1. 研究目的与意义（文献综述包含参考文献）

Literature review

In the field of Second Language Acquisition (SLA) research, language competence can be studied from different aspects. Many scholars do research in the use of chunks and fluency of output, especially in oral English. In this chapter, the author will discuss four sections. Section one states the two key terms, chunk and fluency. Section two study the theoretical issues and section three displays empirical studies. Finally, the author addresses the research gap of this study.

2.1 Key terms

2.1.1 Chunk

The conception of chunking was first put forward by American psychologists Miller and Selfridge in 1950s to explain the process of unit information into chunks. From 1970s, the noun ”chunk” has been introduced into linguistics teaching filed. It refers to block structure which is between grammar and vocabulary, fixed or unfixed, and macular.

Many scholars at home or from aboard do many researches on the chunks.

In her (2009) paper The characters, Properties and Functions of Chunks, Yang Jinhua#8217;s defines chunks as speech unit, combining of several words. The relations between the component parts, simple or complex, flexible or fixed, tight or loose, are clear and understood easily.

In her (2012) paper The Use of Academic English chunks and the Research of the Development Characters of Learners in China, Xu Fang#8217;s finds Chinese students are prone to use noun-centered phrases, compared with those, verb-centered phrases are used less frequent.

In her (2010) paper Follow-up Research: The Use of Chunks in Oral Spoken English of English Major students, Qi Yan classifies the chunks into the categories as showed in the following chart.

Among these, discourse chunks embody connection function, which connect the two components of internal parts of sentences and the sentences themselves. Interpersonal relationship chunks show speakers#8217; attitudes toward certain things. Topic chunks relate to experience function, having something directly to body or abstract entities and reflecting some kind of experience module. Topic chunks can be divided into three parts, which include verb chunks, noun chunks and adjective, adverb as well as preposition chunks. Verb chunks express the process itself, noun chunks the participant and adjective, adverb as well as preposition chunks the environment.

2.1.2 Fluency

2.1.2.1 Definition of L2 oral fluency

Oral fluency is an important factor for researchers to study second languages speech. Defining oral fluency is difficult as no agreement reached on this definition.

Leeson is the first researcher to define oral fluency. He defines oral fluency as ”the ability of the speaker to produce indefinitely many sentences conforming to the phonological, syntactical and semantic exigencies of a given natural language on the basis of a finite exposure to a finite corpus of the language” (Leeson,1975).

Fillmore presented another important definition of oral fluency. He holds that oral fluency should include four basic abilities: the first one is the speaker should produce more words, fewer pauses in limited time. The second one is the speaker should pronounce speech in a coherent, meaningful and reasonable way. The third dimension is the ability to have appropriate things to say in a wide range of contexts. Finally, there is the ability some people have to be creative and imaginative in their language use (Fillmore, 1979).

Lennon classifies the oral fluency into broad and narrow sense. In the broad sense, oral fluency represents the highest point on a scale that measures spoken command of a foreign language. The latter refers to isolatable components of fluency such as correctness, idiomaticness, relevance, appropriateness, pronunciation, lexical range, and so on.

2.1.2.2 Measurement of oral fluency

Towel et al. provides five temporal indexes to measure oral fluency. It can be calculated with five indices as follows: speech rate (SR), phonation time rate (PTR), mean length of runs (MLR), articulation rate (AR), average length of pause (ALP). All of the calculation about the time index is based on the definition of pause, which refers to all intervals inside or between sentences last for at least 0.3 seconds (including 0.3 seconds). All filled pause and unfilled pause over 0.3seconed are included (Towel et al., 1996)

1) Speech rate (SR): refers to the ratio between number of all syllables of the target speech samples and the time (expressed as Seconds) for the production of the speech (including the time of pauses). The result get by multiplying the ratio with 60 means the syllables spoke in every minute.

2) Phonation time rate (PTR): refers to the ratio between the time spent on articulation and that of the production of the target speech.

3) Mean length of runs (MLR): refers to mean length of the runs between every two pauses which both lasts for at least 0.3 seconds. It can be expressed by the ratio between the number of syllables of the target speech (fore and aft excluded) and the number of times of all pauses lasting for at least 0.3 seconds.

