Towards a Learner-Centred Approach to Designing Role-Play Instruments for ILP Studies: A Study Based On Complaints

. Role-play tasks have been widely used in pragmatic research to explore spoken interaction. This instrument consists of situational scenarios purposefully designed to make participants elicit specific pragmatic data in controlled situations (Kasper & Youn 2017, Félix-Brasdefer 2018). Notwithstanding the widespread use of role-plays, some drawbacks have been identified concerning their design (Hudson et al. 1995, Trosborg 1995, Youn 2015) and real-life consequences (Al-Gahtani & Roever 2012). Against this backdrop, this study presents a learner-centred approach to designing situational scenarios based on participants’ examples of complaint situations. Specifically, an exemplar generation task and a likelihood questionnaire (Jianda 2006a, 2006b) were used to develop the role-play task. The study reports on the implementation of the learner-centred approach and its effectiveness to construct a role-play task. Furthermore, based on information gained from retrospective verbal reports, this study discusses whether participants’ engagement in the design of the role-play task encouraged them to act out the situations, as they would in real-life contexts. The study evidences the usefulness of adopting a learner-centred approach to design the role-play task. In terms of performance, it seems that, in general, the participants would exhibit similar pragmatic behaviour in a real context. However, they were aware of the lack of real consequences that role-play tasks carry. , enfoque centrado en el alumno, quejas. Contents. 1. Introduction. 2. Gathering spoken data in second/foreign language pragmatics. 3. Methodology. 3.1. Participants. 3.2. The investigated speech act: complaints. 3.3. Instruments and procedure. 4. Constructing a learner-centred role-play task. 4.1. Constructing the role-play task. 4.2. The role-play task. 5. Discussion. 6. Conclusion.


Introduction
Researchers can choose from a variety of instruments to conduct experimental research in cross-cultural pragmatics and interlanguage pragmatics (ILP). Determining the appropriateness of a given instrument may depend on the purpose of the research. As literature shows, research instruments in cross-cultural pragmatics and ILP may fall into three main categories: (1) examining spoken interaction data like authentic discourse, elicited conversation, and role-1 Department of English Studies, Universitat Jaume I, vbeltran@uji.es. https://orcid.org/0000-0003-3139-6629. ARTICLES plays; (2) questionnaires such as Discourse-Completion Tasks (DCTs), multiple-choice questionnaires, and scaled response; and (3) self-report data including interviews, verbal reports (retrospective and introspective), and diaries (Kasper & Rose 2002). Among these, the instruments most frequently used to gather data, either written or spoken, are DCTs and role-plays, respectively.
Since the present study focuses on spoken interaction, the following section provides a review of the three well-documented methods used to explore spoken data, with particular attention paid to role-play tasks, the domain of interest here. Please note that the questionnaires and self-report instruments are not reviewed here for reasons of space (see Taguchi & Roever 2017).

