Other-Initiated Repair Strategies in Solving Understanding Problems in EFL Learners Conversations

The aims of this study were to investigate the employment of Other-Initiated Repair Strategies (OIR Strategies) in solving understanding problem in EFL learners' conversation and to examine the kinds of trouble sources that prompt the employment of OIR Strategies. The participants were nine EFL learners participating in a speaking class in small university in Indonesia. To elicit the learners' conversations two communicative tasks, Spot the Difference and Desert Island were used. The learners' conversations during task performance were video-recorded, then, transcribed using the conventions proposed by Markee (2000) and analyzed qualitatively using Conversation Analysis method. The results showed that EFL learners managed to employ eight types of OIR Strategies comprising of 62 instances in total. The strategies are unspecified repair, interrogative repair, partial repeat plus a question word repair, partial repeat repair, understanding check repair, request for repetition, request for definition, and correction repair. Three different types of trouble sources triggered the use of OIR Strategies, namely linguistic-related problem, interactional-related problem, and meaning-related problem. The study demonstrated that by employing different OIR Strategies, even EFL learners with limited proficiency in English managed to take initiatives to overcome the understanding problem in conversations as part of their learning process. Therefore, the EFL teachers need to consider incorporating the teaching of OIR Strategies as part of their speaking class to improve the students' fluency.

