The word before the words

Recursion the Piraha, and Thomistic settlement of a quarrel that was misframed from the start

A woman of the Tiboli tribe in Mindanao, Philippines

By Christopher S. Carson, J.D., M.A.

I. A bird on the bench

In a laboratory in Tübingen a few years ago, a carrion crow was shown a row of colored brackets and asked, in effect, to copy their pattern. The pattern was recursive, brackets nested inside brackets in the way a human reader meets when a sentence folds one clause inside another, as in the man who saw the dog that chased the cat. Children manage this with some effort. Macaque monkeys, given the comparable task, mostly did not. The crow, on its first exposure, produced center-embedded sequences of its own, at a level the monkeys in the parallel study had never reached. The result sits in the journals beside an older and rougher report that European starlings can be trained to tell recursive patterns apart, and the two findings together pose a small, hard problem that the experimenters did not invent and could not quite dissolve. Recursion, the embedding of structures within structures without any natural limit, had been named for half a century as the one thing human language does that no animal mind can. Here was a bird doing something that looked a great deal like it. I want to leave that problem open for most of this essay, because the impulse to settle it quickly, in either direction, is precisely the impulse that has wrecked the larger debate into which the crow has flown.

II. The missionary and the Pirahã

That larger debate is a war between Noam Chomsky and Daniel Everett, and it has the shape of a theological controversy, which is fitting, since it began with an actual missionary and ended by deciding nothing about God and everything about the soul. To follow it the general reader needs first to meet the people at its center, and then the man who carried them into the journals.

The Pirahã are a community of a few hundred souls living along the Maici, a tributary in the Brazilian Amazon, hunter-gatherers who have kept their own counsel through more than two centuries of contact with Portuguese-speaking Brazil and who remain, almost uniquely among the peoples of the earth, monolingual in a language belonging to no known family but their own. Their speech can be sung, hummed, and whistled as readily as spoken, its prosody is famously rich, and it has been claimed, by the one outsider who came to know it best, to do without number words, without fixed color terms, and without the embedding of one clause inside another. They are, in short, the sort of people a theory comes to the jungle to find, and the man who found them had gone there looking for something else entirely.

Daniel Everett arrived in 1977, a young missionary trained at Moody and sent under the Summer Institute of Linguistics, carrying a wife, three small children, and the intention of learning the language well enough to give the Pirahã the Gospel in their own tongue. They received the translation with a courtesy that withheld assent. They wished to know whether Everett had himself seen the Jesus he described, and when he admitted that he had not, the matter was closed, for these are people who credit what a man or his named witnesses have directly seen and set the rest aside. Decades of patient labor produced not a single convert. Everett drew the conclusion few missionaries would draw, which was that the defect lay on his own side of the encounter, and somewhere along the Maici he set down his faith and did not take it up again. He stayed for the language. He stayed long enough to suspect that this language had no recursion in it at all, that where English folds the man who saw the dog into a single breath the Pirahã prefer two plain sentences laid side by side, and in 2005 he wrote it up and lit the fuse.

III. The quarrel and what the fieldwork showed

Set Chomsky’s claim at its full height before the crow is allowed to settle anything, because the thesis is genuinely deep and a refutation that begins by belittling it has already lost. Human beings produce and understand sentences they have never met, in numbers without ceiling, from a finite vocabulary and a finite memory. Wilhelm von Humboldt had called this the infinite use of finite means; Chomsky supplied the means. A single operation, which his later work named Merge, takes two syntactic objects and forms a third, and the operation may apply again to its own output, and again, without natural end. From that one recursive device the discrete infinity of human speech follows, much as the whole of arithmetic follows from succession. The famous paper of 2002, written with Marc Hauser and Tecumseh Fitch, was more careful than its disciples remembered, distinguishing a broad faculty of language from a narrow one and offering, as a hypothesis for testing and hedged as such, the proposal that the narrow faculty might contain recursion and little else. That is a scientist’s sentence. The trouble came in its afterlife, where the hedged hypothesis hardened into a creed and recursion was crowned twice over, as the essence of language and as the badge of the species.

Chomsky’s reply to Everett came, as ever, from a distance. He called the man a charlatan, a word he supplied to a Brazilian newspaper in 2009, and held in any case that the report was beside the point, since recursion is an innate capacity of the mind whether or not a given grammar troubles to display it. He had never gone to the Maici and would not go, his standing among Amazonianists was nil, and he savaged the fieldworker through intermediaries and in the press while keeping his own hands clean of the river. Around him a guild took up the work. Everett was called a liar and an exploiter, and, in the turn that disgraced the affair, a racist, the charge being that to set the Pirahã outside a claimed human universal was to set them beneath the human norm. Everett came to believe that one of his critics, the Brazilian co-author of the principal rebuttal, had written to FUNAI, the federal agency that licenses entry onto indigenous land, and that the agency thereafter declined to let him return to the people he had given his life to studying. The charge was never proved and remains his allegation, muddied by reports that the agency may itself have solicited the letter and that the Pirahã may have grown indifferent to his return. What is certain is that a dispute about subordinate clauses had acquired the machinery of an inquisition, and that the man holding the field data was the one kept from the field.

