Abstract

In less than two decades interpreting corpora have covered a wide range of interpreting settings and language pairs. Corpus-based studies have also been enriched by multimodal data, addressing both verbal and embodied dimensions of interpreter-mediated interactions. This paper contributes to expanding this corpus-based multimodal analytical approach by examining the dynamics of chunking, the practice of dividing extensive turns in consecutive interpreting. In a speech delivered to a foreign audience, chunking is essential in facilitating the rendition in the target language. Yet, it may potentially interrupt the guest speaker’s speech flow on stage. Underpinned by a small-scale corpus featuring Chinese–Italian consecutive interpreting delivered in public events, the analysis investigates how the interpreter and the guest speaker navigate this conflict and how the situational and cultural contexts impact their actions. The paper begins with an overview of the multimodal approach in interpreting studies, focusing on chunking and consecutive interpreting in public events. Subsequently, data collection and transcription methods are outlined. Drawing on the multimodal approach of Conversation Analysis, we explore how chunking is managed by participants who employ various semiotic resources. Our results reveal an overall collaborative framework characterized by mutual monitoring and signalling between the speaker and the interpreter, with diversified approaches adopted by the participants. The reasons behind their choices are also discussed in relation to the contextual features of the communicative event.

Keywords: consecutive interpreting, chunking, interpreting for public events, corpus-based interpreting studies, multimodal analysis

1. Introduction^[1]

Anecdotal evidence of the interpreter’s performance has long been replaced by a different approach, namely the corpus-based approach advocated by Shlesinger in her 1998 seminal paper. Drawing from the Corpus Linguistics (CL) methodology already embraced in Translation Studies (Baker 1993), she envisaged the potential of the ‘creation of parallel and comparable corpora comprising discourse which is relevant to interpreting and of the use of existing monolingual corpora as sources of materials for testing hypotheses about interpreting’ (Shlesinger 1998: 486) to advance Interpreting Studies. The availability of large interpreting datasets could enable the identification of general trends, thus contributing to overcome narrow approaches in interpreting research fields, such as research on quality in interpretation carried out back then and stigmatised by Francesco Straniero Sergio (2003: 135) as follows:

Research on quality in simultaneous interpretation seems polarised, if not paralysed, in the following two directions:

1) Speculative studies that indicate how a text ‘can’, ‘should’ or ‘should have been translated’

2) Experimental or case-study studies analysing individual performances in the light of pre-established quality criteria.                                         

The access to millions of written, published texts has certainly favoured the earlier development of corpus-based translation studies (CTS) vs corpus-based interpreting studies (CIS), given the lack of accessible large sets of source and target speeches and the extremely time-consuming transcription of oral data. It was not until mid-2000 that the first interpreting corpora based on CL methodology and tools – for example, Part-of-Speech tagged, lemmatised, tokenised, and indexed corpora for semi/fully-automatised queries – appeared (Bendazzoli 2018; Setton 2011). This can be considered the final innovative step of the evolution of interpreting data collections stored in digital format ranging from one single/some interpreter-mediated events to Straniero Sergio’s CorIT Corpus including the largest ever compilation of 1200 interpreters’ performances on Italian television (Straniero Sergio 2007; Straniero Sergio and Falbo 2012).

The number of machine-readable interpreting corpora have increased by the dozens in the subsequent years and several of them are presented in co-edited publications, such as Straniero Sergio and Falbo (2012), Bendazzoli, Russo and Defrancq (2018), Russo, Bendazzoli and Defrancq (2018), Kajzer-Wietrzny et al. (2022). Their authors have investigated a wide range of research topics such as disfluencies (Bendazzoli et al. 2011), cognitive load (Defrancq and Plevoets 2018), ‘interpretese’ (Aston 2018; Kajzer-Wietrzny 2018,), fluency parameters (Chmiel et al. 2022) and so on. More recently, a review of interpreting corpora revealed the existence of 125 parallel corpora in 38 and 8 signed languages (Liu and Russo 2025).

