I repeat myself when I say that effortless understanding across languages and cultures is an age-old dream of humanity. We have searched for it in many places: in miracles, in futuristic imaginary technology, in real-life technology, and in imaginary biology, too. This post will give an example of each, with further resources to explore.
The best-known linguistic miracle is the Pentecost:
“All of them were filled with the Holy Spirit, and they began to speak in other languages as the Spirit enabled them. Now there were devout Jews from every nation under heaven residing in Jerusalem. When this sound occurred, a crowd gathered and was in confusion, because each one heard them speaking in his own language. Completely baffled, they said, “Aren’t all these who are speaking Galileans? And how is it that each one of us hears them in our own native language?” (Acts 2:4-8)
This totality of understanding goes beyond language: it is a symbol of redemption. On the one hand, understanding is a prerequisite of peace; on the other hand, in the Jewish tradition, the confusion of languages is a consequence of human sin [see the story of the Tower of Babel, referenced in the opening post].
In the millennia that followed this event, many people at various levels of cultural sophistication shared one obsession: to reach redemption by finding God’s language and returning to it. Some of them went beyond speculations, and conducted experiments at varying degrees of cruelty.
I plan to elaborate more on this storyline, but impatient readers – and those who wish to go deeper – can turn to a book called The Search for the Perfect Language (The Making of Europe) written by a favorite author of mine, Umberto Eco.
I’m not being entirely fair with all these researchers, probably because when it comes to learning about the world, I’m in favor of science rather than other approaches. But there are scientifically valid attempts at finding a common conceptual foundation that all languages share. Some of them succeeded in representing a subset of thoughts by abstracting them from language, and using language-agnostic symbols to describe them. Such things are the modern symbol systems of mathematics or formal logic. In linguistics, scholars have been giving serious thought to finding universals, characteristics that are true to all languages.
On this note, in later posts I plan to talk about Aristotle and Descartes, and the systems they conceived – both of them were serious scientists in their respective times.
Here we go again: human progress is quite unpredictable. The yearning for total understanding gave us a lot of scientific benefits, but not the way most dreamers originally intended.
Modern dreams follow. Imaginary technologies fall in the territory of science fiction. These are usually computers or robots performing speech translation – this is the most realistic approach, easy to grasp for the audience. The most famous professional interpreter of this sort is probably C-3PO (See Threepio) from Star Wars, who is, as he boasts, “fluent in over six million forms of communication”. However, the Star Wars stories don’t tell us how he acquires new languages, that is, new forms of communication.
The first story I read that featured an ‘interpreting machine’ was a Polish novel by Czesław Chruszczewski called ‘Fenomen kosmosu’ (Cosmic phenomenon, 1975) in the Polish original. This book has a highly symbolic storyline, and although it more or less follows the ideals of the Communist regime, it is a fascinating read. Unfortunately, it was never translated into English. (I can’t boast to be able to read Polish – I’ve read this book in the Hungarian translation.)
‘Fenomen kosmosu’ had an interpreting machine that could learn a new language in a matter of hours, apparently without human assistance. The method is not described, but there are plot elements that are included because of the training process.
The technology that I find the most ingenious was included only to help with the continuity of the plot. I’m talking about the Universal Translator in Star Trek, of course. A lot of episodes in each series have humans in a first encounter with an alien species – talking to them in plain English, with the aliens responding in plain English. This is naively unrealistic. To make it more credible, screenwriters conceived technology that performs translation using (wireless) neural transmission rather than text or speech.
The result? Everyone speaks their native tongue, and hears everyone else speak in their own language. This is the language miracle of the Pentecost, in a futuristic, high-tech, man-made version.
We get the most insight into the Universal Translator from the lowest-rated series of the franchise, Star Trek: Enterprise. This series has episodes that feature fairly realistic linguistic technobabble, and we also get to peek into how the UT gets programmed by a highly accomplished linguist.
