New York Times and Wall Street Journal and to figure out the structural and functional distribution of lexical bundles used in two corpora. In this chapter, research questions are firstly put forward. Then, it elaborately gives an explanation of procedures of bundles identification and data analysis.

3.1 Research questions

The present study aims to investigate the lexical bundles in two corpora. It is noteworthy that only three-word lexical bundles are examined.

The research questions are specified as follows:

1. What are the most frequently used lexical bundles in New York Times and Wall Street Journal?
2. What are the structural characteristics of these lexical bundles?
3. What are the functional characteristics of these lexical bundles?

3.2 Corpora Used

The research subjects are China-related reports of New York Times and Wall Street Journal in 2016. Each corpus contains over 300,000 words and their capacities are similar. Therefore, these two corpora are comparable.

Corpora collections are divided into two steps: the collection of corpus of New York Times (hereafter called the NYT corpus) and the collection of corpus of Wall Street Journal (hereafter called the WST corpus). The former one is comprised of 296 texts and the latter one is comprised of 424 texts. These are China-related reports including economy, politics and so on.

请支付后下载全文，论文总字数：57246字