Interpreter-mediated events collected in interpreting corpora cover many different settings and interpreting modes. The vast majority of them investigate conference interpreting (consecutive and simultaneous) based on data coming from different sources, in particular the European Parliament sessions, a formidable multi-lingual resource providing both source and target speeches, as well as from other political assemblies, such as the Chinese/English Political Interpreting Corpus (CEPIC) counting millions of words (Pan 2019) or  political press conferences, especially from China (among others, Wang and Zou 2018). All these venues provide publicly accessible interpreting data to an unprecedented extent. Not only are conference interpreting corpora increasing in number, thus providing material for a growing number of publications, but dialogue interpreting corpora too are steadily developing, thus expanding CIS (among others, Mason 1999, 2001; Angelelli 2004; Pöchhacker and Shlesinger 2007; Gavioli 2012; Angermeyer, Meyer and Schmidt 2012). These shed light on different aspects of the interpreter’s role and agency, such as interaction with the other participants, participation status, meaning co-construction, role-conflict or gate-keeping (see further on). The analysis of dialogic settings has been further enriched by multimodal corpora investigating embodied interactions which complement verbal exchanges. Elements of multimodal analysis include the impact of gaze, gestures, facial expressions, body posture and object manipulation, which have become fully-fledged ‘tokens’ of manually or machine-readable dialogue interpreting corpora (for example Pasquandrea 2011; Davitti 2013; Dong 2022). Furthermore, to better account for such embodied interactions, new theoretical and methodological frameworks have been proposed to analyse how verbal and nonverbal resources influence shifts between source and target texts in machine-readable corpora (Gao and Wang 2017).

This paper contributes to expanding this corpus-based multimodal analytical approach by providing evidence of some relevant features of chunking, that is, the practice of dividing extended turns into manageable units, in consecutive interpreting for public events (CIPEs). These emerge from a small-scale corpus featuring non-Eurocentric embodied interactional data in a ritualised setting, where two markedly different cultures, discursive practices and mindsets come together, namely the Chinese and Italian ones.

The paper begins with a brief overview of the interactional and multimodal approach to study an interpreter performance (Section 2), with a special focus on consecutive interpreting in public events (Section 3); it then presents the data and methodology in Section 4, the analysis in Section 5, the discussion of results in Section 6 and some conclusions in Section 7.

2. Multimodality and chunking

In the history of Interpreting Studies, investigations on interpreting activity have undergone several paradigm shifts (Pöchhacker 2004). Of particular interest in this respect is the rise of the interactionist and discourse-analytical approach, which started to gain popularity some twenty years ago (Mason 1999; Roy 2000; Straniero Sergio 1999; Wadensjö 1998). Informed by Sociolinguistics, Discourse Analysis and Conversation Analysis, which emphasise the situated and co-constructed nature of communication “(Goodwin and Duranti 1992; Sacks, Schegloff and Jefferson 1974), more and more studies have addressed the interpreter’s role and participation in ever-changing communicative situations “(Baraldi and Gavioli 2012; Diriker 2004; Mason 1999, 2001).

The context-sensitive and jointly-achieved features of interpreting activities have been further highlighted in the multimodal analysis of video data “(Davitti 2013, 2018; Salaets and Brône 2020), a more recent direction of study drawing on the multimodal approach of Conversation Analysis (Deppermann 2013; Goodwin 2000; Mondada 2016). The interplay of verbal and nonverbal semiotic cues, or the ‘triple audiovisual reality made up of verbal language, paralanguage and kinesics’ as Poyatos (1996: 249) puts it, pervades interpreting activity. Gaze, for instance, plays a fundamental role in monitoring the ongoing interaction (Tiselius and Dimitrova 2023). Shifts in gaze and body direction are proven to be effective turn-regulatory cues in face-to-face dialogue interpreting. At turn transition-relevant places, the primary parties and the interpreter may establish eye contact to hint to each other at the timing of turn-transition (Davitti 2013, 2018; Krystallidou 2014; Mason 2012; Pasquandrea 2011, 2012; Vranjes and Bot 2021; Vranjes and Oben 2022; Wadensjö 2001). Moreover, gaze and body orientation are also indicators of addressee selection and engagement (Krystallidou 2014, 2016; Mason 2012; Vranjes, Brône and Kurt 2018; Vranjes et al. 2019). Thus, when turning towards or looking at the interpreter, the primary parties recognise the latter as a ratified co-participant in the interaction (Davitti 2016; Pasquandrea 2011, 2012). The interpreter may directly look at primary participants to draw their attention and show reciprocity (Davitti 2013, 2016) or provide affiliative backchannels, such as minimal responses (Gavioli 2012) and nods (Vranjes, Brône and Kurt 2018; Vranjes et al. 2019). However, Wadensjö (2001) also noticed that the seat arrangement affects the participant’s gaze behaviour, underlining the necessity of analysing the specific configuration of the communicative event.

