Performance Goal-Setting and Feedback for Second Language Tasks. An Empirical Study of TBLT Group Discussions

TABLE OF CONTENTS

CHAPTER 1. INTRODUCTION
1.1 Thesis focus and aims
1.2 Background and research motivation
1.3 Research questions
1.4 Thesis outline

CHAPTER 2. LANGUAGE LEARNING WITH TBLT DISCUSSIONS
2.1 Introduction
2.2 SLA in English communication classes
2.2.1 Oral communication and SLA
2.2.2 Orally interactive tasks and SLA considerations
2.3 The group discussion approach to language learning
2.3.1 Potential learning and teaching benefits of group discussions
2.3.2 Challenges for learning and teaching with group discussions
2.4 Task-Based Language Teaching (TBLT) group discussions
2.4.1 Communicative Language Teaching (CLT) and TBLT discussions
2.4.2 TBLT versus Present-Practise-Produce (PPP) for group discussions
2.4.3 Challenges for learning and teaching with TBLT group discussions
2.4.4 Challenges in Japan for TBLT group discussions
2.5 Chapter summary

CHAPTER 3. DETERMINING ORAL GROUP DISCUSSION PERFORMANCE
3.1 Introduction
3.2 Participation and CAF measures
3.2.1 Participation
3.2.2 Fluency
3.2.3 Accuracy
3.2.4 Complexity
3.3 Additional performance considerations
3.3.1 Group interactions
3.3.2 Clarity of communication
3.3.3 Discussion outcome
3.4 Chapter summary

CHAPTER 4. GOAL-SETTING AND FEEDBACK (GSF) FOR GROUP DISCUSSIONS
4.1 Introduction
4.2 Goal-setting and learning
4.2.1 Task goal-setting, motivation and engagement
4.2.2 Interpersonal and intrapersonal task goals
4.3 Formative Assessment (FA)
4.4 Task performance scoring rubrics
4.5 Group discussion GSF design
4.5.1 GSF focus
4.5.1.1 Individual Process GSF
4.5.1.2 Group Product GSF
4.5.2 Performance self-scoring method
4.5.2.1 Performance rating scales
4.5.2.2 Performance counting systems
4.6 Chapter summary

CHAPTER 5. METHODOLOGY
5.1 Aims of the study
5.2 Rationale for the research methods
5.2.1 Mixed-method approach
5.2.2 Use of classroom observations
5.2.3 Use of surveys
5.2.4 Use of peer-interviews
5.3 Participants
5.3.1 Teachers
5.3.2 Students
5.4 Research procedure
5.4.1 Classroom-based study preparation
5.4.2 Semester-long classroom-based study
5.4.3 GSF class procedure (Weeks 4-7 and 9-12)
5.5 RQ1 data collection and analysis
5.5.1 Teacher surveys
5.5.2 Teacher follow-up interviews
5.6 RQ2 data collection and analysis
5.6.1 Classroom observation data collection
5.6.2 Discussion performance measures selection
5.6.3 Discussion transcript coding and analysis
5.7 RQ3 data collection and analysis
5.7.1 Attitudinal surveys about classroom discussions, tests, goal-setting and feedback
5.7.2 Student peer-interviews
5.7.3 Student response coding and analysis
5.8 Ethical issues
5.8.1 Data collection
5.8.2 Teaching considerations

CHAPTER 6. RESULTS AND DISCUSSION
6.1 Chapter Introduction
6.2 RQ1: Appropriate discussion performance goals
6.2.1 Introduction
6.2.2 Teacher survey and interview results
6.2.2.1 Giving opinions
6.2.2.2 Taking speaking turns
6.2.2.3 Reacting to speaking turns
6.2.2.4 Clarifying turns
6.2.3 GSF pilot and teacher journal results
6.2.4 RQ1 results summary
6.2.5 RQ1 discussion
6.2.5.1 Performance rubric considerations
6.2.5.2 GSF and learning considerations
6.2.6 Limitations
6.3 RQ2: Changes in observable discussion performance
6.3.1 Introduction
6.3.2 Overview of performance measure changes
6.3.3 Specific performance measure changes
6.3.3.1 Participation
6.3.3.2 Fluency
6.3.3.3 Accuracy
6.3.3.4 Complexity
6.3.3.5 Task process-focused performance
6.3.4 Additional performance considerations
6.3.4.1 Outcome-promoting, on-task and off-task turns
6.3.4.2 Clarifications
6.3.4.3 Turn-taking strategies
6.3.4.4 Possessive pronoun usage
6.3.5 RQ2 results summary
6.3.6 RQ2 discussion
6.3.6.1 Overall discussion performance changes with a TBLT approach
6.3.6.2 Discussion performance changes with Product and Process GSF
6.3.6.3 LP/HP performance changes
6.3.7 RQ2 key findings summary
6.3.8 Limitations
6.4 RQ3: Student self-reported feelings towards the GSF and discussions
6.4.1 Introduction
6.4.2 Discussion feelings survey results
6.4.3 Test difficulties survey and peer-interview results
6.4.3.1 Initial reported discussion test difficulties
6.4.3.2 Similarities in ProdS and ProcS final reported test difficulties
6.4.3.3 Differences between ProdS and ProcS final reported test difficulties
6.4.4 GSF survey and peer-interview results
6.4.4.1 Overall reported feelings about the GSF and performance
6.4.4.2 Reported feelings about the GSF sheets
6.4.4.3 Reported feelings about the GSF diaries
6.4.5 RQ3 results summary
6.4.6 RQ3 discussion
6.4.6.1 Reported feelings about discussion performance
6.4.6.2 Similarities in the reported effects of Product and Process GSF
6.4.6.3 Differences in the reported effects of Product and Process GSF
6.4.7 RQ3 key findings summary
6.4.8 Limitations
6.5 Summary of research question findings

CHAPTER 7. CONCLUSIONS
7.1 Contributions to research
7.2 Recommendations for language teaching
7.3 Thesis limitations and future research directions

REFERENCES

APPENDICES

ABSTRACT

This thesis draws on a classroom-based empirical study to explore the actual effects that Task-Based Language Teaching (TBLT) has on students’ performance, when applied to group discussions, and the impacts that different forms of Goal-Setting and Feedback (GSF) have on their learning. In doing so, it challenges the assumptions in the research literature that TBLT will necessarily improve multiple aspects of performance within group discussions with low-level students, and reveals that applying GSF can lead to very different outcomes.

