Effects of communication strategy training on EFL students ’ performance in small-group discussions

In recent years, a number of studies have been conducted with regard to communication strategy training and performance on communicative tasks (Lam, 2009; Nakatani, 2010; Naughton, 2006). This study aims to add to the literature by examining how two strategies, clarifying/confirming and extending a conversation, and two methods of teaching the strategies, affected the interactional sequences and overall group discussion performance of EFL students at a university in Japan. Pre and posttreatment small-group discussions were recorded for assessment, and a stimulated recall interview was administered to determine the participants’ perceptions of their learning and language use. Posttest results reveal that the experimental groups Stuart Benson, Danielle Fischer, Joe Geluso, Lucius Von Joo 246 that were taught predetermined phrases aimed at clarifying/confirming and extending a conversation employed such phrases more frequently than the control group. However, this employment of phrases did not lead to higher gains in group discussion skills as the control group enjoyed the largest gains from pre to posttest. The researchers consider the findings in light of previous research, and conclude with recommendations for future research on the topic with special regard to research design.

This article reports on research that explored the implementation of communication strategy instruction for in-class small-group discussions.The research had multiple purposes: to explore how strategy training affected students' performance on an institutionalized oral communication skills test, to measure the effects of two different styles of strategy training, and to record and analyze student perceptions of their own strategy learning and usage.To begin, three principal studies that influenced the current research will be reviewed, followed by an overview of the conversation strategies chosen for the present study, and descriptions and rationale behind the implementation of two different instruction styles.Methodology with regard to data collection and analysis will then be reviewed for the two teaching styles for the experimental groups and the control group (test scores).Attention will then be turned to the findings of the stimulated recall which aimed to shed light on student self-perceptions of their strategy use.Lastly, limitations of the present study will be identified and briefly discussed before concluding with suggestions for future research and research design.

Group Interaction and Language Development
Support for small group interaction in language learning is based on years of research in SLA.Long (1983) showed how learners negotiate the meaning of language used in conversation with interlocutors both by modifying their own output based on their interlocutor's reactions and requesting modification of input based on their understanding of what their interlocutor says.When engaged in conversation, language learners have a chance to test their knowledge of language by observing the reactions listeners have to their own output (Swain, 1985).Language learners can store in their memory language that met with positive results and reform language that did not.The meaning of the language used in conversation becomes clear and language use consequently improves through this process.
Despite the promising research on language learning through group discussion, there is also a downside to using group discussions to improve language skills.In some cases, conversation is simple enough that there is no need for negotiation of meaning and thus no new language learning takes place (Foster, 1998).This dearth of opportunities for miscommunication is even more likely in a classroom where all speakers share the same L1 (Buckwalter, 2001).In addition, using group discussion specifically with the intention of improving grammatical accuracy may be ineffective.If the main goal of a task is communication, listeners may feel there is no need to correct the words of a speaker if the meaning of the utterance is clear, bringing the unfortunate result of no negotiation of meaning (Pica, 1994).Even if accuracy is an explicit goal of a language task, language learners may not be able to focus on meaning and accuracy at the same time.According to Skehan (1998), students may even feel overwhelmed when they are trying to concentrate on both what they want to say and how to say it accurately.Even when accuracy is set aside and communication set as the primary goal of a group discussion task, some individuals may be more successful at conversation than others by utilizing their strategic competence to avoid or repair communication breakdowns caused by lack of knowledge of the target language (Canale, 1983;Canale & Swain, 1980).
In spite of these pitfalls, the researchers of the present study believe that group discussions are beneficial to language learning as well as necessary to facilitate other activities in a content-based curriculum.Dörnyei (1997) suggested that if a teacher is well trained in the effects of group dynamics and chooses appropriate tasks for group interaction, there will be a greater possibility for the negotiation of meaning and language learning opportunities in small-group discussion.However, strategic competence is not an innate skill and can be developed.Learners can be trained in various communication strategies in order to improve their language development in group interactions, as supported by research discussed in the following section.Naughton's (2006) research on training for communicative tasks focused on four strategies: follow-up questions, requesting and giving clarification, repairing non-target-like utterances, and requesting and giving help.The experimental groups were given explicit training in the strategies, while the control group received no training.All classes in the study were given the same group tasks, which were designed to naturally bring out the opportunity to use the communication strategies whether or not the students were trained to use them.The results showed a slight increase in overall strategy use in the control group and "a pronounced increase" (p.174) in the experimental groups.Naughton concluded that in order to improve students' strategy use and, consequently, their group conversation skills, it is not enough to just give students the opportunity to work on group tasks together; it is also necessary to teach cooperative communication strategies explicitly.Implementing the explicit instruction of communicative strategies, specifically in an EFL setting like the current study, may be an effective solution to the lack of natural opportunities for negotiation of meaning.

