Robinson Crusoe revisited
Peter Bakker
I’ll always remember that long-ago evening around the fire on Molokai Island when Derek, Yvonne and I, stoned out of our gourds, agreed to go ahead with the Island Project. It’s going to be revolutionary, we agreed, unprecedented. But more to the point, Derek observed, it was going to be fun.
- Tom Givón
Many generations of young boys have dreamt of being stranded on an uninhabited island, and needing to survive there. The source of this desire is perhaps inspired by a multitude of cartoons or the successful adventures of Robinson Crusoe, written more than three centuries ago but still widely read. On the other hand, the dystopian reality as depicted in the book and film, Lord of the Flies, in which a group of children are beached on an island and mob each other, may have shown a more realistic picture. These kids “degenerate into savagery”, as one reviewer described it.
The only person I personally know who spent time alone on an uninhabited island, and here I do not mean a small island in a lake, but a real, far-away island in the Pacific, is the linguist, Derek Bickerton. His wife Yvonne, did not join him on this trip. Bickerton ended up all alone on the island of Ngemelis, in the archipelago of Palau. So few people have been there, that the only photos available are strictly copyrighted. He spent not more than a day and a night there, some 36 hours in all. How did he end up there?
Derek Bickerton (1926-2018) is well known among linguists, especially with those who study so-called creole languages. He is also famous for his work among those who study language origins and language evolution. In both areas he was praised as well as criticised for his bold and original ideas. Not all people liked his self-assured style, but most were captivated by his excellent presentations and writings. His past as a journalist and novel writer can be felt on every page of his academic writing, formulated and built up in such a way that one wants just to turn to the next page. His reasoning and argumentation in his writings are also strong. Not all like such an approach because they often want to disagree if his ideas contradict their own image of the world.
Bickerton was born in England, but he had lived most of his adult life with his wife and later his children in a range of tropical locations, including: a Caribbean island (Barbados), Ghana in Africa, Guyana on the South American coast, and, from 1972 until his death on the island of Mānoa in the Hawaiian archipelago. Bickerton had worked as a journalist in the Caribbean, taught literature in Africa, and, in the early 1960s, he wrote several novels set in the tropics. In those regions he was confronted with the existence of creolised languages. These are languages that are newly developed on the basis of the words from a colonial language and a completely different grammatical system.
Apparently, the grammatical system of the source language had been lost, and ancestors of the speakers of creole languages had to create a new grammatical system on the basis of the building blocks—words—but they had to find the nails—connective elements—and cement—grammar—themselves. Therefore, creole languages are newly developed languages. Almost all creole languages are spoken on islands or in coastal regions, as these were the locations of European colonial greed and the involuntary habitat of their displaced and enslaved workers from other parts of the world.
How different are creole languages from their colonial source languages? Take, for instance, the English sentence “I went home to eat.” In Saramaccan Creole of Suriname, the creole most deviant from English, one would say: mi go wosu go njan (me go house go eat). Here, the last word is from a language spoken in Senegal and the rest is derived from English. There is only one form, mi, for the personal pronoun (not I, me, my, ormine like English), no past marking on the verb (no -ed ending orgo/went distinction), and the action of eating in this sentence implies that the actor actually did eat, as the verb is preceded by go.
If it were only a plan, fu (from English for) would be used to indicate this unrealised action: “I went home with the intention to eat”. If you want to express “I would have been walking” in the same language, one would say, mi bi o ta waka (me been go stand walk). The words in this utterance are all from English, but the structure is radically different. In Sranan Creole English of the same country, one would say mi ben sa e waka (me been shall there walk). In Haitian French Creole one would say li t’av ap maché(French words: lui été va après marcher, but in French one says j’aurais marché; here, only the verb root is the same).
Such evidence demonstrates that the grammatical system of French is completely gone, and a new system of building blocks has been created bottom-up. Creole languages show similarities, not only in the absence of many properties from the source languages, but also in the ways in which the new grammatical distinctions are expressed. The three example languages show that tense (time), mood (realisability of the utterance, like “want”, “can”, “must”), and aspect (internal temporal constituency, like “used to do” “repeatedly do”, “busy doing”) are all expressed with brief markers, particles, and short words that precede the verbs.
All creole languages have such so-called preverbal particles, even when the source languages do not have them, and express these meanings with other means. It cannot be chance that creole languages with words from Arabic, Dutch, English, French, Portuguese, and Spanish develop similar yet different systems in Africa, South America, the Caribbean, Australia, India, the Philippines, and Papua New Guinea. Creolists try to look for reasons why creole languages are similar. Bickerton played a major role in research on common traits of creole languages from around the world. He was not the first scholar to put forward such ideas, though he was one of the first to develop and systematise them.