The rebuttal itself was formidable, and honesty obliges me to correct a thing that is often repeated and that I had half-believed myself. Nevins, Pesetsky, and Rodrigues never went to the Amazon, never fought the mosquitoes, never gathered a fresh syllable. They worked at their desks from Everett’s own earlier grammar, the descriptions he had published in 1986 and 1987 before his thesis hardened, and there they found what he had once recorded plainly: relative clauses, complement clauses, and a construction marked by the morpheme -sai that his own dissertation had read as a nominalized, embedded clause. The reanalysis is the more telling for its source, since it convicts the later Everett from the mouth of the earlier one. Everett’s answer was that his first understanding had been incomplete and bent by the very Chomskyan training he later threw off, and that -sai marks old information and performs no embedding. The point is contested still, with tonal evidence read one way by one team and the other way by the next, and the man who might settle it by returning to the river is the man forbidden the river. Yet the weight of it leaves Everett’s strong claim badly undercut and leaves standing the modest conclusion that Pirahã uses recursion, sparingly and perhaps primitively, and uses it. Even Ted Gibson, the MIT cognitive scientist whom Everett himself recruited to test him and who holds no brief whatever for universal grammar, did not deliver the clean, recursion-free language the strong thesis required.

Then comes the datum I find most beautiful in the whole quarrel, the one that points past it. The Pirahã who have left the village and made their way into Brazilian society, learning Portuguese as they went, begin to embed. A study of the Portuguese spoken by a group of Pirahã men found at least one speaker producing a complement clause that sits, in its structure, between the juxtaposition of his mother tongue and the full embedding of his second, a man caught in the very act of acquiring recursion. Everett himself never claimed the Pirahã could not think recursively. He said the opposite, that the gap lived in the grammar while the mind ranged freely, and that human cognition plainly reaches beyond what any one language encodes. Put the pieces together and the conclusion is difficult to avoid. The capacity for recursion was latent in the Pirahã all along, resident in the rational soul and waiting on a grammar that would call it forth, expressed faintly or not at all in the speech of the village and flowering the moment its speakers entered a world that asked it of them.

Here the old ghost of Edward Sapir and Benjamin Whorf walks back into the room, and here it must be made to sit down. The strong form of their hypothesis, the claim that the language a man speaks fixes the thoughts he is able to think and walls off the remainder, cannot survive the Pirahã who learns to embed in Portuguese, for his thought was never walled; it was waiting. A man whose grammar lacked recursion proved able to acquire it, which is the one thing linguistic determinism forbids. The weak form survives, and deserves to, claiming only that the habits of a tongue incline the habits of its speakers, so that a language wholly turned toward immediate experience will school its people in a certain attention to the present while leaving wholly intact the mind that could, given another tongue, think otherwise. Language shapes the path of thought. It does not build the walls of a cell.

What strikes me, after years of watching this, is that Chomsky and Everett disagree about nearly everything two men can disagree about and stand together on the single premise that governs the field. Both hold that the question of what sets man apart can be answered by inspecting the embedding depth of his sentences. The perennial philosophy would have denied that premise before the first datum was gathered.

IV. Where the difference actually lies

Recover the order in which the older tradition arranged language, an order it took from Aristotle and refined for fifteen centuries. At the opening of the De Interpretatione Aristotle says that spoken sounds are symbola of the affections of the soul, and that those affections are homoiōmata, likenesses, of things. Aquinas, commenting, compresses the architecture into a formula a schoolboy once knew by heart: voces significant res mediantibus conceptibus. Words signify things through the mediation of concepts. Language stands in three tiers, descending from the thing, to the concept by which the mind holds the thing, to the sound by which the concept is uttered. The vox, the audible sign, is the outermost and most material layer. Beneath it lives the verbum mentis, the inner word, the conceptus, which Augustine in the De Trinitate had called the verbum cordis, the word of the heart belonging to no language and preceding every tongue.

Place recursion in that architecture and its rank is plain. Recursion is a combinatorial property governing how the sound-signs of the outermost tier may be nested within one another. It works on the vox and its scaffolding, at the very floor of the structure, beneath which the labor of meaning has already been done by the intellect. To install a floor-level property as the essence of language is to read the nature of the whole off a disposition of its most material part. The essence of language is signification, the intentional relation that orders sign to thing through concept, and signification is the intellect’s achievement long before any clause is folded into any other.

