Studying indirect translation: a conversation with and between L. Davier, M. Marin-Lacarta and F. Pöchhacker

ABSTRACT Focused on key issues in researching indirect translation, this article presents a dialogue with and between three scholars from different branches of Translation Studies: news translation, interpreting and literary translation. The aim of this dialogue is to address fundamental questions concerning the history of indirect translation studies, the methodological challenges faced by researchers investigating this practice, and the impact of English as a lingua franca and recent advancements in machine translation on the quality and future of indirect translations. This intra-disciplinary exchange allows us to identify unique perspectives and potential blind spots in approaches to indirect translation within specific branches of Translation Studies. At the same time, the dialogue sheds light on the various overlaps and complementarities between these approaches. We hope that the insights gained from this dialogue will not only deepen our understanding of indirect translation as a field of practice and research but also foster opportunities for future intra-disciplinary collaborations.


Introduction
The guest editors of this thematic collection -Hanna Pięta, Laura Ivaska and Yves Gambierinvited three scholars who look at indirect translation from the perspective of three different domains -Lucile Davier (news translation), Franz Pöchhacker (interpreting) and Maialen Marin-Lacarta (literary translation)to engage in this online dialogue, which took place via Google Docs during the spring of 2022. As knowledge of the authors' background can be useful in interpreting their insights on indirect translation, we open this publication with biographical notes on the invited scholars.
Lucile Davier is an assistant professor at the Faculty of Translation and Interpreting (University of Geneva). Her research interests span news translation, indirect translation, translation ethnography, nonprofessional translation and convergent media. She jointly convened the panel Advancing Intradisciplinary Research on Indirect Translation at EST2022 with Hanna Pięta and Ester Torres Simón and published an article about indirect translation in Target 34 (3). In addition, she is the author of Les enjeux de la traduction dans les agences de presse [The role of translation in news agencies] (2017) and coedited the collective volume Journalism and Convergence in the Era of Convergence (2019) with Kyle Conway. She was a postdoctoral fellow at the University of Ottawa, Canada (2016 and a visiting scholar at the University of Leuven, Belgium (2012-2013. In 2013 she earned a joint doctoral degree in translation studies and communication studies (University of Geneva and University of Paris 3).
Franz Pöchhacker is Professor of Interpreting Studies in the Center for Translation Studies at the University of Vienna. With professional training and experience in conference interpreting, his interests have expanded over the years to include issues of interpreting studies as a discipline, media interpreting and community interpreting in healthcare, social service and asylum settings. His more recent work focuses on technology-based forms of interpreting such as video remote and (interlingual) speech-to-text interpreting. The latter area of speech-to-text interpreting is of particular interest as a form of cross-modal indirect translation accomplished through humanmachine interaction. Professor Pöchhacker has lectured and published widely, with his English-language books including The Interpreting Studies Reader (2002) (2008,2017,2018) as part of her research on the history and reception of Chinese and Sinophone literatures in Spain. Her more recent work focuses on ethnographic research in translation studies and digital publishing of literary translations. She has published in major journals such as Translation Studies, The Translator, Perspectives and Meta and is currently co-editing a special issue on ethnography for The Translator. Marin-Lacarta is also an award-winning Chinese literature translator into Spanish and Basque.
Our questions and the answers of the three scholars take us through a number of historical perspectives, current challenges, work-arounds and solutions, as well as towards meaningful future avenues.
Our hope is that after reading this dialogue, researchers will have greater insights into areas that are worthy of exploration in terms of indirect translation, into the conceptual and methodological pitfalls that might await them, and into ways in which these pitfalls could be (or are being) addressed in different subfields.
As a disclaimer, we want to stress that this dialogue is not meant to propose one-sizefits-all solutions or to make an exhaustive survey of the current state-of-play. Rather, we believe that the dialogue will yield food for thought that will help indirect translation develop into a fully-fledged research area in its own right. We are convinced that such a dialogue on indirect translation between researchers from different strands of translation research will be illuminating.

Question no. 1 (On the history of indirect translation research)
The dialogue begins … Hanna Pięta (HP), Laura Ivaska (LI) and Yves Gambier (YG): Our first question is about the history of research on indirect translation (including relay interpreting). Indirect translation is an age-old and fast-evolving practice but as a subfield of research it is still in its infancy. Systematic studies are recent and rare. They are mostly productoriented, limited mostly to literary translation in book form (and, to a lesser extent, conference interpreting) and focused on issues of quality. There is a shortage of well-established hypotheses or comprehensive typologies of various instances of indirectness. If you look at your domain in particular (news translation, interpreting, literature), what might the causes for this be? Why is research on indirect translation lagging so far behind other subfields that focus on adjacent phenomena (e.g. retranslation or self-translation, which both have been buzzwords for quite some time now)?
