The Impact of Teaching Chunks on Speaking Fluency of Iranian EFL Learners

[1]Amir Mahdavi-Zafarghandi

Associate Professor of Applied Linguistics

[2]Abdorreza Tahriri

Assistant Professor of Applied Linguistics

[3]Maryam Dobahri Bandari

MA. Student

Abstract

Research on multiword clusters (chunks) is based on the assumption that native speakers use plenty of chunks in their everyday language and they are considered as fluent speakers of language. Therefore the present study was an attempt to investigate the impact of using chunks on speaking fluency of Iranian EFL learners. In the first phase of the study, the students of two intermediate classes sat for a general proficiency test and then were interviewed for their speaking ability. Next, the two groups were statistically compared in terms of their general proficiency and speaking fluency which indicated that they belonged to the same population. The 18-session instruction of the control and experimental groups included the same content and skills, but the experimental group received training on how to use chunks. At the end of the instruction period, the participants were interviewed once again in a posttest to track possible differences in their speaking improvement with respect to the frequency of chunks. The findings showed that the treatment had significantly improved speaking fluency of the experimental group and that there was a direct correspondence between the number of the chunks used and the listeners' perception of the participants` speaking fluency.

       Keywords: speaking, fluency, chunk.

1. 1.       Introduction

Speaking is one of the four language skills used for many purposes, for example, when we are having a causal conversation, the purpose might be to have social contact with the other people; or when we discuss over a subject, we may aim to express our opinion (Richard & Renandya, 2002). The context is essential to be understood and accounted for. Levinson (1983) argued that the term context has two aspects: social and interpersonal. Social context is important in studies that focus on sociolinguistic aspects of language. However, interpersonal context should be studied in sub-disciplines such as pragmatics, conversational analysis, and discourse analysis. In the present study the conversational analysis plays a critical role.

In the process of communication, the speakers of a language should have knowledge of the forms of language they use. They must know how to use this knowledge to negotiate meaning. In order to clarify meaning, the speakers and hearers or writers and readers should be able to interact. The quality of interaction influences many aspects of the speakers` life. One factor that increases the quality of interaction among nonnative speakers is fluency.

Speaking activities in language classroom have various goals including learning for the content matter and language items from other participants, learning communicative activities, developing skill in discourse, and finally developing fluency (Nation, 1989).  Language learners wish to develop fluency and be native like. Native speakers tend to use a great deal of prefabricated chunks of language which are not composed each time by the rules of syntax (Pawley & Syder, 1983; as cited in Conklin & Schmitt, 2008). McCarthy and Carter highlight the fact that ''many chunks are as frequent as or more frequent than the single word items which appear in core vocabulary'' (2006, p 46). If clusters are so important to language use and are so widespread in discourse, they should be paid special attention.The reason why chunks are so widespr            ead is because they can be processed more quickly and the mind can store these ready- made chunks in the long term memory to be used in language production (Conklin & Schmitt, 2008). Chunks could be used in a wide variety of ways. Some are used to show pragmatic integrity (e.g. one of the …), some have discourse- making function (e.g. you know), some may be used to keep face and politeness (e.g. I don’t know if…), and some mark vagueness and approximation (e.g. and things like that) (McCarthy and Carter, 2002). Research on chunks is based on the assumption that native speakers use plenty of chunks in their everyday language and they are considered as fluent speakers of language (Foster, 2001; cited in Leedham, 2006; McCarthy& Carter, 2002; Boers et al., 2006; Conklin& Schmitt, 2008).

Most researches works conducted in the field of chunks have focused on native speakers (Foster, 2001; cited in Leedham, 2006; McCarthy& Carter, 2002; Boers et al., 2006; Conklin& Schmitt, 2008). Unfortunately, little is known about the use of chunks by nonnative speakers. Further investigation is needed to be conducted on nonnative speakers and in EFL context. Therefore, the main aim of conducting this study was to assess the extent to which teaching ready-made chunks to learners might affect speaking fluency of intermediate learners. Moreover, the participants` usage of chunks in their speech was studied to see if the chunks had appropriately been used and the relationship between listeners` judgments on the participants` level of fluency and the frequency of chunks was also examined.