A longitudinal mixed-method approach was adopted using surveys and peer-interviews with 10 teachers, and observations, surveys and peer-interviews with 132 low-level students in a Japanese university. Students used product or process GSF alongside TBLT group discussions across a semester. Findings showed improvements in fluency and accuracy, positive feelings towards learning, and larger improvements for lower performers. Furthermore, product and process goals influenced students' focus differently in terms of individual performance, collaboration and discussion outcome. These findings create a clearer picture of the impact of TBLT, when applied to group discussions, and show how students' focus within learning can be greatly influenced by task goals. Resultant recommendations for course design, student and teacher training, and implementation of TBLT and GSF are given.

LIST OF FIGURES

Figure 5.1. Research summary

LIST OF TABLES

Table 5.1 Study preparation

Table 5.2 Study procedure

Table 5.3 LP and HP group distribution

Table 5.4 Class procedure

Table 5.5 Discussion coding and analysis process

Table 5.6 Survey and interview responses coding and analysis procedure

Table 6.1. Teacher ratings of individual process measures for individual assessment in discussions

Table 6.2. Finalized individual process measures for students during discussions

Table 6.3. Repeated measures ANOVA results

Table 6.4. Participation paired-sample t-test results

Table 6.5. Fluency paired-sample t-test results

Table 6.6. Accuracy paired-sample t-test results

Table 6.7. Complexity paired-sample t-test results

Table 6.8. Task process-focused paired-sample t-test results

Table 6.9. Mean total group off-task and outcome-promoting turns

Table 6.10. Summary of significant ANOVA repeated measures and follow-up t-test results

Table 6.11. Summary of significant differences between LP and HP performances

Table 6.12. Student self-reported feelings towards discussions

Table 6.13. Initial (W3) and final (W13) self-reported difficulties for discussion tests

Table 6.14. Week 13 student self-reported usefulness of sheet/diary

Table 6.15. Week 13 student self-reported future usage preferences for sheet/diary

Table 6.16. Final (W13) self-reported helpfulness of discussion sheet

Table 6.17. Final (W13) self-reported helpfulness of discussion diary

Table 6.18. Summary of main final reported difficulties for discussion tests

Table 6.19. Summary of student perceptions of benefits and problems with discussion sheets and diaries

Table 6.20. Summary of RQ1-3 findings

LIST OF APPENDICES

Appendix A. Summary of Module One findings regarding factors affecting group discussion participation for Japanese university students

Appendix B. Summary of Module Two findings regarding the effects of group discussion planning for Japanese university students

Appendix C. Teacher journal notes

Appendix D. Teacher survey and interview content

Appendix E. Product GSF sheet and diary screenshot

Appendix F. Process GSF sheet and diary screenshot

Appendix G. Teacher and student study information sheet

Appendix H. Teacher and student ethical content form

Appendix I. English versions of 1st (Week 3), 2nd (Week 8) and 3rd (Week 13) student attitudinal survey and interview content

Appendix J. Discussion topics

Appendix K. Week 3 product group discussion transcript and coding example

Appendix L. Week 13 product group discussion transcript and coding example

Appendix M. Week 3 process group discussion transcript and coding example

Appendix N. Week 13 process group discussion transcript and coding example

Appendix O. Student discussion test difficulty open-ended responses coding examples (ProdS – Week 13)

Appendix P. Student GSF sheet usefulness open-ended responses coding examples (ProcS – Week 13)

Appendix Q. Student GSF diary usefulness open-ended responses coding examples (ProdS – Week 13)

LIST OF ABBREVIATIONS

CAF - Complexity, Accuracy and Fluency

CLT - Communicative Language Teaching

FA - Formative Assessment

GSF - Goal-Setting and Feedback (task performance focused and self-regulated by students)

L2 - Second Language

LPs/HPs - Low Participators/High Participators (half of students who spoke the least/most in discussions at the start of the study)

ProcS - Process Students (used Process GSF sheet/diary during class across the semester)

ProdS - Product Students (used Product GSF sheet/diary during class across the semester)

SLA - Second Language Acquisition

SRL - Self-Regulated Learning

TBLT - Task-Based Language Teaching

CHAPTER 1. INTRODUCTION

1.1 Thesis focus and aims

This thesis is the third part of a Modular PhD investigating the use of Task Based Language Teaching (TBLT) oral group discussion tasks for language learning with low-level learners. The overall aim of the PhD is to investigate and report on ways to improve the learning for students. This was done by firstly determining key factors affecting low-level Japanese university students’ oral participation within discussions in the first module (see Appendix A for a summary), and then by examining the short-term effects on participation of pre-discussion planning (a significantly reported factor in the first module) with low-level Japanese university students in the second module (see Appendix B for a summary). The main finding was that when the students undertook such additional planning, they would speak more and with more fluency during discussions immediately afterwards.

Three of the other task design factors reported to potentially improve participation in the first module were related to 1) having a scoring system for performance, 2) getting feedback on performance, and 3) seeing measurable progress of performance over time. As a result, I decided to focus this thesis on these three factors by investigating the effects on TBLT group discussion learning of self-regulated performance Goal-Setting and Feedback (GSF) via a semester-long classroom-based study. Data in this thesis considers observable changes in performance by Japanese university students due to the use of a TBLT approach to group discussions, changes observed with the use of two different types of GSF (task product versus process focused), and self-reported feelings of the students towards the learning undertaken. The findings contribute to TBLT and goal-related research by examining the suitability of TBLT group discussions as an approach to improving language use with low-level learners and how GSF may support the learning or not.

1.2 Background and research motivation

Upon entering university, most Japanese students have studied English since an elementary school age, most recently with five years of mainly grammar-focused English instruction in Junior and Senior High School involving translating between Japanese and English, known as the yakudoku method (Gorsuch, 1998; Nishino, 2008; Nishino & Watanabe, 2008). Such classes have often not involved Communicative Language Teaching (CLT) approaches to second language learning, such as Task-Based Language Teaching (TBLT), resulting in limited chances for students to interact orally with each other in English. The Japanese Ministry of Education, Culture, Sports, Science and Technology (MEXT) laid out plans in 2013 to enable students to hold conversations in English by the time they leave High School in preparation for the Olympic Games in Tokyo in 2020 (MEXT, 2013). If such ambitious goals are to be met, they require careful consideration with regards to the teaching of conversation skills during high school and into courses at the university level. However, because of the pressure placed on high school students to pass university entrance exams in Japan (Aspinall, 2005) classroom learning focuses mainly on the content of such tests via the yakudoku method. As a result, little time is left for orally interactive tasks, resulting in university students' oral English communicative competence being often limited to simple exchanges at best (King, 2012, 2013, p. 72).

