AI may be our best hope for talking to aliens
Back in 2008, I started writing a novel about alien languages. I was watching House a lot at the time so I crafted the story around a…
Believe it or not, there is a substantial amount of academic literature on alien languages, despite there being no examples of extra-terrestrial communication. We have many examples of terrestrial non-human languages. The languages of bees are a good example but also prairie dogs. I think it is reasonable to say, however, that no species on this planet comes close to the variety and complexity of human language. Nor is it clear what level of consciousness animals have of their languages. Consider that computers can use highly sophisticated languages, including ours, without any conscious understanding of it at all. Thus, language and what we think of as conscious intelligence are separate.
A technological species on another planet, we expect, would have to have at least as complex a communication system as us. That communication system would have to be capable of communicating abstract concepts sufficient to achieve not only a functioning society, but also engineering, mathematics, and science.
Most science fiction languages are not created to be truly alien languages. Studies of languages in science fiction and fantasy such as Klingon, Elvish, Dothraki, and so on find that they are sound symbols. That is, they are a form of sonic symbolism that evokes the character of their speaker. Klingon is a warrior’s language and is also deliberately constructed to sound non-human with difficult to pronounce phonemes. Elvish communicates the sophistication and beauty of the elves. Indeed, Tolkien was a master at this art and would form languages that he would then adapt the species to rather than the other way around.
Created languages are distinguished from constructed languages like Esperanto and miniature languages used in scientific research. Created languages frequently do not follow any linguistic logic and are not designed to convey information. Their only purpose is to tell the reader or listener something about the beings that speak the language. Occasionally, linguists have attempted to construct languages out of created languages such as Klingon, fixing many problems and providing a logical grammar that may be absent from the few examples in fiction.
When I wrote The Xenological War (* see note), I wanted to emphasize the problem with attributing human-style language to aliens, so I deliberately did not include any examples of alien languages in the book! In places I would describe some aspects of how the languages were communicated and some aspects of alien culture, but I felt that alien communication would have to be so different from human language, that to include it in the novel would have immediately humanized the aliens, falling into the sound symbolism of created languages. To me the very concept of words and grammar are innately human, and, while I can imagine an alien evolving these for sure, I could also imagine that they might communicate concepts in an entirely different way. A lot of their language would be based on how they actually parse the world with their senses and what makes sense to their normal means of interacting with it and each other.
Indeed, linguist Noam Chomsky, who pioneered the theory of a universal grammar, has suggested that aliens would see all human languages as essentially dialects of the same language because they are so similar to one another compared to their own. While an alien language might have words and a grammar, it is possible that it bears no resemblance to any language on Earth and no human baby would be able to learn it even if they grew up hearing it. That is why in the book I assume that computers, meaning AI, will ultimately talk to aliens for us much as the universal translator does in Star Trek or C3PO does in Star Wars.
My fictional universe recognizes this real problem that linguists and even NASA have struggled with. In a true alien encounter, whether face to face or, more likely, via radio, how would we recognize and understand what the aliens are saying? Carl Sagan’s 1985 novel, Contact, goes into this problem in detail, suggesting that the “universal” language of mathematics would provide a basis for communication. As a mathematician, I am not so sure, for, while mathematics may be universal, our way of representing it is just as human as any other language, and we would be hard pressed to get past basic numbers to more advanced concepts without considerable “human” language or at least some formal logical notation such as predicate calculus. Also radio waveforms are a language in themselves. To go from waveform encoding to mathematics concepts to true communication would require code breaking skills that I think even the world’s top spy agencies would struggle with.
Given that even human languages of ancient times such as ancient Etruscan are still largely undeciphered, how likely is it that we would be able to decipher an encoded alien message from lightyears away?
Of course, there is no requirement that either we learn the alien’s language or that they learn ours. We just need a means of communication that is mutually intelligible and sufficient to meet our needs whether that is exchange of technology or saying “please don’t blow up our planet”.
In a paper published in the journal Nature, the following questions are posed:
assuming we want to communicate meaningfully with other cognitively advanced beings, how can we possibly do this without any pre-established system in place? In other words, how do we bootstrap a shared, mutually intelligible language from nothing?
Studies of humans often use a process called Artificial Language learning (ALL) in which participants are asked to learn constructed languages in order to understand the process of learning languages with certain features. These languages are taught by matching made up words with images. Could aliens teach us their language by matching words to images too?
I have my doubts because while humans tend to parse images in similar ways, even non-human terrestrial beings do not. What is clearly a hammer to us may just be an abstract drawing to an alien who has never seen one. What is required, it seems, is first a common frame of reference. That frame of reference could be things like star charts. After all, presumably both the aliens and we have access to the stars.
With the Voyager Golden record, Carl Sagan and others hoped that such line drawings and recordings on it would be decipherable to an alien culture. And of course, we can only hope that the aliens have ears that decipher sounds in the same registers as ours, as that is what is contained on the minuscule bumps on the record.
A few features of language are probably present in alien languages, we can assume, because they are important to enabling complexity. One feature is called combine or merge and repeat, which allows language to be built up from the combination and repetition of clauses. Mathematical notation uses the same feature as do computer programming languages.
Then again, it is possible that alien languages are like nothing we can imagine.
Consider the case where Facebook AI trained two bots to negotiate with one another. At first the bots used English but, having no incentive to use it, they quickly diverged into developing their own language. These languages, while artificial, are completely alien and the researchers had no way of understanding what the bots were talking about.
As scary as that sounds, it is a fascinating insight into the difficulties of understanding alien language, especially from afar. Unless the aliens take it upon themselves to teach us their language, we have no hope of understanding theirs and vice versa.
One fascinating experiment, that might remove the universal grammar constraint in human studies, would be to try to get a set of two groups of AIs to develop their own internal languages with no commonality between them, and then see what happens when they are forced to communicate with one another. What approaches would they take to teach each other their languages? What if they could only communicate through radio waves?
I propose that they would probably find a way simply based on information theory, that with sufficient information packed into the waveform, the small grain of meaning could be extracted, but what would be more interesting is how long it would take, i.e., how much would they have to communicate to understand each other.
That is a Ph.D. thesis for somebody, I’m sure.
And perhaps mathematics, particularly geometry, is still the best subject matter for communicating with aliens. Yet, I think that, while mathematics may be the offering, language still has to be the currency, and solving that problem requires more research into the only alien surrogates we have that can speak human languages: computers.
So, it may not be us talking to aliens directly in the future but our own version of C3PO. We’ll just have to hope he is on our side.
* Note about this article: Back in 2008, I started writing a novel about alien languages. I was watching House a lot at the time so I crafted the story around a gifted expert in alien languages who was also something of a misanthrope named Tolan Smith. I made the narrator one of his young assistants who is, in all honesty, based on my 20–something self. It is called, The Xenological War, and you can still find it on Amazon. With the serialization of its sequel, The Xenological Society, here on medium and on my substack, I wanted to talk about alien languages both in science fiction and what they might be like in real life.
Cheyne, Ria. “Created languages in science fiction.” Science Fiction Studies (2008): 386–403.
Vakoch, Douglas. “Can we speak alien?.” (2018): 22–23.