Gathering Spoken Data in Second/Foreign Language Pragmatics
Authentic discourse or naturally occurring data refer to spoken utterances found in real contexts. Through audio/video recordings and field notes, typically taken in the presence of participants, researchers have access to authentic pragmatic data in a variety of communicative events (Golato 2017). Hence, a clear advantage of naturally occurring data over other research instruments (e.g. role-plays) is that data emerge from authentic interactions. Nevertheless, this technique has rarely been used to explore second/foreign language (SL/FL) pragmatics (e.g. Bardovi-Harlig & Hartford 1993, Shively 2011, Al-Gahtani & Roever 2014a, 2014b. This is so probably because this technique may present some drawbacks. For example, drawing comparisons across contexts (Kasper & Rose 2002) and compiling a representative corpus (Byon 2015) can be difficult. Moreover, researchers cannot control sociolinguistic variables, and it may be challenging to explore specific pragmatic features (Kasper & Dahl 1991). The presence of researchers on-site may also influence participants' interaction (observer's paradox, Labov 1972). Even if researchers are not present, participants can be aware of the research situation, a fact that may influence their pragmatic behaviour (Culpeper et al. 2018).
Elicited conversation is another type of research method used to gather interactive spoken data in SL/FL pragmatics. As in role-play tasks, researchers arrange participants' involvement in the experiment and request them to follow specific instructions to complete a given task (Schneider 2018). Unlike in naturally occurring data, where language emerges naturally, in elicited conversation, language production depends on the researchers' interests. Therefore, participants elicit spoken utterances prompted by specific situations and under specific conditions (Schneider 2018), a fact that may influence their overall spoken production (Taguchi & Roever 2017). In elicited conversation, participants can be asked, for example, to discuss particular issues (Baumgarten & House 2010) or to get to know each other (Taguchi 2015). Elicited conversation operates differently from role-plays since participants maintain their identity and do not assume new roles, thereby somehow limiting the scope of analysis. Notwithstanding some drawbacks, elicited conversation is a promising research instrument for studying spoken interaction in SL/FL pragmatics (Taguchi & Roever 2017).
Although naturally occurring data and elicited conversation methods serve to gather spoken pragmatic data, most studies have used role-plays. A role-play may be defined as "a social or human activity in which participants 'take on' and 'act out' specified 'roles', often within a predefined social framework or situational blueprint (a 'scenario')" (Crookall & Saunders 1989: 15-16). In role-plays, learners are engaged in simulated spoken communicative tasks in which they perform specific roles and say what they would say in such situations and circumstances in authentic interaction (Kasper 2000, Crookall & Saunders 1989. Role-plays are mainly employed for teaching, assessment, and research purposes in a variety of areas, including applied linguistics (Kasper & Youn 2017). In cross-cultural pragmatics and ILP, role-plays refer to a research instrument consisting of purposefully designed situational scenarios intended to make participants elicit specific pragmatic data in controlled situations (Kasper & Youn 2017, Félix-Brasdefer 2018. Kasper and Dahl (1991) identified different types of role-play tasks. According to these authors, role-plays can be either spontaneous (participants maintain their identities) or mimetic-replicating (participants take on a different role). Moreover, role-plays are also distinguished according to the level of interaction: closed or monologic role-plays and open or interactive role-plays. In closed role-plays (or oral DCT), participants produce one spoken turn to a particular situation without having the response of another interlocutor (no interaction). In contrast, in open role-plays (or discourse role-play tasks), participants can elaborate as many turns as necessary to fulfil communicative purposes.
Role-play tasks allow contextual variables, roles, and settings to be controlled and manipulated according to the researchers' interests (Félix-Brasdefer 2010). Therefore, when designing role-play tasks, researchers can specify the different characteristics of each scenario according to the purposes of the study. For instance, researchers can decide on the contextual variables of the situations. These variables provide sociopragmatic information about the situation, for example, different levels of power (i.e. high, equal and low), degree of social distance (i.e. stranger, acquaintance and intimate), and rank of imposition or severity of offence (i.e. high and low) (Brown & Levinson 1987). In doing so, researchers can examine how specific contextual variables influence participants' interaction in different contexts (Culpeper et al. 2018). Moreover, when designing situations, researchers may take into account other aspects such as participants' degree of familiarity with both the roles (Trosborg 1995, Youn 2015 and the settings (Hudson et al. 1995), and whether the scenarios are appropriate for participants' socio-cultural knowledge (Beltrán-Palanques 2013). Participants' lack of familiarity with particular roles and settings in situations that may not be socio-culturally appropriate for them could influence the construction of their utterances. For example, asking participants to act out the role of a salesperson in a business negotiation meeting may be challenging if they do not have some prior experience or belong to that particular field (e.g. learners of Business English). This possible situation can be resolved, for example, by taking into consideration the above recommendations as regards roles, settings and participants' socio-cultural knowledge.
Nevertheless, even when researchers may want to adapt situations to participants' characteristics and knowledge, another issue may arise, that of data authenticity. The fact that role-plays are not authentic, and therefore do not have any real-life impact (Félix-Brasdefer 2007, Al-Gahtani & Roever 2012, can prevent participants from eliciting what they would say in real contexts. In line with this, Ewald (2012) compared authentic and elicited data (role-play) in direction-giving. The author found that the data that emerged from authentic interactions tended to be more elaborated and precise as compared to the data elicited through role-play tasks. These differences may be attributed to the lack of real-world consequences carried by role-play tasks. Despite this, some researchers have made efforts to engage participants in authentic-like situational scenarios. For instance, in the field of English for Specific Purposes (ESP), Padilla-Cruz and Martínez-López (2017) explored learners' pragmatic behaviour in service encounters using mimetic pretending role-plays in which learners took the roles of informer and tourist. The task consisted of authentic situations that learners would probably face in their future professional duties. Thus, role-plays, when carefully designed, may be of great pedagogical and research value.
In spite of the drawbacks, role-plays are extensively used in SL/FL pragmatic research. They not only allow comparison and replication (Kasper & Dahl 1991) but also relatively systematic implementation. Moreover, although data elicited through role-plays are not authentic, the spoken data that emerge from those interactions can serve to exemplify how spoken discourse is constructed (e.g. Kasper & Dahl 1991, Kasper 2000, Kasper & Roever 2005, Félix-Brasdefer 2007, 2010, 2018, especially when analysing data from a conversation analysis perspective (Kasper & Youn 2017). Therefore, through role-play tasks, researchers can have access to, for example, the structure of talk exchanges, turn-taking features, conversational sequences, coordination between speaker and listener, learners' achievement of transactional and interpersonal skills, production of communicative action, as well as comprehension or miscomprehension of interlocutors' contributions (Kasper 2000). Furthermore, from a multimodal perspective, researchers may explore how interaction is constructed and deconstructed using different semiotic resources (e.g. gestures, facial expressions) (Beltrán-Palanques & Querol-Julián 2018, Beltrán-Palanques 2019).
Bearing in mind the potential of role-play tasks as a research instrument to elicit spoken data in ILP, this study discusses the design of a role-play task from a learner-centred perspective. This approach is adopted to provide participants with scenarios based on their own experience. In doing so, the scenarios designed are expected to consist of familiar roles (Trosborg 1995, Youn 2015 and settings (Hudson et al. 1995), and to be socio-culturally appropriate (Beltrán-Palanques, 2013). Furthermore, using retrospective verbal reports, this study explores whether the participants' engagement in a learner-centred role-play task may prompt them to perform the role-play situations as they would in real life. Thus, in terms of specifics, this study intends to explore: (1) the effectiveness of adopting a learner-centred approach to create scenarios based on authentic examples, and (2) whether participants' collaboration in the design of the situations of the role-play task would encourage participants to perform them as they would in authentic contexts.