Bring the four causes to bear and the diagnosis sharpens. A theory that specifies the combinatorial mechanism of the sign has given a partial account of formal and efficient causality and has said nothing of the final cause, which Aquinas calls the causa causarum, the cause of the causality in all the rest. The end of language is communicatio, the manifestation of the inner word to another intellect, ordered finally to the communio of minds in a shared hold on the real. Read any living power from its end, as one must with everything that lives, and recursion takes its proper and modest station as one serviceable means by which the inner word travels outward.

Now the difference itself, the differentia of the animal rationale, the zōon logon echon of the Politics. The logos that defines man is the intellect’s power to abstract the universal from the particular, to seize the quiddity, to know being as being. From that power flow the two operations the tradition never tired of marking: simplex apprehensio, which forms the concept, and compositio et divisio, which joins and parts concepts in the act of judgment. The sensible flowering of that second operation is predication, the enuntiatio, the proposition that binds a universal predicate to a universal subject and stakes the result against the world as true or false. Aristotle reserved truth and falsity for the logos apophantikos alone, the declarative sentence, setting aside the prayer and the command as speech that does not assert. Here, in the order of language, stands the human difference, exact and irreducible: the power to mean man the universal above this man the bare particular, to predicate a universal of a universal, to utter a sentence that can be true. The form of the known thing comes to exist in the knower according to esse intentionale, the mode of being proper to a sign, a thing wholly other than the esse naturale by which a stone weighs and a fire burns, and the word ferries that intentional existence across the air into another mind. Hold on to that phrase, esse intentionale. The whole resolution of the crow will turn on it.

It bears saying plainly, because the quarrel turned ugly on this very point, that the Pirahã possess this power whole and entire, as every man does, and that the perennial philosophy holds the firmest possible ground for their dignity. Whether their grammar embeds a single clause is a question about the village’s vox. That its speakers grasp universals, weigh the immediate against the absent, and acquire the embedding the moment Portuguese asks it of them is the visible shadow of an invisible certainty, that the rational soul is equal in them and wanting in nothing. The critic who called Everett a racist for setting the Pirahã apart might have reflected that the surest defense against the charge was the very tradition his master spent a lifetime deriding.

From this height the three positions sort themselves. Chomsky holds recursion to be the essence of language and the mark of the species, universal among tongues and absent from the beasts. Everett holds recursion missing from at least one tongue, hence contingent and cultural, so the universal falls and the innate organ with it. The third way grants and refuses on both sides. On the fact it stands against Everett, since a primitive recursion is very probably present in Pirahã and the trait is widespread after all. On the meaning of the fact it stands against Chomsky, since even a perfectly universal recursion would remain an instrument of the logos, a property of the vox, downstream of the intellect that gives the sign its meaning. Were a language found one day wholly without recursion, its speakers would forfeit nothing of their rationality, for they would still abstract, still predicate, still mean. Recursion is sufficient for the human difference in no case, since a thing may nest brackets and grasp no quiddity, and necessary to it in no case, since a man would reason without the nesting.

The surest sign that the third way rests on firmer ground is its composure. Everett’s position trembles at every reanalysis filed from a desk in Cambridge; Chomsky’s trembles at every dispatch carried out of the Amazon; each has staked the dignity of the species on a contested point of field syntax and must hold his breath whenever a linguist boards a plane. The perennial account holds no breath, because it never wagered the human essence on a property of the sound. A theory of man that must be renegotiated each time a fieldworker returns from the jungle has confessed by its nervousness that it was searching in the wrong place, and the calm of the older theory is the repose of a position that has found its foundation and can let the data fall as they will.

John Searle, whose death last year prompted my first thoughts on all this, was the modern witness who reached the right verdict against Chomsky by a shorter road. He insisted that words perform acts, that a promise and a marriage vow carry an illocutionary force no tree diagram can hold, and that symbols shuffled in a sealed room yield no understanding however well they are shuffled. He had the verdict and wanted the grounds for it. A self-described biological naturalist, he meant intentionality to be real while declining the metaphysics that alone secures it, and esse intentionale is the very account he needed and would not take. He spent the perennial philosophy’s capital with great skill and disowned the bank that issued it.

There remains the summit of the analogy, which is why the order of signification hangs together at all. In the thirty-fourth question of the Prima Pars Aquinas teaches that the Son is the Verbum, the inner Word proceeding in the divine intellect, in whom the Father utters Himself and, in the one utterance, every creature He has made. Human signification is a participation in that procession, dim and analogical and real: the inner word proceeds from our intellect as the expressed likeness of what we know, and the spoken word bears it outward to another. The line of meaning descends from the divine Word, through the human inner word, to the human vox, and every act of speech faintly traces the first and infinite act of self-expression. A mind that cannot see the Verbum at the summit will hunt for the essence of the word along the floor, in the mechanics of the sound, and will end by mistaking the rule that nests the brackets for the life that means. Chomsky’s atheism and Chomsky’s linguistics are one gesture performed twice, a single refusal to read upward from the sign to the meaning to the Word, and his lifelong horror of dogma in politics never stopped him founding one in grammar, where the reductive temper stays the same temper whatever it reduces.

