Abstract

For some time now, research on Dialogue Interpreting (DI) has relied on data collected in real life situations, recorded and transcribed in order to analyse their discoursal and interactional features. A number of studies (e.g. Wadensjo 1998; Davidson 2000, 2002; Mason 1999, 2006) have highlighted that interpreting in interactions does not necessarily occur turn by turn, may include sequences involving the interpreter and only one of the interlocutors, may take forms different from the turns it refers to, by e.g. expanding or reducing their content. These findings have recently been taken up in DI teaching and training, in which authentic data - recorded and transcribed - are being increasingly used with learners or trainees to highlight relevant interactional features or for the students to engage in similar situations. The debate on the use of authentic data in teaching and learning is however not new and raises a series of problems which need to be considered when engaging in DI teaching. In what follows, I summarize the debate on the relationship between authentic data and authentic learning (section 2), and then illustrate some issues deriving from analysis of authentic DI interaction which may be relevant in DI learning and training (section 3). Through the observation of a roleplay activity, I then show that translating these insights into learning tasks is not easy and needs ad hoc research (section 4). Section 5 offers some concluding thoughts.

Keywords: interaction, dialogue interpreting, learning interpreting, interpreter training

1. The debate on authenticity in language pedagogy

After the surge of the so called “communcative method” in the late 1970s and throughout the 1980s, authentic texts like newspaper extracts, film trailers, interviews or agony aunt columns made their entrance into the language classroom like a breath of fresh air. Teaching the foreign language was no longer based only on the illustration and reproduction, in various forms, of grammar rule mechanisms: instead, students were engaged in e.g. listening to or reading “real” news, or in speaking or writing to get a “realistic” hotel booking abroad. This new teaching approach was found to be much more effective, and defintely more motivating, for students to eventually use the foreign language in communication (Widdowson 1983). Increasingly, language tasks aimed to contextualise or recontextualise ‘real-world’ materials in order to give learners opportunities to use the foreign language by engaging in plausibly authentic communication. The 1980s’ emphasis on communicative authenticity had some follow-ups relevant to the debate on authenticity in DI learning/training.

First, the birth of spoken language corpora. While authentic oral or written “monologic” texts were readily available via newspapers or TV, authentic interactions (apart from news interviews) were not easily found and needed to be collected in situations where a recorder was normally “not naturally there”. The PIXI (Aston 1988a) and the Cancode project (Carter 1998) had the explicit pedagogic purpose of informing and possibly providing materials for spoken language teaching and learning. The PIXI corpus, first made available through the Oxford Text Archive and then printed a year later (Gavioli & Mansfield 1990), and early work on these data (e.g. Aston 1988b, Zorzi 1990), represent important achievements in research foregrounding spoken language teaching/learning.  

Second, the birth of electronic corpora. The possibility of storing larger and larger quantities of electronic text and interrogating it gave dictionaries and grammars (including language learner resources) a new dimension and impact. For the first time, language description and language use seemed to go together and features of lexis and grammar could be made “more authentic” in the language classroom. However, developments in corpus linguistics increasingly seemed to pay attention to formal aspects of language, while the exploitaton of corpus resources in a communicative language learning environment was debated to a lesser extent (though see John, 1994; Aston, 1995, 2001; Gavioli & Aston 2001; see Seidlhofer 2003: 77-123; Boulton, 2017 for an overview).

Third, while on the one hand it became clear that samples of authentic language are not enough to engage students in “real” communication and that “reality does not travel with the text” (Widdowson 1998: 711), authentic texts and materials and authentic language activities do not always go hand-in-hand. It is interesting to note that in a recent account of task-based language teaching by Ellis (2017), little mention is made of authenticity - or, indeed, of quality issues in general - in discussing materials’ choice and design. Aston (e.g. 1997, 2002, 2009, 2011), Ciliberti (e.g. 1994) and Zorzi (e.g. 2008, 2011), to whom this book is dedicated, devoted time and energy to bridging the split between research on authentic (spoken) language and research on its impact on teaching/learning. They showed that describing authentic interaction is only the first step of a process that involves learners, teachers and materials designers.              

