Abstract
Although pair and small group activities are commonly used in second language (L2) classrooms, there are very few studies which can inform teachers about whether it is best to have students work in pairs or in small groups. In this study, conducted in a junior high school in Indonesia with learners of English as a Foreign Language (EFL), we compared the interaction of these students when they worked in pairs and then in small groups. All talk was recorded and transcripts examined for language production and for deliberations about language. In terms of language production, we investigated how much of the target language (L2) the learners produced compared to their first (L1) language and where the L1 was used, what functions it served. In terms of deliberations about language, we used Language Related Episodes (LREs) and investigated the quantity, focus, and resolutions of LREs. The study found that pairs generated more language than small groups and more LREs. However, students working in small groups tended to use less L1 than pairs, and resolved a greater proportion of LREs correctly. We conclude with some tentative pedagogical implications based on our findings.
Introduction
Assigning students to work in small groups, ranging in size from two to four learners, is common practice in today’s communicative second language (L2) classroom. Small group activities offer a number of benefits for students. Small group activities offer students opportunities to practise using the L2 more than in the teacher-led classroom (Long and Porter, 1985; Ohta, 1995; Storch and Aldosari, 2013). Richard (2006) also points out that small group activities offer a low anxiety context within which to use the target L2.
Leading theories of Second Language Acquisition (SLA) and of learning support the use of small group work in the language classroom because they view interaction as playing a key role in language learning. These theories have also informed a growing body of research on peer interaction. For example, from a psycholinguistic theoretical perspective, Long’s (1996) interaction hypothesis and Swain’s (1985) output hypothesis propose that second language acquisition is facilitated by negotiations for meaning (e.g. clarification requests, comprehension checks) which make the linguistic input comprehensible and which encourage learners to modify their output to make it more target like. Research informed by this theoretical perspective has investigated the impact of various variables such as task type, learners’ L2 proficiency, and context on the quantity and quality of negotiations and modified output (e.g. Eckerth, 2009; Foster and Ohta, 2005; Fujii and Mackey 2009; Izumi and Bigelow, 2000; Mackey et al., 2007; Mayo and Pica, 2000; Muranoi 2000; Nakahama et al., 2001). Some research has also provided evidence that when learners engage in negotiations and output modification these processes result in language acquisition (e.g. Ellis and He, 1999).
Another theoretical perspective on language learning which is also supportive of the use of interaction for learning is sociocultural theory. This theory is based on the work of Vygotsky (1978) and given prominence in the field of SLA by the work of scholars such as Lantolf (1994, 2000) and Swain (2000). From this perspective, language learning is perceived as a socially co-constructed process between an expert and a novice (learner). The assistance provided by the expert is internalized by the learner and this enables the learner to eventually function independently. However, for learning to occur, the assistance needs to be graduated and carefully attuned to the learner’s needs (scaffolded) and capabilities. In the classroom, such assistance is generally provided by the teacher, but as a number of researchers have shown, such assistance can also be provided by peers when the learners work in small groups (e.g. Donato, 1994) or in pairs (e.g. Ohta, 1995; Storch, 2002).
From this theoretical perspective, language – whether it is the learners’ shared L1 or the target L2 – is perceived as a psychological tool that enables learners to deliberate and solve cognitively challenging problems. When working with their peers, these deliberations and the pooling of linguistic resources result in language learning. For example, Swain and Lapkin (1998) found that when learners worked in pairs and resolved linguistic problems that they had encountered, the solutions corresponded to their responses on a post-test. Similar results were reported by Donato (1994) about L2 learners working in small groups. Researchers, informed by sociocultural theory, have also investigated the use of L1 in pair work in terms of the frequency of its use and the functions it serves. Studies (e.g. Antón and Dicamilla, 1999; Storch and Aldosari, 2010; Swain and Lapkin, 2000; Villamill and de Guererro, 1996) have found that the L1 is used sparingly when learners work in pairs. The L1 was found to serve three main functions. The L1 enabled learners to better understand task requirements and thus to complete the task successfully. The L1 also facilitated deliberations about the target language, about both grammar and vocabulary. The shared L1 also helped the learners to build social relations, an affective dimension of interaction that has been shown to be important for language learning (see Storch, 2002).
Most of the work on peer interaction from both theoretical perspectives has tended to be conducted with adult L2 learners and with learners assigned to work in pairs rather than small groups. Research on whether pair work is more conducive to language learning than small group work is scarce. The two exceptions are a study by Foster (1998), informed by the interaction hypothesis, and a study by Fernández Dobao (2012), informed by sociocultural theory and Swain’s work on collaborative dialogue (Swain, 2000). These studies have yielded mixed findings and thus very different pedagogical implications.
