In the Business of Translation: Human-Computer Interface
This is part of a series of articles that take a multidisciplinary look at content creation, policies, and strategy within interconnected industries in ways that you would not find in typical articles on marketing, business, branding, and strategy.
Fact and Science Fiction
About almost 14 years ago, around 2009, during the early years of graduate school, I had to watch a Ted talk in YouTube as part of of classwork. At that time, Web 2.0 was just in nascent development, AI startups were still in the cradle, and YouTube was nothing like what it is now despite there being plenty of tech nerds uploading videos about science and tech. I no longer remember the name of the speaker nor the exact content of the talk. However, I remember he was demonstrating an immersive eyewear (reminiscent of today’s Apple Vision Pro) but with sensors attached to the fingertips (maybe there were gloves involved) so that you could swipe the virtual screen as you navigated your environment to access the information you needed to help assist you in ‘localizing’, which, in other words, is to allow you to develop contextual familiarity with an otherwise alien environment.
The prototype wearable ‘computer’ (and this was before smartphones were even ubiquitous) was aimed at helping you navigate even the most unfamiliar environment. At that time, such goggles were not very comfortable (especially for spectacles wearer as myself) and they tend to create a sense of vertigo with prolonged use. IoT was not that much of a thing outside the wireless network of a laboratory or self-contained room. Neither Google translate nor DeepL were available at that time.
A dozen a years after that, in the spring of 2023, I saw another headgear, in the form of a multi-lingual translator headphones. The scene where this took place was in a few episodes of the ‘Three-Body Problem’, a Chinese drama series based on the first book of Liu Cixin’s hit science fiction trilogy, although not much detail on this scene was given in the novel. In the drama series, world leaders from around the world were gathered together to discuss a problem relating to preventing an alien invasion and it was remarkable that they were speaking their own languages while wearing headsets that could interpret said language to the preferred language of each delegate . In other words, nobody had to struggle through an unfamiliar ‘lingua franca.’ However, the series did not dwell further on the technology so its function and precision was left to our imagination.
Imagine if we were to combine the IoT like goggles of the first example with the science fictional example. As far as I know, there is no AI-controlled simultaneous audio translation headsets in the market at this point, although there are human interpreters who would perform that function. While I have seen variations of the goggles in fictional dramas and movies (mostly from the US, China, and South Korea), and have even experienced the wonders of present generation of AR/ER goggles that are not only more attentive to the need of spectacles wearing users but also more delightful to the senses, much work still has to be done before such a combination is possible. But given how far research into AI has progressed, including in the arena of voice recognition and chat/talkbots, as well as the development of algorithms that could exploit large databases of language corpus in an unprecedented manner, one may see even a primitive version of such a gear in the next 5 years. The mushrooming of .ai startups will hasten the technological process in ways unprecedented.
Human/Machine Translation and Why the Tech Industry Should Care
As an avid film and drama viewer across different language groups, including in languages I have minimal to no comprehension of, I am a frequent user of subtitles. As a language learner, I would sometimes try to use a subtitle of the language I am learning that differs from the language of the show, even if I have native comprehension of the show’s language. For some of the streaming platforms I have used (mostly Asia-based), I found that these platforms are dependent on fan subtitlings, unless the drama or film being streamed come with their own selection of subtitles. For one of the Chinese platforms I use, it allows you to select more than a single subtitle, which is a gift for language learners. However, most streaming platforms these days are resorting to the use of automatic machine translation when they could not get human translators to do the job. But as most linguists and translation studies scholars will tell you, contextual translation between different language groups sharing minimal to almost no overlap is still a challenge for ML translators, especially when colloquialism or argots are deployed.
More recently, I have seen how entertainment content platforms are making room for the hiring natural language processing experts. However, this is insufficient in itself because of the politics of language, and the issue of accessing languages that are low-resource. Moreover, working with minority languages whose users refuse to cede the sovereignty of their language corpus to big tech firms require a different and respectful approach beyond the business-as-usual practice. Beyond expertise in computational linguistics, one would still require the expertise of other linguists, including those who specialize in translation studies. However, big-tech firms may still decide that these minority users do not constitute a sufficiently big enough user-group for their investment of time and resource, but such line of thinking is problematic for reasons I will explain.