4) Articulation rate (AR): refers to the syllables articulated in each second among the whole time of pronunciation of all syllables. It can be expressed by the ratio between the number of all the syllables and the time used for the articulation of these syllables (pauses excluded).

5) Average length of pause (ALP): refers to the ratio between the time of all pauses lasting for at least 0.3 seconds and the number of times of all pauses (fore and aft excluded).

2.2 Related theoretical issues

2.2.1 Levelt#8217;s model of speech production

After years of hard work, Levelt (1989) put forward ”speech production model”, which is quite explanatory. In his model, Levelt divided knowledge into two kinds. One is declarative knowledge and the other is procedural knowledge. Declarative knowledge means ”what”, while procedural knowledge refers to ”how to do”. Due to the limitations of the capacity of working memory, Levelt holds that the more fluent the speech production is, the higher requirements of procedural knowledge.

There are five processing components in Levelt#8217;s model of speech production: the articulator, the formulator, the conceptualizer, the acoustic-phonetic processor and the parser system. Among this model, three parts are closely connected with speech production. They are conceptulizer, formulator and articulator. These three parts are the most relevant to measures of oral fluency by working independently, collaboratively and simultaneously.

Utterances begin as conceptualizer, in which speakers choose the declarative language that is related to their intentions, arrange them in order and organize them into pre-verbal message. The preverbal message generated by conceptualizer is maintained in Working Memory and fed into formulator in which the lemmas or lexical items are selected. The formulator contains two parts: one is grammatical encoding and the other is the phonological encoding. The articulatory plan of an utterance emerges from the formulator. However, prior to entering the articulator, where the vocal musculature is engages for producing an utterance, the articulatory plan may be monitored internally with the support of sub-vocalization.

According to his model, Levelt originally intended to propose this model to explain the mature, fluent L1 speech. But this model can also be applied to explain second language speech production. He presumed that during the speech production, the adult native speakers deal with the grammatical encoding subconsciously as well as automatically.

2.2.2 Anderson#8217;s ACT theory

Anderson#8217;s Adapted Control of Thought (ACT) theory has been well recognized and explanatory. It is a general theory of cognition that focused on memory processes. Three types of memory structures are included in ACT theory. They are declarative, productive and working memory. As for declarative memory, it takes the form of a semantic net linking propositions, images, and sequences by associations. Productive memory (also long term memory) displays information in the form of productions. Each production has a set of conditions and actions based on declarative memory. To some degree, all nodes of long-term memory have activation and working memory is most highly activated. Different knowledge focuses on different aspects: declarative knowledge is about ”knowing that”. Procedural knowledge is about ”knowing how,” the ability to apply rules to carry out functions.

According to ACT theory, declarative information is the first step while procedural knowledge is learned by making inferences from already existing factual knowledge. Productive knowledge is different from declarative knowledge in that it can be applied to task directly. Acquiring language knowledge, like any other type of skill learning, involves much development of procedures that transform declarative knowledge into a form that make performance easier and more efficient. The transition from declarative knowledge to productive knowledge takes place in three stages: cognition, association and unconscious automaticity. As a result, this process facilitates learners#8217; abilities to apply language resources to output more quickly, which helps learners fulfill self-achievement at the same time.

2.3 Related empirical studies

Empirical studies on lexical chunks start from Zhang#8217;s doctoral dissertation (1993). He examines the correlations between EFL students#8217; knowledge of lexical chunks (i.e. measured by a fill-in-bank test), the use of lexical chunks (i.e. frequency of accurate produced lexical chunks recorded), and English writing fluency (i.e. determined by a paper-and-pencil written test).

In Zhang#8217;s study (Zhang, 1993), he tested 60 speakers of English at Indiana University of Pennsylvania. 30 of them are native speakers and 30 non-native. The results show that: a) non-native writers perform significantly worse than native writers; b) native writers outperformed the non-native writers in terms of the use of accurate lexical chunks in writing. Zhang summarized that collocational knowledge is a source of fluency in written communication among college freshmen. On the other hand, the variety and accuracy of lexical chunks can indicate the writing quality of freshmen (Zhang wenzhong, 1999).

Oral fluency is also an essential objective in spoken English teaching in China. Many scholars do researches on fluency. And Zhang presents a discussion of theoretical and empirical investigation on L2 oral fluency. His findings provide the basis for further studied on this subject (Zhang, 1999).