A peculiar practice often managed through gaze and bodily conduct is the chunking of extended talks. It involves two actions (Licoppe 2023: 159): the first consists of the splitting of the source talk into smaller units to provide a chance for the interpreter to deliver the rendition; the second occurs when the interpreter concludes their turn, as a speaker is expected to gain the floor. The interpreter has been found to take the initiative in this process (Licoppe and Veyrier 2020; Vranjes, Brône and Kurt 2018). However, their attempt may also fail because of conflicting communicative needs of the primary parties (Vranjes and Bot 2021). Therefore, as a complex activity, it has to be ‘tacitly negotiated on a moment-by-moment basis through verbal, prosodic and embodied resources’ (Davitti 2018: 14).

Chunking is not confined solely to the dialogical form. Licoppe (2023: 159) indeed refers to it as ‘characteristic of consecutive interpreting’. In public events, the stage is generally reserved for one guest speaker at a time. However, this does not imply that floor management is unnecessary in CIPEs; on the contrary, the speech may be viewed as an exceptionally extensive and multi-unit turn that must be ‘chunked’ to allow for its rendition.

3. Interaction in public events: what corpus-based multimodal analysis can reveal

Parallel to the interactionist research on dialogue interpreting, a relatively small body of literature has adopted a similar approach to address consecutive interpreting delivered in public events, a one-to-many form of communication.

Based on data collected in an academic conference interpreted in simultaneous mode, Diriker (2004: 82) argued that interpreting is an ‘activity in context’: even when the interpreter was confined to the booth, their renditions and behaviours were still linked to the real-time dynamics in the conference hall and to the broader social context. Napier (2007) provided evidence of cooperative actions carried out by a deaf guest speaker, an active interpreter and a support interpreter to deliver a speech to an English-speaking audience. The author highlighted how the participants mobilised nonverbal cues to coordinate the talking pace and check on comprehension.

Turning specifically to the consecutive mode in CIPEs, Errico and Ballestrazzi (2014: 378) reported the challenges that the interpreter faced when the speaker displayed ‘a markedly dominant interactional status’ and a playful expressive approach. In the highly dynamic unfolding of the public speech, the interpreter had to continuously alternate between the translating activity and the co-acting task (Straniero Sergio 2012) to comply with the ‘ethics of entertainment’ (Katan and Straniero Sergio 2003). In another case of a public literary event presented by Poignant and Wadensjö (2020), the interpreter did not simply translate the content to the audience but reframed it in a more engaging contour through rhythmically marked prosody and gesture. The boundary between the original and the rendered utterance was blurred: the speaker’s and the interpreter’s contributions co-formed a unique onstage performance. These studies shed light on the situated dimension of mediated communication and the interpreter’s complex role in CIPEs. The relevance of nonverbal resources such as gaze and prosody shift in the speaker-interpreter interaction was also stressed. However, despite its rich and in-depth analyses, research in this direction has been mainly focused on single case studies and language pairs within the European cultural background.

In a recent investigation, Li, Liu and Cheung (2023) explored the multimodal interactions between speakers and Chinese-English interpreters in institutional press conferences. The analysis provided further evidence of the collaborative actions jointly undertaken by the speaker and the interpreter through verbal and nonverbal cues. In particular, the interpreter acted as an ‘icebreaker’ or ‘rectifier’ (ivi: 10) by reacting to the speaker’s involvement or rephrasing the rendition according to the speaker’s prompting. Yet, it is worth noting that the study involved a peculiar group of speakers who were also fluent in the target language. By adopting a corpus-based approach, this study demonstrated the interpreter’s contribution to the co-construction of meaning in this type of setting.

