Thursday, January 7, 2010

Alien Languages - how foreign would they really be?

This post was requested by CWJ, my friend from the forum over at Analog - thanks so much for the question, CWJ! It also strikes me that this may be a timely topic for people who are considering the Na'vi language that was used in Avatar.

CWJ asked:

Juliette, I'd like to hear more about (constructing) non-human languages. In particular, if Chomsky's idea of universal innate grammars is correct, does that mean there are only certain avenues down which humans can go, which might be different from aliens? That is, maybe there are some concepts or constructs that would be difficult for humans to truly conceptualize. Or the other way around. In short, I am interested in the possibility that communication may be very difficult.

This is a complex question, so I'll take it a bit at a time.
First, the Chomsky question. Chomsky proposed the idea that there was some basic sense of grammar universal to all humans, that was passed on as an instinct.

Now, human languages are very diverse. The most thorough article I've seen on this topic was recently published in the Economist, and you can check it out here.

In fact, it's hard to say how much of human language is innate and how much is learned. Humans are oriented towards language from birth or even earlier; this is well known, as newborn infants prefer to listen to language sounds over non-language sounds, and their mother's native language over other languages (studies measured strength of sucking response!). They also go through a number of language development stages, like early babbling, even if they don't have any auditory language input (say, with non-hearing babies). Non-hearing babies are also known to babble with their fingers. People have also looked at pidgin languages, which tend to take on grammatical structure - and very similar grammar structure - when they're passed on to the second generation, and used this as evidence for a more extensive innate language faculty.

On the other hand...

I've been quite impressed by research which looks at language acquisition from a neural-network point of view. Neural net computers have been shown to learn language patterns like English past tense -ed in much the same progress trajectory as human kids. I've also heard about research that says re-occurrence of phrases may play a larger role than we thought in language acquisition. Certainly human brains are very good at tracking the frequency of occurrence of things (sounds, words, etc.) - a critical skill for language learning. So in the end I'm not personally convinced that grammar is really what's innate and at the root of the drive for language acquisition.

I suppose if there were some kind of universal grammar beneath all human language, then it might be restrictive for the learning of alien tongues. Since I'm not really in the innate grammar camp, I don't think that the primary restriction on learning alien tongues would come from that department.

I think it would come from perception problems. More on this below.

The evolution of language is not simply a matter of brain evolution, but of the co-evolution of the brain, the ear, and the vocal tract as language developed. As a result, they are all well-suited to one another, and babies can hear any language sound from any language in the world, and learn it natively, given the opportunity and a normal course of development. We listen for audible building blocks from the vocal tract that are put together sequentially. Our brains are excellently tuned to process them, and we associate them with physical, temporal and social context in lots of complex ways. But alter these very basic prerequisites, and the problem becomes much harder (even if we assume maximum language-learning ability like that of a child). Sign language shows that the language stream need not be auditory. But what if the language stream is not sequential but simultaneous? Or what if the language producing organs of the alien create a language stream that our eyes and ears are unable, or only partially able, to perceive?

Sheila Finch's stories about the Guild of Xenolinguists speak more directly to the kind of language problems that CWJ mentions - basic problems of auditory vs. visual, processing in the brain and such - than my own. In fact, CWJ, if you haven't read them, you might find them very interesting, because she takes a very Chomskyan approach to her concept of language. She has humans being able to understand all sorts of things, but requiring special drugs to make them forget their existing categorizations of perception, for example. Speak with her in person, though, and she'll tell you - as I will - that any communication with aliens would be next to impossible.

In the realm of animal communication on Earth, we're still discovering things, like the super-low sounds produced by giraffes. We're working hard on the communication of dolphins, too (see a very interesting article on dolphin intelligence, here). We've taught some creatures how to interpret basic signals on a behavioral basis (I remember Mike Flynn having an interesting post on the nature of communication - I'll see if I can find the link). But we haven't really cracked any codes. One of the things that can cause misunderstandings between humans is differences in categorization of concepts - places where the two languages file things differently, as when the Dutch say a picture is "up the wall" instead of "on the wall." A creature that lives underwater and perceives its world through sonar signals will have a totally different way of perceiving the world. It may not even conceptualize the separation of objects as we do. What does it do to language concepts when the means of producing language (sound) is the same as the means for perceiving one's surroundings?

So effectively I think it would be hard to recognize alien language as language at all, and probably harder to try to "break it down," especially in a situation where the physiology of the aliens in question, and their environmental context (not to mention social context) were unknown.

This doesn't stop me from designing alien languages, obviously! As far as constructing the languages goes, I think it really depends on the author's intent with the story, and the nature of the primary language problem in the story. If decipherment is your primary problem, then you can really embrace problems of channel (auditory/visual etc.) and the identification of structure. If your primary problem is one of first contact and code cracking, then you can do some channel stuff, or you can focus on grammar or phonetic dificulties, pronunciation difficulties, etc. If your primary problem has to do with cultural issues and misunderstanding, then it's helpful to create a language and assume that humans have already cracked the code, which allows you to place the focus where it really belongs.