Sung directed her focus on the possible connection between chunks and speaking fluency. She tested 24 natives and 72 non-natives in two tests, one chunk test and one speaking test. The chunk test assessed the subjects#8217; knowledge of lexical chunks while the speaking test elicited the subjects#8217; use of lexical chunks and measuring their speaking fluency. His findings pointed that between EFL learners speaking fluency and their knowledge of lexical chunks exists a strongly correlation (Sung, 2003).

2.4 Research gap

This empirical study conducted in Chinese context to find out the relationship between the number of chunks used and the oral fluency of EFL learners. The research studied here is focused on English major students. Though the sample is not large, it is typical since all the subjects having the same learning background and similar language level.

Oral fluency is an important factor which affects people#8217;s communication abilities. However, we also find that many EFL learners can pass all kinks of written exams of English while their speaking skills should be more polished. If the learners can speak fluently and involve themselves actively in conversation, they will have greater motivation and pleasure than those who cannot. Therefore, we should attach more attention to this practical skill. And many scholars have done some research on oral fluency. But few studies have been concerned with the application of lexical chunks and its influence on oral fluency. All in all, the author wants to complete the chart with the hope that this research can provide an added perspective of the importance of lexical application in oral fluency and arouse teachers#8217; and students#8217; awareness of the application of lexical chunks.

2. 研究的基本内容、问题解决措施及方案

Methodology

3.1 Research questions

The current study investigates the developmental patterns of L2 learners#8217; use of chunks and the fluency when they spoke about a given topic. The specific research questions are as follows:

1. How do the Chinese EFL learners use chunks in their oral English?

2 How is the Chinese EFL learners#8217; oral fluency?

3. What is the relationship between the usage of chunks and the oral fluency?

3.2 Participants

Participants are 29 second-year college students from English major of Nanjing University of Technology. Among them, there are 25 female students and 4 male students. All of them are native Chinese speakers with similar English training experience. Before entering the university, they have studied English for 6 years with 3 years at middle schools and another 3 years at high schools. According to the subjects#8217; intelligence and academic achievements, all these factors mentioned above are used to guarantee the subjects#8217; homogeneity.

3.3 Indices

In chapter 2, the author has mentioned temporal indices to measure the continuity and rapidity of the speech production. In this study, the researcher adopts the temporal indices proposed by Towell et al. (1996), which includes speech rate (SR), phonation time rate (PTR), mean length of runs (MLR), articulation rate (AR), average length of pause (ALP) with the help of COOL EDIT Pro 2.0.

3.4 Working definition of lexical chunks

In this study, the categories of lexical chunks are measured by the method proposed by Qi yan (2010). For sake of clarity, these categories are now displayed as follows:

3.5 Data collection

The research focuses on the students of English major and concludes twenty-five girls and four boys. The corpus source is the tape records of their TEM-4 oral tests. The students are given a specified topic and they are required to make speeches in three minutes after three minutes for preparation. The records are transferred to written materials and marshaled to research source.

The data collected in the study include two parts: one is oral fluency variable and the other is chunks variable. The subjects#8217; speech production is collected in computer hard disk. The author calculated the number of chunks according the categories above and transformed the documents intoMP3 format which is more convenient for data collection and analysis. The time spent on articulation and pause was calculated by employing COOL EDIT Pro 2.0. The software has a function of noise reduction. The raw material is collected in a studio room as 29 subjects making speeches together. It is important to filter all the noise so that the researcher can achieve high quality recordings. This also gives very accurate indications of pauses and enables the researcher identified all the pauses clearly, which, to some degree, reduces the risk of making mistakes. In the wave file, fluctuating waveforms indicates the time spent on articulation and the horizontal ones reflecting the time of pauses. The time length indicator is used to measure the pause length (rounded to 0.001 seconds).

3.6 Data analysis

According to the definition of chunks on Longman Modern English Dictionary and the judgments of language intuition and sense of Chinese readers, the standards are specified as follows:

1. length, at least two words;

2. phrase level, deprived of all subordinate clauses and sentences;

3. pause, no pause between the words of chunks;

The analysis is done in three steeps: 1. Account all the chunks in each speech of every student. 2. Do analysis on the relationship between the five indices and the use of chunks 3. Find out the relationships between the use of chunks and the fluency.