Studies on Strategy Instruction
In another study, Lam (2009) investigated the impact of metacognitive s t r a t e g i e s ( M C S s ) o n g r o u p t a s k s .T h e c o n c e p t o f M C S w a s t a k e n f r o m O' Malley and Chamot (1990).Lam explains that the goal of MCSs is "to oversee the general learning process by enabling the learner to think ahead of the goal and demand of the learning task, to plan for some action to tackle the task, and to assess how well one has done the task" (p.130).As in Naughton's study (2006), both groups were engaged in the same classroom tasks, but only the experimental group received explicit strategy instruction and how to implement them for the task.Seven strategies were employed: identifying the task problem, planning the content, planning language to use, evaluation of performance, asking for help, giving help, and positive self-talk.The study sought to investigate if these strategies would improve overall task performance and if students would utilize the strategies in future tasks.While Lam's (2009) research was not unique in relation to strategy training in group tasks, it was unique in its implementation of a multi-method approach to assessment.A total of four assessment methods were used: audio recordings, questionnaires, task session observations, and stimulated recall interviews (Gass & Mackey, 2000).The stimulated recall interviews were conducted to record the students' strategic thinking.The compiled results of all four assessment methods showed a statistically significant increase in the self-perceived use of strategies, the actual frequency of strategy use, and an increase in performance ratings in the experimental group.However, the increased frequency of actual strategy use did not correspond to that of the perceived use.Lam concluded that while the study was successful in fostering awareness of strategy use in a more immediate task, more practice and more time spent on the instruction of each strategy might help automatize students' declarative knowledge to be carried over successfully to future tasks, a point that we will come back to in the present study.
Finally, Nakatani (2010) administered communication strategies training to groups of Japanese university students in an EFL setting.In comparison to Naughton (2006) and Lam (2009), Nakatani (2010) used a more diverse set of communication strategies; the use of fillers and shadowing were considered a subcategory of communication strategies referred to as communication enhancers.Similarly to the other two studies, the strategies in Nakatani's study were taught explicitly.Nakatani also utilized multiple methods of assessment to obtain data on learners' cognitive processes and their perceptions of strategy use.The participants received a five-phase strategy training that consisted of review, presentation, rehearsal, performance, and evaluation of a task.Learners were presented with a list of strategies and could freely choose which ones to use during each training session.The results showed a correlation between students' proficiency levels; length of utterances typically produced, or words per c-unit; and type of strategy use, with maintenance strategies being the most significant predictor of performance.The research also reiterated the point that strategy training and negotiation of meaning in general is not automatically conducive to improving accuracy.Thus, the ideal use for strategy training in the classroom may best be reserved for tasks that focus primarily on meaning, reducing miscommunications, and maintaining the conversation flow.

The Study
The environment of the current study is unique in that the learners, while all sharing the same L1 of Japanese (or categorized as fluent speakers of Japanese), are expected to use the target language they learn in class, English, with foreign exchange students and teachers for a multitude of projects outside the classroom such as conversation activities and interview projects.It is thus necessary to define which communication strategies would serve them best in this capacity and train them accordingly.The present study investigates how communication strategy training can be used in group discussion activities to both reduce misunderstandings through clarification and confirmation, and to extend conversations through asking for more information.Thus, the strategies that are the object of this study are asking for clarification and extending a conversation.In addition, while the aforementioned studies that inspired the present study all focused on the explicit instruction of such strategies, the current study has the additional aim of investigating the outcome of an additional experimental group featuring a student and teacher coconstructed approach to strategy training to see if students are able to draw on their own preexisting knowledge of communication strategies.Lastly, in light of Lam's (2009) implementation of a multi-method approach to assessment, the current study analyzes the effectiveness of the conversation strategies in three ways: video-recorded observations, test scores on an oral communication test, and stimulated recall sessions.