Foster’s study (1998) compared the amount of language (L2) produced, the quantity of negotiations and of instances of modified output when adult ESL learners worked in pairs and in small groups on oral tasks. Foster found no significant difference in terms of language production between pairs and small groups. Although overall the total number of negotiations for meaning and instances of modified output found in the data was very small, pairs generated more negotiations than small groups. Foster noted that in the case of small groups, some students did not speak at all. The findings suggest that it is better to have students work in pairs than in small groups.
However, Fernández Dobao’s (2012) study reached a different conclusion. Fernández Dobao compared the number and nature of LRE resolutions of adult L2 learners when working in small groups of four, in pairs, and individually on writing tasks. The study found that small groups produced more LREs than pairs and that more of the LREs negotiated in small groups were resolved correctly than was the case for pairs. The researcher concludes that small groups may be more conducive to language learning because they encourage a greater focus on language and because in small groups there are more knowledge sources to draw on, such deliberations are more likely to be resolved correctly.
The present study builds on this limited body of research comparing the language learning benefits of pair and small group work, but also extends this research to a context and a cohort of learners which have been under-investigated. Our study was conducted with adolescent L2 learners of EFL in an Indonesian school context. We set out to investigate the optimal participant structure (pairs versus small groups) for L2 practice and for deliberations about language. We were also interested in whether these young learners used their shared L1 to a greater extent when working in pairs than in small groups, and if so for what purposes. In foreign language contexts, language teachers seem reluctant to implement small group and pair activities in their classrooms because of the concern that their learners will tend to use their shared L1 rather than target language (e.g. see McDonough, 2004).
Thus the three research questions guiding our study were:
How much language (L2) do learners produce in pairs compared to small groups?
How much L1 is used by learners working in pairs compared to small groups and what functions does the L1 serve?
What is the frequency and nature of learners’ deliberations about language (L2) use when working in pairs compared to small groups?
The Study
The study was conducted in an educational institution attached to a large university in East Jakarta, Indonesia. A group of 10th grade students from two parallel classes, aged 16 to 17, were invited to participate. Participation entailed allowing the interaction (in pairs and in small groups) to be audio recorded. Data collection took place in a regular class and participation in the study was voluntary. A 30 minute time slot was allocated during two 90 minute classes for these tasks but students generally completed the tasks within 10 minutes. The students’ English proficiency was assessed by their teacher to be in the low to intermediate range.
Due to erratic attendance, the data of seven pairs (14 students) and seven small groups (21 students) were used in the study. The small groups consisted of three members each and, where possible, included two students who previously worked in pairs. Thus of the 21 students who participated in the group work, 13 (three males and 10 females) participated in pair work as well. The pair and group arrangement is presented in Table 1 (all names used are pseudonyms). The names of students who participated in both pair and group work are shaded. Their data were analysed in more detail.
Participant Arrangement.
The tasks used for the pair and small group activity were oral communicative tasks. They were developed in consultation with the classroom teacher and took into consideration the English language curriculum and course book. Both tasks required learners to produce descriptions. The task used in the pair work activity required the learners to describe a picture that depicted a certain type of occupation (e.g. soccer player). Each pair was given a different occupation card. The students were given instructions guiding them to describe what the person does, their skills, and the setting for this kind of occupation. The task used in the small group activity was a jigsaw task, where each learner received three pictures showing body parts (e.g. head, legs). The students had to describe their cards to each other in order to complete a composite picture. All students’ talk was audio recorded and observation notes were taken of the students’ interaction.
Data Analysis
The audio recordings were transcribed verbatim. Words in L1 were written in italics and translated into English as closely as possible (translation is shown in square brackets […]). Any non-verbal communication was identified in brackets (…). Instances of mispronunciation were transcribed verbatim and shown within slants (/…/).
The transcribed data were analysed for the following: language (L2 and L1) production, functions of L1 use, and deliberations about language.
Language Production (L2 and L1)
The first stage of analysis was language production. We first identified all turns, and then counted manually the number of total words per turn, the number of L2 and L1 words per turn. A turn is defined as a string of speech by a single interlocutor. A turn consisting of a backchannel (e.g. yeah) was not considered as a turn unless it was uttered in response to interlocutors’ questions (Nakahama, et al., 2001). False starts, fillers (e.g. um…), and incomplete words (e.g. sal…) were omitted from the count. Any overlapping talk was counted as a turn for each of the learners. For example, when Aufar and Hanifah (Pair 5) both said simultaneously ‘Today we are going to describe er…’ this was counted as a turn for each participant composed of six words in the L2 and none in the L1. The filler ‘er’ was not counted as a word.
