XENOLINGUISTICS
According to David Graeber in his anthropological study Bullshit Jobs(2018) the highest human achievement is the capacity for make-believe. To imagine aliens arriving and wondering what, if any, language they speak or use, must therefore rate as a very worthy project. However in recent times the alien has taken on the form of the beast already being here, made by humans, Frankenstein-like AI robots that evoke fear and loathing lest they turn out to be as stupid as the humans that make them. To discuss the language of the alien is therefore a two-sided affair.
While in the first instance humans and aliens can imagine everything and anything, they are unlikely to be able to escape the laws of nature as we know them, and as such the universe (or multiple universes) and everything and everyone in it, are absolutely constrained by the laws of nature (e.g. E=mc2). Language is no exception, unless we invoke the ‘following the rule paradox’ (via Wittgenstein and Kripke) that states that we cannot apply known rules to unknown phenomena.
So, at the level of present human knowledge we do not know if we know all the laws of nature (e.g. the particle responsible for gravity was only recently confirmed). The ones we do know certainly define and constrain our existence (language included).
Since human language(s) seem/s to have the unique ability to reflect upon itself/themselves, we are tempted to ascribe to language a super-natural aspect simply because we can express such sentiments. Following Chomsky and bio-linguistics we assume that languages and thoughts are generated by biological processes in the brain, hence there is no constraint on output, at least in terms of infinite expressions that can be generated via MERGE and iteration. That many of these expressions may not be reflections of the real world is well attested, both as pathology and as poetic licence (something similar confounded Wittgenstein).
Science fiction (a hybrid between reality and fantasy) exploits the grey areas between what are known laws of nature versus scientific hypotheses and speculations.
AI in its various forms claims to model human learning (via computer models) and thereby surpass human efficiency and speed of calculation. Popular imagination is fired by ill-informed journalism that tends to sensationalise the scope of AI. More circumspect opinion claims that AI is as good as its human designs, so there cannot be a scenario in which AI robots take over human intelligence. AI may be able to enhance human languages, thoughts and communications, thereby further enhancing human intelligence but AI cannot ever catch up to or surpass human intelligence.
AI as a refined form of technology can be very useful in applications that aid human endeavours (medicine, engineering, transport) and of course communication in as far we now have social media channels that both enhance and stupefy communication.
Assuming that AI is truly intelligent – at least in the sense of human intelligence attributed to great scientists, philosophers and revolutionaries – how would we as humans communicate with AI? Given that AI is far superior to us in terms of efficiency and speed – not having to contend with limits of memory and processing bottlenecks – AI would have to translate their thoughts into human language, slowing down if not dumbing down content. While the language of the AI robot would be constrained in the same way as human language (via the synthetic biology of its computational brain) and use MERGE and iteration as starting blocks, we humans could not keep up with the sheer volume of output, now already known as information overload.
So what would two AI robots communicate about? Space exploration? Sub-atomic particles? Evolution? Love and procreation? Or something beyond our comprehension? If the latter were the case, say on scientific principles that we have not yet figured out, say time travel or any such science fiction ideas, then we would be simply amazed. Simply amazed as the knights of the realm were when a 20thcentury American (àla Mark Twain) landed in their castle, his time travel having gone haywire. In fact our literary thought experiments may well be on the level of what AI robots talk about – or maybe not.
We could equally imagine that any AI alien would talk to us as we talk to our pet animals, or maybe as human parents talk to their babies, i.e. in a very simplified form that allows the cognitive apparatus to get the message.
That leaves the question as to what kind of mechanisms AI aliens would use to convey their thoughts. How would their language be designed (to use Hocket’s inquiry)? Direct thought transmission, known by humans as telepathy? A form of sign language? The parable of St. Francis may be useful in this context: as he tried to figure out on how to communicate with the birds in the trees in his monastery, he obviously took the view that he should learn how to sing like the birds, as that seemed to be the way they communicated. However as he mastered their singing no communication took place. It finally occurred to him that the birds communicated by that characteristic hopping on the ground, and as he imitated that hopping (which in the movie version is of course very funny) he could finally communicate with the birds and the bees (with the latter by way of flying in particular formations to indicate where the flowers are).