One of the approaches used in studying different kinds of strategies employed by EFL learners in their attempt to overcome understanding problem in their conversation is Conversation Analysis. Interest in the application of Conversation Analysis in the fields of language learning and language teaching has grown considerably over the last two decades. Hutchby & Wooffitt [02] explicate that Conversation Analysis is the analysis of talk, specifically a systematic analysis of talk produced in everyday situation on human interaction (talk-in-interaction). The principal aim is to discover how the participants understand and respond to each other, with the central focus being on how sequences of actions are generated. Its relevance in EFL interaction is also asserted by He & Young [3] saying that CA provides a powerful tool for understanding and analyzing problem in communication with its description of repair organization. When trouble in conversation occurs, it is noticed and then corrected, either by the party whose turn contains the source of trouble or by another party. In addition, CA can be adapted in SLA studies as a tool to analyze the conversational data that allow the researchers to understand the sequence of events in the conversation among learners and their relevance to the construction of turn-taking and repair organization [4] and could benefit the teachers and the methodologists in revealing the format and interactional design of language learning tasks in the classroom, as well the teachers' interactional practices [5].
There are four basic interaction practices to combine to form conversation as a system (Wong & Waring [6], namely turn-taking practices (ways of constructing a turn and allocating a turn), sequencing practices (ways of initiating and responding to talk while performing actions such as requesting, inviting, story-telling, or topic initiation), overall structuring practices (ways of organizing a conversation as a whole as in openings and closings), and repair practices (ways of addressing problems in speaking, hearing, or understanding of the talk). This study focuses on conversational repair which is defined as action taken by conversants to treat trouble sources in the accomplishment of talk. Trouble is anything which the participants judge is impeding their communication and a repairable item is one which constitutes trouble for the participants [7]. Furthermore, Liddicoat [8] adds that repair refers to the processes available to speakers through which they can deal with the problems which arise in talk. Repair is relevant to all levels of talk from the turn-taking system to sequence organization and preference. Repair is a mechanism of conversation: a set of practices designed for dealing with the sorts of difficulties which emerge in talk. Conversational repair can take different forms and only limited to the modification of linguistic errors, but also includes confirmation checks, comprehension checks, or requests for clarification or repetition, and deals with the problems in interactants' understanding, speaking, or hearing [7].
The basic sequence of repair practice is trouble source in the first position, followed by repair initiation in the second position, and repair outcome in the third position [9]. Trouble source is a word, phrase, or utterance treated as problematic by the participants. Repair initiation refers to the practice or strategies of signaling or targeting a trouble-source. Repair outcome refers to the solution to the troublesource or abandonment of the problem. I will use this technique in analyzing the data on other-initiated repair strategies in the EFL learners' conversation.
Studies on Conversation Analysis and repair have been done in several different settings. Some researchers focus on studying the conversation between nonnative speakers of different language backgrounds (For examples: Wang & Wu [10]; Khodadady & Alifathabady [11]; Fotovatnia & Dorri, [12]; Liebscher & Dailey-O Cain. [13]: and Buckwalter, [14]). Fotovatnia & Dorri [12] investigated the repair strategies used by EFL learners whose native language is Persian. The study found that the participants used more strategies in single-gender class than mixed-gender class and there is no relationship between the gender of students and the use of repair strategies. It was also found that there were no significant differences observed when the students and the teacher were of the same or different gender. In this study, Fotovatnia & Dorri [12] incorporated nine different types of other-initiated repair strategies. They are unspecified repair, interrogative repair, partial repeat plus a question word repair, partial repeat repair, understanding check repair, request for repetition, request for definition, nonverbal resources repair, and correction repair. The EFL learners with Persian language background were also the participants in Khodadady & Alifathabady [11] study. The study focused on the difference in the frequency of other-repair used by the high group and low group. It was found that the high group used more repair strategies in their conversation than the low group. The other-initiated repair strategies included in the study were partial repeat, understanding check, and correction.
In the studies by Wang & Wu [10] and Buckwalter [14], the participants are nonnative speakers of the target language. Wang & Wu's study used nonnative speakers of Chinese coming from different nationalities and language background who learn Chinese as a foreign language, while Buckwalter used English speaker learners who learn Spanish as a second language in America. Both studies focused on the occurrence of both self-repair and other-repair (self and other) and the interactional devices to initiate repair actions. Results show that both self-repair and other-repair are ubiquitous in the interaction of the leaners regardless of their levels of proficiency. It was also demonstrated that the two types of repair are dominant in  [13]. The participants of their study were native speakers of English who learned German as a Foreign Language in Alberta, Canada. The study investigated the ways in which students and teachers in a content-based German as a foreign language class used repair in order to negotiate meaning and form in their classroom. The main finding of the study indicate that repair initiation in the classroom is clearly different from repair initiation in mundane conversation, both in terms of types of repair initiation and in ways how those types are used.
Another line of study analyzed the conversation between Native Speaker and Nonnatives Speakers (e.g. Tudini, [15]; Lilja, [16]; Yasui, [17]). Lilja [16] analyzed other-initiated repair sequences in the conversation between first and second language speakers of Finish. Specifically, the study focused on the employment of partial repetition as repair strategies. It is found that partial repetition is used frequently as strategy to overcome specific language-related understanding problem. Tudini [15] focused her study on the use of codeswitching as repair strategy in the conversation between Italian speaker and English speakers, either as L1 or as L2. A study on the conversation between Japanese native speaker and nonnative speakers by Yasui [17] investigated the conversational repair as a corrective activity of troubles in conversation in an informal language learning setting. The analysis reveals that the non-native speakers orient toward their language proficiency or deficiency through their repair processes. While the advanced learner's language proficiency is displayed through his preference for self-repair, the beginning learner exhibits his limitation in language competence through his preference for other-repair.
A group of repair studies investigate conversations between nonnative speakers in mundane noninstitutional settings. The studies by Galina B. Bolden (Bolden,[18]; Bolden [19]) examined a way in which the presence of more than two interlocutors matters for the organization of repair. The participants of the study were native speakers of Russian and nonnative speakers of Russian whose L1 was English. This case study used corpora of previously recorded conversation as data source. The analysis shows that, "other"selection in other-initiation of repair is a systematically deployed practice [18]. Bolden's study in [19] shows that the interactional construction of language (in)competence and the organization of repair are two communicative practice to solve understanding problems.
Two articles by Dingamanse & Enfield [20] and Dingemanse, Roberts, Baranova, et.al., [21] reported the same topics, namely about the typology of other-initiated repairsin several languages. Dingamanse & Enfield [20] reports on studies on other-initiated repair in 10 different languages.The data consist of corpora of naturally occurring conversations, collected in field sites around the world.Methodologically, they combine qualitative analysis with a comparative-typological perspective and formulate principles for the cross-linguistic comparison of conversational structures. It is presented that formats for repair initiation form paradigm-like systems that are ultimately language-specific, and that comparison is best done at the level of the constitutive properties of these formats. Dingemanse, et al. [21] conducted a systematic comparison of conversation in 12 languages of 8 language families to find out the universal system for the real-time resolution of breakdowns in communication. Data for their study were taken from a corpora of conversations recorded from informal interaction in 12 languages in a spontaneous, naturally occurring conversations between families and friends. They found that a system of 'other-initiated repair', where the recipient of an unclear message can signal trouble and the sender can repair the original message, is frequently used in all the languages being studied.
Studies on other-initiated repair in conversation that have been done in different settings reveal an insight that the employment of OIR Strategies in spoken discourse as a valid and normative communicative practice. It is to say that the interactional phenomenon of doing repair is normal practice that takes place in mundane conversation, institutional setting, education setting, even in language learning setting. The studies also show that in order to achieve mutual understanding in conversation, interactants need to employ certain strategies so that communication breakdown could be avoided. The studies presented here used Conversation Analysis approach, both as theoretical foundation and as methodology. Since Conversation Analysis is one type of qualitative approach, most of the studies employed qualitative research design.
One of particular interest of the studies is repair acts performed in a setting where the interlocutors are nonnative speakers of the language used in conversation.Specific case is for the ELT setting, where the participants are non-native speakers of English.Given the fact that non-native speakers are limited proficient users of English, other-initiated repair is usually rich in occurrence in the EFL leaners' conversations. Studying conversational other-initiated repairs of EFL learners will shed light of the strategies employed by the EFL learners in overcoming understanding problems, as well as how the learning process is evidenced in the learners' conversations.
The current study investigated how EFL nonnative-speaker interlocutors employ OIR Strategies to resolve communicative difficulties and misunderstandings in conversation. More specifically, the study focuses on the other-initiated repairs used by nonnative speakers of English, situated in classroom-setting when they perform communicative language tasks. The studies specifically answer the following research questions: a. What are the types of Other-Initiated Repair Strategies employed by the EFL learners in resolving understanding problems in their conversation? b. What are the types of trouble sources that trigger the use of Other-Initiated Repair Strategies in the EFL learners' conversations?
This study is expected to shed light on how language learners with limited proficiency of English use different repair strategies in resolving communication problem. Hopefully, the findings will inform teachers to incorporate repair initiation strategies in their speaking classes in order to improve EFL learners' proficiency.