I have been teaching English within Japan for ten years at the time of writing this thesis, having taught English communication skills at the elementary, high school, university and business-level. Of specific relevance to the focus of this thesis, I taught English communication courses at Kwansei Gakuin University in Kansai, Japan, between 2013 and 2016, and have been teaching similar courses at Hosei University in Tokyo since 2016. From my own experience of working within universities in Japan, students undertaking group discussion tasks have seldom experienced goal-setting for discussion performance, nor been provided with specific feedback to help focus their efforts on improving their performance related to such goals. However, a large amount of recent research, including some of my own, suggests that helping students focus on specific task performance goals and feedback can improve their motivation, efforts made, participation within classwork, and performance across time (Bargh, Gollwitzer, Lee-Chai, Barndollar, & Troetschel, 2001; Hart & Albarracín, 2009; Moskowitz & Grant, 2009; Stroud, 2017).

The number of choices available to teachers for implementing performance goals and feedback for oral tasks are vast (Lai, 2015; Leung, 1999; Leung & Lewkowicz, 2006; Norris, 2008) and are often subjective scale ratings of measures such as 'fluency', ‘accuracy’ and ‘complexity’ (such as in the TOEIC, TOEFL and IELTS speaking tests). A focus on such scoring can often leave students without an understanding of how to focus their efforts to improve in the future (Orsmond, Merry, & Reiling, 1997; Price, Handley, Millar, & O’Donovan, 2010). From what I have seen in Japan, feedback on classroom discussions also often comes in the form of such subjective, non-specific scale ratings from classmates or the teacher. I do not believe that this helps students understand their performance with enough detail, nor provide them with any measurable progress on that performance over time to understand how to focus future efforts to improve. If Japanese students are expected to improve their performance across courses, they require specific and measurable goals to become motivated to take part in classwork (Moskowitz & Grant, 2009), as well as clear, specific, and ongoing feedback which provides them with what is called 'assessment for learning' (Dann, 2002) via a 'formative' style of feedback (Black & Wiliam, 2009; Harlen & James, 1997; Sadler, 1998; Tunstall & Gipps, 1996; Wiliam, 2018).

Several challenges exist for the implementation of goals and feedback into classroom group discussions. Firstly, it can be unclear for teachers and students how they should focus efforts within performance, such as goals related to individual speaking turns, interactions between speakers, or the outcome of the discussion itself. Secondly, there may be a lack of time for the use of goals or feedback within class. Such extra workload may take away from the time required for practising the use of the language. Also, English communication class sizes can sometimes be too large for the teacher to be able to spend time observing individual students across a course, in order to give them detailed individual feedback. Goals and feedback may need to be self-regulated by students themselves to avoid such issues with time. Thirdly, individual differences, such as learning preferences and English-speaking ability, can make the use of goals and feedback more difficult for some students than others. Lower-level students may already be struggling to perform within discussions alongside higher-level English speakers, and the additional workload of goals and feedback may actually have negative effects on their performance. Therefore, any goals or feedback used should be as quick and simple to use as possible. Lastly, any performance feedback provided to students needs to be clear and specific, but this may be difficult to do in a limited amount of time within classes.

Detailed research projects which investigate the development of student performance and feelings towards group discussions across time are scarce, even though this data would prove very helpful for teachers who are struggling to improve English oral interactions within their classes. Due to the extensive positive research which exists about the use of goals and feedback to improve classroom learning (see Chapter Four), as well as my own research and the findings in the first module (Appendix A), I decided to focus this study on how combined performance GSF might be self-regulated by students in typical English communication courses within Japanese universities to improve the learning with TBLT group discussions. I believe that such an approach is an important topic of future language learning research, as it can potentially help students understand their ability better (as determined by the goals and feedback used) and focus more on improving across time.

1.3 Research questions

The research questions within the study were selected to help improve the understanding of the potential effects of using a TBLT approach and GSF to support learning undertaken during classroom group discussions. With regards to the GSF used in the study, goals were those focused on discussion task performance which were set by students themselves within their electronic diaries (Appendices E and F) prior to each classroom discussion. Feedback for students referred to that which was provided by 1) audio recordings of group discussions, 2) notes which students took on their own discussion sheets and 3) the excel tables showing performance over time within the electronic diaries. The main overall RQ addressed was:

Main RQ: ‘What are the effects on learning of using Goal-Setting and Feedback (GSF) with TBLT group discussions across a semester?’

I decided to approach this RQ by breaking it down into three separate RQs. The first RQ was used to specify what type of goals should be used with the students in the study. The second RQ then addressed observable changes in student performance over time with a TBLT approach, as well as with two different types of GSF used, including differences between students who spoke less (Low Participators) or more than others in discussions (High Participators) at the start of the study. The third RQ addressed self-reported student feelings towards undertaking TBLT discussions and the two types of GSF used (ProdS using Product GSF and ProcS using Process GSF). Within the study, Product GSF focused on goals related to the outcome at the end of discussions (the final group choice, reasons, examples and other possible choices and reasons), while Process GSF focused on goals related to the interactions which took place during the discussions (the number of opinions, reasons, examples, questions, answers, agreements and disagreements). The three RQs were:

RQ1: What are appropriate discussion performance goals for the Japanese university students in this study?

RQ2: (a) How does observable discussion task performance change for the students across a semester using a TBLT approach (regardless of the type of GSF used)?

(b) What different effects do Product and Process GSF have on observable performance across a semester?

(c) Are these effects the same for Low and High Participators?

RQ3: (a) How do ProdS and ProcS report feeling about performing in discussions across the semester?

(b) How do they report feeling about the support the two types of GSF provided for their learning (or not)?

The findings for these RQs make important contributions to research by providing original data on the longitudinal effects which a TBLT approach to group discussions can have with low-level learners (in and out of Japan), as well as the impact which the addition of GSF has on learning. This is beneficial to both researchers and teachers currently using or wishing to apply such approaches to their own language courses.

1.4 Thesis outline

In line with the RQs above, the theoretical background discussed in this thesis addresses three main themes within Chapters Two, Three and Four. Firstly, a background to the current use of group discussions within language learning classrooms is examined, with particular attention given to the common use of a Task-Based Language Teaching approach. Secondly, a discussion of the literature connected to appropriately measuring group discussion performance for students is provided. Thirdly, the potential effects on performance and learning of the design of goal-setting and feedback for group discussions is discussed using current research and theories related to goal-setting, formative assessment and performance rubrics.