Methodology
Research questions.This study investigated the potential correlation between various instructional methods used to promote the application of communication strategies and the frequency of use of the said strategies in small-group discussion tasks.In addition, the overall effect of communication strategy training on group oral discussion performance was evaluated.For this purpose, the following research questions were addressed: 1. What is the extent (frequency) with which students use communication strategies before and after training?2. Is there a relationship between the frequency of use of the oral communication strategy taught for the purposes of this study and the overall group discussion performance as determined by posttest scores?3. Does co-constructed or explicit instruction of communication strategies result in more student-employment of the said strategies and/or improved discussion performance?
Participants.The participants in this study were 76 native speakers of Japanese at a private university in Japan.They were from three sections of an English for International Communication course for first-year students; there was a total of seven sections at the time of the study.The participants' age ranged from 18-19, and all the participants had studied English as a foreign language for a minimum of 6 years before entering the university.The TOEIC test was administered one month before the study and the mean score was 439.7 across the seven sections of the course.The study was conducted within the first three weeks of the academic year with the intent being to minimize the effect students' other English courses and interactions with native English speaking international students at the university would have on their learning, and also to lessen the influence of teaching styles, as each section had a different teacher.
Teaching procedure.The sample consisted of three groups: Experimental Group 1, the group that received explicit instruction on conversation strategies from the instructor (n = 26); Experimental Group 2, which went through "co-constructed learning" with the instructor (n = 26); and a control group (n = 24).The treatments took place over two 90-min class periods in which the communication strategies were taught.The communication strategy in Lesson 1 was clarifying and confirming and the conversation strategy taught in Lesson 2 was extending the conversation.The instructors were given a carefully scripted lesson plan with the goal being to lessen the effect of the teaching style for each treatment.In order to further compensate for teacher effect, and in turn increase validity, the experimental groups were formed by dividing two intact classes in half and matching them with each other for the treatments.That is, half the participants from each of the two intact classes formed Experimental Group 1, and the other half formed Experimental Group 2. For treatments, Lesson 1 and 2, the respective experimental halves of the classes were brought together for treatment from instructor A, and the remaining students from each group were put together for treatment from instructor B. For treatment 2, the instructors taught the opposite group.
Experimental Group 1 was taught the two communication strategies in a teacher-fronted explicit manner.Predetermined formulaic sequences or phrases that were deemed useful in realizing the target communication strategies were taught to the group lecture style.The lecture was followed by an instructional video, produced in-house, that showed examples of how the predetermined phrases could be used in a small-group discussion to facilitate communication.After viewing the video, the participants were given a discussion question and had 7 min to discuss it, along with the chance to use the phrases.The participants were explicitly told to try to incorporate the phrases into their discussion.
Experimental Group 2, the co-constructed group, did not receive predetermined phrases explicitly from the teacher.Instead, they were given a transcript of a discussion and instructed to identify phrases that they believed facilitated the conversation.The teacher asked the participants to describe the strategy being realized by the phrases in their own words, and after they had done so, to highlight the formulaic sequences in the transcript that matched with the respective strategy.Lastly, participants brainstormed further phrases to realize the target strategies and collected them via a student-generated list on the whiteboard.After the participants exhausted their knowledge of phrases, the teacher made any necessary grammar corrections.They were then given the same discussion question as the explicit group and had 7 min to discuss it with the chance to use the phrases co-constructed with the teacher in class.In the treatment the participants identified the phrases included in Table 1.
For the duration of the 3-week study, the control group did not receive any instructional training targeting communication strategies.They did, though, have two 7-min discussions using the same discussion questions as the experimental groups as a warm-up activity to their daily lessons.They then proceeded with the usual tasks and activities in the course unit packet.

Assessment Procedures
Conversation test.The pre and posttreatment discussion tests were used to investigate whether participants' overall oral proficiency had increased by receiving communication strategy training.The pre and posttest consisted of a 7-min discussion in the same style as the training session discussions.For the discussion, groups of three or four learners sat at a table and discussed a set topic.The prompts used in both the pre and posttest were taken from the preexisting in-house test (Bonk & Ockey, 2003) and designed specifically to be answerable without requiring special topic knowledge.Each discussion group was taken into a private room for the pre and posttest and both test sessions were video recorded and backed up with an audio recording.Participants were told they were being recorded for research purposes and all participants signed a consent form.They were not informed of the experimental or control group conditions.In order to use the videos for pedagogical purposes beyond that of research, the videos were later uploaded onto a private Youtube station.Participants were given access only to the videos in which they were featured and they could reflect on their own language use in group discussions by comparing their pre and posttreatment performances.
Video stimulated recall interviews.Immediately following the posttest, a video stimulated recall interview was performed.Using the notes and the video recording taken during the posttest, two raters used a 3-staged approach to interview learners on conversation strategies that were used during the posttest.The 3-staged approach allowed raters to achieve an appropriate answer without influencing the participant's reflection.The questions asked were: 1. Why did you say that?/ what made you say that? 2.Where did you learn '_________'?3. Have you used that phrase before?These questions were asked in the order outlined above as necessary.So, for example, if a student's answer to the first question led him or her to answer Questions 2 and 3 without being prompted, no further questions were asked.
Rating of conversation test.The participants' pre and posttest performances were scored using the conversation skill band of a rubric that was developed and used at the time of this study for the university in-house oral assessment test: KEPT (Bonk & Ockey, 2003).Three raters, all of whom had had training in the KEPT test, rated both pre and posttests.Each rater watched the video-recorded tests on a private website which housed the videos and digital scorecards.The raters were able to assess the conversations online at their leisure within a 3-week period.This method of rating was chosen to avoid burnout the raters might have suffered had they watched the live pre and posttests as they were administered.After watching the videos of the small-group discussions, the raters assigned scores to the participants' conversations.The raters were not given any information on whether a video was pre or posttreatment and a few of the videos were repeated to test intrarater consistency.All the results were tabulated and the interrater reliability was calculated.