L1 Functions
The second level of analysis was for the functions served by the L1 when used. To code for L1 functions, pair and group talk data were categorized into episodes. An episode is defined as a segment in which L1 was used in a single turn or several turns (Storch and Wigglesworth, 2003). Each episode was then coded for its function. Two L1 functions were identified. These were based on categories established in previous research on L1 use (Storch and Wigglesworth, 2003; Swain and Lapkin, 2000) as well the transcripts of our learners’ talk:
Task management: discussion on how to carry out the task, including clarifying task instructions and the pictures presented in the cards.
Vocabulary: discussion on lexical choice and meaning of words.
The following examples illustrate these functions of L1. Example 1, taken from the pair talk of Aufar (AU) and Hanifah (HN), is where the learners used the L1 to clarify the task requirements and suggest a rehearsal.
71 AU Kita presentasi ini di depan kelas. [we will be presenting this in front of the class]. 72 HN Kita latihan! [Let’s practise]. We are going to describe about err apa tadi? [what was that?]
Example 2, taken from Group 7, a group composed of Astari (AT), Diva (DV) and Fitria (FI), illustrates the use of L1 for vocabulary. Diva described the body part she had, but did not know how to say nail polish in English so she used the Indonesian word ‘kuteks’ (turn 46) and sought help also using her L1 from her friends in the group.
46 DV I think I have the body of what you need. The body I have is slim, and a woman body, and…and the finger… the finger wears…kuteks [nail polish]. Kuteks apa? [what is nail polish?] 47 FI Nail polish? 48 DV I think nail polish, and…and the body wears a dress with the pattern of the dress is flower. Is it the body what you mean?
Deliberations about Language (L2)
Deliberations about language, whether in L1 or L2, were coded using Language Related Episodes (LREs) as units of analysis. An LRE is defined as an episode when learners talk about their language production by questioning it, self-correcting or correcting others (Kowal and Swain, 1994; Swain and Lapkin, 1995, 1998). These episodes are seen as representing opportunities for language learning (Swain, 1998, 2000). LREs were classified according to whether they focused on lexis, grammar, or pronunciation.
Lexis-based LREs were instances when learners negotiated word choice and meaning. Example 2 above was coded not only as an L1 episode, but also as an LRE where the learners deliberated about vocabulary. Example 3 is a further example of a lexis-based LRE. It comes from the data of Bagas (BG) and Diyah (DY). In this episode Diyah suggests using the phrase ‘work out’ instead of the word ‘practice’ (turn 50). Bagas agrees and Diyah provides the justification for why the phrase should be used.
49 BG They work in football fields, they play in a football field, they practise in a football field. They may go to the gym and practise their muscle so they … 50 DY They work out at the gym 51 BG Yes 52 DY They work out at the gym. They not practise their muscle. That means.
Grammar-based LREs focused on a wide range of grammar points, such as word forms, tenses, word order and choice of articles. Example 4 is of a grammar-based LRE, where one learner corrects the other for word forms. It comes from the pair talk of Pair 6 composed of Fransisca (FR) and Hidaya (HI).
11 FR emm…gimana ya? [emm…how is it?] Emm…I think he needs a great muscle because er…they have to push their shoulder to get the ball. And they’re…they’re… 12 HI They’re to be stronger. 13 FR Strong…
On some occasions, mispronounced words led to some misunderstanding. In this case, learners’ generated pronunciation-based LREs. Example 5 is of a pronunciation based LRE where the learners clarify the pronunciation of the word ‘beard’. It is taken from the talk of Group 1 composed of Dita (DT), Syafira (SY), and Diyah (DY).
11 SY I think I have a body of that man err…for the man. Err…it’s…err…he have…he have a… apa [what]? (makes a gesture around her chin) 12 DT Beard /bɪəd/ 13 SY Bird? /bə:d/ 14 DT Beard /bɪəd/ 15 SY Beard /bɪəd/ And …and small eyes. Err…it is…
LREs were also coded for whether they were resolved correctly, incorrectly, or left unresolved. Examples 3–5 were coded as resolved correctly. Example 6 is of an incorrectly resolved LRE, taken from Group 6 composed of Aliya (AY), Hihayah (HI), and Fransisca (FR).