In Chapter Five, the methodology of the semester-long classroom study undertaken is explained. This includes a rationale for the mixed-methods approach taken for the data collection and analysis, details of the participants and procedures, specific details of the data collection for the three separate RQs, and the ethical considerations within the study.

In Chapter Six, the results, discussion and limitations for all three RQs are given. The first part discusses the observational data collected from a classroom pilot, as well as self-reported survey and interview data from teachers, which were used to create the two types of GSF in the study (RQ1). The second part summarizes the changes in student performance across the semester with TBLT groups discussions using the two different types of GSF via classroom observations (RQ2). The third part explores reasons for the changes seen in RQ2 by using data regarding student feelings towards their performance in discussions over time and the two types of GSF used with data from self-reported surveys, interviews and my own observations during classes and tests (RQ3).

In Chapter Seven, conclusions are reached about the use of TBLT group discussions as an approach to language learning with low-level learners and the effects of GSF. Based on these findings, the contributions made to research, recommendations for language teaching, overall limitations for the thesis, as well as recommended future research directions are explained.

CHAPTER 2. LANGUAGE LEARNING WITH TBLT DISCUSSIONS

2.1 Introduction

This chapter is made up of three main sections which gives an overview of TBLT group discussions as an approach to the learning and teaching of spoken English. The lack of classroom-based research to understand the actual effects of such an approach on student performance is highlighted and later analyzed using longitudinal data within the study. This chapter is focused mainly on how students and teachers may be benefiting or not from such a TBLT approach, and also how it may compare to a more traditional alternative approach called Present, Practise, Produce (PPP). The first section discusses relevant literature for understanding how students may acquire a second language through oral use of the language during tasks. The second section discusses the potential benefits and challenges to both learning and teaching with the use of group discussions within English communication courses. The third section gives an overview of the Task-Based Language Teaching (TBLT) approach to classroom group discussions, as well as the potential benefits and problems for both students and teachers with using it in general and specifically within Japan.

2.2 SLA in English communication classes

2.2.1 Oral communication and SLA

The processes through which students may acquire a second language needs careful consideration, so that courses can be designed to assist that acquisition. The two main, but contrasting, perspectives for this are the nativist and interactionist viewpoints, which will now be discussed. Within this thesis, I do not make any conclusions as to which of these viewpoint is more likely to be correct for SLA. I explain how they both relate to orally interactive tasks, as well as highlight the lack of empirical data which currently exists to show how language use can develop across time within interactive tasks (such as group discussions), which the data in the study provides.

The nativist viewpoint within SLA is that it is the natural internal mechanisms of a student working on the language they hear and prepare to say which leads to SLA and resultant communicative competence. A specific example is Krashen's (1985) Input Hypothesis, which suggests that learning a language is more about acquiring it through input, rather than learning it through interacting and responding to others. In addition to this, the Output Hypothesis (Swain, 1985, 1995) states that it is important for students to orally produce language in order to improve at it, because it promotes noticing, experimenting, and becoming more structured and accurate at speaking through self-reflection of mistakes made and difficulties experienced. A more in-depth discussion of these theories is beyond the scope of this thesis. However, they both suggest that improving the oral communicative competence of students is mainly about having them practise listening to and understanding the speech of others, as well as producing their own speech by going through the internal processes of language production described as conceptualizing, formulating and articulating (De Bot, 1992; Levelt, 1989). In a second language communication course, this could involve a high focus on listening tasks and monologue speeches for example.

However, the interactionist viewpoint of SLA is that language acquisition occurs as a result of social interactions between speakers, rather than just the internal processing of input or output of speech. Sociocultural Theory (Vygotsky, 1978, 1986) highlights the need for speakers to use the language to interact with others in a social context, in order for meaning-negotiation and hypothesis-testing to be present and drive SLA. Furthermore, Long (1989) stated that Krashen's input hypothesis (discussed above) is only practical for SLA if the input is comprehensible and if interactions between learners help clarify misunderstandings. The Interaction Hypothesis (Hatch, 1978; Long, 1996) states that interaction created by tasks helps students improve their language use, as opportunities are provided to attend to problems using the language within specific contexts. The modifications which take place in the negotiation of meaning and new utterances which are used to clarify meaning between speakers as a result, are believed to lead to SLA. This theory of learning has also been referred to as the Interactionist Approach (Gass & Mackey, 2007), proponents of which hold that learners who use meaningful and functional dialogue in an interactive way will be practising a more 'authentic' style of language use than within individual tasks and will become better at using the language (Larsen-Freeman & Anderson, 2013, p. 157; Savignon, 2002). Such language practice has been shown in studies to lead to better performance in future interactions (Gass & Varonis, 1994; Pica, 1994). In addition, the Socio-Interactionist view of SLA (Firth & Wagner, 2007; Sun, 2011) is very similar to the interaction hypothesis discussed above, as it suggests that interactions between speakers are key for SLA, not only because they practise using the language, but also because it is done within a social context and that the social interactions which students have are more supportive of the learning than their cognitive processes of producing sentences of speech. More classroom-based research is required at this time to see how/if students who practise second-language speaking skills through interactive tasks (such as group discussions) will acquire the language, as the above theories suggest, as little empirical evidence exists to show this (Keck, Iberri-Shea, Tracy-Ventura & Wa-Mbaleka, 2006).

2.2.2 Orally interactive tasks and SLA considerations

Following on from the section above, and assuming that interaction plays an important role in SLA, teachers also need to consider some other important cognitive processes involved in the learning. Firstly, the working memory of students is limited and will determine how much pre-formulated language they can draw upon whilst interacting with others (Baddeley, 1986, 1993; Baddeley & Hitch, 1974). Skehan (1996, 1998, p. 97) suggests that students have a limited capacity for learning, and that tasks which require attention on certain elements in performance leave students without enough 'attention resources' to focus on other elements. However, Robinson (2001, 2003, 2007) opposes this belief, saying students can draw on different 'pools' of attention at the same time. For example, students trying to speak with higher accuracy, by making less errors in speech, may speak less fluently, by speaking more slowly (or vice-versa). Whether students can focus on and improve different aspects of performance at the same time or not requires further research, to better understand any cognitive limitations within learning for students.