1. 2.   Theoretical and research background

Traditionally unit of analysis in language teaching was based on sentence but nowadays we are concerned with spoken utterances that can range from one word to an extended monologue. ''Language is used to negotiate and achieve meaning in social context and so cannot be divorced from those contexts'' (Schmitt, 2002, p.211). Burns (2001) stated that for many years spoken language was considered too disorganized, ungrammatical, and formless to be analyzed systematically. However, with the growth of communicative teaching methods it became important to investigate speaking and its influential factors. Regarding the corpus data on spoken language, it can be seen that spoken language contains many clauses which are independent of each other while written language consists of more dependent clauses (Schmitt, 2002). According to commentary on speaking fluency, the complexity of speaking task and also familiarity of the topic have significant role on second language speakers’ ability to form lexis, syntax, and access chunks without false starts and undue hesitations (Pawley & Syder, 2000; cited in Gorsuch, 2011).

As existing literature in language asserts, the term fluency carries a wide range of definitions. Hartmann and Stork (1976) pointed to the issue of automaticity and normal conversational speed. This issue is one of the important factors that need to be considered. However, it is not sufficient. Fluent native speakers use language effortlessly and this increases their speed; Hartmann and Stork defined fluency regarding this fact and neglected the point that there might be some fast speakers who are not considered fluent. McCarthy (2005, p.5) states that ''speed is not everything, at least not constantly rapid talks; some parts of the conversation may be uttered rapidly, but it may often be desirable to slow down in crucial parts of one`s massage''. Fillmore, Kempler, and Wang (1979) defined fluency as the ability to talk coherently at length without pause about a broad range of contexts and also the ability to be creative in language use. This definition has something in common with Hartmann and Stork`s. Fillmore pointed to the issue of automaticity implicitly. Brumfit (1984) and Schmidt (1992) also included automaticity as one of the characteristics of speaking fluency. Lennon (1990) noted that the term 'fluency' can be seen in two different senses. In broad sense it is often used as a general term to refer to proficiency and in its narrow sense it is considered as one component of oral proficiency that is often assessed in examining oral language skills. Lennon (1990) separated fluency from other scores in speaking exams and argued that fluency is different in a way that it is a performance phenomenon and the role of the listener is also important. Therefore, he pointed out that “fluency reflects the speaker’s ability to focus the listener’s attention on his/her message by presenting a finished product, rather than inviting the listener to focus on the working of the production mechanisms” (pp. 391-392). MacCarthy (2009) highlighted the point that the equivalent word for fluency in most languages, is typically based on a metaphor of ‘flow’ or ‘smooth delivery' and Persian is not an exception (e.g. Persian  equivalent ''ravani'' i.e. flowing). A research was carried out by Salmani (2008) on speaking fluency on schizophrenics. In this research fluency included three structural characteristics: a) rate of speech b) continuity c) articulation facility. In this research the third characteristic was focused on since the core of research was studying schizophrenics. However, issues such as rate of speech, filled or unfilled pauses and false starts are mentioned in various parts of the same medical research. Salmani (2008) pointed out that normally all people have filled and unfilled pauses in their speech; filled pauses occur due to many reasons like misperception, getting time to think, getting listeners` attention, emphasizing a point or a way to show your disagreement. Therefore, it can be concluded that having reasonable pauses in speech does not mean that the person is not fluent.

Fluency is also one of the criterions carefully examined in the standard proficiency tests such as IELTS. Such standard tests measure all skills and every skill is assessed by some criteria. Grading speaking has four criteria including fluency, coherence, pronunciation, lexical resource, grammatical range and accuracy; the first criteria in IELTS speaking test is fluency defined as the combination of a) speed of speech b) length of answer and c) pausing correctly (Cloe, 2011). Riazantseva (2001) stated that controlling the pauses and hesitations are important factors in fluency and concluded that second language learners should know how to pause and hesitate since these are some characteristics of fluency. Koponen and Riggenbach (2000; cited in Negishi, 2011) argued that there is no single all-purpose definition for fluency which includes all its aspects.