METHOD 2.1. Participants
The participants involved in this study were 9 EFL university students who join the "English for Everyday Conversation" speaking class at an English Education Study Program in small university in Indonesia.These students were in the early stage of their study at the second semester of the English Education Study Program.Except for Intensive Course Subject, all the other subjects that they took in semester one were taught in Indonesian.The Intensive Course focuses on consolidating the students' mastery of English grammar, vocabulary, and pronunciation.It means that the students almost have no experience in speaking class.These participants were conveniently chosen for the present study and their English proficiency was rated as low by the lecturers who taught them in the previous semester.Thus, the rate of their language proficiency was just based on the lecturer's judgment in the Intensive Course. There were four female (f) and five male (m) participants of the study, and they were pseudo-named as Haru (m), Fran (m), , Anch (f), Alga (m), and Efan (m).

Data Collection
The main source of data for the present study was obtained from the participants' dyad conversations when they participated in task-based communicative activity. The use of language tasks was aimed at providing the participants an arena for real communication, during which time they may have the opportunity for language practice, meaning and form negotiation in a real communicative exchanges. In doing the tasks, the students were given ample opportunity to express themselves in talking about the topic of the tasks. The tasks are not made rigid to follow strictly pre-specified procedures.I only provided general instructions on how to carry out the tasks. In fact, I have been trying to make the group work as free as possible, in the sense that my involvement was very limited to only facilitating and managing the group work. Two communicative tasks developed and used for the study were Spot the Differences and Desert Island.In Sport the Differences task [22], the participants worked in pair and each was assigned as Student 1 and Student 2.They were given a set of pictures which are almost the same, but actually small number of details was different. The task of each of the participant was to describe the picture he/she was holding to his/her interlocutor to find the differences between their pictures.They were asked not to show the picture to their partner but solely rely in their oral description to find the differences.
The second task called Desert Island [23] is a kind of discussion and decision-making task.The participants worked in pairs to do the task. The following is the instruction for the Desert Island Task.
"You are stranded on a desert island in the Pacific.All you have is the swim-suit and sandals you are wearing. There is food and water on the island but nothing else.Here is a list of thing you may find useful.Choose the eight most useful items and rank them in order of usefulness". [23] Before doing the task, the student worked individually to be familiar with the task and the items to be discussed and chosen, but not knowing who their interlocutor to work with were.