Secondly, the cognitive load (Candlin, 1987) which orally interactive tasks put on students may cause problems with their learning. Asking students to interact in English requires consideration of what Skehan (1998, p. 99) describes as 'code complexity' (how difficult the language required for the task is), 'cognitive complexity' (how complex the task is to undertake), as well as 'cognitive processing' and 'communicative stress' (the amount of organizing and processing of language required within the task time available). Teachers must ensure that these demands on students are not so high that they do not prevent interactions between students which are expected to lead to SLA.

Thirdly, there is a clear lack of current research data to link theories of interaction (Section 2.2.1) to SLA in group discussions across time. The studies mentioned by Gass and Varonis (1994) and Pica (1994) were only for pair interactions and only considered improvements in language use across very short periods of time. It cannot be assumed that the same effects of interactional tasks will occur within larger group sizes or in the longer-term within classes. As discussed above, some studies (such as Keck et al., 2006) suggest that there is no proven connection between oral interaction and SLA. This study provides more data related to this by making connections between the oral interactions which take place during group discussions and how performance within those discussions changes over time (Section 6.3). This helps teachers see more clearly if time invested in learning through group discussions is in fact leading to improvements in language use or not.

2.3 The group discussion approach to language learning

2.3.1 Potential learning and teaching benefits of group discussions

Orally interactive tasks, involving two or more students, are a commonly used approach to promote language learning, as they are believed to create the interactional setting necessary for SLA to occur (Section 2.2). The interactions which occur between multiple students are also believed to be of a higher 'quality' than within individual or whole-class tasks because of the variation in language use amongst speakers which will lead to comprehensible input necessary for SLA (Long, 1990; Long & Porter, 1985), although more empirical data is needed to show this. Through working with other students to practise negotiating meaning by clarifying, questioning, responding to questions, disagreeing, and giving opinions, students are more likely to become communicatively competent than practising giving opinions through monologue-style speeches for instance (Lynch & Anderson, 1992; Rignall & Furneaux, 1997). Also, having two or more students discuss topics is believed to be more beneficial for learning than one-to-one with a teacher. This is because discussions between peers are more representative of authentic communication between speakers of a similar level, compared to discussions which are controlled and supported by a teacher (Johnson, 2001).

Other potential pedagogical benefits exist for the use of groups (involving three or more students) rather than individual, pair or whole class tasks. Discussions within groups are believed to offer a more positive affective climate than with a teacher or in front of a class for example (Long, 1985, 1990; Long & Porter, 1985). They can be a more intimate, private and supportive setting for students, where making mistakes and receiving feedback on errors creates less anxiety amongst students. Also, group discussions offer students more individualized speaking time and feedback compared to whole-class tasks (Foster, 1998; Long, 1977). They give students more individual freedom in their choice of speech content and language skills to focus on improving, as well as feedback from students within the same group. In addition, if students practise in groups of three or more, teachers will be able to watch a larger percentage of classroom discussions during classes, as there are fewer discussions taking place at the same time (as opposed to a higher number of pair or individual discussions for the same class). This more frequent Teacher-Based Assessment style of feedback during class is expected to lead to better learning over time than feedback given on only test performance (Davison & Leung, 2009). One more pedagogical benefit of group discussions is that they encourage both collaboration and cooperation in the learning. Collaborative learning involves students working in tandem on shared goals and is believed to enhance social and cognitive skills, as each student is an accountable team member (Ahmadian & Tajabadi, 2017; Johnson, Johnson, & Holubec, 2013; Oxford, 1997). Cooperative learning requires each team member to work independently on their separate role and responsibilities for the group outcome, and encourages support between group members to do so (Donato, 1994; Lantolf, 1993). If a discussion task requires students to reach an agreed outcome together, such as an action plan related to the topic, both collaboration and cooperation would be expected to be present. Furthermore, leadership skills may develop among some students, due to the collaboration and cooperation which group discussions require (Forsyth, 2000, 2016, p. 255; Hackman & Johnson, 2013, p. 203). Such leadership is also viewed as an important part of completing tasks in groups and valuable as a learning experience for students (Ehrman & Dörnyei, 1998, p. 154).

2.3.2 Challenges for learning and teaching with group discussions

Despite the potential benefits of using group discussions, several potential restrictions to learning with them also exist. When students are asked to undertake work in groups, factors related to the relationships and interactions between speakers can have large influences on the learning. Many of these were reported by students as 'barriers to participation' related to group set-up within the first module of this PhD (Appendix A), such as the personal relationships and differences in language levels between group members. A recently published paper by Poupore (2016) helps add to this discussion by explaining the importance of a positive and comfortable feeling amongst a group in order for participation and resultant learning to take place. Poupore explains how Group Work Dynamics (GWD) are important to consider if students are learning in groups, as the actions of each member can influence the actions of others depending on how collaborative or cooperative, and dominant or passive they are (Storch, 2002). For instance, if students decide to dominate the discussion in a non-cooperative way, then the others may not participate or improve their ability to perform in discussions as a result. The larger the group becomes, the more chance there could be of this happening. Also, even though larger groups may give a teacher more time to observe a lower number of groups with their available class time (as discussed above), larger group sizes may cause two problems. Firstly, the students may find it harder to take speaking turns within their groups, as speaking time will be shared between a larger number of group members. Secondly, the teacher will then have less chance of seeing all of the students speak, as there will be fewer students speaking at the same time within class compared to a when there are a larger number of groups. These points may be especially important to consider for students with lower speaking abilities within groups, as it is believed that having a lower level within a discussion may be enough to prevent students from speaking at all (Foster, 1998). If groups within classes contain a variety of speaking abilities, then this may be a problem for the use of group discussions. Thus, the learning undertaken and changes in performance over time by both low and high performers needs consideration. This is addressed in the study in this thesis by categorizing the participants as low and high performers, based on the words they say in discussions (Section 5.4.2), and analyzing differences and similarities between the performances of the two groups across time.

2.4 Task-Based Language Teaching (TBLT) group discussions

2.4.1 Communicative Language Teaching (CLT) and TBLT discussions

Communicative Language Teaching (CLT) is an approach to teaching which is focused on developing the communicative competence of students within a second language (Hymes, 1972, 1974) by helping them develop their ability to use the language both ‘interactionally’ (establishing and maintaining contact with others) and ‘transactionally’ (using language referentially to exchange information) (Brown & Yule, 1983; Ellis, 2003, p. 27). It can be used with what are described as a 'weak' or 'strong' approach (Howatt, 1984). The weak approach is focused on identifying and teaching specific components of communicative competence. This may mean having students learn about the notion and functionality of disagreements before actually trying to disagree with each other using a second language in discussion tasks for example. The strong approach is focused on the belief that language is acquired through communication. For discussions, this would mean that students would learn about disagreements within discussions through simply disagreeing with others when opportunities arise to do so.