11 AY Yes…I’d love to get it. Thank you. FR, I have a body of a girl. She’s using a stripe…rok? [skirt] 12 HI Skirt … skirt… 13 AY Stripe… 14 FR Cloth! 15 AY T-shirt…T-shirt. I think this girl is thin. Do you have the leg part of this girl?
In this example, the members of the group negotiated the English equivalent of an Indonesian word (rok). Hidayah (HI) suggested that the word in English is ‘skirt’ and Fransisca (FR) that it is ‘cloth’. Although Hidayah suggested the correct word, Aliya (AY) was not satisfied with either suggestions, and settled on the word T-shirt instead, which is in fact incorrect.
There were also a few instances where LREs were left unresolved. Example 7 illustrates such an LRE, coming from Pair 6 composed of Fransisca (FR) and Hidayah (HI).
16 HI They run everywhere. 17 FR Lincah…lincah Bahasa Inggrisnya apa? [agile…what is it in English?] 18 HI Lincah apa lincah? [what is it?] (looking up her dictionary) 19 FR Yang ini gue gak bisa. [ I don’t know this] Sama aja Inggris – Indonesia, sama punya gue! [It’s the same with mine English-Indonesian dictionary!] 20 HI Kan ada Bahasa Indonesianya juga [There is Bahasa Indonesian too, isn’t it?] 21 FR Nggak ada tau! [I don’t think there is!] 22 HI Nggak ada [ I don’t find it] (gives up) 23 FR Ya udah gitu aja. Pokoknya nanti lincah aja. [yeah.that’s it. It’s just agile]
The LRE in Example 7 begins with Hidayah suggesting that a football player usually runs everywhere. Fransisca offers an adjective (turn 17) to describe this behaviour in L1 ‘lincah’. Both searched for this word in English unsuccessfully. Hidayah then looked up this word in her dictionary (turn 18), but because the dictionary was an English-Indonesian version, it was not helpful. Eventually, the learners gave up on the word (turn 24), and continued their description to complete the task.
Inter-rater Reliability Check
The entire data set was coded by another rater to check for inter-rater reliability. The second rater was instructed on the coding categories. Inter-rater reliability (calculated as proportion of agreements of all coding categories) was very high: 92% for L1 functions, 98% for LRE identification, and 100% for LRE resolutions.
Findings
The findings are presented in the same order as the stages of analysis for pairs and for small groups.
Learners’ Language Production
Table 2 summarizes the amount of language the 14 learners produced when working in pairs and of the 13 students who then worked in groups (Bagas was absent in the second session). Language use was quantified in terms of turns and total words. The table also shows how many L2 words each of the learners used and what proportion L2 formed of total word production (rounded off). We also show the average per participant, and the range for turns, L2 use, and proportion of L2 of total talk. From these figures we can also infer the amount of L1 used.
Language Production for Students in Pairs and Small Groups.
As the table shows, the number of turns per student was smaller when the students worked in small groups. When working in pairs, most of the students (with the exception of Astari and Diva) generated well over 10 turns per student, averaging 23.5 turns. When working in small groups, most learners (again with the exception of Astari and Diva) produced a smaller number of turns. The total number of words produced in pairs was also greater than for small groups (174 versus 139). The large range in turns and in the total number of words produced reflects the large variations between pairs and between groups. It is interesting to note that in the case of some learners (e.g. Hidayah) more turns and more words were produced when the learner worked in small groups than in pairs.
However, despite the greater number of total turns and words, the number of L2 words produced when the learners worked in pairs was similar to that produced when working in small groups (147 versus 137). This is because there was far less use of L1 in group work compared to pair work. Whereas when they worked in small groups, the students spoke almost entirely in the L2 (an average of 98%, with a range of 92 – 100%), when they worked in pairs, 83% of the total words produced were in L2, with a range of 59-100%. It is interesting to note that students who used L2 in 60 – 70% of total talk when they worked in pairs (e.g. Aufar, Fransisca) used L2 in over 90% of their talk when working in small groups.
L1 Functions
As shown by Table 2, the learners generally used very little L1 in small groups but that there was some L1 use when they worked in pairs. Further analysis found that the functions served by the L1 were slightly different in pairs than in small groups. The functions of the L1 in pairs and small groups are presented in Table 3. As the table shows, L1 was used by the pairs about equally for task management (55% of L1 episodes) and for vocabulary (45% of L1 episodes). In the case of small groups, L1 was used mainly for vocabulary (75% of L1 episodes).