Methodology
Since this study addresses the construction of a role-play task, the methodology section briefly describes the participants, the investigated speech act, the instruments used to construct the role-play task and the procedure followed. Section 4 provides further details as regards the application of the research instruments to construct the learner-centred role-play task.

Participants
The participants in the study were 64 first-year university students (mean age: 19.7) who were taking an ESP subject in two different university degree courses. All the participants were informed of the procedure of the study and were asked to sign a consent form. The study was conducted at a Spanish university during the academic year 2015/2016.

The investigated speech act: Complaints
Complaints are complex speech acts (Geluykens & Kraft 2008) that fall into two main categories: direct complaints (D'Amico-Reisner 1985) and indirect complaints (Boxer 1993(Boxer , 1996(Boxer , 2010. On the one hand, direct complaints involve situations, commonly based on a dissatisfactory event or experience, in which the speaker expresses displeasure or annoyance because of a particular action that affects him/herself (Olshtain & Weinbach 1993). On the other hand, indirect complaints refer to "the expression of dissatisfaction to an interlocutor about oneself or someone/ something that is not present" (Boxer 1996: 219). Indirect complaints differ from direct complaints in the sense that they typically seek agreement and may have the potential to build rapport (Boxer 2010).
This study focuses on direct complaints, although throughout the paper they are referred to using the term "complaints". Complaints of this kind typically represent face-threatening acts (FTAs) (Brown & Levinson 1987) and, like other FTAs, are commonly performed using indirect forms to soften the illocutionary force of the utterance. This is so probably because complaining involves the speaker (the complainer) expressing his/her disapproval, dissatisfaction or feelings towards a particular offence for which he/she considers the listener (the complainee) to be responsible (Trosborg 1995). However, complaints, unlike other speech acts such as requests, do not have a prototypical set of strategies that guide their realisation (Laforest 2002, Geluykens & Kraft 2008 and there is no common adjacency pairs structure, but instead extended sequences (Drew & Walker 2009). Therefore, performing appropriate complaints may be difficult for language learners, even at high proficiency levels.

Instruments and procedure
The instruments of the study consisted of an exemplar generation task and a likelihood questionnaire (Jianda 2006a(Jianda , 2006b. These instruments, administered using Google Forms, were used to construct the role-play task. Additionally, a retrospective verbal report, carried out face-to-face, was employed to gather qualitative data after implementing the role-play task. Figure 1 illustrates the procedure followed to construct the role-play task. The exemplar generation task served to gather examples of complaint situations based on the participants' experience. The examples derived from the exemplar generation task were analysed, and then methodological decisions as regards the selection of the situations to be included in the likelihood questionnaire were made. After that, the participants ranked the likelihood of occurrence of the situations. The results of the likelihood questionnaire were then examined and further methodological decisions were made to choose the scenarios of the role-play task. The role-play task was then implemented (not discussed in this study) and, immediately afterwards, the participants undertook a retrospective verbal report to obtain qualitative data as regards their performance in the task. The retrospective verbal report was conducted in Spanish. Although these reports contained several questions, this study centres on just one: "¿Actuarías de manera similar en una situación del mismo tipo en la vida real?" (Would you act similarly in a real-life situation?). The following section presents further information concerning the construction of the role-play task.