Task-Based Language Teaching (TBLT) is an approach which adheres to the belief of the strong approach of CLT discussed above. In particular, it follows the belief that students will improve their communication within a second language if they practise completing what are called 'tasks'. Research in the last four decades has been unable to agree on precisely what a task should involve for students, but does agree on some fundamental points for the learning. In short, tasks should be used to promote meaning-focused language use in situations which students carry out in the same way as they would in the 'real-world' (Ellis, 2003, p.6; Long, 1985, p.89; Nunan, 1989, p. 6; Skehan, 1996). This means that teachers should not interfere with the 'authenticity' of a task by focusing on or planning specific forms of language (such as lexical items or sentence grammar) before or during students undertaking them, but only after the task has been completed. For group discussions, this would mean that students do not focus on any specific language points or discussion skills beforehand, but only after they have used them with others to communicate within a meaningful discussion involving processes, interactions and outcomes similar to a discussion held outside of a language classroom.

2.4.2 TBLT versus Present-Practise-Produce (PPP) for group discussions

Using a TBLT approach of focusing on negotiating through the 'meaning', rather than the 'form', of the language during tasks is promoted as an effective way to nurture SLA (Bygate, Skehan & Swain, 2013; Ellis, 2003; Littlewood, 2004; Long, 2014; Robinson 2007, 2011; Thomas & Reinders, 2015; Willis & Willis, 2008). However, more traditional communicative approaches to language teaching are often preferred by teachers in Asia (Iwashita & Li, 2012). The most common of these is the Present-Practise-Produce (PPP) approach. The important difference with PPP is that students begin practices by being introduced to specific language forms (Present), practise them in specific contexts chosen by the teacher (Practise) and are then asked to use them within a final practice stage designed by the teacher (Produce) (Gower & Walters, 1983). Such practices are believed to help students improve their ability to use specific language forms more than with TBLT, as the language which is practised can be controlled more by the teacher and matched up more closely with the assessment performed later on, compared to the more ‘open choice’ use of language with TBLT. This has been recognized as a useful routine for practising important parts of speech (Swan, 2005), but although the practising of such language points may be expected to improve the ability of students to say them in class, no evidence exists to show that a PPP approach can improve language use during a group discussion.

Thus, TBLT may be a preferred approach to PPP for discussion tasks, because if students use language to undertake communication with much less external direction or restrictions on the language they must use within PPP, they will be able to notice the difference between their own ability and that of grammatical rules being used more (Batstone, 1996, p. 273). This is because the freedom of language use which TBLT gives students makes them more aware of what they can and cannot communicate, rather than just repeating language points already provided to them by PPP, and be able to focus more on improving those gaps in their abilities afterwards. This stage in learning is called Consciousness-Raising (CR) and helps students internalize grammar structures in language by drawing their attention to it (Ellis, 2002, p. 168). It is argued that other more form-focused approaches to learning, such as PPP, cannot raise the awareness of students’ own language gaps which need improving in this way, but only the awareness of specific forms which have been selected by a teacher and practised beforehand (Ellis, 2003, p. 29). For classroom group discussions, the vast number of language points which may be required for students to exchange ideas and possibly reach decisions together may make PPP style practices impractical. It is my belief as a teacher that identifying gaps in ability and improving those weaknesses in language use for many different students within the time available in language courses requires the freedom of TBLT to do so.

Despite all of these potential benefits for learning of using a TBLT approach to group discussions, little evidence exists of the actual improvements in performance which it may result in within classrooms. Thus, the study in this thesis addresses this using an analysis of discussion recordings across a semester to see how performance actually changes to help researchers and teachers better understand the effects which TBLT has.

2.4.3 Challenges for learning and teaching with TBLT group discussions

One of the potential problems with the freedom in language use discussed above in relation to TBLT is that it creates a large amount of possible focuses for students within discussion tasks. Although such freedom in language use may promote noticing, consciousness raising, and resultant SLA (Section 2.4.2), it may also leave students confused about what content, actions or language they should focus on within tasks (Burrows, 2008). It is clearly expressed in the literature that for improved language use to occur with a TBLT approach, students must reach an 'outcome' (Prabhu, 1987, Skehan 1998) or obtain an 'objective' (Bygate et al., 2013, p. 19) for the task. Ellis (2003, p. 8) helps clarify what this means for tasks by describing an outcome as something that students arrive at after completing a task, such as a story or a list of differences. He compares this to the aim, described as the pedagogical purpose of the task, which is eliciting receptive and productive meaning-focused language use. However, without any outside guidance or influence on what language students should be using or how they should interact with the language (which PPP can provide), teachers cannot be certain that students will undertake the processes which TBLT expects to lead to such SLA. For oral tasks, students can decide which performances to focus on or ignore, such as 'sacrificing' accuracy in oral tasks for fluency, (Cuestra, 1995; Foster, 1999), and can even choose to switch to their L1 to complete discussions away from the view of a teacher (Carless, 2007a). In such cases, students may still be able to produce the same discussion output (a list or spoken summary of their agreed discussion outcome in the L2 for example) as those who did not. In addition, studies analyzing which of these focuses will actually result in improvements in oral language use over time are scarce. An important question for group discussion tasks is how important is the focus on the outcome (what students decide upon) and the aim (the processes of language use they go through during the task) for promoting learning? The study in this thesis addresses this question by examining how Japanese university students’ focus on either the outcome (product goals) or aim (process goals) of their group discussions will improve their task performance over time (Section 2.4.3 and RQ2).

