Issue 11 / April 2016
a portrait of Luc Steels, Dorit Bar-On, Holger Diessel, Peter Gärdenfors by Manuela Lenzen
Four researchers have gleaned diverse evidence regarding the origin of language
“Bolima,” says the little robot and his neck grates faintly in turning to his neighbor who is also a robot. This latter peers past the oversized building blocks that lie before them and finally points to a blue cube. The first robot nods – the language game has proven a success. From now on “Bolima” is the word for “blue cube” in robot-speak. “What’s happening here is somewhat magical,” says Luc Steels, a linguist and computer scientist at the University of Barcelona and convener of the Focus Group Biological, Cultural and Social Origins of Language at the Wissenschaftskolleg zu Berlin: “The two robots don’t see the same way, each views the world from his own perspective. And yet they are able to agree on the designation for a certain object.” Steels is convinced that it is only when robots themselves generate the terms by which they communicate with one another that they are able to develop intelligence – and this likewise is aiding us to understand just how it was that human beings learned to speak.
The animal kingdom has numerous and divergent forms of communication – from the chameleon’s play of colors to the dance of bees to birdsong – yet the diversity and flexibility of human speech is quite singular and an essential building block in our intelligence. Because there is no record of the first words uttered by our species, researchers have investigated a wide array of fields for indications as to the development of human language. Four of these researchers have pooled their work in the Focus Group – these include Luc Steels, who calls his experimental robot method “synthetic linguistics” and which he bases on synthetic biology; Holger Diessel, professor of English Linguistics at the Friedrich-Schiller-Universität in Jena; Dorit Bar-On, philosopher at the University of Connecticut; and Peter Gärdenfors, a cognitive scientist at Sweden’s Lund University. Holger Diessel is primarily interested in how the structures of language develop when children learn to speak. Peter Gärdenfors is seeking an “ecological” answer to the question why only humans learn to speak – and in answering this question the factors of environment and human interaction are as important as their cognitive abilities. Gärdenfors believes that the evolution of language is about how the human brain perceives the world and how that same world then emanates from the brain. Dorit Bar-On is an expert in expressive behavior; it is only when we come to grips with animals’ non-verbal communicative behavior – for instance the fact that the rooster not only crows in a certain way when he has discovered food but that he also indicates the finding place with certain head movements – that we can succeed in bridging the postulated gap between human language and communication in the animal kingdom. Furthermore, in the dynamic debates of the Focus Group, Dorit Bar-On critically examines hitherto unquestioned premises and the tangled confusion of terms. As is typical of interdisciplinary collaboration, in the clash of diverse nomenclatures there are myriad little gremlins whose main preoccupation is to provoke misunderstandings.
But it is in one respect that the four researchers with their various approaches are indeed unanimous: there is no such thing as an inborn instinct for language – so the extremely influential linguist Noam Chomsky is wrong. “No one would contest that language and speaking are dependent on innate abilities,” explains Holger Diessel while squinting into the winter sun which streams warmly into the Wissenschaftskolleg’s restaurant. “Our concern, though, is whether or not these abilities are only there for language. And we’re in agreement that this is not the case. We’re not born with any particular sense for subjects, objects, verbs and relative clauses. The capacities that we employ so as to learn to speak are the same ones that we use to walk and drive cars.” The emergence of language was thus a process that built on already existing faculties and which greatly accelerated when a critical mass of cognitive prowess became available. “With the first beginnings of culture there are ever more things which you have to talk about, for instance the preparation of food, look how much we talk about that!”, says Gärdenfors. “This is a process that accelerates of its own accord.”
If human beings have no particular aptitude for speaking then the question as to why humans learn to speak and other animals don’t would seem to be all the more exigent. For Gärdenfors the answer lies in the dual evolution of communication and cooperation: “Animals communicate through the here and now whereas humans plan what they will undertake together in the future. To this purpose they need an idea of what the others want, they must find joint goals, and in order to descry future objectives they may perhaps not require words but some kind of symbols.” He has found evidence for his thesis that language emerged with human cooperation in – among other disciplines – one that cognitive researchers have otherwise little concern for, namely archeology. Archeologists are certain that the manufacture of stone tools of the Oldowan culture, which created the oldest known such implements, would not have succeeded had there not been an instructor who could demonstrate the working steps. In order to teach a neophyte just how to properly manufacture a hand ax along lines of the Acheulian culture, the teacher had to speak some words. Gärdenfors is persuaded that Homo docens, the teaching man, was precursor of Homo sapiens, the wise man.
Showing someone something – Holger Diessel underscores that this must have been of crucial importance in bringing humankind to utter its first words. Very young infants can have their attention diverted through the gaze of other people. They look in the direction that the other is looking. And before they even speak their first words they start to point and direct the attention of others as well. Holger Diessel says: “And when they finally begin to speak they are more frequently uttering demonstratives such as ‘There! There!’ than they are in exclaiming ‘Mama!’ and ‘Papa!’” These homely demonstratives play a large role in the further development of human grammar. Diessel: “These words are simple, but in examining grammars around the world one sees that many important words are based on these indicatives, for instance the article in English and German. From the start of communication to the most complex of linguistic structures, everything has its start in showing.”
Showing also plays a central role in the language games of Luc Steels’ robots. “There are new indicators that ravens point with their beaks,” declares Peter Gärdenfors, and Dorit Bar-On adds: “The alarm calls and feeding cries of guenons are likewise linked to where they cast their gaze and point their body, that is to say in direction of the threat or the food.” There is little doubt – the Focus Group has taken a shine to showing.