V. Back to the bird

I left a crow waiting on a laboratory bench, and it is time to go back for it, because the height we have climbed makes the bird at last legible.

What did the crow do. It arranged sign-tokens in nested order, brackets within brackets, in a pattern that satisfied the experimenters’ grammar. What did the crow grasp while doing it. No quiddity, no universal, nothing. It meant nothing by the arrangement, because meaning is the carrying of a form in esse intentionale, the existence of the known thing in a knower after the manner of a sign, and a knower of universals is exactly what the crow is not. The bird performed the syntax and skipped the signification, and the gap between them is the whole of the matter. It is the Chinese Room with feathers, a clever traffic in tokens with the meaning subtracted, and it confirms the tradition it appeared to threaten.

This is the escape valve, and the pressure it releases had been building since the first paragraph. A recursing crow alarms only a theory that placed the human difference in recursion. To Chomsky’s program the bird is a scandal, one more result for the disciples to parse away with talk of transitional probabilities and counting strategies, because if a crow can recurse then recursion is failing its assignment as the wall between man and beast. To the perennial philosophy the same bird is a gift. The tradition never built the wall there. It built the wall at intellectus, at the grasp of being and the power to mean a universal, and that wall no laboratory has breached or will. Every clever corvid that nests its brackets supplies one more proof that recursion is an instrument lying about in the order of sensible signs, available to be drilled into a creature that understands not one word of what it arranges.

So the small hard problem we agreed to leave open turns out to have an answer that neither Chomsky nor Everett could give, since both were looking in the brackets and the answer was never in the brackets. The crow on the bench, sorting its colored tokens with a competence that shames the monkeys and unsettles the linguists, grasps no more of meaning than the brackets themselves do. It is the living proof that one may possess the whole apparatus of recursion and remain entirely outside the country of the word. The Pirahã, who may embed little and who embed more when a wider world asks it of them, dwell at the center of that country and always did. The Beloved Disciple, writing in Ephesus, put first what the reductionist puts nowhere, and named at the beginning a Word that was alive and dangerous and divine, the source from which all lesser words borrow their power to mean. One can only imagine his expression were he shown the crow and told that here, in its nested brackets, lay the glory of the human creature. He would have known better. The crow does not.

References

• Aristotle. De Interpretatione, 16a3–8.

• Augustine. De Trinitate, Book XV.

• Aquinas, Thomas. Expositio Libri Peryermenias, lect. 2. Summa Theologiae I, qq. 27, 34, 85. De Veritate, q. 4.

• Poinsot, John (John of St. Thomas). Tractatus de Signis (1632).

• Hauser, Marc D., Noam Chomsky, and W. Tecumseh Fitch. “The Faculty of Language: What Is It, Who Has It, and How Did It Evolve?” Science 298 (2002): 1569–1579.

• Everett, Daniel L. “Cultural Constraints on Grammar and Cognition in Pirahã.” Current Anthropology 46 (2005): 621–646. Don’t Sleep, There Are Snakes. New York: Pantheon, 2008. “What Does Pirahã Grammar Have to Teach Us About Human Language and the Mind?” WIREs Cognitive Science 3 (2012): 555–563.

• Nevins, Andrew, David Pesetsky, and Cilene Rodrigues. “Pirahã Exceptionality: A Reassessment.” Language 85 (2009): 355–404.

• Frank, Michael C., Daniel L. Everett, Evelina Fedorenko, and Edward Gibson. “Number as a Cognitive Technology: Evidence from Pirahã Language and Cognition.” Cognition 108 (2008): 819–824.

• Sakel, Jeanette. “Transfer and Language Contact: The Case of Pirahã.” International Journal of Bilingualism 16 (2012): 37–52.

• Gentner, Timothy Q., et al. “Recursive Syntactic Pattern Learning by Songbirds.” Nature 440 (2006): 1204–1207.

• Ferrigno, Stephen, et al. “Recursive Sequence Generation in Monkeys, Children, U.S. Adults, and Native Amazonians.” Science Advances 6 (2020): eaaz1002.

• Liao, Diana A., et al. “Recursive Sequence Generation in Crows.” Science Advances 8 (2022): eabq3356.

• Searle, John R. Speech Acts. Cambridge: Cambridge University Press, 1969. “Minds, Brains, and Programs.” Behavioral and Brain Sciences 3 (1980): 417–457.