Another potential problem for TBLT group discussions is that some students may benefit less than others from this style of learning. Students with a higher proficiency in the L2 have been found to benefit more than those with a lower proficiency from a TBLT approach to oral tasks (Burrows, 2008; Tseng, 2006). This is believed to be due to the complexity of tasks and the freedom of language use required during task time. Therefore, it is important for teachers using group discussions with students of mixed levels to consider them on an individual basis for learning, performance and progress over time, as the study in this thesis does by looking at differences in performance change for low and high participators (Section 5.4.2). With a meaning-focused TBLT approach, where students do not receive specific guidance on or support with their language use prior to or during discussions, lower-level students may struggle to participate in groups with higher-level students. This may lead to motivational issues for those students who may require more guidance and support on an individual-basis to improve over time. A planning stage has been found to help students improve their overall performance in tasks (Guará-Tavares, 2011, 2013, 2016) with improvements made within oral task participation and fluency (as was concluded in the second module, as shown in Appendix B), accuracy (Bygate, 1996, 2001; Bygate & Samuda, 2005; Lynch & McClean, 2000, 2001), and complexity (see Ellis, 2009 and Javad Ahmadian, Tavakoli, & Vahid Dastjerdi, 2015 for recent summaries of related research). By allowing students to plan for discussion tasks, the approach then becomes more of a 'Task-Supported Teaching' approach (Ellis, 2003, p. 28), where students can practise language items already practised before a task (more of a PPP approach). One concern with this is that students may not then interact as much in a meaningful way as the tasks become what Long calls 'synthetic', as opposed to 'analytic' in nature (2014, p. 7). This may lead to a reduction in 'real-world' interactions which occur, without a prior focus on language forms, which is described above as important for the pedagogy of TBLT.

TBLT also faces challenges with finding the required time to explain and train students and teachers, so that is becomes understood, accepted and used in the intended way for learning (Lai & Lin, 2015; Waters, 2009). TBLT has been reported as an inappropriate approach to learning on a world-wide scale, mainly because of the need for more time for tasks than other approaches (such as PPP), and the under-prepared feeling that some teachers report (Ogilvie & Dunn, 2010; Sparks, 2010; Van den Branden, 2006, p. 217). The requirement to 'facilitate' discussion tasks, by giving students control of the process of discussions, rather than guiding them in the language they should be practising, can cause confusion and power-struggles for some teachers (Ellis, 2003, p. 271; Stroud, 2013). PPP has often been chosen as an approach to teaching over TBLT as it offers teachers more control than TBLT, and PPP is believed to have an 'excellent relationship with teacher training and teachers' feelings of professionalism', and 'lends itself very neatly to accountability' (Skehan, 1998, p. 94). TBLT often does not get used in classrooms, as the PPP approach is believed to create clearer language use goals which better match assessment criteria and exam results (Butler, 2011, Thornbury, 1999, p. 64). The study in this thesis uses the addition of GSF with discussions to see if it can help with these potential weaknesses of TBLT, compared to PPP, by providing more clarity and connections between learning and assessment for both students and teachers (Section 4.5).

A final question for a TBLT approach is can students who practise discussions with such freedom of language use then transfer what they learn to new discussions afterwards? In general, skills practised in one task (whether it be related to language learning or not) would be expected to be transferred to another, if the goals, method and approaches are similar (Blume, Ford, Baldwin, & Huang, 2010). Although much discussion of transfer is only theoretical (such as Barnett & Ceci, 2002; Haskell, 2001), some transfer has been shown for low-level speakers of English from one orally interactive language learning task to another (Benson, 2016). Students with lower-speaking abilities within a study group were able to use conversation skills learnt for everyday situations (such as giving directions) in subsequent practices. However, no data exists to show that such transfer can take place across time for group discussion tasks, involving three or more students, especially if the topics are changed each week. The study in this thesis addresses this by analyzing performance changes across time (using different weekly topics), which is crucial for teachers who use such an approach to understand what improvements may or may not be occurring in performance.

2.4.4 Challenges in Japan for TBLT group discussions

As discussed in the introduction, the overall intent of communication courses within Japan is to educate a generation of students who can exchange their thoughts and feelings through using the English language (MEXT, 2009, 2013). The use of a TBLT approach to improving the oral communicative competence of students can be argued as an appropriate approach, due to the expectation that students will ‘notice’ gaps in their language use and improve upon them with a higher understanding of their own ability (Section 2.4.2). However, for Asian students, TBLT faces not only the difficulties mentioned above (Section 2.4.3), but additional challenges related to expectations for learning from students, teachers and institutions (Carless, 2004; Lai, 2015; Littlewood, 2007).

The Japanese students within the first module of this PhD reported many 'barriers' to participating within TBLT-style discussions (Appendix A), including a 'fear of making a mistake' during speech which can cause students to feel anxious and remain silent (Chang, 2011; Tsui, 2001; Williams & Andrade, 2008; Woodrow, 2006). Such problems with participation are not uncommon among students in other Asian countries. Asian students often report feeling underprepared to talk within TBLT tasks and switching to using their first language to avoid making mistakes (Burrows, 2008; Carless, 2007a). Because TBLT is an unfamiliar approach for Asian students, studies have shown that they often prefer to practise the language forms before oral tasks (with a PPP-style approach), so that they will be more accurate in their language use (Lai, Zhao & Wang, 2011), rather than being concerned with learning through interactions, noticing and consciousness-raising, as TBLT is designed to promote. Therefore, an important challenge for the use of TBLT discussions within Japanese classrooms is to make students feel relaxed and confident to undertake discussions through being aware of what is expected in terms of interactions and/or outcome of tasks (which may not always be recognized as the same by teacher and their students), as well as feeling prepared to interact with others in the L2 to do so. This was a benefit I expected to come from students using the GSF in the study and is analyzed for its specific effects on performance and learning (RQ2 and RQ3).

As discussed in the previous section, TBLT is often rejected as a teaching methodology, due to a lack of clear connections between the learning undertaken and the assessment criteria which most students are focused upon. This has also been reported as a problem in Asia, (Deng & Carless, 2009; Lai, 2015), where students, such as Japanese, are often asked to undertake knowledge-based, summative, vocabulary and grammar based practices which teachers feel will prepare them better for their exams. Such a PPP-style approach to learning is often preferred, as it directly addresses grammar points which are tested at the end of courses and teachers report feeling more confident that they can direct the learning of the students to pass such tests (Carless, 2007b). One other problem for TBLT is that teachers in Japan often report not being comfortable or clear about their own role within the learning, and become uncomfortable with the power-shift from a 'teacher', who may control the language being practised, to a 'facilitator', who leaves the contents and outcome of tasks up to the students (Stroud, 2013). Thus, if TBLT discussions are to be used within language courses in Japan, it is important that teachers can see clear connections between 1) the processes and outcomes of discussions, and 2) the assessment criteria for the course (Butler, 2011). This would help them understand how they should (or should not) direct student behavior and language use to nurture improvements in test-related performance over time. I believe that the GSF used in the study can help teachers in this way by providing important feedback across time (Section 4.3) which can be used by teachers to support the learning (see Section 7.2 for recommendations).