Descriptive data.
As can be seen in Table 2, the participants in this study who received instruction on how to incorporate conversation strategies into their small-group discussions demonstrated higher rates of strategy employment in terms of raw frequency than the control group.Experimental Group 1 employed the taught conversation strategy phrases for extending a conversation and asking for clarification a total of 49 times.Experimental Group 2 had the second highest rate of strategy employment at 45 instances, and the control group had the lowest at 37 instances.Table 3 outlines the average raw gains between pre and posttreatment performances for each group.As can be seen, the control group enjoyed the highest average raw gain in score from pretest to posttest with a gain of 1.11.The explicit instruction experimental group enjoyed the next largest gain with an average increase of 0.87 points.Finally, the group that participated in the coconstructed approach to instruction demonstrated the smallest gains at 0.23.Inferential data.In order to determine the statistical significance of the groups' performances in relation to one another, an analysis of covariance (ANCOVA) was carried out.The ANCOVA was chosen over the more common analysis of variance (ANOVA) in order to tap the difference in pretest scores between the groups.That is, the pretest scores are the covariate and the posttest scores are the response variable.As Table 4 illustrates, despite the more frequent employment of the formulaic sequences representative of the conversation strategies in experimental groups as shown in Table 2, the control group outperformed both experimental groups.The difference between the control group and Experimental Group 1, however, did not reach levels of significance.

Discussion
The results show that both experimental groups, explicit and co-constructed, employed the expressions in conversations more than the control group.This increased frequency of employment of phrases reflects the results of Naughton (2006), Lam (2009) and Nakatani (2010).However, judging from the ratings participants received on the in-house proficiency test, this increase in employment of expressions did not result in an overall improved score of conversation skills that outshone the control group.On the contrary, the control group demonstrated the largest gain from pre to posttest.Although this was unexpected, several conclusions as well as limitations and suggestions for future research may be drawn from the data.Firstly, the treatment sessions were short with only 90 min for each of the communication strategies.With the learners receiving little strategy instruction, it could be argued that students' declarative knowledge about the strategies had increased, but instruction had yet to affect the participants' procedural knowledge (DeKeyser, 2003).This means that the learners were knowledgeable of the strategies, such as their effectiveness in conversation and speaking proficiency, yet were still unable to effectively use the strategies in conversation, resulting in lower scores given by the raters.Recall that Lam (2009) observed as much in her own study: It is through repeated practice that declarative knowledge of strategy use may be automatized to become observable, procedural knowledge of strategy use . . .Hence, while the instruction effect may not yet be observable, the value of strategy instruction may lie in explicit learning.(p.144) Aston (1986) noted that the intentional incorporation of strategies into conversations may disrupt smooth interaction, resulting in exactly the opposite of the desired effect.More time spent on the acquisition of the strategies, in terms of introducing the strategy, the phrases and explicit practice with them, might have allowed participants to effectively go from the knowing stage to the using stage of the strategies (DeKeyser, 2003).This is a question that could be addressed via a delayed posttest.Future studies may consider taking a more longitudinal approach to data collection to address this issue.
With respect to Experimental Group 2, the co-constructed group, it could be that not only did they need more time to learn the phrases associated with the speaking strategies, but that the participants needed more time to become comfortable with the central tenets of co-constructed learning.Research has indicated that the Japanese secondary schooling system rarely uses co-constructed instruction; the prominent approach being explicit instruction (Kikuchi & Browne, 2009;Nishino, 2008;Taguchi, 2005).The fact that the participants had not studied English in a co-constructed manner prior to the study could be detrimental to their learning of the expressions.As stated earlier, in order to minimize teacher effect and outside influence, the study was conducted in the first three weeks of the school year.Therefore, we should consider the timing in the academic year in which the study was conducted.Perhaps replicating this study at the end of an academic year would yield higher gains as it would afford students more time to get accustomed to a more co-constructed style of learning.This is also an issue that would benefit from a more longitudinal study.
In addition to providing the students with more time to use and learn the target phrases, teaching the students fewer phrases or limiting or pruning the number of phrases they generate in the treatment may have increased student mastery of the phrases and thereby increased conversation proficiency.Perhaps future studies can prioritize limiting the number of phrases to be studied.
Finally, the function of the strategies taught may also have an impact on the scores doled.Nakatani (2010) found that the most significant predictor of positive posttest scores was the use of maintenance strategies, such as providing an active response, for example, I see or that sounds good, and shadowing, or repeating what the interlocutor just said to show that you are engaged in the conversation (pp. 122-123).Recall that the conversation strategies focused on in this study were clarifying and confirming and extending a conversation.Future studies may find more powerful results by focusing on maintenance strategies, or simply focusing on a wider gamut of conversation strategies in general over a longer period of time.