Data Recording and Transcription
The students' conversations were video-recorded using Sony HDR-CX405 Handycam every time for each pair the tasks. Each pair was recorded separately, and the conversation and recording started after I pushed the record button on the video camera.When the students proceeded with the conversation, I left the conversation area and returned only to push the pause button.In this way, the students may have ample opportunity to explore that tasks and express themselves without being threatened by my presence in their proximity. The video recorder was positioned in such a way that it could capture the speech of every

Data Analysis
The data analysis in the present study focused on the discursive practices of the EFL learners engaging in conversation in order to get a better understanding of how understanding problems were resolved using different OIR Strategies. Using CA approach, data analysis was focused on the instances in the conversations where OIR Strategies occurred by examining sequences of conversational turns. This microanalysis of the learners' turn sequences allowed the researcher to examine how the resolving of understanding problem was accomplished in moment by moment of conversation course.
CA was selected as method of data analysis because of its characteristic that considers talk production and meaning as something accomplished by members of conversation participants through mutual orientation and collaboration to achieve orderly and meaningful communication [2]. In this study, I focused on instances in the conversations that reflect the sequence of other-initiated repair which consist of trouble source in the first position, followed by repair initiation in the second position, and repair outcome in the third position [2]. This was achieved by reading the data transcripts repeatedly in conjunction with the videorecorded data. As a guideline to determine the types of other-initiated repair strategies, the researcher used the list of OIR Strategies found in the study of Fotonania & Dorri (2013). The strategies are unspecified repair, interrogative repair, partial repeat plus a question word repair, partial repeat repair, understanding check repair, request for repetition, request for definition, nonverbal resources repair, and correction repair.
To establish the credibility of the findings, the procedure of peer debriefing or investigator triangulation [24] was employed. This procedure was done by asking a doctoral candidate from Nangyang Technology University in Singapore to verify the findings, especially the identification and classification of other-initiated repair strategies based on their types. This person was conducting a study on membership categorization analysis using Conversation Analysis as both the method and the theoretical basis, which is also employed in the present study. Out of the 62 transcribed conversational excerpts, fifteen percent were emailed for investigator triangulation. The result of triangulation shows that there is total agreement in the identification and classification of OIR types between the researcher and the triangulator.

RESULTS
This section presents the result of data analysis in terms of types of OIR strategies and types of trouble sources that trigger the employment of OIR Strategy in the conversation. From a cyclical review of the data transcript and repeated listening of the recorded conversation, it is revealed that the EFL learners employed eight different OIR strategies in resolving understanding problems in their conversation. The eight strategies occur in all the ten conversations produced by each conversational dyad. The analysis will start from the presentation of the types of OIR Strategies across the ten conversations, followed by the presentation of different kinds of trouble sources being resolved by the OIR strategies.

Types of OIR Strategies in EFL Learners' Conversation
There are eight different OIR Strategies used by the leaners in solving understanding problems in their conversation. The strategies are presented in Table 2 below, together with their frequency and percentage of occurrence in the conversation.
The most frequently employed strategy is interrogative repair with the total number of 18 (29.03%) out of 62 total strategies used. The second in order of frequency is partial repeat strategy, with the total number of 14 (22.58%), followed by 10 (16.12%) understanding check repair strategies. The Other-Initiated Repair Strategies with lowest number of occurrence is Request for Definition (1 or 1.61%). Further data is shown in Table 1. In terms of the occurrence of strategies across the ten conversations, Conversation 2 of the Desert Island task occupies the highest (ten times or 16.12%) in number, followed by Conversation 5 of the Desert Island task with 9 (14.51%) strategies. Conversation 1 of the Spot the Different task and Conversation 1 of the Desert Island task rank 3 in number of occurrence, both comprise 8 (12.58%) strategies. The conversations with low number of OIR strategies happen in Conversation 3 of the Desert Island task (only 1 or1.61%) and Conversation 2 of the Spot the Difference task (2 or 3.22% strategies). When these two conversations are reviewed, it is revealed that in fact there are many trouble sources in their conversations but these are not repaired by the interlocutors.