Nowadays, collections of recordings and transcriptions of DI encounters are increasingly available. Such collections provide multiple instances of comparable events which constitute potential authentic materials for DI teaching and learning. To make the move from DI samples to DI teaching/learning is, however, not a straightforward task.

2. The reality of DI data: some relevant issues in DI teaching and learning

In the second half of the 1990s, on her suggestion, Daniela Zorzi and I started collecting recordings of DI encounters. This was done initially through students’ theses, which provided us with recordings of interpreted business meetings and trade fairs, guided tours in local food productions and judicial hearings. In later years, a more systematic collection of DI data was carried out in collaboration with healthcare services (the AIM corpus of mediated doctor-patient interaction, Baraldi & Gavioli 2012) and an immigrant support centre (Gavioli 2016). Over the last 20 years, I have analysed these collections of DI data with Daniela Zorzi (see Gavioli & Zorzi 2008; Zorzi & Gavioli 2009), Anna Ciliberti (Ciliberti 2009) and a number of colleagues,  many of whom are contributors in this volume. In what follows, I discuss three aspects of DI which have emerged from discussion with these colleagues, as especially useful from a learner’s point of view.

1.2 Interpreters “cannot not interpret"

Studies on conversation show that participants in the interaction display to each other their understanding of what is going on. In DI, interpreters make sense of what participants say in response to each other. To do so, they negotiate their and others’ understanding by asking for and providing clarification and explaining what may be hidden behind the words. Let us have a look at two sequences. The first is from a conversation at a bicycle trade fair. The client’s request in turn 1 is rendered in turns 2 and 4. Immediately after starting her rendition in turn 2, however, the interpreter stops, switches into English and asks for confirmation about which precise product the client is asking about. When completing her rendition in turn 4, moreover, she provides her guess about which model the client is presumably referring to: “penso intendano i BR settecentouno credo”, which guides the interlocutor to identify the object called for. The plausibilty (and correctness) of the interpeter’s guess is confirmed by the seller in turn 5.

1. [Natural_Data][1]

The second extract comes from the immigrant support desk. Here the applicant is asking whether his son’s birth certificate, issued years before, has expired. The mediator, an expert and trained Ghanaian professional, asks the applicant several clarification questions before relaying the latter’s request to the assistant in turn 9. It is interesting to note that the assistant asks a couple of questions (turns 10, 12) before answering. Turns 14 and 15 are interesting too, in that the assistant’s answer, that birth certificates do not expire, is only made explicit in the mediator’s expansion of the assistant’s response (turn 15).

2. [Natural_Data]

While data show that interpreters participate in the talk they are interpreting with more than just renditions, sequences like those above reveal that this participation is oriented to making sense of talk, first and formost, for the sake of accurate rendition.  Since participants’ understanding is negotiated in talk with reference to what they want to achieve, interpreters’ understanding is negotiated in talk with the aim of making such achievements possible. So while interpreters may, in fact, influence what is said, for the very sake of rendering sensibly, they cannot “not interpret” (Pöchhacker, 2012) and this interpreting activity is constantly visible in ongoing talk.

2.2. Dyadic sequences are managed in the interaction

Dyadic sequences, i.e. sequences in which interpreters talk with only one of the interlocutors, are generally considered risky in the interpreting profession, as they appear to exclude the other of the two interlocutors. Yet, dyadic sequences are rather frequent in interpreter-mediated talk and in many cases do not seem detrimental to the course of interaction, in that interpreters and mediators alike show they can coordinate participation. Extract 1 and 2 above, for instance, show how dyadic sequences can lead to effective provision of the required information.

Looking at the function of minimal responses in doctor-patient mediated talk in an earlier study (Gavioli 2012), I noted that mediators’ “mhms” seem to have a double function: on the one hand, they serve to invite the speaker-in-turn to go on talking; and on the other, they put the other participant “on hold”, with the communicative result of making it clear that not providing a rendition immediately is functional to providing an appropriate rendition of what was said. Mason (2012) and Pasquandrea (2011) reach similar conclusions on the basis of videorecorded data, where gaze and posture seem to confirm the coordinative function of this suspending mechanism.