When acquiring a new language, the Universal Translator does not need a linguist to enter all linguistic data from scratch – it is capable of ‘analyzing’ the language, and set up initial structures.
Expectations toward real-life machine translation have always been unrealistic – to a varying extent, depending on the period.
To make themselves more at home with new technologies, humans use a clever strategy: they anthropomorphize machines, as if they had human characteristics. Thus computers became ‘electronic brains’. That’s how early articles referred to them. As a result, people attributed cognition to computers, while many researchers used the assumption that human brains work by largely the same logic as computers.
This is why early human thinking about computers qualifies as a dream, not unlike those that appear in legends and in fiction.
Early machine translation was also influenced by a rationalist view of human language, assuming that languages (and their underlying cognitive structures) are rational or logical to a great extent, and also that there are sufficient linguistic universals to create a common foundation where one can go back from the source language, and build the target-language structures from the same foundation.
This is very similar to the quest for the perfect language, although its motivation must have been more practical. The translation logic had to be very simplistic and very structured to fit into the very limited resources computers had in the 1950s and the 1960s.
The output of early machine translation was quite poor: it simply didn’t work to produce any of the expected result. The ALPAC report of the US Department of Defense points this out quite harshly.
It isn’t enough to say that the early models were inadequate, and that we only need to find a better model. Language is more deeply intertwined with human cognition than we first thought. And human cognition is not entirely rational.
Because language is intertwined with human cognition, units of language cannot be tackled independently. The ambiguity of words is the most obvious sign that language goes deep in our minds, as Yehoshua Bar-Hillel pointed out in 1952.
It turns out that the human brain is still more powerful than any computer or network of computers in the world. Although machine translation does make progress, even the disambiguation of the sense of one single word requires an enormous amount of data and processing power. No matter what partial results we see today, this puts truly high-quality natural language processing in the distant future.
Disambiguation means the capability of distinguishing between various meanings of a word, depending on its context – the other words around it.
In fiction, and to a lesser extent, some human beliefs, translation is performed by humans with extraordinary powers or fictitious organisms.
In the simplest case, there are people who can learn languages and translate like no-one today. An example is Hoshi Sato, a fictional character from Star Trek: Enterprise. We might get there someday – we know that humans’ language capabilities have evolved over time, and who said it’s over?
Many stories feature telepathy. In fact, telepathy is believed to exist by many humans, but in most cases, what we might perceive as telepathy is actually the result of extra-linguistic communication like intonation and body language, which are essential to human communication in addition to speech.
The most important work of fiction that includes acquired telepathy is the Foundation series by Isaac Asimov. (In another story of his – called Nemesis –, the protagonist can consciously read body language, which, as far as the result is concerned, qualifies as a sort of telepathy.) In other stories, humankind also meets telepathic aliens. One of the most interesting books of the kind is Ceux de nulle part (Those of nowhere), written by Frenchman François Bordes (going by the pseudonym Francis Carsac) in 1954. An English translation is painfully missing.
Non-sentient telepathic organisms also help with translation. The most famous example is the Babel fish, conceived by Douglas Adams in the Hitch-Hiker’s Guide to the Galaxy series. You slip the Babel fish in one of your ears, and it will re-scramble the brainwaves around you, so that you will understand whatever aliens speak around you. I believe Douglas Adams used the Babel fish to mock other ridiculous ideas in some science fiction works.
In Douglas Adams’s universe, they no longer manufacture a lot of things because everything grows naturally on a planet or another. For example, a tree yields ratchet screwdrivers, and mattresses grow in a swamp in another place.
Another interesting idea appears in the TV series called Farscape, where you get injected with bacteria that translate for you, pretty much the same way as the Babel fish.
The history of technology tells us that humans mean business when they dream about the future and strange lands, and those dreams fuel the development of new technologies in one a way or another. And when this happens, failures are often more interesting than successes because they can spawn unexpected novelties with unexpected benefits. The history of translation technology is an outstanding example – many stories remain to be told in future posts.