L1 Functions in Pairs vs. Small Groups.
Deliberations about Language
Table 4 summarizes the findings about the LREs found in the pair and small group work data. The table shows the total number of LREs for each pair and small group and whether they focused on lexis (Vocab), pronunciation (Pron), or grammar (Gram).
The Number and Distribution of LREs in Pairs vs. Small Groups.
The analysis suggests that interaction in pairs promote a greater focus on language deliberations than interaction in small groups. As Table 4 shows, more LREs were generated by pairs (33) than by small groups (21). Furthermore, it is interesting to note that whereas most of the pairs showed some focus on language, in the case of group work, most of the small groups generated no or very few LREs (between one to two LREs). The bulk of the LREs in the group work data were generated by Group 1 (12 of the 21 LREs). The pairs focused mainly on vocabulary (76% of LREs); the small groups focused on vocabulary and on grammar. Table 5 shows the nature of LRE resolution in pairs and in small groups.
The Nature of LRE Resolution in Pairs vs. Small Groups.
From Table 5, we can see that a large proportion of LREs were resolved correctly by both pairs and by small groups. However, the proportion of correctly resolved LREs was smaller in pairs than in groups. In the case of pairs, a sizeable proportion of LREs were not resolved (21%) whereas in the case of small groups, most of the LREs were correctly resolved (90%) and only a small number of LREs were incorrectly resolved (5%) or unresolved (5%).
Discussion
In this section we summarize and discuss our findings in response to the research questions which guided this study.
Language Production (L2) in Pairs Compared to Small Groups
Our study found that when learners worked in pairs, they had more opportunities to engage in the tasks, evident in the greater number of turns and total words produced in pair work. The greater number of turns per student in pair work than small group work is not surprising because in pair work turns are shared by two people. In small group work there are more people to share turns. These findings are similar to Foster’s (1998) findings. Foster also found that pairs generated more language than small groups and that students working in pairs shared the turns evenly. However, Foster also found that many students in the small groups were quiet, not contributing to the task. In our study, both pairs and small groups shared the turns evenly. We observed variations between pairs and small groups (signaled by the large range) rather than within pairs and groups. In our study, all pairs and small groups seemed to collaborate on the tasks, with roughly a similar number of turns per student in each small group.
This difference in findings may be explained by the size of our small groups and the nature of our tasks. In Foster’s study, students were in bigger groups of four to five students in each group and the task was a discussion. In larger groups, especially when using discussion tasks where individual contributions are not obligatory for task completion, it is more likely perhaps that only the active and more confident students will talk. In our study, the groups contained only three students and the jigsaw tasks required information exchange; that is, each member of the group was obliged to speak in order to complete the task. Such smaller groups and jigsaw type tasks seem to encourage student participation.
In terms of L2 production, learners’ contribution in pairs and in small groups was similar. This is because the learners tended to use more L2 as a proportion of their total talk in small group work than in the case of pairs. Students who used L2 extensively when working in pairs (e.g. Pairs 3, 4 and 7) continued to do so when working in small groups. In the case of students such as Aufur and Hannifah (Pair 5), it was interesting to observe that when they worked as a pair, just over half of their talk was in L2 (59% and 60% respectively). However, when they worked in small groups, the proportion of L2 use was closer to 100%. The same upward trend in L2 use can be observed in the talk of members of Pair 1 (Dita and Syifa) and Pair 6 (Hidayah and Fransisca) when they worked in small groups (Group 1 and Group 6 respectively). The greater use of L2 in small group work may be attributable to groups per se, where perhaps learners exert a greater peer group pressure to use the L2. However, it may also be due to the sequencing of the tasks in our study. The small groups work took place in a session after the pair work session, when learners were more familiar and at ease perhaps of using the L2 in task talk. This also explains, to some extent, the functions served by the L1 in pair and small group work.
L1 Use and Functions in Pairs Compared to Small Groups
Our study found that although overall students used L2 more than L1 in pair and small group work, pairs tended to use the L1 to a greater extent than the small groups. In pair work, the L1 was used for two functions: task management and for dealing with unfamiliar vocabulary. In small group work, there was minimal use of L1, and when it was used it was mainly to deal with vocabulary. One reason to explain the greater use of L1 in pair work was related to the clarity of the pictures and the occupation the pictures described in the task used in pair work. For example, in the case of Pair 5, (Aurfar and Hanifah) the picture described an occupation that was unusual (a deodorant tester). It took the two learners many turns (over 30) and the use of the L1 to work out the occupation that the task card described. Thus the shared L1 enabled the learners to gain a shared understanding of the picture and complete the task successfully. In the small group work, the students were already accustomed to working with peers on a description task, and although it was slightly different, the learners no longer needed to use their shared L1 for task management. In the case of the small group work, L1 was only used to clarify the meaning of unfamiliar vocabulary.