This more nuanced way of looking at dyadic sequences is supported by data showing that when participants are truly excluded from participating in the interaction, this exclusion is considered unacceptable in the interaction and is resisted. To illustrate this, let us examine Extract 3, from the doctor-patient mediated corpus. Here, the patient, a Moroccan woman undergoing pregnancy screening, complains of her belly bloating and tells the mediator that the pregnancy is causing intestinal gas. The mediator - here, as in many other cases - responds to the patient by asking for more specific details: in particular, she asks whether gas is eventually expelled or not. While in many cases, such queries have the effect of putting the patients at ease and get an account of their symptoms (Baraldi 2012), in this case the patient insists on her claim that intestinal gas is caused by pregnancy. In the extact below, we see the mediator’s feedback to the patient in turns 1-5 and the patient’s insistence in turn 6. In turn 7, the midwife intervenes signalling she is ready to listen and the mediator renders the patient’s claim. The midwife’s confirmation that this is a likely symptom during pregancy finally closes the sequence (turn 12), suggesting that the patient’s insistence was directed at selecting the midwife, not the mediator, as her primary interlocutor.

3. [Natural_Data]

In conclusion, while it is true that dyadic sequences create some risk of excluding one interlocutor from the interaction, as extract 3 shows, this risk can be managed in the interaction itself. The data show, in short, that participants in DI interactions do deal with the (non) relevance of dyadic sequences and accept or do not accept them accordingly. Learning how to momentarily suspend translation and how to return to it are thus important aspects of DI training.

2.3. “Giving voice” to the participants

One problem in DI is that language mediation may end up by substituting, paradoxically, active participation by one or the other parties to the interaction. This point is highlighted in Davidson (2000), who notes that patients’ narratives of their worries may be cut short in interpreting, and by Bolden (2000), who shows that in translating from a lay to technical register, interpreters sometimes overinterpret. In DI, in particular, the distribution of access to talk and recognition of other participants’ rights and obligations to talk is complex: while other participants are institutionally expected to bring their own contents and purposes in the interaction, mediators and interpreters are not, their institutional goal being that of conveying the other participants’ contents and purposes effectively. It follows that mediators and interpreters are institutionally required to give the participants voice by acting in their place in another language. DIs must thus learn to recognise contributions by other participants that invite them to refrain from substituting others or, instead, legitimise them to do so. An example of the former was the patient’s insistence, in 2.2, above, on obtaining a response directly from the health professional. An example of the latter are mechanisms that authorize interpreters to speak on behalf of another participant. One such mechanism is “diglielo” (“tell him/her”). In the doctor-patient interactions described above, this expression is used to introduce issues of particular complexity or delicacy, for example when health professionals are about to alert patients to particular problems or explain complicated procedures (see Gavioli 2015). “Diglielo” is also found in other types of Italian-English mediated data when one speaker is asking the interpreter to render what is being said with particular attention, highlighting the purpose of what is said. In the following extract (the continuation of the trade fair sequence shown in extract 1), the exhibitor is alerting potential customer that brake pads need to be used only with appropriate wheels, as otherwise they may be very dangerous:

4. [Natural_Data]

So “diglielo bene”, which the seller uses in his turn, seems to authorise the interpreter to take responsibilty and translate the contents in a way as to make the point absolutely clear: it is dangerous to utilise these brake pads on bicycles for which they were not designed. Through this authorisation mechanism the interpreter and the exhibitor align “as a team” and the interpreter is allowed to choose what s/he considers the most appropriate way to highlight particular information. Consequently, learning to recognise these mechanisms and responding appropriately to them is important in DI learning. The three aspects highlighted above show how fundamental  coordinating participation in DI is, and thus constitute possible key learning outcomes to be pursued in constructing teaching/learning materials. Creating data-informed materials is, however, not easy, and requires further exploration.