Firstly, catering only to dominant language users increases the chance of misinformation, disinformation, and malinformation — all of which connects to how information could be unintentionally and intentionally misused; or misinterpreted due to contextual error. The reason is because of the reification of echo chambers and silencing of a vast majority of different minority language users of varying population size. We are only able to get news from those who could speak the dominant language, and potential misrepresentations are elided over. You might have noticed that some social media platforms, and their apps, allow you to perform automatic machine translation of posts from another language — the translations perform better in some languages than others, with no translations available for some of the less widely used languages on digital platforms.
Secondly, when you exclude non-dominant language users and/or disrespect their sovereignty, you are also discarding the principles of universal design. This principle suggests that when you cater to users facing the most challenges and obstacles in using available infrastructures and facilities, you will design a system that is not only user-friendly in all respects, but also adheres to principles of social justice and equity. Moreover, one of the biggest issues relating to platform technologies are privacy and data management — respecting the data sovereignty of various language users is one of the important steps toward mitigating that end.
Thirdly, any kind of exclusion is just bad business in the long run. I have myself done casual preliminary studies into the uptake of popular cultural content emanating from particular geographic zones by those living in other geographic zones — globalization is indeed a force as I was amazed to find the most unlikely participants in these cultural content despite bearing almost no overlap with each other (within the present day context, historical circulation being excluded from consideration here). For instance, as someone who observes how Asian content is being taken up by non-Asians outside of the English speaking world, I have seen constant requests for more subtitles in their own languages even as they strive to make sense of a culture alien to their immediate sensibilities (outside a few Asian delectables they could access in their home countries). In an age of prosumerism, we are beginning to see the production of content outside the present dominant languages on social media, and the production of such content will have a longer term impact on the consumer industry. Over time, these new adopters of the ‘alien’ content will also become avid consumers of products from the originating culture, whether through the purchase of consumer goods , merchandise, and even cultural content.
A Passion for Translation
Translating is not just the immediate interpretation of a spoken word, but also the entire translation of need and desire. In other words, when translation makes an object accessible, it creates new desire for the objects. In my own country of Malaysia, professional translations of cultural products into the national language which is Malay, outside of on-screen subtitling, are fairly limited. However, for those who could read other languages such as English or Chinese fairly well, translated works are more readily accessible.
In another a country using an official language that is familiar to, even if not entirely similar, to Malay, which is Indonesian, the practice of professional translation is more active. This means that for the longest time, prior to Google Translate, it was easier to find Indonesian-to-another-language and vice versa dictionary online than was the case for Malaysian Malay. Even to this day, if one uses Google translate, even the most dominant language, being English, fails to translate properly into Malay although Malay could not be classified as a low-resource language. Hence, for the longest time, native Indonesian cultural content have circulated more widely than native Malaysian cultural content. The same holds true when it comes to the localization of foreign cultural content. Even if you were to go to a small community library in Indonesia, you would find Indonesian translations of novels out of which popular K-dramas have been adapted from!
Translation also involves the fulfillment of a need and desire. In a previous article, I have mentioned how merchandise could be created out of popular cultural content (i.e. trash fiction). The production of a merchandise is itself an act of cultural translation — the successful promotion and dissemination of a merchandise across the globe is a mark of successful translation. How does the translation take place in this regard? What is translated here is a common dream and common desire for a particular fantasy or product. In other words, one has to show to someone in another continent why their desires and fantasies have strong resonances with the desires and fantasies of the source of the cultural/linguistic content. Opinion leaders and influencers do play a role in this kind of translation. However, all of translations begin with the medium of communication: language.
Coda
We look forward to the day when the machine and human could interface with each other seamlessly, not merely as individuals to machine, but as a community of humans to machine(s). After all, translating an alien environment to familiarity is the consequence of crowd-sourcing available tacit and explicit knowledge surrounding that environment. Therefore, the technology created should not stop us from engaging with another human more directly, but instead, encourage that engagement from the outset. Hence, when you listen and you sense, you desire to commune and to understand, rather than to isolate yourself by constructing a wall of unchallenged monolingualism.