Once words enter the world then something magical happens. As Holger Diessel puts it: “The representation creates distance and therewith flexibility.” Words place humans at a remove from the world and create a space for them to ponder before acting. The psychologist and primatologist Sarah Boysen has demonstrated just what that entails. Chimpanzees cannot count well, they only know numbers one and two and thereafter they can only distinguish which pile is the larger one – they cannot be compelled to point to the smaller of two piles even if they receive a greater reward for doing so. But after Boysen taught them signs for counting up to seven they pointed to the smaller number so as to receive the greater reward.
So can we only think through language? “Language and thinking is like the chicken and the egg problem,” says Luc Steels. “I believe that language is decisive in the development of thinking, for when we speak with someone we must generate terms, create categories, order the world. It is like a motor which drives us forward. And it is only when you have a terminology that you can have conversations with yourself - and what that means is to think.”
Human beings cooperate, they instruct, they plan and at some point in this process they begin to speak. A nice rounded story. If only it weren’t for the ravens. Not just humans but elephants, dolphins, birds and in particular ravens perform complex cognitive acts – without speaking of them. This is above all a concern of the Focus Group’s philosopher. “I don’t know how we should conceptually grasp the cognition of animals,” admits Dorit Bar-On. “Should we assume that these animals have complex thoughts and draw conclusions but that for whatever reason they cannot communicate them? Or should we go on the assumption that the problems we as humans ponder are solved by them in another way?” Perhaps with the assistance of mental maps or sub-symbolic representations? Ms. Bar-On does not believe in the existence of a mental language, that philosophical twin sister of the language instinct which is the postulate of Jerry Fodor. She says that, “We still don’t have a language for describing non-verbal thoughts.” One must first understand what it means to have something in one’s mind that is not completely objectifiable but is also not mere behavior. In short it is not at all clear what is to be understood by the term “thinking.” She laughs – “For a philosopher there’s a lot of room at the table!” – and this yields unanimous and somewhat acquiescent nods from the others.
Once language has entered the world it forms an evolutionary system with its own rules. “That is a cultural evolutionary process, fast and open-ended,” says Luc Steels. An evolutionary process that is not about survival, fitness and reproduction but successful communication. New words arise, old words vanish, sound combinations that are difficult to enunciate go missing, case markings emerge and are then abraded. Which might make for a simpler language but not necessarily simpler communication. “That which is not made explicit in language must be introduced through more background information, one must have more in his head,” explains Steels. “It is not easy to say whether this is simpler or more difficult for speakers and auditors.”
Equating evolution with progress or optimization is a misnomer. That goes for the world of nature as well as for the development of language. “This is a much discussed point among us,” says Diessel. One thing is clear – languages change with the world and the communicative needs of speakers. New technologies bring new words along with them. But according to Diessel: “One could say that the move from a proto-language to a structured one constitutes progress – progress in terms of communicative possibilities, which for its part has an impact on human cognitive faculties. But once you have such a language, say English or French, it is useless to ask which is the better of the two.” Well, interjects Bar-On, it does make a real difference as to whether a language has numerals for instance or whether there is only one, two and then “lots.” The question whether there is also progress in language in the description of wines cannot be settled spontaneously. Gärdenfors asks: And who can describe a face so that the person in question might be recognized? Holger Diessel gives his own summary: “In comparing the various languages you above all see large overlaps in the functions and great differences in the structures. In all languages there are expressions for things and actions, for present and future, but not necessarily declinations and conjugations as we know them from European languages.” And the Focus Group then enters into a lively discussion regarding what it is precisely that languages are able to describe . . .
But the four linguistic researchers are unanimous that the assorted structures of the various languages do indeed impact thought. And they share a regret over the disappearance of languages, for in the next hundred years up to 90 percent of the languages currently spoken will become extinct. Yet in their view there will not be a single unifying tongue in a globalized world since languages not only serve to promote communication and understanding but function as demarcations. “The older generation inveighs against the linguistic usage of the younger generation – that is a universal given,” declares Diessel; one of the driving forces behind the development of languages is to delineate one’s own group through a particular use of language.
But Luc Steels’ robots still have not reached the stage where they delimit themselves through their own language – they first work to find a common language. And they are thus more rigorous than philosophers can ever be where it concerns those certain unquestioned assumptions, unclear concepts and forgotten premises of their creators and programmers; if something doesn’t fit then they simply don’t function or they behave strangely. Steels explains: “With the robots we can reenact seminal steps in the evolution of language, develop new ideas and test our assumptions.” In so doing there is no need to prescribe certain terms or vowels or consonants to the robots, for they generate these themselves. The most important thing is that these artificial systems have a body and sensory organs and that they learn to speak about what they see. And a single robot alone is insufficient. “Language is always about communication with others,” Steels reminds us. “If you were alone then you wouldn’t have a language.” At the Wissenschaftskolleg he is writing a book which summarizes his experiments in language evolution, and he is very pleased with what he has achieved thus far. The robots have succeeded in forming words, categories and even the beginnings of a more abstract syntax and recursive communication structures – i.e. someone thinks that the other person thinks . . . What is still lacking here is a meta-language, namely the ability to speak about language itself. Yet Steels is optimistic: “That is like the search for the origins of life – you have to bring the right ingredients together and shake them and just see what develops. It’s the same approach.” Bolima!
Images: © Maurice Weiss