In the present contribution we set out to investigate a different yet representative situation in which the speaker and the audience do not share a common language. Here, the interpreter enjoys a bilingual advantage, and therefore, they could potentially take on coordinating responsibilities such as chunking. However, in the light of the fact that in public events, the guest speaker enjoys a privileged position, or better, the ‘speakership’, chunking could be seen as an interruption of their speech flow. This raises some key questions: How do speakers and interpreters manage this conflict? Who takes the initiative? Are there any factors that could impact their actions?

To our knowledge, there are still limited datasets in the literature on this issue. Therefore, underpinned by a multimodal frame and a small-scale corpus of CIPEs data, the present paper attempts to provide a preliminary account of the chunking dynamics and to reflect on the implications of the situational and cultural contexts, as advocated by Gao and Wang (2017).

4. Materials and methods

Even though there are mixed conceptualisations of representativeness in corpus design (cf. Egbert et al. 2022), corpora as large as possible are generally considered representative as they enable reliable estimates of frequency and distribution of an investigated feature in a given larger population, which is what an anecdotal case-study approach cannot possibly deliver. Egbert et al. (2022: 11) provide an operational definition of representativeness: ‘[it] refers to the extent to which a corpus permits accurate generalizations about the quantitative linguistic patterns that are typical in a target language or discourse domain’. As a continuous construct, representativeness is high when corpus designers tackle (1) domain considerations – (1a) define the (operational) domain boundaries (for instance ‘ZH-EN dialogue interpreting in the Chinese context’ vs ‘dialogue interpreting’) and major categories (such as facework strategies in interpreter-mediated cross-examinations), (1b) use a sampling method (for example random, convenient or purposive) to select from a list of mediated situations in the sampling frame –, and (2) distribution considerations, namely identify the corpus size required for the precise measurement of a feature’s frequency (Egbert et al. 2022). Given the well-known difficulties of finding, accessing and analysing real dialogue interpreting data (Bendazzoli 2017), the literature on corpus-based interpreting studies also features small-scale corpora – among which Liu and Hale (2017), analysing five bilingual moot court cross-examinations, or Määtta (2017), analysing one telephone interview of one hour 46 minutes –, which have the merit of being in-depth, theoretically and methodological sound studies providing insights, evidence-based trends and viable constructs for further investigations.

In our case, the research data was extracted from a larger corpus of CIPEs between Italian and Chinese, in which subjects’ verbal and nonverbal resources were transcribed in ELAN to be queried using CL tools (Wang 2022). We delimited the domain boundaries (1a) as follows: ZH-IT consecutive interpreting in public events posted on YouTube and Weibo. Next, we applied the following sampling method (1b): to highlight the nonverbal communication between the speaker and interpreter, we excluded events in which the speaker was holding a script and was reading from it, as the interactional gaze movements and gestures could not be compared to naturally occurring interactions. The speaker and the interpreter had to be positioned side-by-side or at a close distance so that one could have visual access to the nonverbal features of the other. A fair number of formal events with politically relevant speakers failed to correspond to these criteria. Moreover, events with more guest speakers were also excluded, since the co-presence of multiple speakers onstage can significantly alter the chunking and turn-transition dynamics. As we had no pretentions to providing reliable generalizations of the features’ frequency given the limited dataset we managed to collect in this domain, we did not consider identifying the corpus size required (2).

Six events meeting the above-mentioned criteria were selected to build the final research corpus. The speeches were delivered and interpreted by different speakers and interpreters, with a total duration of 175 minutes. Occasionally, a moderator was present onstage during some events, but their interventions, generally minimal, were not included in the analysis or in the overall video duration count. The name and general information of each event are provided in Table 1.

For practical reasons, source speakers and interpreters are named in the following sections by the number assigned to the respective event. For instance, ‘Speaker-I’ and ‘Interpreter-I’ refer to the participants of Event-I, ‘Meeting with film director JZK’.