Lucile Davier (LD): Journalistic translation research is still a young subfield of translation studies. According to Valdeón (2015), news translation started to emerge and develop in the 2000s. Publications about news translation significantly increased from 2010 onwards (Davier, 2022b;Schäffner, 2018). Therefore, it is not surprising that journalistic translation research has not yet given much attention to indirect translation: it was too busy developing as a separate subfield.
Journalistic translation research is strongly product-oriented (Valdeón, 2020, p. 326). Research published in the twentieth century focused mainly on news reports as a source in answering linguistic, textual or ideological research questions (Valdeón, 2015). However, scholars need triangulation with field data to trace indirectness (Davier, 2022c). It is thus possible that product-oriented research does not have enough tools to identify indirect translations. Even in process-oriented research, the 'hunt for traces of indirect translation still entails much time-consuming guess work' (Davier, 2022c).
Integral and explicitly-mentioned translations are 'extremely rare in the media' (Davier & van Doorslaer, 2018, p. 245), which might explain why scholars have not turned their attention to the complex travels of a news item across languages. News items are often considered as raw material that does not have to be cited: their status as a source text is low. In addition, news stories build on pre-existing public relations materials such as press releases and press conferences (Davies, 2009), which are even more short-lived than news stories. The challenges of identifying possible sources are even greater for historical research (Navarro, 2020, pp. 355-356;Valdeón, 2015, p. 637).
Finally, given the low status of source texts in news translation, concepts such as retranslation or self-translation have few chances of emerging either.
Franz Pöchhacker (FP): While interpreting studies is several decades older than journalistic translation research, scholarly interest in indirect translation, known as relay interpreting, has been similarly scarce, though for different reasons. One fundamental issue is the diverse nature of interpreting as a social practice and object of study. Various professional domains, such as conference interpreting, signed language interpreting and community interpreting, are associated with their own research traditions, or paradigms (Pöchhacker, 2015), which lack a shared perspective on the object of study. The forces inhibiting a shared research interest in relay interpreting differ according to domain and range from practical constraints and methodological challenges to professional ideologies.
In conference interpreting, the former Soviet system of relay-based multilingual communication via Russian as the pivot language (Denissenko, 1989) was challenged (and superseded) by the 'Western' model of (simultaneous) interpreting into the interpreter's 'A' language, directly from as many acquired languages as possible. The so-called Paris School as thedistinctly profession-linkeddominant school of thought in the formative decades of interpreting research (Lederer, 2015) saw relay interpreting as something to be avoided and not meriting systematic studyparadoxically, given its regular use in institutions such as the United Nations and given Seleskovitch's (1978, p. 102) claim of 'total accuracy' achievable by suitably-trained professionals. Relay interpreting as a 'necessary evil' is all but taboo in research on conference interpreting, with entrenched ideologies lingering in major institutions.
In community-based domains, on the other hand, relay interpreting is a rare, if not marginal, practice and has therefore attracted little scholarly attention. Its use is fraught in legal settings, particularly because of the verbatim standards aspired to in common-law jurisdictions; it is limited by lack of resources in healthcare and social settings, and is professionally delicate in Deaf relay interpreting, where a native signer adapts a hearing signed-language interpreter's output to specific user needs.
Maialen Marin-Lacarta (MML): Although the history of research on indirect translation of literary texts is fairly recent, there is now a significant and growing body of literature in this subfieldunlike in journalistic translation and interpreting studies. Hanna (Pięta, 2017) presents a review of publications on indirect translation published between 1963 and 2000, the majority of which are devoted to literary translationand to historical research in particular. The special issue in which Hanna's article was published (Assis Rosa et al., 2017) was key in giving visibility to indirect translation research (together with other subsequent initiatives, such as specific panels in conferences and forthcoming special issues in major journals).