Researchers pointed to different aspects of fluency such as pausing and hesitation (Ellis & Barkuizen, 2005), automaticity and conversational speed (Hartman & Stork, 1976), and ready- made chunks (McCarthy, 2005). McCarthy (2005) has modified the definition of fluency and de-emphasized monologic performance in favor of the skillful use of chunks and also the cooperative construction of meaning among the speakers in dialogue. He also suggested three significant aspects of fluent conversation which include speakers as the central criteria, formulaic chunks that increase speech rate and conversational flow, and scaffolding which refers to the idea of confluence. He argued that '' chunks by their nature, are retrieved whole; they are not created anew each time; they are part of that automaticity which enables effortless accuracy'' (2005, p. 4).

Multi-word units make up a large proportion of any discourse. Erman and Warren (2000; cited in Conklin& Schmitt, 2008) calculated that formulaic sequences of various types included 58.6% of the spoken English discourse and 52.3% of the written discourse that they analyzed. It was judged that 32.3% of the unplanned native speech which was analyzed included formulaic language. Foster (2001; cited in Nattinger & DeCrrico, 1992) concluded that English speakers must master the formulaic language to a certain extent since they are so widespread in English discourse and these multi word units must be included in teaching materials as well.

Wood (2002) stated that formulaic sequences might have two roles in language acquisition. One role is to facilitate communicative ability and awareness and the other one is facilitating the cognitive skills of analysis since formulaic sequences can be broken down and analyzed. Therefore, by using this analysis, the original formulas, the pieces and rules are retained.

In general, a wide range of language features can be considered formulaic, from proverbs and sayings, the more traditional idioms to patterns of collocational choice based on frequency of co-occurrence (Huttner, 2005). In this study the focus is on the latter which can be easily accessible through corpus data.

Multi word units are mentioned in various research works by different terminologies including: pre formulated language, formulas and lexical phrases or bundles each emphasizing different aspects of how multi word units can be used. However, one feature is common in all that the words in each unit cannot be freely substituted with another word. Multi word units have a strong connection called collocation (Schmitt, 2002). The computer frequency counts are recorded in corpus linguistics for different purposes. One purpose may be for the material writers who want to present vocabulary or grammar in books for language learners; therefore, they choose the material with higher frequency to be presented first (Nation & McAlister, 2010).

While most speech production research and corpora investigations have shown the importance of multi word chunks, there has been few research into teaching of the formulaic sequences to nonnative speakers. Most researches (Boers et al., 2006; Conklin& Schmitt, 2008; Foster, 2001; Leedham, 2006) have focused on the noticing, identification and the process of making multi word chunks. Therefore research needs to be done with the major focus on teaching chunks in order to get advantage and employ the results in the classrooms.

1. 3.       Method

3.1 Participants

The present study includes two different categories of participants. The first category consists of students who have been observed during the semester to collect the necessary data. The second category of participants consists of two teachers, with fifteen years of teaching experience on average, who have acted as examiners measuring the students speaking fluency. Each of the participant types will be described thoroughly in the following paragraphs. 

3.1.1    Students` background

The students who participated in this study belong to two different classes. The first class consisted of twelve female students among whom five were selected randomly to be the participant of the experimental group. The second class included thirteen female students among whom five were assigned to the control group.

Both groups of the students were studying the same book with the same teacher and also the same class time on different days. The experimental group had classes on even days, but the control group on odd days. They were all intermediate students, between the age range from 25 to 35 with the mean age of 23 and the number of years they spent learning English ranged from 1 to 3 except for the time they studied English at school. They all shared the common experience of 2 years (on average) of learning English. They all took part in the class with same previous learning experiences. Therefore, they already had a good command of the basic structures of English and they were gaining more knowledge of English. The last point that should be mentioned about the participants is that none of them had ever been in an English speaking country.

3.1.2. Teachers background

The two teachers assigned as examiners to measure the students` speaking fluency, shared similar backgrounds. That is, they had the same native language, the same length of EFL learning and teaching experience. They have been teaching English for fifteen years in average in different institutions. Moreover, they have been professional, highly motivated, and experienced teachers.

3.2         Data collection instruments

Four different instruments were used in the present study: (1) Oxford placement test, (2) IELTS speaking evaluation sheet with band core descriptors, (3) Tables of two, three, four, five and six word clusters, and (4) Interviews. A description of these instruments appears below.