Constructing a Learner-Centred Role-Play Task
This section reports on the learner-centred approach followed to construct the role-play task and the various methodological decisions taken throughout this process.

Constructing the role-play task
A learner-centred approach was adopted to construct the situational scenarios of the role-play task. This approach, as detailed here, involved using an exemplar generation task and a likelihood questionnaire.
Through the exemplar generation task, the participants elicited examples of complaint situations based on their experience. The participants did not receive instruction as regards the speech act of complaints before administering the exemplar generation task. Nevertheless, the researcher provided an overview of the basic requirements of the task, i.e. social distance, participants' relationship. The task included instructions and an example to facilitate comprehension. Table 1 illustrates the exemplar generation task. Table 1. Exemplar generation task.

Exemplar generation task: Instructions
As you know, we usually make complaints to express that a particular situation annoys us.
In this activity, we would like you to think about 3 situations in which you made a complaint or saw someone making a complaint.
Please consider the following aspects: -you and the other person should be of similar ages -you and the other person should be of equal 'rank' (e.g. two students, two friends) -how familiar are you with him/her?
(1) a stranger (2) someone you know (3) someone you know well (4) a close friend (5) someone you have an intimate relationship with -how offensive was the situation?
(1) a little bit angry

Degree of familiarity: someone I know
Describe the situation: I had to do a project assignment with one of my new classmates but he was so late that I had to start writing it on my own.

How offensive was the situation? very offensive
How did you feel? extremely angry As shown in Table 2, the task required the participants to (1) indicate the place where the situation took place; (2) specify the participants involved in the situation; (3) select the degree of familiarity, from stranger to someone you have an intimate relationship with; (4) describe the situation; (5) indicate how offensive the situation was; and (6) rank from 1 to 4 how angry they were.
The task was designed taking into account the variables of power, social distance and severity of offence (Brown & Levinson 1987). The power relation between the interlocutors was equal. Therefore, the participants had to provide examples of complaint situations involving the same power. The degree of social distance was adopted taking into account the following threefold classification: stranger, acquaintance and intimate (Brown & Levinson 1987). The interlocutors' degree of social distance consisted of (1) a stranger; (2) someone you know; (3) someone you know well; (4) a close friend; and (5) someone you have an intimate relationship with. The participants also had to rank their perceived level of offence according to the following items: (1) a little bit angry; (2) angry: (3)  (2) the roles were familiar for the participants (Trosborg 1995, Youn 2015; (3) the contexts were familiar for the participants (Hudson et al. 1995); (4) the content of the examples was appropriate for the participants' socio-cultural knowledge (Beltrán-Palanques 2013); and (5) the examples did not contain any inappropriate information.
In general, the main reasons for discarding examples were: (1) incompleteness of the task (e.g. missing information), inappropriateness of examples (e.g. roles, context); (2) repetition of examples (some situations were quite similar and only one was accepted), and (3) unclear description of the complaint situation. Interestingly, some scenarios were perhaps too oriented towards a particular area, specifically video games, probably because some participants were pursuing a Bachelor's Degree in Video Game Design and Development. Some scenarios involving playing video games or playing online were accepted if they did not require much specific knowledge.
Some discarded situations are shown below. Please note that all examples are written verbatim and may contain some grammar and/or spelling mistakes: Example 1: I went to the supermarket to buy printer ink, and when I arrived home I realised that the cashier had charged twice the printer ink, at first I fell irriated but I calmed and I returned to the supermarket and I explained the situation to the responsible and he, very friendly returned me the money.
Example 2: I told my mum that I wanted water, but she gave me wine and a I don't like wine!! So I was angry with her about 2 hours because I told it several times...

Example 3:
We was playing League of Leguends, a MOBA game. My friend stole all the team kills and after that 50 min of mach they left the game and we lost the mach.

Example 4:
We are at the night club and my friend started to smoke next to me. I started cough but he continued smoking. I started to stink of smoke. I asked him to stop but he continued.
Example 5: We were playing online and talking through Steam chat, and he told me that in the last match I played so bad. But he was wrong, I did my work correctly. I was pretty sure that he didn't kill the necessary number of enemies, not me.

Example 6:
We ask for some food in a restaurant and when about half an hour they give us our food it was absolutely cold and the meet was so hard. they didn't change the food and they make us pay all the bill of the lunch. but finally they invite us to another lunch in this restaurant.