3. Learning DI: a pseudo-authentic task

3.1. Preamble

Corpus-based studies have amply shown that intuitions about language use do not match actual language use. Work within the PIXI project has shown that conversation mechanisms are complex and suggested that language users are seldom aware of the work it takes to accomplish actions and goals interactionally (see Aston 1992). And this is certainly the case for DI interpreting, where what students intuitively expect is often very different from actual DI occurrences. So some of the tasks I utilise with my students are oriented to helping them compare their expectations about DI with suggestions provided by authentic occurrences.

The following is a roleplay I use in my 3rd year undergraduate class. I provide the students with a detailed, written description of what is taking place in a DI sequence I select for them from my corpora. I ask them to work in groups of three and act out a dialogue, each of them taking the role of a participant: the English speaker, the Italian speaker or the interpreter. Their roleplay is recorded with one of the students’ smartphones and the file is shared by the three participants. After completing their roleplay, the students are given the transcript of the original DI event to compare it with what they did; they are asked to discuss the main differences or similarities they can see. This phase shows the students that the expectations that they had about the encounter, even if based on a very detailed description of it, do not match with what occured, as evidenced in the transcript.

I then ask the students, individually (and usually as a homework), to listen to the recording of their roleplay and transcribe a sequence that they find interesting in comparison with the original transcript. They can select any segment, provided that they explain why they found it interesting. Finally they send me their selection, together with their written explanations of why they found the selected sequence interesting. Normally this gives me more than enough material for a followup lecture, in which I choose and discuss some of the students’ transcripts. In the discussion, I highlight: a) that although all the groups had the same instructions, none of them acted out the same conversation; and b) that none of the conversations recorded by the groups were the same as the original conversation, even though all are based on a very detailed description of the sequence. This allows me to show that intuitions about how DI encounters may occur, and indeed most briefings given to interpreters, do not help much when interpreters are actually engaged in a specific, real-life DI encounter. Below, I deal with some reflections raising from the students’ comments and the transcripts of their roleplays.

3.2. The Balsamic Vinegar case: observation of a roleplay

The roleplay transcripts produced by my students, as well as simple observation of the groups at work, in which three speakers of English and Italian pretend not to share the languages used and rely on interpreting, clearly show that playing out a real situation provides absolutely no guarantee that the reality of that situation will be reproduced; quite the opposite, it brings home to students  that “reality does not travel with the text”. The roleplay and the work that was conducted on it are thus interesting from a teaching/learning perspective. Here, I provide an account of my observations of the task described in the appendix, based on a transcript taken from guided tours to facilities in which balsamic vinegar – an important export product in the Modena area – is produced. I will first share reflections based on the students’ comments and, second, on their transcribed roleplays.

3.2.1. The students’ comments

A first reflection relates to  the students’ comments about  differences between their interaction and the “original” one. In comparing the two, the students observed a number of characteristics of the “original” which contrasted with their expectations. Some included the organization of talk and of interpreting. For instance, in contrasting their roleplays with the transcript of the authentic guided tour, the students noted that the exhibitor’s and the interpreter’s turns are much longer than they expected, that the questions from the visitors are many fewer (even though the task instructions mentioned that there had only been a couple of questions and that they had all occurred at the end of the presentation). The students thus realised that these product presentations were not organised as they had expected, but rather conjointly managed by the exhibitor and the interpreter, who worked “as a team” and seemed to deal “together” with the visitors’ questions. Other observations on the students’ part concerned the competence of the participants, particularly the interpreter’s. For instance, they noted that the interpreter seemed to be appreciated by the guide, who says “è giusto” o “esatto” when she speaks. They additionally noted that the interpreter explains “very well”, sometimes better than the guide, by clarifying the details and adding some of her own. They observed that the interpreter sometimes asked the guide questions in order to get clarification or confirmation and that she sometimes answered to the visitors’ questions without getting back to the guide. Although finding these actions communicately effective, the students commented that they had not expected them to occur in DI interaction.