Table 1: List of events

Drawing on the multimodal approach of Conversation Analysis “(Deppermann 2013; Mondada 2018; Stivers and Sidnell 2005) , the data were transcribed using the software ELAN^[2] (Wittenburg et al. 2006) (for the transcription conventions, see Annex 1). Four tiers were created for the speaker (SP) and the interpreter (INT) respectively (Fig.1). The first tier included propositional content and the main paralinguistic features (for example pauses and stressed tone). The second tier contained the English translation. In the third one, sideway gazes were annotated; therefore, a blank space in this tier indicated that the participant was looking frontally towards the audience. The fourth tier included other body movements, such as head movements and gestures. In total, 975 instances of chunking were identified and coded in ELAN, among which 487 initiated by the speaker and 488 initiated by the interpreter (Table 2).

Fig. 1. Example of the multi-tier transcription

Table 2: Chunking count

5. Analysis

In line with the communicative norms in public events, our video data confirmed the leading role held by the guest speaker in chunking. Namely, the speaker stopped when deemed appropriate and left the floor to the interpreter. Yet, the micro-analysis of the floor-transition moments disclosed a collaborative framework, in which both the speaker and the interpreter were inclined to monitor each other and signal the end of their turn to the other. Such hinted chunking actions were seen in over fifty per cent of the total transitions. Both verbal and nonverbal resources, such as the Chinese discourse marker ng, eye contact, gestures, body movements and object manipulation, were exploited in this collaborative framework.

In the following analysis, we selected the most representative and relevant examples of each distinct practice, seeking at the same time to provide at least one example from each event.

5.1 Speaker-to-interpreter transition

The speaker tapped several resources to signal the end of the ongoing speech chunk to the interpreter. At times, speakers slightly turned their heads towards the interpreter, but without eye contact. On other occasions, they directed their gaze to the interpreters until eye contact was established or feedback was provided.

Excerpt 1 exemplifies the second case in which Speaker-III turns to the interpreter when she is about to leave the floor. She kept her eyes on the interpreter until the latter nodded back and began her rendition (Fig. 2).

Excerpt 1 – Sideways gaze and head movement as the speaker’s cue

Figure 2. Sideways gaze and head movement as the speaker’s cue

Signalling practices were not restricted to the kinetic domain, but speakers were also found to pass the floor with vocalised signals. Both in Event-III and IV, speakers often closed their turns by uttering ng (yeah), a Chinese discourse marker which adds a conclusive tone to the utterance. This was particularly relevant in Event-IV, as the guest speaker’s body movements were restrained: gaze and head turned towards the interpreter were observed only three times out of all the 94 turn-transitions from the speaker to the interpreter, whereas the vocalised token, the ending ng, occurred 48 times.

Excerpt 2 represents an example of Speaker-IV, who concluded his talk with a final ng and then moved away from the microphone (Fig. 3). At almost the same moment, the interpreter stopped taking notes: she nodded and then took the floor.

Excerpt 2. Discourse marker as the speaker’s cue

Figure 3. Discourse marker as the speaker’s cue

On numerous occasions, speakers did not turn towards or gaze at the interpreters at the end of their chunks of speech. However, some behaviours repeatedly performed before leaving the floor, such as moving away from the microphone or gesture retraction to return to a ‘standby’ posture, may also have served an indirect signalling function.

In Event-V, the speaker has to manage a multimedia presentation from a laptop on his left. As illustrated in Excerpt 3, as a sign of chunking, he looked away from the audience to focus visually on and manipulate the laptop (Fig. 4). Though such cues were not addressed to the interpreter, they seemed somehow ‘choreographed’ with the interpreter’s actions, as the latter took the floor without hesitation.

Excerpt 3. Object manipulation as the speaker’s cue

Figure 4. Object manipulation as the speaker’s cue

Speaker-VI showed another peculiar signalling mechanism. When concluding his chunks of talk, he was often seen to hold the last gesture or posture performed instead of returning to a relaxed position, as if he literally put the speech on pause. Excerpt 4 illustrates that, in introducing Leonardo da Vinci’s code to the audience, Speaker-VI held a notebook in his hand to show the approximate aspect. Right before he left the floor, he first stressed the holding gesture by slightly shaking the notebook and then continued to keep his hand aloft (Fig. 5). What is also worthy of notice is the backchannel of the interpreter: almost at the same time, she uttered an affirmative ng (well, yeah) as a token of her readiness to take the floor.