However, I agree that indirect translation is a subfield with a shortage of well-established hypotheses and typologies. More importantly, literary translation scholars still tend to mention it only in passing. One reason for this is that translation often continues to be seen through the lens of a binary epistemology as reflected in the notions of 'source' and 'target' texts. Micro-level research (involving translation comparison) rarely pays attention to indirect translation and, when it does, it is often to demonstrate the inferiority of such practice. The disdain towards translation, which is frequently considered to be derivative, has been reproduced in a disdain for indirect translation, and this seems to be the case across all domains. On the other hand, macro-level research with a historical focus tends to pay attention to the mediating role of cultures in the diffusion of literature, culture and knowledge, but in these contexts an amalgam of practices (such as pseudotranslations, retranslations, adaptations, indirect translations and combinations of all this) are examined and researchers are rarely interested in distinguishing these practices and examining them in detail. I believe that indirect translation research has huge potential to bridge the gap between microand macro-level research, which are essential and compatible dimensions in the study of literary translations. Question no. 2 (Methodological issues: identifying indirect translation) HP, LI, YG: One of the oft-mentioned obstacles hindering systematic research on indirect translation is the lack of efficient methods to identify indirect translations. Indeed, it is often difficult to distinguish direct translations from indirect translation and to establish which languages and texts were used as a source for translation. Our impression is that this concern exists across the board but to varying degrees. To what extent is this an issue in your domain and how could it be efficiently tackled (e.g. what type of sources, archives, bibliographies are particularly relevant in this respect)?
FP: Generally speaking, research on interpreting is beset with methodological challenges much more so than studies of written or audiovisual translation are. The evanescence of spoken (or signed) language, the multitude of contextual variables, the constraints on documenting multimodal utterances through transcription and restricted access to interpreter-mediated communicative events (except for media settings) constitute major obstacles to research in this field. For the study of relay interpreting, however, the situatedness and immediacy that give rise to these basic challenges put interpreting researchers in a rather favourable position, at least as far as the identification of the source language and of the pivot are concerned. While relevant distinctions must be made according to interpreting modes and settings, it is unlikely that the source of a relay-taker's interpretation could not be traced. In community settings involving less widely-used languages, such as Mixtec-English via Spanish in California (Mikkelson, 1999), and in the case of Deaf relay interpreting, the interpreter recruited as pivot would of course be known. In multilingual meetings of the European institutions, by contrast, the use of relay may be less obvious and source identification may be cumbersome in retrospect, or even impossible once the team sheets for a given session can no longer be retrieved. On the other hand, the number of pivot languages tends to be limited and in some international organisations, such as the United Nations (which boasts a long tradition of relay interpreting), the booth from which the other team members will take relay when needed is largely predetermined: the Arabic booth, for instance, would provide retour into English or French, whereas the pivot language for Chinese would invariably be English. And since English has practically become a sine qua non in a conference interpreter's language combination, it is increasingly predominant for relay in conference settings.
MML: It is interesting to read that researching relay interpreting is in principle less challenging as the source language and pivot can usually be easily identified. That is not the case for literary translation. Researching indirect translations of literary texts involves stimulating detective work: paying attention to missing pieces and contradictory information in the sources we consult, trying to find answers through a variety of methods and triangulating results. Mistrust towards information found in paratexts, book reviews, catalogues and databases is a prerequisite because texts that are presented as direct translations might actually be hidden indirect translations. When studying certain historical periods, it is no different from analysing other translation phenomena, such as pseudotranslations. For example, one of the first Chinese literature translations in early twentieth-century Spain (various poems by T'Zin Pa-O, published in the literary magazine Prisma) is presented as an indirect translation from French but is in fact a pseudotranslation, a fact that can only be elucidated by tracking down information such as possible mediating texts and source texts, verifying the authorship and getting to know the historical context, in which, in fact, indirect translations from French were wellaccepted. In this sense, the historical context is key and often leads us to a limited number of mediating languagesas is the case in interpreting, as Franz just noted. I have previously written about how to study and document indirect translations, the various contributing sources (such as bibliographic databases and catalogues, paratexts, book reviews, sources about translators and sources about contexts and translations) and methods (close reading, translation comparison and interviews) (see Marin-Lacarta, 2017). Although some progress has been made in the application of machine learning to corpora to identify indirect translations since the publication of that article, such research is still nascent and close reading and archival research continue to be the main methods.