As Chomsky famously noted, an alien from Mars would figure out human languages in five minutes, and speak all of them accordingly. Would the alien be equally quick in figuring out why humans seem singularly unable to live in peace and harmony? Or would the alien treat us like a benevolent farmer who talks to his cows? Or simply use his/her/its superior technology to enslave us even though our language and thoughts are on par?
What do we want to communicate to aliens? The content of the gold plated data disc placed in the Voyager spacecraft is based on the assumption that the human species’ greatest achievements in science, music and art – and language itself – are what defines us and is worthy of communicating to aliens who may then consider us as worthy neighbours in space. But what about the real human condition? Why not communicate our worst moments in history?
Maybe the point is not to communicate anything at all but try to explain our means of communication the best we can – an endeavour that seems paradoxical because an explanation (even in the form of a law of nature) can be considered a type of communication. Nevertheless there is a well-established dichotomy in terms of language, i.e. the Saussurean langueversus parole, or in Chomsky’s terms competence versus performance. Analogies are pure versus applied science and theoretical versus applied linguistics. The former categories are generally considered value-neutral while the latter subdivide again into good versus bad applications.
So how can we best explain human language to the alien?
Since there is no agreed procedure amongst linguists, philosophers and others who have an interest in the subject, we might have to select representatives from at least the currently dominant schools of thought – which itself should be of interest to the alien in that we have no common understanding of how we communicate, let alone of what we communicate.
In the first instance the representative of generative syntax/minimalism/Chomskyism/etc. will assert that communication in itself has nothing to do with what language is all about, i.e. it is not the functional use that determines its design. A radical stance might be that language equals thought – a process that occurs in the brain and predates the externalisation of thoughts via speech. A less radical approach might be that cognitive processes include language together with other modules. Interdisciplinary research comprising neurolinguistics and neuroscience in general has produced various insights like mirror neurons and quantum alignments that can explain some fairly basic cognitive feats like navigation in birds and memorization of conceptual words in humans but as Chomsky has observed, we are still at the stage of Galilean mechanics or Cartesian philosophy when it comes to understanding human language.
Maybe the alien will be much surprised at this level of humility given our technological and scientific advances that allow us to land a small explorer craft on a meteorite, suspended from a satellite that travelled some three years through space to get there. On the other hand our relative lack of understanding of linguistic and cognitive processes might explain our dreadful history of human degradation, assuming of course that the alien society has achieved a much higher state of development, perhaps something akin to what humans can imagine only in terms of a paradise.
Even so a binary design of language in terms of MERGE and iteration may well impress the alien as many of the laws of universal nature appear to be manifested in such a way. Take the computational power achieved by humans: the very idea of an artificial intelligence is predicated on such a design, so why not imply that language is artificial intelligence squared (L=AI2).
Maybe the alien will have perfected the attempt that Wittgenstein made in investigation the construction of human language as a purely logical system that cannot but speak the factual truth at all times. Of course Wittgenstein famously failed and suggested that language is better explained via game theory, i.e. a functional approach whereby the rules of the game are finite but the output is infinite and mainly predicated on winning the game.
At this stage of the game then we hand over to the representative of the functionalists, pragmatists, behaviourists, generative semanticists and others who hold that the use of language (mainly as a communicative tool) determines its design. Our alien will carefully listen to this argument, as it sounds quite interesting, especially in evolutionary terms. What was it that turned a bunch of primates into our human ancestors? Many animals have communicative abilities but never on par with the human ability, hence is it this very extension that made us human? Did the brains accommodate this evolution by growing neural networks that allows us to communicate in a fairly efficient manner? Darwinian theories of selective adaptation sound probable, as much for language as for finches’ beaks. Then again the study of evolutionary processes has not yielded a straight progression of adaptive advantages but rather a punctuated process in which random mutation is as important as steady adaptation. Here Chomsky and Co. might interrupt the proceedings and tell the alien that MERGE is one of these random mutations that happened some 80,000 years ago, and voilà, human evolution took off in a hyperbolic curve that might yet lead us to the promised land (current setbacks are only due to adaptive behaviours that favour short-term individual gain).