This co-managed practice was also spotted on several other occasions. At times, the interpreter’s reaction was so prompt that it became difficult to establish whether it was the speaker who left the floor first or the interpreter who took the initiative. Nevertheless, we still consider it an occurrence of indirect signalling activity by the guest speaker, as the ‘on-pause’ practice occurred naturally and routinely throughout the event.  

Excerpt 4. ‘On pause’ chunking

Figure 5. ‘On pause’ chunking

When looking at Table 3, it becomes evident that in all six events, more than half of the instances of chunking were either directly or indirectly hinted. Speaker-I, II and III relied primarily on shifts in gaze and head orientation. In contrast, the primary chunking cues observed in Event-IV, V and VI were less apparent and did not seem to be explicitly addressed to the interpreter. Yet, these actions still disclose the speaker’s awareness of sharing the stage with a co-performer, that is the interpreter – and the intention to leave the floor to them. Despite being indirect and subtle, these cues were effective, largely due to the interpreter’s close and consistent monitoring. The physical proximity between them also facilitated the reception and perception of nonverbal features.    

Table 3: Speaker-to-interpreter transitions

5.2 Interpreter-to-speaker transition

A common practice adopted by interpreters was orienting their heads towards the guest speaker at the end of their rendition. Excerpt 5 from Event-I provides a case in point: when concluding the rendition, the interpreter directed his head and gaze towards the speaker, who immediately took the floor.

Excerpt 5. Sideways gaze and head move as the interpreter’s cue

Figure 6. Sideways gaze and head move as the interpreter’s cue

Another aspect worth noting in this example is that, during the rendition, Speaker-I kept his head oriented towards the talking interpreter and closely looked at him. Eye contact was also briefly established when the interpreter turned towards the speaker at the end of his rendition (Fig. 6).

What we witness here is an instance of the speaker’s monitoring. As the speaker was unfamiliar with the target language, he needed more than linguistic elements to perceive the end of the rendition. Therefore, it was to be expected that the speaker would look for visual clues from the interpreter. However, publicly performed monitoring behaviour by the speaker, as shown in Excerpt 5, was observed only 18 times and limited to three of the six events. Conversely, it was much more common to observe the speaker’s prompt reaction to chunking cues without directing their sightline to the interpreter.

This is illustrated in Excerpt 6. In this event, the Event-II, speaker and interpreter were standing next to each other. Thanks to this close interpersonal distance, when the interpreter turned towards the guest speaker at the end of his rendition (Fig. 7), the latter took the floor without hesitation even though she was looking at the audience.

Excerpt 6. The interpreter’s cue and the speaker’s floor-taking reaction

Figure 7. The interpreter’s cue and the speaker’s floor-taking reaction

Such discreet monitoring is not easy to discern and prove. Nevertheless, in Excerpt 7, it is possible to flag the speaker’s attention on the interpreter due to a late chunking cue from the latter. After closing her rendition, the interpreter remained gazing at the audience without retracting from the last gesture performed (Fig. 8). It became challenging, at this point, to distinguish whether this was a pause or a turn transition-relevant place, especially for the speaker who did not speak the target language. The speaker, therefore, hesitated to gain the floor: he briefly opened his mouth but then closed it immediately. Only after 1.2 seconds, when the interpreter turned towards him, did he finally take the floor. In this case, despite the speaker’s diverted gaze, his turn-taking actions still seemed to be influenced by the interpreter sitting beside him.

Excerpt 7. The interpreter’s late chunking cue

Figure 8. The interpreter’s late chunking cue

Akin to the indirect signalling practice observed in speaker-to-interpreter transitions, the end of the rendition may also be pinpointed by other non-interactional cues, such as changing one’s posture, which turned out to be equally effective in prompting the speaker’s reactions. Excerpt 8 shows an instance in which Interpreter-V assumed a notetaking posture at the end of her rendition (Fig. 9).