LD: The analogy used by Maialen above very much applies to indirectness in the news: finding evidence of indirect translations equates with detective work. Firstly, as already mentioned in passing in my reply to Question 1, this methodological problem is rooted in the fact that source texts are evanescent in the news: mentions of source text, source language or source author are very rare (Davier & van Doorslaer, 2018). If translation is passed over in news reports, one cannot expect to find relevant information about it in news archives. In contrast to literary translation, scholars cannot use bibliographies. Secondly, compilative writing is characteristic of news writing because news is 'what others speak and write about' (Catenaccio et al., 2011(Catenaccio et al., , p. 1844. In other words, news stories rely very much on reported and recontextualised speech. Compilative writing goes hand in hand with compilative translation. However, compilative translation in the news differs from that observed for other text types. Fragments of various source texts are translated, compiled and mixed with fragments of text that were originally written in the language of the journalist (Davier, 2022c). Under these conditions, finding sources for every translational fragment would appear to be unrealistic. There is hope here of benefitting from scientific advances in the fields of machine learning and digital humanities, although, as Maialen writes above, we may have to resort to other methods until these technologies are well-developed enough. Thirdly, these methodological problems can partly be solved by investigating the sociolinguistic and pragmatic context of a news outlet and doing fieldwork. In the field, scholars can witness, for instance, oral interlinguistic explanations or capture ultimate oral source texts in a different language (Davier, 2022c). In these cases, scholars are in a situation close to that described by Franz in the context of relay interpreting: they are more likely to identify the source or mediating language. Nevertheless, fieldwork should not be idealised since interactions are increasingly digitalised (Cronin, 2016): in the field, the researcher cannot access everything journalists are doing on their phones and computers. In addition, they cannot know whether materials that reporters receive were translated beforehand and whether their stories will be translated by colleagues from other media organisations once they are published.

Question no. 3 (Other methodological issues)
HP, LI, YG: What other issues still hinder the development of systematic research on indirect translation in your domain? Who seems to be most affected by these issues? What can be done to overcome (or at least partly address) these issues? MML: Systematic research on indirect translation in the field of literary translation remains fairly invisible and fragmented, which is reflected in a lack of standardised terminology and conceptual agreement, and this is an issue for scholars. The lack of visibility of the phenomenon has been partially mitigated by the inclusion of entries in handbooks and encyclopaedias, such as in Baker andSaldanha (2009, 2020) and Gambier and van Doorslaer (2010), but the conceptual and terminological confusion persists. In addition, the fact that Translation Studies has traditionally focused on central languages has been an obstacle for the development of indirect translation as a subfield (Marin-Lacarta, 2018;Pięta, 2016). This has, however, changed in recent years, with a growing number of historical studies on indirect translation focusing on cross-cultural exchanges involving peripheries (see, for example, Marin-Lacarta, 2018; Pięta, 2016;Pokorn, 2013). In this sense, Translation Studies scholars would also benefit from a dialogue with other disciplines, such as comparative literature, in which the indirectness of cross-cultural exchanges has been studied (see, for instance, Cho, 2016;Liu, 1995;Prado-Fonts, 2022;Thornber, 2009). Another issue is the strong negative stigma attached to the indirect translation of literary texts, which is reflected in the fact that when indirect translation is analysed it is often for the sole purpose of proving its inferiority and distorted nature. This negative stigma results in a lack of process-oriented research and of teaching materials that address the skills, strategies and good practices needed when translating indirectly. The forthcoming publication by Pięta et al. (2022) will hopefully be a first step to addressing this gap. This issue affects not only researchers but also practitioners and instructors. Another consequence of the mistrust towards indirect translation is that there is no research involving practitioners, a field that deserves attention. LD: In contrast to what Maialen has to share about the realm of literary translation, indirect translations do not carry an academic stigma in news translation, since news reports are often considered to be raw material that is not meant to last and where there is no named author (this situation causes other problems, as already mentioned in Question 2). However, a narrow understanding of 'indirect translation' can probably explain why interest in this concept has emerged only recently in journalistic translation research. This issue can easily be addressed by intradisciplinary collaborations and discussions at international conferences.
The main issue hindering the development of research on indirect translation in the news is probably the fact that the focus has been on central languages (or languages originally spoken in Western Europe), exactly as described by Maialen. The growing number of publications presenting case studies with Arabic (e.g. Haj Omar, 2019), Chinese (e.g. Xia 2019), Persian (e.g. Hajmohammadi, 2005) or Russian (e.g. Spiessens & Van Poucke, 2016) certainly brings hope in this respect. This trend can be reinforced by publications applying indirect translation as a conceptprovided these publications are freely available (open access). Indeed, subscription journals are less likely to be accessible in countries in which so-called peripheral languages are used.