3.2.1    Oxford placement test

To make sure the participants in the two groups belonged to the same population in terms of proficiency level, Oxford placement test was administered. The test was divided into two parts containing 60 language use items and students were asked to answer the questions in 30 minutes as stated in the test (see appendix). Moreover, there was an assessing table at the end of the test which the researcher used to score the tests.

3.2.2    IELTS speaking evaluation sheet

This evaluation sheet was given to the raters to assess the fluency of the learners based on the issues outlined in the form as to increase inter- rater reliability and consistency of the judgments (see appendix). In order to measure its reliability, a pilot study was conducted.

3.2.3    Tables of two, three, four, five and six word clusters

Five tables presented in the Irish yearbook of applied linguistics were used in presenting the chunks to the learners. The tables include some pragmatic categories. By pragmatics, it is meant ''the creation of speaker meanings in context'' (McCarthy& Carter, 2002, p. 18). These pragmatic categories consist of discourse making clusters e.g. you know what I mean, the preservation of face and politeness e.g. I don’t know if/ whether, and the act of hedging and purposive vagueness e.g. to be honest with you (ibid).

3.2.4    Interviews

At the simplest level, interviews can be described as “the elicitation of data by one person from another through person-to-person encounters” (Nunan 1992, p.231). There are different types of interview that have been identified and differentiated by their degree of explicitness and structure, ranging from very open interviews to very structured ones (Shohamy, 1989).

One popular type of interview called semi-structured interview, was used in this study. Drever (1995) stated that the term “semi-structured” interview means that the interviewer sets up a general structure by deciding in advance what issue is going to be covered and what main questions are going to be asked; the person interviewed can answer at some length in his or her own words, and the interviewer responds using follow-up questions to get the interviewee to clarify or expand the answers.

In this study, the 10 EFL students were interviewed twice individually at different time intervals (at the beginning of the semester and at the end) by the raters. Each interview took about 8-10 minutes. The participants` speech was recorded in order to count the number of the chunks. It is worth mentioning that the second interview consisted of the same questions as the first one. This way participants` performance could be compared more accurately.

3.3     data analysis

Inferential statistics allows decisions to be made about the population based on the information about samples (Salkind, 2006). In order to measure the reliability of OPT test   Cronbach`s alpha, a coefficient of internal consistency was used. Pearson correlation was also used to measure the inter rater reliability between the two raters. Due to nonparametric features of this study Mann Whitney test was used to compare the score of the participants in the two groups.

1. 4.       Results and interpretation

The first step was to pilot the placement test (quick placement test, version 1) which was to be used to make sure that the two groups were equal and belonged to the same population in terms of their general proficiency. Therefore, prior to the study, the test comprising of language use and cloze test each with 25 and 35 items respectively, was administered to eight participants of the pilot study with the same qualities as those of the main study and then the reliability of the test was calculated using Cronbach's Alpha formula which turned out to be satisfactory at 0.87.

Table 4.1 Reliability Statistics

Next, the inter-rater reliability of the speaking section was computed for 8 participants who took part in the piloting process. In order to measure the inter rater reliability between the two interviewers, Pearson correlation coefficient which is a measure of the linear correlation between the two variables wasused.

Table 4.2 Correlations

As it can be seen in Table 4.2, the correlation between the two raters was .80 meaning that the average score of their marking could be safely used as the speaking score of every individual for the pre and posttest of speaking. Moreover, common variance turned out to be 0.64 which confirmed the significance of the correlation overlapping in terms of reporting accurate research results.

The general proficiency test after the process of piloting and inter-rater reliability measures was administered to the two target groups of the study. The score were subjected to Mann-Whitney which is a non-parametric test to ascertain the equality of the two groups in terms of their general proficiency.  Table 4.3 presents the results.

Table 4.3 Mann Whitney of the general proficiency test

The significance level of 0.84 being greater than 0.05 in the table above shows that the two groups were at the same level at the beginning of the study and belonged to the same population regarding general English proficiency.

The next step was to analyze the speaking section of the above mentioned general proficiency in isolation in order to make sure that the participants of the two groups were at same level in terms of speaking skill as well.