Excerpt 8. Interpreter’s indirect signalling

Figure 9. Interpreter’s indirect signalling

Overall, the interpreters’ chunking cues are less diversified. As evinced by Table 3, Interpreter-I, II and III are more inclined to sideways gaze and head movement, while the subtler signalling, that is the adoption of the note-taking posture, is commonly found in Event-IV, V and VI. These actions provide helpful clues to the speaker to identify the conclusion of a chunk of talk in an unknown language and, thus, are key to keeping a smooth communication flow.

Table 4: Interpreter-to-speaker transitions

6. Discussion

Thanks to the multimodal approach and a CIPEs corpus between Italian and Chinese, the present study has revealed some of the main features of chunking, a collaborative activity that speaker and interpreter carry out.

To different degrees, all the speakers analysed have employed sideways gaze and head orientation in chunking, demonstrating their recognition of the interpreter as a validated participant. This practice, already observed in dialogue interpreting (Davitti 2016; Mason 2012; Pasquandrea 2011, 2012), becomes reminiscent of a theatrical performance (Goffman 1981) when it occurs on a stage: when the speaker shifts their sightline and body towards the interpreter, the audience is likely to follow suit, directing their attention to the interpreter. Therefore, the speaker’s gaze shifting serves as a spotlight, casting the interpreter as a co-performer. What appears to be characteristic of CIPEs are indirect chunk-ending cues, such as computer manipulation in Event-V. While not explicitly directed at the interpreter, it still delimits the boundary of the ongoing chunk of speech and facilitates floor-taking by the interpreter. Therefore, we regard it as having collaborative significance in any case. Its subtle form also aligns with the behavioural pattern in this setting.

The interpreter has also contributed to the collaborative outcome. Their close and consistent monitoring of the speaker is pervasive throughout the corpus. Such monitoring not only enables them to grasp visually transmitted information in the speech but also maximizes the possibility of seizing the speaker’s indirect cues, which could otherwise be overlooked. The interpreter’s signalling practice is relevant, too. It assists the speaker in regaining the floor smoothly by providing helpful clues of the end of the rendition in an unknown language. Although the speaker’s monitoring is less apparent due to the context and the participants’ respective roles, their actions are closely and timely related to the interpreter’s cues.

Despite the general collaborative frame, the analysis also disclosed diversified approaches among participants. Given the limited size and small sample groups of our current corpus, we do not consider our findings to be representative of the larger population of this domain, yet evidence-based reflections on the possible reasons of the observed phenomena may contribute to a better understanding of the participants’ behaviours in CIPEs.

We noted that the Chinese guests, Speaker-I, II and III, are decisively keen to using publicly performed cues, while the chunking behaviour of the two Italian guests, that is Speaker-V and VI, is more subtle. This pattern differs significantly from those reported in previous studies, according to which members of a Chinese cultural group are likely to adopt a discreet conduct (Hlavac and Xu 2020: 50-52; Wang 2024) by avoiding eye contact (Vargas-Urpì 2013) and animated bodily conduct (Wang 2024: 183-186), while Italians tend to engage in an expansive way of speaking (Katan 2004: 298). Firstly, such a marked contrast may result from some situational factors. The Chinese speakers were invited to talk about their works and creations (a product, a book or a film), which may lead them to see the interpreters as their aids and allies and, therefore, enhance their connection with them.  The two events involving Italian speakers were organized by the Italian Embassy in China to promote Italian culture. The speakers were likely to align with the organizer by focusing on the audience and cooperating with the interpreters more discreetly. Secondly, a different setting and participant profile distinguish the present study from previous ones. Namely, in CIPEs the guest speaker is generally a distinguished figure with experience in public speaking. The practices and the strategies they adopt are arguably different from those observed in the community setting. Finally, personal inclination and prior working experience with the interpreter could also have contributed to shaping the participants’ not-so-typical behaviours.