Unlike literary translation, news translation does not benefit greatly from research conducted in journalism studies or communication studiesdisciplines in which translation has generally been ignored (Valdeón, 2018). Conceiving of news production as free from translation or free from indirect translation can lead to inaccurate research in communication studies and in translation studies. These misconceptions may be due to the way news stories are presented in translator training as one ultimate source text which should result in one ultimate target text. From this perspective, the discussion of news translation in Pięta, Bueno Maia, and Torres Simón is to be lauded (2022).
FP: As mentioned above, research on relay interpreting has been hampered by a longstanding negative bias (mainly in the conference-interpreting community), by its infrequent use in intra-social settings, and by limited access to real-life interpreter-mediated events. Institutional users (organisations) therefore hold the key to generating fresh momentum, but this would in turn require researchers to make a convincing case for the benefits and relevance of such studies. For most settings, the issue of quality is bound to loom large and employers and interpreters may or may not want to see the potential shortcomings of relay interpreting exposed. One way of moving beyond such an 'applied', evaluative orientation might be to reframe relay interpreting as a complex form of communication in its own right rather than as a less reliable variant of an established practice. In an analogy with the labelling of dialogue interpreting as 'triadic' (Wadensjö, 1998) and its investigation as co-constructed communication, a label such as 'quadrangular interpreting' could help stimulate interest in the interpreting studies community and beyond. (It should be irresistible, for instance, to sociologists of interaction.) This would suggest a focus on the process, which is generally much more feasible in relay interpreting than in literary and news translation, as the agents, tasks and conditions in relay interpreting can be clearly identified.
The methodological challenges for such a research programme are considerable but could nonetheless be met. They include accounting for the multimodality of interlocking acts of interpreting, as analysed in Pöchhacker (2022) with regard to speech (verbal and paraverbal), kinesics, images and writing. Where the focus is on the cognitive microprocess of, say, the relay taker, the use of eye-tracking can help to capture the interpreter's visual attention patterns (e.g. on source-speaker kinesics, slides or a script). Productoriented studies, on the other hand, would require corpus-based approaches, especially in multilingual organisations.

Question no. 4 (Status of the English language as lingua franca)
HP, LI, YG: In the eighteenth century, French was a default pivot language for other, globally less powerful, languages (for instance, between English and German). What about English today? With fewer students learning languages other than English as foreign languages, is it possible to avoid a growing use of indirect translation (including relay interpreting) in the foreseeable future?
LD: Indeed, English is an important lingua franca in the media as well. Journalists working in languages that are not covered by the global news agencies (e.g. Dutch) often choose to translate news dispatches from English even where French is an official language in their country, as happens in Belgium (van Doorslaer, 2009). As a consequence, indirect translation is more likely to happen with languages that can be considered 'minor' from the point of view of news agencies. English does influence other languages through news translation (Gottlieb, 2010;Stenvall, 2001), including French (McLaughlin, 2011), even though this latter is one of the languages of the global agency AFP.
However, this linguistic influence should not reduce the complexity of the situation. English very often appears as a source or target language (when the source is identifiable at least) in news translation studies but this dominance may be explained by academic rather than journalistic reasons. One can hypothesise that scholars very often analyse English news reports because of their proficiency in English and because they do not have to provide back translations when they publish in English.
English is not necessarily a pivot language in the media in general or in news agencies in particular. Firstly, if possible, local media generally have subscriptions to news agencies that circulate news dispatches in their own language, such as EFE for Spanish-language media or Lusa for Portuguese-language media (Lisboa & Aguiar, 2017). Secondly, news agencies do not systematically translate in the same direction (Bielsa & Bassnett, 2009;Davier, 2014). For instance, in the Geneva bureau of AFP, Anglophone and Francophone journalists produce content in parallel. In other words, journalists cover important events in both languages at the same time. In other situations English is sometimes the source and sometimes the target language (Davier, 2014). As argued by media scholars, the idea that English is replacing other vehicular languages in the media is misleading (Nederveen Pieterse, 2015) because of the predominance of national and local news (Quandt, 2008).