Conversation instances based on OIR strategies
In order to provide clear understanding of how OIR Strategies are enacted in the learners' conversation, I will take one conversational excerpt of each type as an illustration. The conversational excerpt will be followed by brief explanatory account of how the process of repair is unfolded. What are we going to do with it? ((with slower speech delivery)) Uh:: we can do:: (2) by the (++) twenty meters of nylon rope. We can tie our nylon tent ....  Efan (at line 2) explicitly asks Indi to repeat her turn by saying 'can you tell me or repeat it to me'. This is responded by Indi at line 4 by repeating her utterance and providing more information to help Efan understand her utterance. 7) Request for Definition. In this type, the speaker initiates a repair of word, phrase, or even the whole utterance that needs to be de defined or explained its meaning by the previous speaker. In the leaners' conversation, there is only one example of this type of repair.

Types of Trouble Sources
In the trajectory of other-initiated repair sequence, trouble source is a word, phrase, or utterance that the interlocutors treated as problematic in the conversation [6]. Trouble source usually appear in the turn prior to the repair initiation turn. After a close look into the learners' conversation it is noticed that the trouble sources produced by the interlocutors in this study comprise three types, namely linguistic-related problem, interactional-related problem, and meaning or message related problem. Linguistic problem means, among others, poor grammatical construction that can cause understanding problem, wrong pronunciation, wrong choice of word, etc. Interactional problem relates to speech delivery, in terms of long pause, sound stretches. Meaning related problem means overall unclarity of meaning of the utterance that makes the next speaker difficult to understand it. The following conversation excerpts illustrate those trouble sources in the learners' conversation.

Linguistic-related trouble source
From all the conversations analyzed, there are three OIR Strategies produced because of linguistic trouble source in the previous turn and all of them deal with wrong pronunciation of lexical item. . It is interesting to observe that the correction is provided by Alga who is rated as having low proficiency in English for the pronunciation error made by Anit, a high-proficient learner.

Interactional-related trouble source
The next type of trouble source that triggers other-initiated repair relates to the way the turn is delivered by the speaker whose turn contains trouble source. The following excerpts illustrate some of this interactionalrelated problem. At line 4 Anit, the high-proficient speaker, initiates an understanding check repair because she noticed that Alga, a low-proficient speaker, was having problem in presenting his idea as indicated by the use of several sound lengthening such as 'uh::" at line 2 and pauses within and between turn at line 3. Anit's repair initiation helps Alga to continue his turn and it is successfully done as seen by Alga's repetition of Anit offer at line 5, although it is given in Indonesian. This excerpt illustrates how repair initiation is done because of poor speech delivery in the form of intra-turn pauses produced by the interlocutors. At line 4, Anch's turn contains several long pauses and sound lengthening indicating that she is having trouble in speaking. This problem is noticed by Alga at line 5, so he initiates a repair to help Anch producing what she wants to say. In this case, other-initiated repair functions as way of collaborative learning between the interlocutors in conversation.

Meaning-related trouble source
Other-Initiated Repair Strategy is also employed because the speaker just does not catch the meaning of what is being said by his/her interlocutor in the conversation. The illustration is provided in the following excerpt.
At line 8, Efan seems to have no understanding at all of what is said by Alga at lines 1 to 7.His repair initiation "excuse me excuse me I don't understand what did you say before. may you tell me what did you say" clearly shows that Alga's long utterances do not make sense to him.A close observation of Alga's utterance proves that it contains several ungrammatical clauses and wrong choice of word ('decorate flower').In addition, it is marked by the presence of unsmooth speech delivery (pauses and sound lengthening).All these create misunderstanding for the next speaker, Efan.