2.5 Chapter summary

This chapter explored the potential benefits and challenges of a TBLT group discussion approach to language learning and teaching, and also highlighted the lack of current classroom-based research data within this area. According to the theories and research discussed, TBLT group discussions may be an appropriate approach to the learning of a second language for several reasons. Firstly, the interactions taking place between students may support more SLA than non-interactive tasks (such as monologue speeches) due to the need to contextualize and use the language to react to others. Secondly, learning as a group was discussed as preferred to learning as individuals, in pairs or as a whole-class, because it provides the greatest opportunity for a variety of language use, individualized speaking time and feedback from others, positive and motivating environments, as well as the practising of collaboration, cooperation and leadership skills. Thirdly, a TBLT approach was described as being more appropriate than PPP, as the negotiation of meaning with authentic language use among students would be expected to encourage higher levels of noticing and consciousness-raising and resultant SLA. However, none of these benefits have been proven with research for group discussion tasks. In particular, more data is needed to see whether TBLT classroom discussions can actually result in improved performance across a course. Also, more data is needed on the effects of having students focus on the outcome (final group decisions) versus the aim (language use during the task) of a discussion. This chapter highlighted several contradictions between the beliefs of TBLT (that task outcome is essential) and interactionalist beliefs (that language use is more important than the task outcome). The study addresses this by analyzing the performance progress and learning of two separate groups of students for each of these two focuses. Furthermore, it is important to analyze performance for individual students, as this chapter also discussed how group work can complicate learning with the added potential barriers of limited speaking time between larger numbers of group members, differences in language levels and Group Work Dynamics (GWD) making it difficult for some students to speak. The study in this thesis also addresses these important points by analyzing changes in group discussion performance across a semester for low and high participating Japanese university students with a low-level of English.

CHAPTER 3. DETERMINING ORAL GROUP DISCUSSION PERFORMANCE

3.1 Introduction

This chapter discusses how group discussion performance may be best judged, using a variety of measures, which can then be used to help guide and assess learning (using GSF in the study). It specifically addresses observable oral performances (rather than non-verbal performance, such as listening skills or body language), which were used to decide measures for use within the study in this thesis (Section 5.6.2). The first half of the chapter discusses the use of commonly adopted participation and CAF measures for oral group discussions. The second half explains other important measures of performance which also need consideration within group discussion settings. A discussion of the contradictions between the measures discussed in the first and second half of the chapter is then given. The chapter concludes with a summary of the use of these measures to evaluate the second language ability of different students within a group discussion, as well as the challenges associated with creating a well-balanced picture of performance.

3.2 Participation and CAF measures

Determining student oral performance within language tasks, based on what the teacher can observe, is of high importance, but also very difficult (Bonk & Ockey, 2003; Fulcher, 2003; Fulcher & Davidson, 2007, p. 24). In addition to student participation measurements, the most common three constructs of oral linguistic performance analyzed are Complexity, Accuracy and Fluency (collectively known as CAF). Many choices exist for measures to represent these constructs, which creates inconsistencies in both the approaches to measurement and findings for research (Housen & Kuiken, 2009). For group discussion tasks, it is important to select appropriate measures to represent performance which connect to past research (which have been mainly focused on monologue oral tasks), as well as any additional conditions which a group discussion set-up may require. These are now explained.

3.2.1 Participation

The number of words spoken is a quick and easy way to assess the level of participation which a student is showing in a discussion which can be counted using recordings and transcripts. The number of turns taken is a useful second measure to complement words spoken, as it reveals more about the number of times a student decided to participate in a group discussion by speaking. These two measures were discussed and selected as appropriate for use with low-level Japanese students in Module Two of this PhD (see Appendix B) and will not be discussed in any great detail here. However, they reveal nothing about the CAF or content of those turns, and so further measures are required to better understand performance.

3.2.2 Fluency

Oral fluency is said to represent the 'ease' and 'smoothness' of speech (Guillot, 1999, p. 14; Koponen & Riggenbach, 2000, p. 8; Lennon, 1990) and is commonly split up into dimensions of speed, breakdown and repair within applied linguistic research (Skehan, 2009; Tavakoli & Skehan, 2005). Firstly, speed is usually measured using speech rates (syllables spoken per minute) and shows how quickly a student is able to articulate their speech when they have a speaking turn in a discussion. Measures used for this are usually either Speech Rate A (syllables spoken per minute, known as SRA), or Speech Rate B (SRB) after removing the repetitions and reformulations to provide perhaps a truer picture of fluency (see Sangarun, 2005; Tavakoli & Skehan, 2005; Yuan & Ellis, 2003). Secondly, even if a student can deliver speech at a fast rate, breakdowns which occur in speech during speaking turns need to be looked at as indicators or fluency. Pauses (more than one second in length) and L1/L2 fillers (such as 'um') are such breakdowns, as they show problems with articulation when explaining ideas or responding to others (see Mehnert, 1998; Skehan & Foster, 2005). Finally, repairs in speech, such as repetitions and reformulations, have generally been perceived negatively in research, as signs of a student struggling to deliver speech (see Bygate, 1996; Elder & Iwashita, 2005; Foster & Skehan, 1996; Gilabert, 2007; Kawauchi, 2005; Skehan & Foster, 2005; Tavokoli & Skehan, 2005). As these measures of fluency were already discussed and determined as appropriate for low-level Japanese student discussions within Module Two of this PhD (see Appendix B), they will not be elaborated on anymore here.

3.2.3 Accuracy

Accuracy within oral language use refers to how much a speaker deviates from the normal usage of that language (Wolfe-Quintero, Inagaki & Kim, 1998, p. 62). This is considered important for second language tasks, as it shows how 'accurate' the language of a speaker is when compared to that of a person performing the same task in their first language. This type of analysis seems straight forward, but one large challenge for such assessment is the definition of an 'error' for discussions. Within second language use, linguistic errors in speech, such as grammatical ones, are perhaps simple to identify within an analysis of a discussion transcript. Linguistic accuracy is often quantified using measures such as errors per 100 words spoken (see Mehnert, 1998) or Percentages of Error-Free Clauses (PEFC) within sentences (see Foster & Skehan, 1996). However, if group members use the target language in a way which may deviate from use by first language speakers, but is understood as normal usage by the group, it is up to the teacher to decide if the error is in fact an error (Housen, Kuiken & Vedder, 2012, p. 4; Pallotti, 2009). For instance, the use of "don't mind" in Japan is intended to mean "please don't worry about that" in English. Although this would be perceived as a type of error in spoken English (semantic or pragmatic perhaps), some teachers may decide to not count it as an error, as Japanese students would be expected to understand its meaning.

[...]