FP: In many settings of transcultural communication, the current status of English as a lingua franca (ELF) affects not so much its use as a pivot language in relay interpreting as the use of interpreting and interpreters in the first place. The instances of UN meetings continuing after hours in English only and of Belgian asylum hearings conducted without interpreters (Maryns, 2014) are salient cases in point. By contrast, the choice of pivot languages in international conference interpreting has always been quite limited and shaped by location and institutional context. Aside from the key role of Russian in Soviet times, the choice has essentially been between English and French, with the latter holding its ground for a relatively long period. The enlargement of the European Union toward the east in the early 2000s made English pivotal in conference interpreters' combinations of working languages, paradoxically facilitating the viability of a fully-multilingual language regime. And yet, the implications of this dominant role of English have not generated substantial research interest, neither with regard to individual interpreters' performance nor in terms of the epistemic impact of 'going through English' on language use in international communication. This would seem particularly relevant for simultaneous interpreting from Chinese at the United Nations, where interpretations into the other five official languages are invariably based on the English retour provided by the Chinese booth.
In other domains of interpreting, the limited use of relay provides even fewer grounds for systematic study. The unique phenomenon of International Sign, a form of signed communication used at events with deaf participants from various countries (de Wit et al., 2021), offers few parallels with ELF but could possibly develop into an alternative to iteither as an increasingly conventionalised lingua franca for deaf users in international settings or for relay interpreting into national signed languages.
MML: Both English and French publishing trends currently play key mediating roles in the dissemination of less-translated literatures into other European languages. For instance, Spanish publishers often discover Chinese works through English-speaking literary agents and through French-and English-language publishers. In addition, American publishers also look at what gets published in French (so English is not the only mediating language). Although publishers' awareness of the importance of direct translation of literary texts has grown in the last decade, indirect translation is a persisting reality. However, I agree with Lucile that the idea that English is replacing other source languages is misleading. Literary translation from non-English texts continues to be a big part of the publishing market and, as a consequence, European translation programmes teach a variety of languages and literatures. In addition, the spread of small presses in the Spanish, French and British markets has favoured the dissemination of less-translated literatures. Besides, our knowledge outside of Western publishing spheres (such as European, American and Canadian markets) is limited and I agree with Lucile that we need more studies of locations where English is not necessarily the principal mediating language; this would certainly show a more complex picture. I should also clarify that the recourse to indirect translation in the current Spanish context is not so much due to the predominance of English as lingua franca (and the disappearance of other languages) but to the fact that it is faster and cheaper to translate from English. At the same time, it is a vicious circle: publishers often face difficulties in finding available Chinese translators, not because there is a lack of these but because since literary translation is a precarious job, Chinese literature translators tend to have other full-time jobs and are therefore not always available to meet short deadlines. I believe that this is also applicable to other less-translated or peripheral literatures.

Question no. 5 (Quality: indirect versus machine translation)
HP, LI, YG: Let's shift the focus to the issue of quality. Here we would like to draw a parallel between indirect translation (including relay interpreting) and machine translation (MT). This parallel seems justified because these two practices are increasingly combined. For example, the French Red Cross uses Aaliatalk, a MT/voice assistant device for healthcare, with 79 languages. Volunteer workers use it between, for instance, Sudanese Arabic and French but it can also be used for Sudanese Arabic-French-Italian (for example, to enable communication involving refugees in Sicily).
Another reason for this MT-indirect translation parallel is that these two practices are often eyed with suspicion. For example, the European Commission recently published a report with recommendations for the translation sector, focusing on 'cultural and creative works ' (2022, p. 6). In this report, MT and indirect translation are both seen as 'solutions that could compensate for a lack of translators, particularly in rare combinations of languages,' but their 'output seen so far is of a quality not deemed acceptable by both academic research as well as publishers' (EC, 2022, p. 10). Curiously, the report's recommendation is that progress in machine translation technology 'should be monitored closely,' whereas indirect translation 'should not be encouraged as a long-term solution.' The recommendations seem to apply to books, films and plays (2022, p. 6) but what is your take on this from the standpoint of your domain? Could this mean that there is hope for higher quality when it comes to MT but that there is no hope for higher quality when it comes to indirect translation? What can researchers do about the notorious poor quality of indirect translations?