Bickerton was one of the pioneers in identifying common traits in creoles. Creole languages were born as a result of the forced displacement of people who had been put together with others in a work force or in a school or army, with no language in common in a different part of the world. How could those people communicate? They developed a new language on the basis of the only common language learned: the language of the colonisers. Many creole languages are spoken by descendants of the enslaved.
In the 1980s Bickerton had developed a rather radical idea about the origins of these languages. As mentioned, he had noted that creoles all show a number of remarkable structural similarities, not only in the verb system, which cannot be ascribed to the original languages of the displaced, blackbirded, or enslaved populations. His explanation for these observations was that it had been children who had created the language structures. The displaced adults spoke broken versions of the colonial languages, often heavily influenced by their mother tongues, and the children had to create linguistic order in this lexical and grammatical chaos. It would have to have been the children who formed the new language and in a short time, one or two generations. Kids were instrumental in this process. And all these groups and languages achieve these processes and outcomes in similar and even the same way no matter what the original language did or was and no matter wherefrom in the world the languages or people came or where they settled. Apparently, he said, humans use the same procedures to create new structures to express, for example, plurality, time, and specificity. Bickerton called this mental device a bioprogram. According to him, it was our common human biological endowment for language that was responsible for the shared structures in creole languages.
In his 1981 book, Roots of Language, Bickerton linked creole languages with child language and with the origin of language more generally. Children would want to use bioprogram features, but they are corrected because people around them use an existing language that has developed away from the bioprogram during many millennia. Humans and their languages have diverged from that biological blueprint. But children make errors in acquiring a language. They may say things like sheeps for “sheep”, goed for “went”, did eatfor “ate” and Kim no eat for “I don’t want to eat”. Many of these errors occur, according to Bickerton, because children apply bioprogram properties to those existing languages. It is only in the genesis of creoles where these properties can be maintained, as opposed to other languages where such ‘errors’ are eventually ‘corrected.’ Therefore, creoles are different from their lexifiers, the languages which provided the word stock, and therefore creole languages are more similar to one another than to their source languages. Thus, when children have no fixed language model but just chaos, then they will impose the bioprogram features on their utterances, and follow those, and create a new language that is in line with the constructions of the bioprogram, hardwired in the brain. That is why creoles show these structural similarities, according to Bickerton.
What evidence is there for these traits, according to Bickerton? First, there are, of course, the creole languages themselves. Second, all children make errors in the acquisition process. These errors may also reveal the bioprogram traits. Finally, and this is the most speculative element, these bioprogram traits would be the most deeply engrained in the human brain, because they were the first grammatical distinctions that have developed in the dawn of humanity, when the ancestors ofhomo sapiens developed language, and therefore, according to Bickerton, hard-wired in the brain. He outlined the idea in Roots of Language in 1981 and in an article published in The Behavioral and Brain Sciences in 1984.
How can we test experimentally whether such a position is feasible? How can we, in modern times, make children create a language? Right: you put some kids with no common language on an island, and let them communicate. That is what Bickerton wanted to do.
Derek Bickerton was in contact during the 1970s with the linguist Tom, previously Talmy, Givón. Givón had shifted from biology to linguistics, and he had contributed a number of important books on functional linguistics and universal traits of grammar. Like Bickerton, he had also published a couple of novels early in his career. Givón had worked and published on many different languages, for instance, Bemba in Africa, the indigenous language, Ute, in their reservation lands in the USA, and he spoke Hebrew and English fluently. In his CV, he lists “language capabilities” in no fewer than 25 languages from Africa, North America, Papua New Guinea, the Himalayas, and Europe, as well as knowledge of two creole languages, Krio of Sierra Leone and Tok Pisin of Papua New Guinea.
In the late 1970s, Bickerton and Givón had developed this rather insane research project that was to take place on a desert island. The goal was to try and replicate the situation in which creole languages had developed, but on a smaller scale: just a few languages, just a few people, and just a few kids. They discussed the proposal at length. Finally, they submitted their application in 1978 titled ‘Artificial Creation of a Natural Language’ or ‘Experimental Creation of a New Language’ (ECOANL) to the American National Science Foundation. I believe they were somewhat surprised that the project was actually approved. But was it actually realised?