Example 7:
Maria and I were lost in the Humans building and I were angry because we didn't find the class.
The examples shown above raised some problems. For instance, the situations described in Example 3 and Example 4 did not explicitly refer to complaint situations. They would probably require further elaboration. Therefore, if these examples were modified, they could serve to make the participants elicit a complaint, as in Example 4. However, the content in Example 4 (smoking), as in Example 2 (drinking wine), was inappropriate. The power relation in Example 1, Example 2 and Example 6 was unequal. In addition to this, it seems that the participants would require some knowledge about video games and online platforms to perform the situations described in Example 3 and Example 5. Finally, the situation described in Example 7 appeared to be an indirect complaint, which was not the pragmatic aspect explored in the study.
Moreover, as indicated, examples that contained similar information were not accepted. These examples were examined to see which ones could be kept. This was the case, for instance, of Example 8 and Example 9.
Example 8: My new flatmate don't clean the kitchen when he cook. It's always the same. It makes me angry because I always clear the kitchen.
Example 9: My flatmate left the plates without cleaning it. I was very exhausted, so I didn't rent my plates also; when my flatmate saw I hadn't cleaned my plates, got very angry and we had an argument.
In short, some situations were not included in the study because they did not meet the specific requirements indicated above. Nevertheless, some of the discarded situations can be adapted for other studies. The researcher of the study and a colleague who was familiar with the study took this methodological decision. Table 2 shows the 20 situations selected for the study. 3. My brother and I are the same age, so we share the same friends. We had a meeting with them and he wasn't ready at all because he was playing videogames, so we were late because of him. I wouldn't be angry if it was the first time, but he keeps doing it all the time.
4. Last day a friend and I were in a club and then another friend came to me and ask me to pay a taxi for her. I complained because she never pays the taxis but then I lent her the money.

5.
A classmate has borrowed my notes, he forgot that they were in his bag and has taken them home, consequently I couldn't study that day. Anyway, it was the beginning of the term but I didn't like that and I told him.
6. My brother has borrow my computer and when I asked him for the computer he didn't want to give it to me. We had an argument but then he told me he had bought a new video game for me and he was installing it.
7. I came home very hungry only to find out that the pizza from last night wasn't in the fridge anymore and the only person in the flat who could've eaten it was my flatmate. So I had to spend more money to buy another pizza and wait more time until it got here.
8. I met my friend at 8 to go to class. We had class at 8.30. He wasn't at the tram stop, I call him and complaint because he always does the same and we're always late for this class.
9. I wanted to have a shower but my flatmate had all of her clothes on the bathroom and I don't like it because the bathroom is unclean.
10. I was at the library to get a book, and another student try to jump the queue. She said she was before me in the queue, but she came after me. I complained because I was there first.
11. My best friend has organised a party with all my friends with a live concert I like but they haven't invited me.
12. I was at the library with some classmates working and then another student came to us to ask us to speak a bit more quietly. I complained because she was talking to her friend loud and we did not say anything. Then, she apologised and we continue working.
13. I arrived late at night at the train station with all my luggage. My friend had to help me with the luggage and when I saw him, he was having a coffee at the bar. I complained but then he apologised and told me he thought my train was late.
14. While we were playing an online game my friend started complaining about my "lack of skill", even thought his score was far worse than mine. The 20 examples shown above were then revised and slightly modified. On the one hand, each example was revised in terms of grammar correctness and style. On the other hand, when necessary, extra contextual information was added to ensure clarity and participants' understanding. After that, the situations were included in the likelihood questionnaire to measure their likelihood of occurrence (Jianda 2006a(Jianda , 2006b. Table 3 shows the likelihood questionnaire.

Likelihood questionnaire: Instructions
In this questionnaire, you should rank from 1 to 4 the likelihood of the situation presented below. Please, take into account the following scale: As shown, the participants were asked to rank each situation from 1 to 4 according to the likelihood of occurrence in their everyday life. The participants did not receive any information about the level of offence (whether high or low) of each situation. The results of the questionnaire showed that although most situations were quite likely to occur, some were more salient. However, since the purpose was to design a role-play task to gather multimodal data, only two situations were chosen. Thus, considering the results obtained, further methodological decisions were made to select the two final situations. These decisions involved: (1) selecting two opposing situations (low and high level of offence) in which the relationship between the interlocutors was that of (close) friends; and (2) using the situation involving a low level of offence as a warm-up to help the participants become familiar with the laboratory conditions of the task.
The rationale behind this decision was based on the following aspects. It was considered appropriate to use two opposing situations because the interaction that would emerge in each situation would differ. The situation classified as a high level of offence was expected to make the participants elicit longer and more complex utterances, and more elaborated moves. Moreover, it was considered suitable to use two opposing situations to prevent the participants from eliciting similar pragmatic strategies in both situations. Regarding authenticity, it was believed that having two situations comprising two close friends (social distance) would be sufficiently representative of the participants' experience, which in turn could make them feel more engaged in the situations. These methodological choices made it possible to design a warm-up situation, which was intended to promote the participants' involvement in the task (especially in the second scenario), and ensure their comfort with the setting and the video cameras (assumed to be somewhat face-threatening for the participants). Therefore, the two scenarios that received the highest score in each specific level of offence involving close friends were selected to construct the final role-play task.