DISCUSSION
This study was intended to analyze the occurrence of other-initiated repair strategies in solving understanding problems in the conversations of EFL learners when they perform two communicative tasks. It shows that other-initiated repair strategies occur in all the conversations with different number. In addition, the study was also aimed at scrutinizing the trouble sources that trigger the employment of other-initiated repair strategies in the conversations. There are three types of trouble sources that trigger repair initiation, namely linguistic-related problem, interactional-related problem, and overall meaning-related problem.
There are eight types of OIR strategies used by the EFL learners in their conversations, namely unspecified repair, interrogative repair, partial repeat plus a question word repair, partial repeat repair, understanding check repair, request for repetition, request for definition, and correction repair. This finding is different from the study of Fotovatnia & Dorri [12], where nine strategies are found, including the nonverbal resources repair strategy. In the current study nonverbal resource is not used as repair strategy but more as sources of trouble that trigger the repair strategy. The prevalence of OIR Strategies in EFL learners' conversation endorses the proposition of Huth [5], Markee [4] and He and Young [3] hat analyzing repair organization in learners' conversation provides us a better knowledge of how learners overcome understanding problem in order to maintain the flow of conversation.
The current study found that the EFL learners with low level of English proficiency managed to produce OIR Strategies. Principally, producing OIR Strategies is prompted by the fact that the previous turn of the repair initiation turn contains some kind of troubles that impede the understanding of repair initiator. The problematic turn may result from the learners' low-level proficiency of English, for example in grammar or speaking fluency, hence requires corrective act from the supposedly more proficient speaker as shown in the study by Khodadady & Alifathabady [11] that show learners from high group produced more OIR Strategies than the learners from low group. It is also different from the study in mundane conversation by Bolden [19] where repairing understanding problem relies on other speaker who has better knowledge of language and topic under discussion. This interaction phenomenon requires deeper examination in order to get better insight on why this has happened.
Regarding the trouble sources that trigger OIR strategies, the current study found that the EFL leaners produced three types of problem, i.e. linguistic, interactional, and meaning-related problem. A closer observation into the utterances in the conversations reveals that actually the learners' conversations are characterized by the existence of many ungrammatical language constructions. Many of the sentences are not grammatically well-formed, such as incorrect use of third person singular pronouns (he for she or vice versa), omission of verbs as predicate, incorrect subject-verb agreement, etc. Normally, the troubles should be noticed and repaired by conversational interlocutors, but in the current study they seemed not to be treated by the participants. However, given the fact that these conversants are language learners and that they are still in a learning process to become better speakers of English, these language-deficient phenomena can be seen as an enactment of their L2 learner identity or as Carroll [25] puts it as their "doing being a NNS". In this analysis, the focus is more on their initiatives to try to solve understanding problem in conversation as a part of a language learning process. Here learning is seen as cooperating enterprise where they assist each other in building up a common understanding though the employment of different strategies when there is a problem in the conversation.

CONCLUSION
The first main finding of this study is that the EFL learners with limited proficiency in English manage to take initiatives to use other-initiated repair strategies to overcome the understanding problems in conversation. In the ten conversations that the learners are engaged in, they employ eight different types OIR strategies to deal with sixty two conversational episodes that might trigger the occurrence of communication breakdown. This finding shows that the EFL learners have the ability to notice, and then repair, the trouble sources in the conversations, although they have not been explicitly taught about these strategies.
Secondly, it was also revealed in this study that the EFL learners' conversations are characterized by the presence of many trouble sources in the form of various linguistic and interactional deficiencies. However, the trouble sources do not constrain the learners to continue their conversations by using the otherinitiated repair strategies. Whether, the understanding problem is evidently resolved, it entails another issue that needs to be investigated further. Therefore, better-designed studies in this area are needed to better enhance our understanding on how learners with limited proficiency deal with understanding problem in conversation. Finally, the result from this study may be useful for the EFL teachers in designing materials in their lesson. They need to take into consideration the incorporation of explicit teaching of OIR Strategies as part of their speaking class to improve the students' fluency. This is particularly crucial in EFL setting since the learners' exposure to English is mostly very much dependent on classroom encounter.