FP: The relationship between relay interpreting and machine translation has changed profoundly over the past 50 years. Once seen as entirely separate phenomena without any point of interface, the two have now become more interrelated, with some shared ground as well as competition. The shared suspicion of delivering sub-par results is very vague, however, even though it applies equally to MT itself and to automatic speech-to-speech translation in the cascade model, with MT as its core between modules for speech recognition (SR) and synthesis. What such 'machine interpreting' could be said to have in common with relay interpreting is the lack of access in the translational process to the full range of multimodal information in the original message. In this regard, the written text delivered by the SR module to the MT module is a kind of intermediate text like the one produced by a pivot. Beyond this relay analogy, MT itself, especially in the form of dialogue interpreting apps, often appears to involve a relay process with English as the pivot language. In a study testing the use of iTranslate Converse in a German -French police interview (Leitner, 2020), several German words, correctly recognised, were rendered with their English 'equivalent' in the French target text (e.g. 'Café Ritter' translated as 'café knight'). MT experts might be able to tell us how this works but the evidence certainly points to some role for English as an intermediate language. Aside from combinations involving English, the assumption that advanced MT could be expected to deliver better results than human relay interpreting therefore seems highly questionable. In particular, the presumed superiority of MT in covering rare language combinations (say, Greek -Flemish) is undermined by the need for ('neural') MT to rely on massive bilingual corpora as training resources.
MML: The situation in my domain is quite different to what Franz has described because literary texts continue to be the greatest challenge to MT. There are no machine-translated literary texts in the market; what we find so far are empirical experiments carried out by academics. On the other hand, indirect translation is very common between peripheral literary systems. For instance, between 2001 and 2010, 25 Chinese novels were indirectly translated into Spanish. Considering this reality, I agree with Franz that the assumption that MT could deliver better results than human indirect translation is not just highly questionable, but pure fiction. It is shocking to read that publishers deem the quality of indirect translation unacceptable, according to the EC report, when they are the ones hiring translators to translate from third languages. I suspect that the EC's recommendation is partly motivated by ethical considerations. Literary translation is a precarious job and even more so for translators working with lesstranslated literatures. Promoting indirect translation could hinder the emergence of translators from peripheral literatures. On the other hand, the recommendation to monitor progress of MT technology could be understood as a way to empower translators (not replace them) by improving their working conditions and tools. At this stage this is, however, highly utopian and what MT and indirect translation have in common is that they are mainly used to cut costs and speed things up at the expense of quality and to the detriment of translators. I am also a Chinese literature translator and I encourage direct translations but, given that indirect translation of less translated literatures is a reality, translators should be trained to translate from third languages and to produce translations that are going to be used as sources for other translations (instead of discouraging this practice completely). There is definitely hope for higher quality when it comes to indirect translation.
LD: To start with, I would like to highlight the fact that as a rule translation is not visible in the media. Translation and languages are barely discussed (Bielsa & Bassnett, 2009;Davier, 2014). In most contexts under investigation, the journalists who are translating are not offered any specific training either in translation or languages (Bielsa & Bassnett, 2009;Davier, 2022a;van Rooyen, 2018). As a consequence, journalists are not interested in quality criteria for translations (which do not exist in their minds) but only in the quality criteria for news stories.
Furthermore, journalistic translation research has not yet tackled the issue of quality but has rather resorted to a framework of risk reduction and mitigation (Matsushita, 2019;Davier, 2022a). This choice can be explained by the constraints of contextor participant-oriented research, in which journalists may refuse to participate in projects assessing the quality of their translations.
Concretely, in the Swiss context journalists I interviewed were not at all shocked by the use of indirect translation (Davier, 2014). Nonetheless, indirect translation was not the main focus of my research. In the Canadian context, some journalists used machine translation as they might a bilingual dictionary but most of my Francophone interviewees acknowledged that the output could not be trusted blindly (Davier, 2022a). Here again, machine translation was not explicitly addressed in my research question. This leads me to anticipate the answers to the last question. There may be a way of addressing questions of quality in news translation by asking journalists about their perception and their uses of indirect translation and machine translation. It is possible that news editors perceive indirect translation less negatively than translators do. Question no. 6 (Future research prospects) HP, LI, YG: Our last question is about looking ahead. What type of topics would you like to see explored in research on indirect translation?
MML: In the specific context of translation history, one of the areas that merits more attention is the role that indirect translation has played in developing cross-cultural contacts between non-European and European cultures from the sixteenth until the twentieth century (St André, 2020, p. 472). The forthcoming book by Prado-Fonts (2022), which focuses on the mediation of French and British sources in the Spanish understanding of China in the late nineteenth-and early twentieth-century, will contribute to this area. 1 This study opens up the possibility of researching cultural artefacts (literary texts, news) that may not necessarily have been translations but that relied on previous cultural representations in English and French sources and that can be seen as evidence of indirectness, complex triangulations and cultural hegemonies.