Table 4.4 Mann Whitney of speaking pre test

As it can be seen in Table 4.4, the significance level of 0.84 being greater than 0.05 indicates that the two groups did not exhibit any significant difference in their speaking means, hence as samples of the same population were considered as homogeneous.

After the 18-session instruction was over, the two groups were interviewed again and their average score measures by the two raters were analyzed.

Table 4.5 Mann Whitney of speaking post test

In Table 4.5, the significance value of 0.008, being far less than 0.05, shows that the two groups changed significantly, when compared with pretest speaking. It is worth mentioning that both groups of experimental and control group had improved in language proficiency owing to getting 18 sessions of training and that this improvement had been typical. However, the experimental group gained significantly higher scores in their second interview which consisted of the same tasks and questions.

4.6 Frequency of chunks in pretest and posttest speaking of the two groups

All interviews were transcribed and the number of the chunks was counted. It is worth restating that the raters were not aware of this part of the study.

Table 4.6 Relative Frequency of chunks in pretest and posttest speaking of the two groups

Table 4.6 shows the frequency of chunks used by the two groups. As it can be seen, there has been a significant growth in using chunks by experimental group since they were trained an also made aware of the importance of chunks in speaking fluency. By taking a look at tables 4.4, 4.5, 4.6, and 4.7, we can claim that there is a direct correspondence between the frequency of chunks and also the scores given by the raters i.e. the more chunks were used by the experimental group, the more fluent they were judged by the raters.

In addition to counting the number of the chunks used by the participant, the appropriate usage of the chunks was examined as well. It needs to be mentioned that the students had mastered the usages of these chunk. However, they overused them on some occasions.

Figure1.  Ten most frequent two to five word clusters

Among all chunks which were taught to learners these ten occurred more frequently, meaningfully, and accurately. As the graph suggests, two word chunks were more frequent than the others, this may be due to participants` familiarity with these chunks since they had heard them before in different contexts. However, these two word chunks were used with different intentions. Sometimes students are afraid of using chunks since they think they must have new words in their sentences; otherwise they will be considered as a weak communicator. However, their idea changed throughout the study by using these chunks in order to show they were good communicators. Moreover, when they were being interviewed, they sounded more confident as if they had always had something to say. As it can be seen in the graph, the last two chunks are five-word chunks used less frequently than the two-word chunks. Although the last two chunks were less frequent, they were used accurately by every participant. The reason of this low frequency may be due to time limitation of the interview.

1. 5.       Conclusion

The findings revealed that chunks did significantly improve the speaking fluency of the experimental group. This finding is along with what McCarthy and Carter (2002) stated about the implications of multi word strings (chunks) for the notion of fluency. Also, mastery of ready- made chunks could help learners to develop their fluency in speaking English. Participants of the experimental group were made aware of the advantages of using chunks in their speech and they were encouraged to use them while speaking throughout the whole semester. According to the comparison made between the pretest and posttest, the number of the chunks used by the experimental group increased and along with this growth, participants` fluency score increased as well.

Since the raters were not aware of the instruction of chunks and their judgments were not influenced by the number of the chunks, it could be concluded that there is a direct correspondence between the use of chunks and the listener's judgments of speaking fluency. The more we use chunks while speaking, the more fluent we will be considered by the listeners.

When chunks and their advantages were presented to the participants of the experimental group, they might have tended to overuse them. It needs to be mentioned that overusing chunks had been observed in some special cases. For example, when students were provided with a new series of chunks, they tended to overuse them in that session. However, the long term effect was rewarding since they got the idea to use chunks properly when needed.

This study has some implications for the different parties involved in the field of ELT. Learners, teachers, syllabus designers and material developers are all requested to appreciate the role chunks can play in both ESL and EFL settings.Both learners and teachers should know that chunks occurring in communicatively meaningful interactions can have greater pedagogical impact on speaking. This paper has found that in interaction, chunks can help learners to be more fluent.Material developers also need to consider the findings of this study when creating language learning materials in micro and macro levels. This requires them to include sections in the teacher’s guide which introduces different types of chunks and gives them a brief training as to when and how they should be used. Though difficult, this job, if done, can hugely contribute to the fluency of learners.