As this is a hypothesis lacking empirical evidence we can dismiss Chomsky for the time being. The alien is nevertheless surprised at the way the debate develops, given that most if not all linguists seem agreed that languages have syntax, be it determined by communication, cognition or specific genetic endowment. So what is syntax? That’s all the alien needs to know so as to appreciate how languages (or language in general) generate sentences, based on syntactic rules. In fact the formulation and description of grammatical rules has been around for thousands of years what with Pāṇini’s grammatical rules of ancient Sanskrit, followed by many a descriptive treatise, covering many of the world’s languages. From dead languages to languages that nowadays have fewer than a 1,000 speakers, there is a rich tapestry of grammars, admittedly mainly from a Eurocentric perspective. Amongst the 6,000 or so languages still extant, there are still many to be described but the descriptive enterprise has lost its mainstream appeal because it appeared that when comparing the grammars of the world’s languages, there is an underlying typology if not universal system of syntax. Even those who reject this approach out of hand will admit to at least a continuum of languages that have very little syntax to those that are ruled by syntax alone.
To formulate, if not to discover, syntactic rules one must at least agree to some building blocks that make up a grammatical sentence. To that end just about everyone agrees that there are lexical items that fill categorical slots, generally known as word categories. How words have evolved is another matter but again it is generally agreed that we have categories like verbs, nouns and various grammatical markers that all interact to make up a sentence. One may again take up polarised positions in this context: generative grammarians may claim that the lexicon arises from syntax while communicative functionalists may claim that the lexicon determines syntax. Be that as it may, syntax is the systematic, rule-based matrix of a sentence. The next polarization is around the fundamental question as to whether syntax has top-down rules or bottom-up. For a long time the top down approach prevailed, i.e. a sentence is made up of a subject and a predicate while the predicate is in turn made up of a verb and an object, drilling down to the word categories and their morphology (if any). The latter received a lot of attention in traditional grammar schools, mixing grammar with learning Latin and Greek (which Hegel thought to be the ultimate education experience), having to rote-learn declension and conjugation tables without much understanding what case and TMA actually mean. As such grammar became a dirty word in modern language learning pedagogies, preferring instead communicative and experiential models that matched the new linguistics of functionalism.
Early functionalism found its guru in Skinner who declared language to be verbal behaviour, which in its crudest form was taken over by the emergent advertising industry, betting on Pavlovian responses. Along came Chomsky who restored grammar (as syntax) to its former glory. Building on phonological feature and rule systems from the Prague school he singlehandedly elevated sentence syntax to the top of language science. Using a dynamic model of generation rules, sentences could now be derived bottom-up. This fitted with the binary design of the new computational sciences, building up complex structures via cyclical rules, evoking mathematical bracketing formulae and eventually changing over to the now ubiquitous tree-structures.
Surely this must impress our alien who in all likelihood has a highly evolved mathematical science at his disposal. Cautionary tales from the remaining descriptivists and newly discovered pragmatists that such a hypothetical language design does not always match the actual language data (especially from newly described languages) is thrown to the wind. Building on Greenberg’s language universals, the Chomsky school of generative syntax declares UG as the ultimate matrix of all human languages. Chomsky himself makes the alien comparison: if and when a Martian alien comes to earth, he will figure out all human languages in five minutes and arrive at an underlying design very similar to UG. Chomsky’s Principles and Parameters set up a flurry of ever more complex rules for language features that otherwise defy traditional, static rules of grammar. Island constraints, local binding (especially for anaphors), move a, trace and a plethora of logical rules turn generative linguistics into a highly sophisticated science of language that can only be understood by experts in the field.
Analogies to psychology and later to biology appear. Humans breathe instinctively but few know how it actually works. So for human language, everyone is a perfect user of language but few understand how it works. Lenneberg’s Biological Foundations of Language eventually convinced Chomsky that this is the way to go by reducing syntax to a system of minimal design, declaring that language is an organ in the brain.
Our alien at this stage may become alarmed at the turn of events: he knows all about biology and the primary building blocks of life but then again even he does not really understand life to the degree proclaimed by Feynman’s adage that if he cannot make it he cannot understand it. While our alien has advanced AI and the most sophisticated robotics at his disposal, he is still light years away from creating life in the test tube.
