Tech, Star Trek, and Language: Are we lost in translation?
Star Trek has both raised and helped us understand philosophical issues. These run the gamut from politics, to games, to identity, to our relationship with technology. The Universal Translator, a device that translates the diverse fictional languages into the characters’ native tongues seamlessly and accurately, has been a core piece of technology in the Star Trek universe from the very beginning. Sort of like the fictional universe’s transporters (technology that converts matter to energy, beams it to a new location, and rematerializes it), it was a practical necessity for creating the show. It wouldn’t make for very good television if none of the fictional characters could understand one another. Unlike the transporters, though, Trek never grappled with the philosophical issues raised by real-time, cross-cultural translation.
Today, many companies are working on real-time translation software. Google, with Google Translate, is the most famous and most used. Other tech giants like Microsoft and Amazon have their own (Microsoft even refers to theirs as a Universal Translator on their blog.) Plus, there are myriad smaller companies, like Poliglu and Pairaphrase, trying to crack the market. They all recognize the potential for real-time translation.
These translators use machine learning techniques to do their translations. They are fed copious amounts of data, analyze the data for patterns, and learn the translations. Google, for example, shifted from a statistical machine approach to translation to a neural machine approach in 2016. Some companies use algorithms that crawl and index the internet for linguistic data. Anywhere people speak or write will work. The trouble is that real language is flexible. Any given word can have many different meanings. That can make the software’s translations confusing. The newest trend - as per the latest Google approach - is to expand the units of translation from words to sentences. AI and machine learning can use context to choose the right definition for each individual word within the greater sentence.
The fictional technology of Star Trek works a little bit differently. Their Universal Translator does not need to compare two languages to translate one to the other. It works on the assumption that there are basic patterns and commonalities in all languages. The translator only needs to hear a bit of the language to learn that language’s patterns, and to effectively produce translations.
As AI advances, the Star Trek-like Universal Translator comes ever closer to existence in the real world. It is not clear how successfully people will adopt the technology. We need to have a broad, public discussion of these issues, and Star Trek can help. Only through an open dialogue can we learn to trust our version of the Universal Translator. The philosophical questions which arise from a Universal Translator are two-fold: first, we must consider whether such a device is epistemically possible; if it is, then, we must consider whether - and in what ways - its use would be desirable.
Any translation is rife with epistemic issues – issues surrounding the production and perception of knowledge, and thereby languages (in fact, it’s an open question for some whether linguistic translation is even possible). Two things make the Universal Translator different. The first is that it is real-time and supposedly seamless. With old-fashioned translation, if there is ambiguity or confusion, the translator can elaborate. They can explain why they chose the word they did and even offer up some alternatives. This happens all the time when reading an English translation of a philosophical work. The Universal Translator cannot elaborate as such; it is designed to facilitate naturalconversations. The only way to do that is to pick a translation and stick with it. The translator can’t consistently stop the conversation to say, “…he said love, but it’s not like romantic love or paternal love. It’s a little like what the Romans would call caritas with a connotation of. . .”. Without elaboration, we will have to decide on an appropriate level of trust when using the translator.
The second related difference is deciding what to do when an object, action, or concept in one language simply does not exist in the other language. The translator could leave the word untranslated. We do that frequently - when there’s no good translation for a word, we leave it as is. We didn’t translate the word quesadilla, for example. The other option is for the translator to coin a term that gives us a sense of the meaning. This happened in the Star Trek Enterprise episode, “Cogenitor”. Enterprise encounters an alien species that has three biological parents involved in the reproductive process: the mother, the father, and a cogenitor. Trip – a Starfleet officer in the fictional universe - befriends a cogenitor and is upset that they are treated poorly by the humanoid species known as the Vissians.
The episode deals with lots of philosophical issues from gender identity to bigotry to the Prime Directive. One thing that it never deals with, however, is the word cogenitor. It would be a shocking coincidence if the Vissian word for third parent is cogenitor, and if their words had roots like the English words for ‘with’ (co) and ‘creation’ (genesis). That likely means the Universal Translator coined the word. It sounds similar to progenitor, so we could assume it has something to do with parenting or creating, but we really don’t know. And so, if the other culture is significantly different from our own, it raises the question of how successful communication can be, even with a translator.
Then, there are the ethical issues with translation. Any time two or more parties have different interests, there are potential ethical issues. For instance, a speaker may wish to be understood, while a listener does not want to be offended. If the speaker says something insulting, which is more important? Is it the fidelity to what was said, or to the listener’s feelings? In the real world, we currently have a scattershot approach to this problem. We sometimes censor things that might be offensive, indicate that an offensive word was used, “he used a slur for Jewish people,” or sometimes simply use the words unfiltered. A lot of the decision-making comes down to context (more on that in a moment).
Star Trek is similarly inconsistent. Sometimes, the Universal Translator leaves words untranslated. The most famous example of this is the word “P’tahk” from one of the universe’s fictional languages. One character described it as a curse word. According to Memory Alpha (a great fan site for all things Star Trek), the original definition was “weirdo,” but it certainly sounds more offensive than that. To my ears, the way it is used sounds more equivalent to when English speakers refer to someone as a non-human (worm, snake, B-word, etc.). Whatever it means, the Universal Translator doesn’t translate it. At the same time, the Universal Translator readily translates another language’s slang term for “humans” as “pink skin.” There is debate whether this constitutes a slur, but it is certainly not a compliment.
Ultimately, as in most ethical consideration, context is king. If a hate group like the KKK, for example, were holding a rally, any translator should probably censor their speech. But if African Americans use the exact same words while talking to each other, it should be unfiltered. An academic discussing hate-speech should be able to say the words being discussed but shouldn’t be able to use them. In other words, intent is huge when evaluating the ethics of a speech act.
A Real Universal Translator
Will we get to a point where we can trust a Universal Translator with recognizing intent? Given how variable people and their feelings, intentions, and speech acts are, the answer is: probably not. People can’t always trust that they know their own intent. Will we reach a point with a Universal Translator where we can be confident that the speaker and listener are understanding the same ideas? Again, unlikely. It is hard enough to do that with two people who speak the same language. Even in the Star Trek universe, the Universal Translator falls short.
In an episode of Star Trek Discovery, there is a scene where the Universal Translator malfunctions and starts translating everyone’s speech into different languages. The crew can’t even read their computer displays. One character, Saru, is able to fix the problem because he is the one crewmember who - as he puts it - “bothered to learn a foreign language.” The scene is a plot point played for laughs, but it is Trek’s best lesson about translation technology. There are many potential benefits to creating a Universal Translator. For example, it could give us unprecedented access to other cultures. It could make migration much easier. It could remove barriers to employment and education. We should pursue the technology. However, unlike everyone except Saru in Star Trek, we should never rely exclusively on the technology. Some things, like treaty negotiations where it is vital for everyone to be on the same page, will still need a sentient, conscious interpreter who can interact with the participants. The bottom line is that there will never be a true substitute for learning another language.