Conclusion
The studies on conversation skills conducted by Naughton (2006), Lam (2009), andNakatani (2010) all found that students increased their frequency of employment of strategies after conversation strategy training when compared to control groups.The present study had similar findings; students who took part in treatments representing both approaches to strategy training, explicit and coconstructed, employed the interactional communication strategies more frequently than the control group.However, the results show that Experimental Group 2 did not receive as high marks with respect to conversation skills as the other two groups.Considering the limitations of the study outlined in the previous section, though, we caution the reader not to be too quick to dismiss the use of co-constructed learning when teaching communication strategies.It could be that students in this context were not yet accustomed to the coconstructed style of teaching and learning, and the approach could be more effective in another context or at a later point in the students' careers at the university in this study.Nakatani's (2010) study was conducted 12 weeks into the course, which allowed learners to get accustomed to the learning style of the institute.As was suggested above for future research in this area, delaying the time period in which data is collected could allow the learners to study in a co-constructed learning setting and in turn not be a factor in the study.However, if the study was delayed to allow learners to study in a co-constructed setting, care would need to be taken in the treatment to ensure that teacher effect was minimal.One way to decrease teacher effect would be to have different instructors from those of the learners to conduct the treatment classes.
Despite the limitations, the results do indicate that students are able to increase their employment of predetermined phrases with the aim of improving the fluidity of their conversations.While this increased use of phrases did not result in outside raters in this study perceiving the participants as having increased conver-sation skills mere weeks after treatment, it remains to be seen if longer-term effects can be found.This in turn could influence the way in which a teacher constructs the class.An example of this could be allowing learners to work together to formulate their own phrases instead of explicitly giving phrases.
Future studies in the area of communication strategies should also consider two important issues when deciding what set phrases to incorporate in the study.One issue would be the number of set phrases to use.This study used 17 phrases from two communication strategies.It could be argued that 17 phrases were overwhelming for the learners to study and produce in a short period of time and thus had an adverse effect on their conversation.One may consider setting a limit on the number of phrases that are produced in the co-constructed learning group.The other issue for researchers to consider is what type of communication strategy to incorporate.Nakatani's (2010) study looked at a broad array of strategies and found that learners who used maintenance strategies, specifically active response and shadowing, received the highest scores.A comparative study involving two types of strategies between experimental groups could provide valuable results as to which strategies are worthwhile to incorporate into a curriculum.With also using two types of strategies in different learning styles, the results could indicate what type of strategy would be best for a particular setting, such as Japan.
Finally, and perhaps most importantly, we wish to reiterate our recommendation that future studies take a more longitudinal approach.Lam's (2009), Naughton's (2006) and Nakatani's (2010) studies were longer than this study, with Lam's (2006) being conducted over a 5-month period.Expanding the instructional period of time could allow learners to further understand the set phrases and in turn produce more substantial results.A delayed posttest would also indicate if learners had achieved the procedural knowledge necessary to accurately produce the set phrases in a conversation in an extemporaneous fashion.

Table 1
Phrases used in the study

Table 2
Raw frequency of strategy employment

Table 3
Raw gains from pre to posttest

Table 4
Statistical significance for ANCOVA comparisons of experimental and control groups