Why Conference Interpreting is the Wrong Model for AI and The World (Part 1)

People worldwide benefit from direct interpreting between languages in everyday life. This professional practice is commonly called community interpreting. Community interpreting is fundamentally different than conference interpreting because it involves real people in real time facing immediate consequences to their lives.

Community interpreting is the kind that happens among regular people in the everyday world trying to accomplish the normal tasks of being a member of society and participating in the typical activities of going to work, taking care of the children, pursuing an education, dealing with health and medical concerns, and addressing legal issues such as buying a house or drawing up a will. These interpreted interactions are often triadic, between two primary interlocutors and one interpreter who switches back and forth among languages to show each other’s meanings and build shared understanding.

The working conditions of community interpreting are the polar opposite of the special type of conference interpreting called diplomatic interpreting that is highlighted in an informative video series from WIRED.com.

This Moment in Interpreting History

WIRED recently hosted a competition between two professional spoken language interpreters (Spanish-English) and an AI called KUDO. The human interpreters, Barry Slaughter Olsen and Walter Krochma, are to be lauded for their courage in going up against the machine. According to Multilingual.com, this is the third in a series between Olsen and WIRED.

It is a kind of common sense for the media to focus on the high profile examples of diplomatic interpreting. These are instances of professional interpreting that often ‘make the news’ because people are curious about this fascinating intercultural communication process. Media stories and discourse about diplomatic interpreting ‘set the agenda’ for discussion about all kinds of interpreting in the public sphere (1). This particular type of interpreting is emphasized in history because it is the child of the first merger of plurilingual social interaction with electronic broadcast technology at the World War II Nuremberg Trials, which some people at the time referred to as the IBM System. Critique enters because this rarefied type has been selected and showcased as the premier example of interpreting, eclipsing the much larger and, I’ll argue, more consequential realm of community interpreting.

Deconstructing to Rebuild

There’s over a million views of the three WIRED videos about interpreting on YouTube. Olsen claims in the Multilingua.com article that if you include the views on WIRED’s website plus “the bootleg copies subtitled in other languages” then the viewership is in the “tens of millions of people all over the world” who have seen (and thus been influenced by) this video series.

First question: Why is this supposedly essential information only available to the non-English speaking world through bootlegging? It does not bring me pleasure to tell you that the producers didn’t take the time to correct the captions (which are not accurate. Read the transcript without sound and see what percent makes sense!) These two facts (uncorrected auto-captions and the need to bootleg community-level interpretations) show the existing language hierarchy at play, both in terms of accessibility and in terms of the value attributed to knowledge sharing.

English speakers are hoarding the resource of information. Only those individuals who are willing — and economically able — to volunteer time, energy and fluency to generate materials in other languages are doing this crucial work of social sustainability and resilience.

The use of AI for purposes of translation is going to fall most heavily in the realm of community interpreting due to the financial logic of ‘saving money.’ Showcasing the threat to diplomatic interpreting as if this is the major or primary risk of AI is a disservice to the broader needs of everyone who requires interpreting to conduct the mundane activities of life in a global economy. Let me state explicitly that I’m not offering a critique of individuals or even of the current practices performed within the specific context of diplomatic interpreting. The target of this critique are the patterns and habits of discourse that perpetuate an unsustainable, unjust system of providing interpreters only within a small set of legally-recognized conditions.

Intellectual reflection and critique about the impact of the discourse of diplomatic interpreting on the rest of the interpreting profession is the mark of a mature discipline. Interpreting is a vast field which demands better of professional interpreters as meaning-makers. Interpreting at the community level represents an actionable field for serious systemic-level evolution of human (and humane) adaptation to the rapidly expanding hazards of global warming.

Two Main “Types” of Interpreting

Community interpreting is typically posed against conference interpreting.

Conference interpreting takes its norms from the same source as diplomatic interpreting. In other words, the entire professional field was spawned from one particular instance with very specific (and excruciating) boundary conditions. Overall, the field of professional interpreting has failed for 80 years to grapple with the social construction of technology that developed spontaneously at the Nuremberg Trials (2). The information necessary for evaluation and re-construction is hardly lacking. For instance, Baynham et al. document “the pervasive precarity of employment of…both the community interpreters and advocates…and their clients” (3) through an extensive case study of a Czech/Slovak/English community interpreter working in Leeds, England.

I, like nearly everyone, am impressed and fascinated by the ability of some multilingual humans to mimic the behavior of mechanical tools and blend seamlessly into the technological flow. This skillset is effective in a diplomatic context where decisions are made over the course of many meetings, with iteration after iteration of nuance and subtlety sorted out by political players with staff and resources to investigate the implications of meaning and correct both the ordinary misunderstandings that always occur between humans in interaction (4) and the more difficult misunderstandings that occur due to what I’ll call intersectional difference — each individual human’s unique blend of autobiography (infused by cultural and social identities) and exposure to information and education (such as via language socialization/development and through academic training in a particular discipline).

The Resilient Culture Potential of Sustainable Interpreting

Unlike diplomatic interpreting, decisions made during each instance of community interpreting always impact quality of life and often impinge on one’s life chances, up to and including the difference between life and death.

How many humans are we talking about? I’m willing to wager that if we accounted, worldwide, for the number of plurilingual interpreted interactions occurring on a daily basis, it would equal the tens of millions of views achieved over four years for the WIRED video series. If we account properly for the actual needs of humans for plurilingual interpreted interaction, it would exceed these numbers dramatically. If we systematically link these humans plurilingually through interpreting, a more resilient and adaptable global culture of working with differences will emerge. The collaborative communication skills required to succeed in interpreting interaction will become widespread and enhance problem-solving and creativity everywhere.

Meaningful Action to Safeguard Sustainable Societies with the Support of AI

Rather than exclusively protecting the structural privilege of interpreters who are already participating in global policy-making, we could be looking to AI to augment the communication of everyday people in the real world by figuring out how to massively expand and extend the availability of human interpreters to all the community situations in which we are needed.

This would be a design for justice that builds social sustainability and community resilience through language-based equity.

As Barry Olsen and Walter Krochma reveal in their competition against KUDO, the AI may be superior in terms of information content, but only humans who are present and patient with each other can generate true understanding.

References:

1. The Agenda Setting Function of Mass Media. McCombs and Shaw, 1992.
2. The Social Construction of Technological Systems. Bijker, Hughes and Pinch, 1989/2012; for a review see Westrum, 1989.
3. Translanguaging Business: Unpredictability and precarity in superdiverse inner city Leeds (Working Paper 4). Baynham, Bradley, Callaghan, Hanusova and Simpson, 2015.
4. Ordinary Misunderstanding. Weizman and Blum-Kulka, 1992.