The role-play task
As reported, the situational scenarios for the role-play task were designed from a learner-centred perspective using two different research instruments (i.e. the exemplar generation task and the likelihood questionnaire). The role-play comprised two scenarios representing opposing situations (i.e. low and high level of offence) and close social distance (i.e. friends) between the interlocutors. Table 4 displays the final role-play task.
The role-play task consisted of two socio-culturally appropriate (Beltrán-Palanques 2013) situational scenarios involving familiar roles (Trosborg 1995, Youn 2015 and known settings (Hudson et al. 1995). The first scenario, classified as a low level of offence, involves an everyday situation in which two friends are having a drink. As seen, the participant who has to complain decides to go to the toilet and when he/she comes back, the other participant, who is his/her friend, has drunk his/her drink. Likewise, the second scenario entails two close friends. One of the participants has organised a party, but he/she has not invited his/her friend. All their friends will go to the party and his/her favourite music group will be playing at that event. As expected, this situation represented a high level of offence because of the participants' relationship and the damage caused. The first scenario (warm-up scenario) functioned as an icebreaker for the participants, which helped them to become familiar with the task, the roles and context, and the different devices used to collect the data. The role-play task was implemented as an in-class activity in two different ESP courses, and the participants' interactions were video-recorded to compile a multimodal corpus. Before implementing the role-play task, decisions about where to carry out the study were adopted. Then, considering the facilities offered at the university and the in-class nature of the task, the experiment was completed in two language laboratories. Likewise, decisions were made regarding the correct position of the video cameras to capture the participants' embodied actions. Moreover, to reproduce an authentic-like setting, extra furniture was included in the room (i.e. a table to simulate that the participants were at a café). Regardless of the artificiality of the context (language laboratory), the audio recorder and the video cameras, and the role-play task, the aim was to create an authentic-like setting in which the participants felt comfortable and interacted as naturally as possible. Table 4. Role-play situations.

Having a drink with a friend
Speaker 1 It is a sunny day and you are having a drink with a friend. Tell your friend that you're going to the toilet.

Go to the toilet.
When you come back, your friend has drunk all your coke. You think that this is unfair. What a cheek! How could he? You complain.

Speaker 2
It is a sunny day and you are having a drink with a friend. While your friend is in the toilet, you drink all his/her coke because you are very thirsty.
You should respond to any comment your friend makes.

Speaker 1
You discover that your friend has organised this great party and you have not been invited. Your favourite music group will be playing there. Your best friends are all going to the party. But you're not invited. You think that this is so unfair and you feel really angry.
You complain.

Speaker 2
You have organised an incredible party with all your best friends and your favourite music group. You just can't believe they are coming! You should respond to any comment your friend makes.