Wow, that became a long post! I hope you find it helpful. I welcome any questions, followup, "what-the-heck-did-that-mean-can-you-explain-this-bit-again-please," etc.


  1. All of what you wrote is thought-provoking and perceptive, I think. My only comment is that this is related to a more fundamental problem, which is that of intelligence in general and its relationship to communication.

    If I meet another human and neither of us have ever heard anything like each other's language, we can still communicate by trial and error, gradually defining a new language as we create common signals and watch for reactions. On the basis of the new, common language we develop, we can start to teach our native tongues (or, if one manages to learn the language of the other directly through such interactions, he can teach the other his own later). However, the ability to do this is based on a number of necessary preconditions:

    1) We must be able to interact (which wouldn't necessarily be the case with aliens, say if we have only a radio signal or a written record to work with).

    2) They must be trying to communicate with us. It isn't necessarily the case that they would try to communicate with us.

    3) We must be able to detect the signals they send as the means of communication, and they must be able to do the same for us. This isn't the same as the signals of their native language, of course. If their language is based on touching cheeks and varying the rhythm of their circulatory system, we might never recognize it on our own, but if we can establish something along a line of perception that we share (say, vision or touch), then a means of signaling may be established.

    4) Our respective intelligences must be able to decipher the patterns used in the signals. Even if the mode of perception is similar, it's an open question whether what we consider intelligence itself follows necessary and universal modes of operation. I think we all suspect that "1111, 2222, 3333, 4444, ...." seems to establish a pattern (even if you don't know the meaning of the numerals, the groups of four would seem to stick out as long as the signaling methodology is clear), and I tend to assume that any true intelligence should be able to decipher the pattern. But who knows? If it's not, our efforts to communicate might be stymied.

    This is probably only a partial list, but I think it's enough to show that the language problem is both easier and harder than just the matter of alien linguistics. Easier, because if we ever met another intelligent life form, we might be able to each come half way; but harder, because the success of even forming a new, intermediate, common mode of communication presumes deeper, far more intractable problems.

    (Sorry, long comment!)


  2. dbp, thanks for your comment! While I didn't put it the same way and in the same detail, I think you'll find some reference to each of your points in my post above - which is to say that I generally agree with you. Interaction can potentially happen at a distance, for example, but without context it's going to be very tough to do any deciphering. There are a lot of very basic ingredients that humans get to take for granted when trying to understand one another's languages.

  3. You're right, of course, about the possibility of interaction at a distance-- in my point #1, I was really referring to one-way communication, such as radio signals originating thousands of light-years from the nearest accessible location. I should have been more clear.

    As to the larger question, I suspected you'd already thought about everything I wrote, but your post was mostly about understanding and learning an alien language (and what influences that). My comment was primarily about the qualitative difference between that and a related but different problem, namely, communicating with an alien. (The latter may be a potential stepping-stone to the former, or at least a way to get around the impossibility of the former.) They aren't the same thing at all, really, but both point back to common, underlying problems like the ones I listed.

    The interesting follow-on topic is, human-alien interaction could potentially produce an entirely new language, utterly new to both us and them, on the basis of what we can find in common. Suppose we meet a species that (unbeknownst to us) communicates by circulatory variation (surely physiologically dissimilar to ours even if we could make our hearts beat at will). Maybe the only thing we can both recognize that we have in common is that a particular cluster of five tentacles appears (to both our visual systems and intelligences) similar to one human foot. On the basis of that, pioneering members of both species create a toe-waggling language, unlike verbal speech, unlike any traditional sign language, and unlike manipulation of circulatory systems.

    The final related possibility is two-way communication by dissimilar languages. A person with no arms could presumably learn to read sign-language, and a person both without functional vocal chords (but otherwise healthy) could presumably learn to hear and write English. These two could learn to communicate with each other naturally, but using completely different "languages" (signing and speaking) for their half of the signaling.

    Sorry if I'm writing a lot. It's just a very interesting topic for me. I'll be quiet now. :)

  4. dbp, you're welcome to comment as much as you like. You make some excellent points. Part of the reason for my emphasis in the post is that I think some of my readers may not be aware of the large degree of natural variation in human languages; part of it also comes from CWJ's comment about Chomskyan models of language acquisition. I've attended two panels about topics related to this: one I was listening to Sheila Finch and other authors discuss the problem of deciphering actual alien language, and the other I was on the panel with her and Dr. Stanley Schmidt of Analog discussing the design of alien languages for use in stories. To me they are two very different questions, precisely because of the kinds of issues you mention. To have two creatures start to communicate already assumes a lot of underlying commonality of perception, channel, and method.

  5. IMHO, the development of language requires grammar, but not in the sense of some "innate grammar," unless that means only "a tendency toward grammar." (Oh, how sneaky the old Stagerite is, getting those final causes in when no one is looking.) No one needs a language to tell someone, "This is an apple," or "It is raining." That we are all getting wet is self-evident. But you do need a language to say, "If we built a lean-to we could be dry, and dry tomorrow, too." The key concepts being "If" and "could" and "tomorrow." These are things you cannot point to and grunt. Where is "if"? How big is "could"? Show me "tomorrow."