Our study, however, did not find that the L1 was used for the purpose of discussing grammatical conventions, as was found by a number of other studies (e.g. Storch and Aldosari, 2010; Storch and Wigglesworth, 2003; Swain and Lapkin, 2000). This could be due to the fact that the other studies have tended to use writing tasks which require more focus on accuracy than oral tasks. Our study used oral tasks. Studies focusing on oral communication tasks (e.g. Foster, 1998; Nakahama et al., 2001) also reported very little focus on grammar.
Our study, as is the case with a number of studies conducted with adult L2 learners (Storch and Aldosari, 2010; Storch and Wigglesworth, 2003) and adolescent learners in immersion contexts (Swain and Lapkin, 2000), shows that learners use the L1 judiciously in task based interactions and that the L1, when used, serves important functions. These findings should serve to allay the concern of L2 teachers about the use of the L1 in pair and group work in L2 classes.
Deliberations about Language (L2)
Our study found that learners did deliberate about language in these oral tasks, but focused on vocabulary more than on grammar and pronunciation. Pairs generated more LREs (33) than small groups (21). Lexis-based LREs predominated over other LREs in both pairs and small groups. Other studies have also reported that lexis-based LREs are more frequent than grammar-based LREs particularly in meaning-focused tasks (Eckerth, 2009; Foster, 1998; Pica, 1994; Swain and Lapkin, 1998; Williams, 2001). Clarification of vocabulary enabled the learners to complete the tasks.
A closer examination of our data showed that grammar-based LREs tended to be generated by the same student, Syafira. Syafira was a participant in Pair 1 and Group 1 who generated most of the grammar-based LREs. These LREs took the form of Syafira correcting her friends’ grammatical errors. Syafira rarely corrected her friends’ vocabulary or pronunciation. Pronunciation-based LREs were rare in the data of both pairs and small groups, in line with other studies who also report few such LREs (e.g. Eckerth, 2009; Foster, 1998; Swain and Lapkin, 1998).
LREs are seen as a potential site for language learning (Swain and Lapkin, 1998). Incorrectly resolved LREs may lead learners to learn incorrect conventions or word meanings (see Storch, 2002) and unresolved LREs represent missed opportunities for learning. Thus we need to note not only the number of LREs generated and what they focus on, but also how they are resolved. Our study found that although the majority of LREs were resolved correctly by pairs and by small groups, groups were more successful in resolving their LREs correctly. Groups resolved 90% of their LREs correctly; pairs resolved 79% of their LREs correctly, leaving 21% unresolved. Our findings are similar to those reported by Fernández Dobao (2012) and support her proposed explanation. In small groups, learners have more resources to draw on when encountering a language problem and this enables them to resolve, and resolve correctly, a higher proportion of LREs than is the case in pair work.
Conclusion
Overall our findings suggest that pair work provides more opportunities for learners to engage in a task (more turns and words) and encourages more deliberations about language. However, in small groups of three, learners are less likely to resort to their L1 when encountering a language problem, and to resolve their deliberations correctly. On the basis of these findings, small groups may be more beneficial for language learning than pairs.
However, we admit that our findings and recommendations based on these findings are tentative. One limitation of our study is its size. Our study was small scale and thus relied on descriptive statistics. Another limitation is that we did not use a counter-balanced design, and thus sequencing of small group work after pair work may have affected our results. The third limitation was the slightly different tasks used with the pairs and the small groups, and that some cards, particularly in the pair condition, described odd occupations. However, we wanted to conduct our study in a regular L2 class, and thus our choice of tasks was influenced to some extent by the type of tasks used in the students’ course book.
Future studies need to be larger scale and employ a counter-balanced research design as well as a more careful choice of tasks. Nevertheless, we feel that such studies need to be classroom based and conducted with younger learners in a range of contexts rather than just with adult learners in ESL contexts. Such a broader body of research will be able to better inform language teachers about how to best group students for oral interaction tasks.
Footnotes
Funding
This work was supported by the Directorate General of Higher Education of Indonesia (Grant number 69508/A4.2/KP/2011).