Discussion
The present study was conducted to investigate whether the learner-centred process adopted to construct the role-play task served to generate situational scenarios that resemble authentic situations. Furthermore, the study also attempted to explore whether the participants' engagement in the learner-centred role-play task made them act as they would in an authentic situation.
Concerning the first objective, this study has shown that, through the learner-centred approach, the participants could not only provide complaint situations but also be engaged in the construction of the role-play task. The adoption of this approach involved the use of two different research instruments: an exemplar generation task and a likelihood questionnaire.
Using the exemplar generation task (Jianda 2006a(Jianda , 2006b) and based on their experience, the participants elicited examples elicited examples of complaint situations. In general, the examples the participants elicited were somehow consistent with the type of everyday interactions university students can have, such as interaction with friends, classmates or flatmates. However, in other cases, the participants elicited examples that did not meet the requirements of the task, and therefore they were discarded. A possible explanation for this could be that some participants found it challenging to deal with the contextual variables specified in the task. Still, it should be noted that the complexity of the contextual variables was kept to a minimum to facilitate comprehension. After revising the examples, 20 situations were considered valid.
Next, the participants used a likelihood questionnaire (Jianda 2006a(Jianda , 2006b to rank the likelihood of occurrence of the 20 situations that were selected. In light of the results obtained, further methodological decisions were adopted. Specifically, two scenarios comprising opposing situations (i.e. low and high level of offence) and intimate social distance (i.e. close friends) were chosen. Therefore, the role-play task comprised two scenarios with familiar roles (Trosborg 1995, Youn 2015 and contexts (Hudson et al. 1995) and with socio-culturally appropriate situations (Beltrán-Palanques 2013). This particular outcome reveals the effectiveness of adopting a learner-centred approach, as it allowed the participants to take an active role throughout the construction of the situational scenarios.
Interestingly, through the learner-centred approach, the participants received implicit sociopragmatic information from the beginning of the process, specifically in the exemplar generation task. The introduction of sociopragmatics from the onset might be beneficial to support learners' sociopragmatic awareness and metapragmatic reflection (Kasper & Roever 2005). However, several participants seemed to have encountered some difficulties in dealing with the sociopragmatic variables of the exemplar generation task. Hence, it seems that, before implementing the task, some explicit pragmatic instruction could have been helpful.
The second objective of the study is discussed drawing on the qualitative data obtained in the retrospective verbal reports. These reports, among other aspects, focused on whether the participants performed the situational scenarios as they would have done in a similar real-life situation. Regarding the two role-play scenarios, but especially the second one, most of the participants tended to perceive this situation as an offensive act that could occur in real life and would require some kind of remedy. The participants also reported that while the first scenario would not represent an uncomfortable situation, the second scenario would require interlocutors to address the issue, mainly because of the shared interpersonal relationship and the damage caused. The participants acting out the role of the complainer perceived the second situation as offensive, and most of them indicated that they would exhibit similar behaviour in a real-life situation because they would feel offended. Likewise, the group of complainees also classified the second situation as offensive, and most of them maintained that they would behave similarly in a real-life situation in an attempt to solve the problem.
The following extracts, taken from the retrospective verbal reports, provide further insights into participants' performance. Please note that some parts of the transcription have been omitted because they are not relevant here.
Example 10 As shown in Example 10, participant B (complainer) did not acknowledge the offensive nature of the situation and suggested that the speaker did not need to elicit complaints. Moreover, she expressed her preference for using other pragmatic strategies to address these situations. When asked whether she would have behaved similarly in a real situation, she responded that she would have acted the same way but remaining calmer and taking the situation more seriously, as she would have felt offended. Thus, although she did not perceive the situation as involving a complaint, she noticed that some kind of remedy was needed, probably using other pragmatic strategies. This is reasonable because there are no specific prototypical strategies to elicit complaints (Laforest 2002, Geluykens & Kraft 2008, and therefore speakers may use a variety of pragmatic strategies to deal with them. Nevertheless, in this case, the use of other pragmatic strategies can be related to the participants' perception of what a complaint entails.
By contrast, participant A (complainee) perceived the situation as a complaint, which requires the interlocutors to address the issue. Participant A also reported that, in a real-life situation, she would have taken the complaint more seriously. In this regard, she indicated that, in a real-life context, the situation would have been more face-threatening because of the damage caused and the shared interpersonal relationship between the interlocutors. These particular comments may raise the issue of the participant's perception of the situation and engagement in the task. Although participant B did not perceive the situation as a complaint, she claimed she would have reacted similarly from a linguistic perspective. However, emotionally speaking, her reaction would have been different. Likewise, participant A would have reacted emotionally differently too. Regarding engagement in the situational scenarios, it seems the lack of real-life consequences could have influenced the participants' performance (Félix-Brasdefer 2007, Al-Gahtani & Roever 2012. Nevertheless, it is interesting to note that the two participants, but especially participant A, seemed to be aware of the contextual variables of the situation and how the sociopragmatic conditions of the situational scenario would influence the construction of interaction.
In addition to this, participant A reported that proficiency level issues prevented her from providing utterances that were more elaborate. In general, the participants could interact in the role-play task, although performing complaints may be complicated. Their realisation and especially their negotiation require great elaboration (Drew & Walker 2009), as well as the use of a variety of speech acts such as requests or refusals (Laforest 2002, Geluykens & Kraft 2008. Therefore, performing complaints may be challenging for speakers, even at higher proficiency levels, since they should demonstrate considerable pragmatic knowledge to negotiate them successfully.
Example 11: B1 level, male participants, Pair 2. In Example 11, both participants seemed to suggest that in a real situation, they would have modified their behaviour. They would perhaps take the situation more seriously due to the damage caused and the shared interpersonal relationship. This reasoning may be related to the participants' awareness of the sociopragmatic conditions of the situation and, more specifically, the level of offence and the interpersonal relationship between the interlocutors. What is observed in Example 11, as well as in Example 10, is that if the situation had had real-life consequences, the participants would have reacted differently from an emotional perspective because of the potential implications for their relationship (Félix-Brasdefer 2007, Al-Gahtani & Roever 2012. Therefore, this particular issue might have somehow prevented them from behaving as they would have done in a real situation. Notwithstanding, since the participants acknowledged the offensive nature of the situation, it could be argued that, in the role-play task, they employed similar core expressions to those they would have elicited in a real-life situation. Example 12: B2 level, male participants, Pair 17. In this case, the participants claimed that they acted in the role-play task as they would have done in an authentic situation in which a friend offended another friend. This may suggest that the participants were fully aware of the sociopragmatic conditions of the situation. Furthermore, they explicitly indicated that they tried to construct their utterances as naturally as possible. Interestingly, participant A reported that what he found artificial was the setting (language laboratory); in the retrospective verbal report, he pointed out that it would have been more appropriate to perform the role-play in a café. Specifically, he suggested that "si esto lo hubiésemos hecho en una cafetería de aquí de la UJI si que se hubiese sido más natural" (if we had done this in a cafeteria here at the UJI, it would have been more natural). This particular comment raises the issue of the artificiality of the setting. Perhaps performing role-play tasks in authentic settings would be interesting since participants would feel more comfortable and engaged with the situations. However, arranging authentic settings to implement role-play tasks would require a significant effort on the part of the researchers.
In short, the retrospective verbal reports seem to suggest that, in general, the participants would have reacted similarly from a pragmatic perspective, especially in the case of the second scenario. However, since role-play tasks do not carry real-life consequences, it seems that some participants did not interact as they would have done in authentic conversations (Félix-Brasdefer 2007, Al-Gahtani & Roever 2012. Not surprisingly, the participants' emotional responses in authentic offensive situations would differ because of the potential implications and consequences for their interpersonal relationship. However, this is not to suggest that role-play tasks are not useful to gather authentic examples of spoken data. Role-plays represent a valid research instrument to gather SL/FL pragmatic data (Kasper 2000;Félix-Brasdefer 2010, 2018.