The idea behind the proposal is widely known among linguists: put six couples with no common language together with their two-year-olds on an island, and see what happens. Do the children make a new language? Until recently, I had thought that it had just been an idea, an interesting idea, indeed, and I was convinced that it was undoubtedly rejected, and that it was because of the rejection that it was never realised. It was probably too expensive, too complicated, and there would be no way that such a project could achieve an ethics approval, I thought. Yet, it was approved. And I learned this fact only a few years ago. Amazing. Still, the plan was never realised.
Yes, the projectwas funded, and that the ethics committee had approved it. Yet, no results ever materialised. Moreover, little has ever been written about it. Bickerton devoted twochapters to this idea and the processes around it in his intellectual autobiography of 2008—Bastard Tongues as he dubbed the book—chapter which was heavily “sanitized” according to Givón, even though Bickerton is normally not afraid of dropping names, as we will see later. These chapters are the main source. I thought that Givón also had devoted some space to it in his academic work, but when I asked him in August 2024, he answered, at the respectable age of 88: “Sorry, Peter, too long ago. My memory doesn't stretch that far back (any more).” Givón did write about it in non-academic contexts. This is what Givón wrote in his In Memoriam for Derek Bickerton in 2018:
In the late 1970s we submitted the soon-to-be-notorious Island Project to NSF [National Science Foundation], proposing to track down in situ the gradual creation of a Pidgin language and eventually a Creole, by isolating several monolingual families on a Pacific island and studying their lingua-franca development, of both the adults’ Pidgin and the children’s Creole. We were soon denounced by the linguistic establishment’s guardians of PC at [sic: as] imperialists, colonialists and racists, with dark visions of Pitcairn Island looming on the Pacific horizon. One of Derek’s greatest regrets was the eventual demise of the project. It was pulled back at the last minute by the highest level of the NSF bureaucracy after a most-encouraging response from the late Paul Chapin, long-time linguistics director at NSF. We were soon saddled with a forbidding committee of social-science luminaries, meeting twice on the So. Ute Rez in Ignacio [Colorado, USA], where I was working at the time.
Gradually, relentlessly, we were nickled-and-dimed to death. Eventually they agreed to fund us–provided we carried it out in an deserted European castle with non-exotic Western subjects. At a campground up the Piedra River drainage, above 10,000 feet, we sat around the fire one evening, Derek, Yvonne and myself. The luster had gone out of the project, we concluded. It was a nobrainer, we declined to pursue it any further. Derek related the story in a later book, “Bastard Tongues”, a somewhat sanitized version. We contemplated a fictional account of the real version, but by then he had given up on his fiction. Alas, he had the wildest imagination.
The project was funded, but only the first year of planning and preparation were realised. Appropriate people were sought speaking appropriate and diverse languages, found through their networks. They ended up with couples speaking the following languages:
Korean, people from the southern more isolated islands of the country, a language with verbs in the end;
Kagayanan, from an outlier island of the Philippines, with verbs in the beginning or in the middle;
Land Dayak of Indonesia/Malaysia, from the island of Borneo, with flexible word order in sentences;
Futunan, a verb-initial language spoken in the New Hebrides archipelago;
Taiwanese Chinese, a verb medial language;
Kiwai, a verb-final language of an island off the coast of Papua New Guinea.
The interview with the ethics committee went well. Bickerton and Givón had the well being of the participants high on their agenda, both mental and physical. They made sure there were possibilities for medical treatment, translation facilities, means of communication, fast transportation in case of emergencies, and the like. The ethics committee approved it. They got the money and the go-ahead. The budget was also approved. Still, there were obstacles.
It was demanded that they needed an advisory board. This group included a psychologist, a specialist in isolated groups of people, and an anthropological linguist. And it seems that it was here all went wrong somewhere, as “concerns over safety and psychological wellbeing” were raised. Several well-meaning linguists also expressed their concern.
Bickerton and Givón had decided it should be an island in the Pacific, and Bickerton checked out islands belonging to Palau and the Marshall Islands. The choice fell on the island of Ngemelis, uninhabited, big enough, not too far from the nearest habited location, and there was fresh water. He stayed overnight on the island and approved it. In the capital of Palau, he found the relevant person from the administration in a local bar, and after sharing some drinks, the traditional leader agreed, and gave permission. The Committee on Human Experimentation had invited Bickerton for an interview, and they had been reassured when it appeared that the applicants had thought of everything: medical assistance, crisis help, consent, and the selection procedure. It was approved at these levels as well.