As regards their own transcriptions, there are two interesting recurrent comments from students about acting as interpreters. The first, probably more obvious, is that it is not easy to select the correct language with the correct interlocutor and that, in acting as interpreters, they had inadvertently spoken English with the intended Italian participant and Italian with the intended English one. The second observation was that the less they understood or were familiar with what was being said by their colleagues, the more they tried to render the previous turn(s) literally – in other words, when their understanding of what was said was low, they tended to translate word-for-word. Two teaching/learning points arose from these comments. The first is that activities which train the students to switch language appropriately may be a real need in DI learning. The second is that the students came to realize that word-for word translation is a weak, not a strong, solution and that it may derive from poor understanding rather than from strong bilingual ability.

3.2.2. The students’ transcripts

A look at the students’ transcripts revealed characteristics of “acted” dialogue, since obviously all of the students understood what was said by the others and all had the same information about the encounter. Some characteristics of their interactions are, however, suggestive from a teaching/learning point of view. The first is that even though the students acted out the dialogue, they dealt with the presence of interpreting in it and showed a recognition that this presence needed to be coordinated in the ongoing interaction. For instance, the transcripts show that the students negotiate when it is time to talk and when it is time to translate. For example, in the following two sequences, taken from the students’ transcriptions of their roleplays, one of the participants (5) and the interpreter (6) claim time to continue their talk or to translate:

(5) [Roleplay]

(6) [Roleplay]

A second point that emerges from these transcripts is that when a speaker hesitates or does not know what to say or how to say it, student-interpreters intervene in help of fellow students in order to allow talk (and the provision of a rendition) to go on. Observe, for example, extract (7), in which the student-visitor indicates that she is unable to continue and the student-interpreter elicits her to produce more questions. This is a type of occurrence which lead to dyadic sequences and that is also frequent in natural data.

(7) Roleplay]

A third observation arising from the roleplay transcripts is that, although there was no real “information gap” in this task since all the students had the same details about vinegar production, the transcripts show that two other “gaps” were present. First, the students did not know precisely how the student who was acting as the guide would present these details. So the rendition was in fact occasioned in the roleplay in response to how the product was  actually presented by the student-guide. The second is that, while all of the students involved knew both languages, none of them was expert in the rendition of a description of balsamic vinegar so they collaborated in producing the description and in rendering it in the other language by adding or asking further details and assessing those provided. This is illustrated in (8) and (9), below. In (8), the Student-Interpreter has obvious difficulties in rendering the previous turn and the Student-Visitor intervenes by suggesting that she specifies a detail that they all know about, but that is missing in the Student-Interpreter rendition:

(8) [Roleplay]

In (9), below, it is the Student-Guide, who, after the Student-Interpreter rendition is signalled as complete (see turns 07 and 08 where understanding is checked and confirmed), provides a further detail about how balsamic vinegar is produced. In this case the detail is rejected as “not a missing one” (see laughter in turns 10-12).

(9) [Roleplay]

Finally, looking at the students’ renditions, it is clear that most are turn-by-turn literal renditions of the previous turn. But there are cases in which information is rendered more explicitly or in which what action is being performed is clarified. In all these cases, the language being used in the rendition turn is the students’ mother tongue. Here are two instances:

(10) [Roleplay]

(11) [Rolpeplay]

These occurrences seem in line with the students’ comments that literal rendition is more frequent when language competence is not so solid. The data suggest/show that when they feel at ease with the language they are speaking (e.g when rendering into their native language Italian, as in the examples above), the students may experiment with types of renditions which are of a more complex kind and which involve explicitation of elliptical forms or of the original speaker’s pragmatic purposes. In short, even in a simple roleplay with basically no information or opinion gap, “doing rendition” is an activity which involves the students and gives them some opportunity to negotiate how to deal with interpreting in the interacion by activating coordinative devices connected to turn-taking, eliciting participants’ talk, contributing to furnishing material for renditions and modifying previous talk to make its pragmatic purposes explicit.