Conclusion
The present study investigated the effectiveness of constructing a role-play task from a learner-centred perspective. On the one hand, it explored the process of generating role-play scenarios based on the participants' experience. On the other hand, it examined whether the participants' engagement in this process encouraged them to perform the role-plays as they would in authentic situations. The learner-centred approach presented here served to design complaint situational scenarios for a specific target group of participants. Throughout this process, the participants completed an exemplar generation task and likelihood questionnaire (Jianda 2006a(Jianda , 2006b. The results obtained in these two instruments served to construct the scenarios of the role-play task. Therefore, this study exemplified how to design a role-play task based on the participants' experience and adapted to them in terms of roles (Trosborg 1995, Youn 2015, settings (Hudson et al. 1995) and socio-cultural knowledge (Beltrán-Palanques 2013). Furthermore, this study has discussed the potential of using this particular approach to engage the participants in the construction of situational scenarios and provide them with opportunities to become familiar with different degrees of social distance and levels of offence from the onset. Additionally, the participants undertook a retrospective verbal report, which was instrumental in revealing information as regards their performance in the role-play task. Qualitative data seemed to suggest that, in general, the participants tended to perform the situations as they would have done in an authentic interaction. However, they were aware of the lack of real-life consequences that role-play tasks carry. This particular aspect might have influenced the participants' engagement in the task as, in some cases, they did not react (emotionally speaking) as they would have done in a real situation. Still, the participants acknowledged the importance of both interpersonal relationships and contextual factors, thereby pointing to the participants' sociopragmatic awareness. For instance, the participants recognised the need to negotiate the situation and find a solution to restore and maintain harmony between the interlocutors. However, since complaints are quite complex speech acts (Geluykens & Kraft 2008) performing them may put high demands on learners, who need to draw on their linguistic competence and especially on their pragmatic competence to achieve communicative purposes.
This study is not without limitations. One limitation is the reduced number of role-play situations. For further research, it would be interesting to construct more role-plays drawing on the results obtained in the likelihood questionnaire. Another limitation concerns the exemplar generation task, which was rather restrictive and guided in terms of contextual variables. Furthermore, it seemed several participants had some problems to understand the requirements of the task. Therefore, it may be advisable to provide explicit sociopragmatic instruction to facilitate participants' comprehension of the task.
Finally, adopting a learner-centred approach may have some pedagogical benefits. For instance, learners can participate actively in the elaboration of role-play scenarios and generate examples that are relevant to them. In doing so, they can start building some knowledge as regards the importance of social variables and how these influence interpersonal relationships and the selection and construction of pragmatic strategies.