They selected six couples who were not speakers of English, from rural regions, and who had children of around two years of age. Thus, none of the pairs had a language in common with the others, but they were living together. The only linguistic elements they would have in common was a shared list of a few hundred words of a fictitious language, so that they could communicate about basic stuff. These had to be learned in advance. Just words, no sentences, no structure, no grammar. At least they could communicate with one another. These words would include objects and life-forms that relate to the local environment, like plants (tree-wood, fruit, grass, seed - nut - kernel, flower) and animals (insect, fish, reptiles, birds, rat - mouse, chicken, duck - goose, pig, goat). One idea was also that they would do some farming to grow food locally.
The theory was that the adults would develop some simplified, compromised language. The technical term for this language type is pidgin. This way of speaking would function as their means of communication. The development of this pidgin could be studied, as there was recording equipment on the island. It is well known that adults are bad language learners, whereas young children can learn a new language within six months.
Maybe the children would all learn the five new languages spoken by the other couples, as children can also learn many languages easily. But the idea behind the experiment was, of course, that the children would also develop their own means of communication among themselves, probably derived from the common words of their parents.
That had been witnessed before. One can think of the case of Colin and Sadiki. Colin was the five-year-old son of an American couple who relocated to Kenya to conduct research on baboons, and Sadiki was the five-year-old son of a local worker on the compound where the researchers were based. The two boys immediately became best friends, and within a short time, they created a new language, on the basis of Swahili and English, but also with innovative traits. Even after Colin also learned Swahili, and Sadiki also English, they continued to speak their private language with each other. People came from close and far to observe the miracle that had unfolded. Thanks to the efforts of Colin’s mother, Perry Gilmore, who recorded their conversations, their language has been documented. She wrote a beautiful book about the boys and their language, Kisisi (Our Language): The Story of Colin and Sadiki.
There are other cases where children create such new languages. Twin pairs sometimes do it. Thus, there was hope that the children on the island would do the same thing, namely create a new language based on the fragments of the languages spoken. This might not necessarily be a creole language, but the children may create structures that were not found in any of the input languages. This would be sensational enough and could point to a universal or biological basis for language. Indeed, some kind of bioprogram. The results of the island project could have changed our understanding of ourselves as humans forever.
The project was praised by world famous linguists like William Labov and David DeCamp and the linguistics officer at the National Science Foundation, Paul Chapin. But not everyone in the world of linguistics was enthusiastic. John Lynch, an Australian-born linguist, specialist in the languages of the Vanuatu archipelago in the Pacific, had done extensive linguistic fieldwork on Pacific islands, and he wrote: ”The project is unethical, racist and exploitative, particularly given the fact that the subjects [some scholars call project participants or person ‘subjects’] are to be from areas with little or no contact with the Western world. It seems that there is no way in which the ramifications of the project can be made known to them. No way in which any real counseling can take place on site. And thus no way in which the experimentees’ rehabilitation can be ensured.
One staff member summed it up very succinctly. The Pacific is not a cultural zoo.” Bickerton responded to this as follows: “No, and I never thought it was, and I'll never know the kind of distorted reasoning that could lead anyone to assume that I thought it was. But that's so-called liberal progressives for you. Professing to be the protectors of underdogs. All too often they secretly (perhaps unconsciously) despise them. Oh, those poor schmucks, they’re far too dumb to understand us hypercivilized Westerners! Well, I've worked closely for years with illiterate and underprivileged people who have had little or no contact with the Western world. And I’d back their judgment and grasp of reality against that of coddled members of academe any day of the week.”
And then the grant givers started to get cold feet. The advisory committee had given their advice. The project had to be modified. The ‘subjects’ all had to be from developed countries. There could only be adults involved, no children. It must be done in three months. It had to take place within continental USA. When Bickerton and Givón heard this, they came to the same conclusion: screw it. The project would be reduced to the kind of experiments with psychology students in laboratories. It would have taken all of the fundamental questions about humans and children out of the project. This was the end of the experiment. We will never know what would have happened with the language-creating capacities of children and communicative creativity of adults.
In the end of his autobiography, Bickerton gets back to the idea. Could it perhaps be done with a dozen or so orphans, with lots of possibilities to play, and loving caregivers who are not allowed to use any combination of words? At the age of 83, he was still thinking about this unrealised project. Another way would be to get a retired billionaire interested in the project, he wrote. Or perhaps now some TV station producing so-called reality programs called after Robinson Crusoe could realise it. There is a podcast about the experiment from 2013, where Bickerton himself also speaks.
https://www.reed.edu/slx-artifacts/artifacts/web/linguistic-island-experiment.php