As far as the interpreters are concerned, it could be noticed that Interpreter-V never directed her eyes to the speaker in all the 54 transitions (cf. Table 3). This ‘detached’ approach is also visible in the translation strategy adopted, namely, the use of the third person singular to report the speaker’s talk and the reference to the latter in her rendition as ‘master photographer’. Considering the age, gender and status differences between she and the speaker, the most pronounced of our whole speaker-interpreter dataset, the choices of Interpreter-V are possibly grounded in the Chinese cultural values of politeness and respect (Hlavac and Xu 2020: 46-52). Nevertheless, such culture-bound behaviour has not emerged in other interpreter’s chunking practices.

Regardless of the in-group variation, when comparing the interpreters’ data with the speakers’, as shown in Table 4, it is possible to recognize a substantial consistency^[3]: in events where the speaker favoured publicly performed cues over indirect ones, the interpreter engaged in direct practice more consistently as well, and vice versa. This finding suggests that the interpreter tends to align with the speaker’s chunking practice. On the one hand, it ensures a consistent performative effect within the ‘coupled turn’ (Poignant and Wadensjö 2020: 3). On the other, it could avoid drawing unwanted visibility on the interpreter in public. The interpreter’s role in public events may seem less prominent than in dialogical forms of interaction. However, their tactful collaboration with the speaker still reflects their professional agency as communication facilitators.

Table 5: Chunking practices: Speakers vs interpreters

7. Conclusions

The challenges in gathering naturally occurring interpreting data (Bendazzoli 2018) and the non-optimal state of the data for research purposes (Mason 2012: 180-182) pose significant obstacles to empirical studies. As a result, our analysis is based on a small-scale corpus composed of video recordings of the only six events sharing direct and indirect chunking activities. Collecting background information on the interpreters and their views on the situation back then also turns out to be unfeasible due to the secondary source of the data.

Nevertheless, with the available resources, we have provided a preliminary account of the collaborative chunking practice in CIPEs. In this ritualised setting, the organisational aspects and participants’ behaviour are utterly influenced by communicative norms and the asymmetrical status between the guest speaker and the interpreter. Our experience has shown the efficacy of multimodal analysis in highlighting the subtle nonverbal dimension of CIPEs. The corpus-based approach has been instrumental in elucidating the main features of chunking in diverse communicative situations. To corroborate our findings, further explorations of CIPEs in this direction are required. This would enable a more comprehensive insight into nonverbal chunking dynamics, hopefully based on larger data collection and triangulations with additional contextual information.

About the author(s)

Han Wang is a postdoctoral research fellow at the Department of Interpreting and Translation at the University of Bologna, Forlì, where she has taught translation and dialogue interpreting between Chinese and Italian since 2013. She holds a PhD degree in Translation, Interpretation and Intercultural Studies from the same university. Her research focuses on the multimodal and intercultural dimensions of consecutive interpreting. In addition to her academic work, she is a practising interpreter, with a client portfolio that includes various public institutions and leading corporations.

Mariachiara Russo is Professor of Spanish linguistics and Interpretation at the Department of Interpreting and Translation (DIT) of the University of Bologna at Forlì, and current Rector’s Delegate for Multilingualism. She holds an MA degree in Conference Interpreting from the University of Trieste, where she taught consecutive and simultaneous interpreting from 1993-2001. Since 2001, she has been teaching at DIT, where she coordinated the Master’s in Interpreting until 2015. She sat on the Steering Committee of the Knowledge Centre on Interpretation of the European Commission (2018-2020) and is a member of the CIUTI (https://www.ciuti.org/) Board since 2018. She coordinated the European Parliament Interpreting Corpus project (EPIC, https://bellatrix.sslmit.unibo.it/noske/epic2/#open) and the Unified Interpreting Corpus platform (UNIC, https://unic.dipintra.it/). She was an active member of the Erasmus+ Project Shift in orality. Shaping the interpreters of the future and of today (https://site.unibo.it/shiftinorality/en). Research fields: aptitude testing for interpreting, conference and liaison interpreting, corpus-based interpreting studies, remote interpreting, contrastive linguistics, and simultaneous film interpreting. https://www.unibo.it/sitoweb/mariachiara.russo

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

•  Previous article
• Next article