In the same vein, a second promising area in translation history is research dealing with indirect translation outside of Asia, Europe and South America (postcolonial African countries, for instance).
A third future research avenue, unexplored to date as far as I know, is that of sociological and ethnographic studies on practitioners of indirect translation (applying methods such as participant observation, interviews, focus groups, etc.), as there is a lack of research on current practices of indirect translation.
A fourth promising avenue of research is that of corpus studies of indirect translation and, more precisely, the application of machine learning to corpora for studying indirect translation. This could potentially allow researchers to automatically identify indirect translations and mediating texts, among other possibilities. An important study in this area is Ustaszewski's (2021) application of replicable supervised machine learning to corpora of indirect translations using proceedings from the European Parliament.
Finally, a crucial research prospect in indirect translation is expanding its study beyond literary texts and developing this subfield even further in relation to news translation, interpreting, audiovisual translation, localisation, technical translation, institutional translation, etc.
LD: Firstly, I simply would like to call for more case studies about indirect translational phenomena in the news. Up until now, only three scholars have applied the concept of indirect translation to journalistic contexts (Davier, 2022c;Valdeón, 2022c;van Rooyen, 2018). Therefore, there is ample room to investigate other geographic areas (South America, Asia, Oceania or other countries in Africa, North America and Europe), multiplatform media (the fact that traditional media combine text and multimedia contents in different formats) and media output on social networks.
Secondly, it may be worth revisiting previously-gathered data from the new angle of indirect translation. Sociological and ethnographic studies on journalists who translate have already developed quite well. Such methods are necessary to track intermediate versions, be they written or oral.
Thirdly, given the growing phenomenon of multiplatform media or convergence (Davier & Conway, 2019), researchers specialising in audiovisual translation and news translation need to work hand in hand.
Fourthly, the historical importance of indirect translation in the news has already been the object of an article (Valdeón, 2022) that shows that indirect translation is ubiquitous even though its presence can only be hypothesised (ibid.). This leaves some doubts regarding the limits of such studies, as already noted by Navarro (2020).
Fifthly, national, international as well as global news agencies play a vital role in the dissemination of news through indirect translation (Davier, 2022c). It may be very stimulating to study how news travels from a global news agency to local media through many intermediate texts and languages. Large-scale studies conducted through the lens of postcolonial and decolonial theory may give insights about the real power of the global agencies. Scholars with skills in peripheral languages will be key in such research projects. In addition, methods used in digital humanities to detect interlingual plagiarism (e.g. Roostaee et al., 2020) will probably be instrumental in identifying indirect translations.
FP: Considering the scarcity of research on relay interpreting to date, there are many promising avenues. Methodologically, the options range from fieldwork in authentic settings to controlled experiments, and the following five thematic focal points could be suggested: Firstly, as mentioned by Maialen, large-scale corpus-based studies could be undertaken to investigate authentic datafrom organisations such as the United Nationsfor any linguistic shifts induced by indirect live translation in language combinations such as Arabic -Spanish via French (or English) or Chinese -Russian via English, using comparable corpora of direct simultaneous interpreting (SI) for reference.
Secondly, SI in relay mode should be the subject of reception studies with a focus on technical meetings involving visual support, such as PowerPoint slides, which may vary in accessibility to relay interpretation users depending on the total time lag between the speaker's presentation and the relay-taking interpreter's target-language rendition.
Thirdly, the way relay-taking interpreters process the complex interplay of multimodal resources in quadrangular interpreting could be the object of simulation studies monitoring visual attention (e.g. using eye tracking), mainly in conference settings but possibly in dialogue settings as well. Aside from on-site interpreting, this could also include remote interpreting scenarios.
A related, fourth research focus concerns indirect live translation in dialogue settings performed in different working modesthat is, consecutive vs simultaneous. The latter would be the default mode for Deaf relay interpreting but spoken-language interpreters could similarly use whispered interpreting (chuchotage) in dialogic interactions involving relay.
Finally, the novel research needs arising from speech-to-text interpreting, which in itself involves a 'relay' between a human agent (respeaker) and a machine (SR software), include comparative studies of interlingual speech-to-text interpreting performed either directly or in relay mode (i.e. spoken-language SI followed by intralingual speech-to-text interpreting). Note 1. A dedicated book review by Ester Torres-Simón can be found in this thematic collection.

Disclosure statement
No potential conflict of interest was reported by the author(s).