Maybe the alien is struck by the apparent language paradox: language cannot explain itself. So what is Chomsky and Co. up to? Are his binary mechanisms of MERGE and MOVE and ITERATION really equivalent to neurological processes in the brain? Is neurolinguistics a branch of bio-linguistics? What about quantum physics in the brain? What about neutrinos from far-away galaxies?
The alien looks around and sees the human condition and is confused. What on earth (literally) is going on? What is language and what is it good for? Scientific discovery of the highest order? To conduct primitive warfare amongst tribes of robotic humans? Where he, the alien, comes from life has evolved in different ways, where ‘freedom’ is not a concept, where ‘love’ is not an emotion, where … but hold on we are now talking language performance not competence!
Let’s get back to syntax. Maybe that’s where the alien will question the absolute distinction between competence and performance. He will know that highly complex organisms, such as his, are subject to Murphy’s Law, as much as humans demonstrate by way of their dismal history. Maybe the human language design is not optimal as it is based on biological laws of nature that can and will go wrong. Going back to Wittgenstein, his failed attempt to design a language that is purely logical, maybe the problem is not in performance but in competence as well.
Investigating the chess analogy a bit closer – will the alien be unbeatable? – it is beyond doubt that the rules of chess are unambiguous, logical and few in numbers, and yet, to win a game means that the clever application of the rules led to an advantage. That many chess games end in a draw is equally informative. It seems counterintuitive to suggest that language rules are equally unambiguous, logical and few in numbers, despite Chomsky’s best efforts to have a minimalist design. Going back to his earlier Principles and Parameters – or even to his Aspects – where highly complex rule systems were designed, it would seem that the possibility of messing up the rules is a real possibility, ending up in at least partially ungrammatical utterances which the listener/reader can recover via pragmatic strategies, i.e. applying the rules correctly. We see this quite often when native speakers recover ungrammatical utterances from L2 speakers. Testing for pragmatic interpretation of anaphora may involve similar conundrums:
The king commissioned his architect to design a palace for himself.
Given that the syntactic interpretation yields an unlikely scenario, the native speaker may infer that the sentence has an error *himself and that ‘him’ was the intention of the speaker, i.e. the sentence will be reinterpreted accordingly. Of course there is the slight possibility that the king did indeed ask his architect to build a palace for himself. Wider context may or may not disambiguate. Since nobody denies that there is an interface between form and meaning – including Chomsky’s former LF – there must be countless ways to end up with utterances that deviate from the initial thought processes (‘think before you speak’) through to intentional obfuscation (infamously attributed to politicians and corporate money men).
Our alien will have to take all of this into consideration in his communications with earthlings whilst the earthlings have to take into consideration that they know very little about the alien’s language design other than it must be based on the universal laws of nature. Perhaps the earthlings will have to bring to bear their pragmatic skills much more than their knowledge of syntax, at least for the initial contact period. Later on an in-depth understanding of the alien’s communicative language design can be arrived at after studying the subject – as is done via the scientific method with all other subjects.
Then again the scientific method is just another way of telling a story, as Nietzsche observed, where nothing is certain except uncertainty. A recent (2018) paper by Sanders, Drexler and Ord re-evaluates the probability of alien life, with the conclusion that ‘our distribution model shows that there is a large probability of little-to-no alien life, even if we use the optimistic estimates of the existing literature’.
Human language as such is the highest point of our evolution, however flawed the language design may be. Perhaps language design can be compared to a highly advanced technology, vulnerable to breakdown, neutral in value with the potential for good and bad when used. As far as AI and language is concerned, one can argue that in fact nothing humans say and do results in anything ‘artificial’ in the sense that nothing in AI goes beyond the bounds of language. AI is a highly developed sub-set of language, designed to parse and compute with speeds unavailable to human brains. Sophisticated algorithms that engender so-called deep learning, based on statistical matching of big data, are equally subsets of subsets of mathematics. Again as a technological tool, AI is already known to be used for both positive and negative1endeavours.