    IOW, language originates not from signs for things, but from symbols for constructions of things.

    And since the Latin word for contruct is "fictio" and the Latin word for knowledge is "scientia," that brings us to science fiction.

  6. Good to see you here! I agree with you about grammar. Really grammar exists to allow us to step away from immediate context - precisely your if/could/tomorrow point. I think there's probably a tendency toward grammar in humans, but it may be the same as, or have evolved from, a general tendency to take fuzzy sensory input and make meaning from it, by identifying separate objects and categorizing them. Useful for remembering which fruits are edible and which are poisonous, and guessing whether an unfamiliar fruit is more likely to fall into one category or the other (for example). I haven't read extensively about creole languages, but I do find myself curious as to what grammar-related information might be perceived in the pidgin spoken by generation 1, and potentially picked up on by grammar-sensitive children.

    Like the science fiction connection, too. Constructed knowledge, indeed.

  7. As a strictly non-professional here, I'd really love to hear what you think of both the language acquisition method and the creating of a language in Suzette Haden Elgin's "Native Tongue". I always found that one to be extremely fascinating.


  8. Sam,
    I haven't read it! Arg... If I can, I'll go out and find it, and read it, and then I can get back to you on that. Thanks for the suggestion.

  9. Juliette, the first one is VERY good, the second interesting, and the third IMHO was not really worth it.

  10. Thanks for this post, Juliette! Sorry I didn't comment earlier, I was out of the country and had limited internet access.

    I happened to be thinking about this question myself a bit while traveling. I know from my own reading that there is enormous variety in what kind of information is carried by grammar. For example, in English we have past, present and future tenses for verbs, but as I understand it such tenses are not as obligatory in some languages as in most Indo-European languages; while other languages can have important grammatical elements that are not emphasized in English, for example, how the speaker knows the information.

    But other things seem close to universal. Most if not all languages seem to be classified by their subject-object-verb order, with english being SVO but other languages SOV and so on. And I began to wonder, could one construct an alien language in which the fundamental blocks are not nouns and verbs, but something entirely else? It's hard to imagine, as I'm prejudiced by my own language to think in terms of things (nouns) and actions (verbs), but maybe I'm just limited by my own Chomskian language modules. :) (I'm agnostic as far as Chomsky, not knowing enough to have an opinion). It'd be really interesting to construct a language along other lines.

    And yes, I've read a number of Finch's stories and always liked them.

    Again, sorry for the late comment, and if this is slightly incoherent I'll blame jetlag... this time.

  11. CWJ, it's great to see you here. I'm glad the post addressed some of the things you'd hoped for. I agree that subject, verb and object seem close to universal. I think that it would be possible, though, to construct a language that didn't use them as we do. SVO etc. are predicated on the assumption that we are subjects, that objects exist, and that we act upon them. An alien who did not perceive its world as being made of discrete objects would probably have a very different kind of language, as would one which did not conceive of itself as acting on its own volition somehow. That would be an interesting language to design, but would probably require a different kind of transmission method from the one we humans use. I'm speculating, of course. Because it would be so entirely foreign, it would be best approached from the outside in a story context - you certainly couldn't write a narrative with it that could be comprehensible to readers.

    What interesting ideas! Thanks again for the question.

  12. I tend to disagree about classification by subject-verb-object *order* since my native language (Latvian) doesn't really bother with that. SVO is slightly more conventional, but a sentence "a dog sees sees a mouse" is valid either as SVO "suns redz peli", as OVS "peli redz suns", VOS "redz peli suns" and all the other variations. Basically, we usually put subject first because it's naturally emphasized.

    I think that understanding a language is much more about how people (or aliens) think than we are aware. I'm thinking about derivative languages like LOLspeak and WrItInG lIkE tHiS or wr1t1ng 4 hax0rz. If I tried to figure out how "the code" works I'd go nuts. But I'm able to read this sort of text quite easily because I drift into another thinking pattern. This is very similar to reading old texts of a foreign culture (say, chinese I Ching, especially academic translations). Or, conversely, reading a story written by a good friend where you can practically hear him reading it out loud. And, most of all, I'm thinking of children and how they learn to speak - they don't "crack the code", they slide into it and start living in it.

  13. Ieva,
    Thanks for the comment. It's really not a question of "disagreeing" with the classification scheme - it's used throughout the world by linguists for a very good reason. Thanks for pointing out that Latvian has flexible word order. It's a good point, and I'm sure there are many other languages out there where the order can be similarly manipulated. Still, in the languages of the world, a default (non-emphasized) word order can generally be determined that falls into one of these categories. Subjects, verbs and objects are universal in their presence in world languages so far as I know.

    I agree with you that language patterns influence how you think. Sapir and Whorf argued for a very strong version of this connection between language and thought; while the relationship may not be deterministic, it is certainly strong. We do language a disservice if we think of it only in terms of code, and message-sending - for children or for adults.

    Thanks again for your comment.