4.Conclusion

Drawing from what we have seen above, I suggest that the use of authentic data for (more) authentic DI learning/training involves choices at multiple levels. First, we have “the data”, authentic transcripts of DI encounters. These collections, often limited in size, are not necessarily representative of “the DI real world”, yet they provide multiple occurrences of mechanisms which are observable across the encounters and across the settings in which the encounters were collected. As shown in this and other studies (Davitti & Pasquandrea, 2014; Cirillo & Niemants, 2017), familiarising novice dialogue interpreters with some of these mechanisms seems a useful learning/training objective.

Second, authentic data are a resource for classroom material design. Selecting “which data” is itself a teaching/learning issue worth exploring. Aston distinguishes between two key modes of approaching interaction in the classroom; observation and practice (Aston, 1988a). The transcript I selected for my roleplay provided useful material for students to engage in both activities. They could see how a product presentation may be handled in DI, how team-work may be achieved between the interpreter and the guide, and they noted some authorizing mechanisms, for instance that the guide praised the interpreter by saying “esatto”, thereby explicitly approving her rendition choices. Moreover, the students directly experienced some problems related to the interactional organization of DI, e.g. that of negotiating sufficient time to translate. The selected transcript did not, however, give learners opportunities to observe, discuss or practice the management of dyadic sequences, for which other data probably need to be selected and activities designed and tried out.

An issue worth further experimentation is how to most successfully design tasks based on authentic data. While giving students some practice in DI, the roleplay discussed above did not give them opportunities to focus on certain issues that seem to characterise naturally-occurring DI data: while communication about interpreting occurred, communication through interpreting was limited. To stimulate communication through interpreting, it is likely that different activities and tasks need to be constructed. For instance, in order to train interpreters to render communicatively, Turner & Merrison (2016) suggest using the well-known map-task. In their version of this task, one student describes a route on a map and the other draws the route on a blank piece of paper, following interpreted instructions. The task, thus proposed, seems to be effective in challenging the “interpreter” to ask for and give clarification. Krystallidou (2015) suggests tasks based on interpreter and medical doctor joint training, where medical and interpreting students are engaged in simulated interactions with patient-actors.

Last but not least, whatever type of communication we are interested in simulating or discussing with students, the fact that simulations and discussions occur within the context of classroom communication should not be overlooked. Aston (1988a) distinguishes between “construction” and “deconstruction” activities, the former engaging learners primarily in practising the interaction, the latter involving them as “observers” of the interaction process. In a classroom setting, like the one described above, the deconstruction component, based on communicating about (DI) communication, is a crucial one in facilitating learning, for at least two reasons. The first is that classes do not - and probably should not - substitute a real working setting and it makes sense in class to train students in skills they are likely to use in their (future) working settings. In this sense allowing them to compare “real work experiences” like those displayed through natural data with their own expectations about these experiences is useful, not to show that the students’ expectations are wrong, but rather to prepare them for noticing in the future how things are or may be done, thereby gleaning suggestions (not necessarily “models”) from real life interpreting. The second is that (dialogue) interpreters are particular types of participants, ones who engage in interaction first and foremost to make sense of what is said by others: they listen, in other words, with the expectation that what they will say will consist in a rendering of someone else’s contribution in the other language. Thus, an acquaintance with the dynamics which generate meaning in the interaction may be of special interest to students of interpreting and mediation.

In conclusion, DI data are an important input for “authentic” DI learning, but on their own are not enough. Research on DI learning needs not only to clarify characteristics of DI talk but also to explore what tasks can make these characteristics “real” for the students, through both practice and discussion.

About the author(s)

Laura Gavioli is Professor of English language and translation at the University of Modena and Reggio Emilia, Italy. Her research work includes the study of spoken language in institutional settings, corpus studies for language and translation learning, and the pragmatics of English-Italian interaction in intercultural settings. Together with Claudio Baraldi, she edited the volume Coordinating Participation in Dialogue Interpreting (John Benjamins, 2012). She has published numerous papers (a recent one in Journal of Pragmatics, 2015). She has contributed to the Routledge volume Researching Translation and Interpreting (Angelelli and Baer eds. 2016) and to the Routledge Encyclopaedia of Interpreting Studies (Pöchhacker ed. 2015).

Email: [please login or register to view author's email address]

•  Previous article
• Next article