Since the time of Socrates and Plato, humans have pondered the origin of our words. Now we're closer to an answer. 

By Daniel Cloud, Ph.D.

How did all the various things in the world get their names?

And out of the ground the LORD God formed every beast of the field, and every fowl of the air; and brought them unto Adam to see what he would call them: and whatsoever Adam called every living creature, that was the name thereof. (Genesis 2:19)

That’s one theory. Who stands behind the lexicographers, or the teachers, and whose laws are they enforcing when they insist that we use words “correctly” rather than “incorrectly”? Perhaps all the meanings were decreed by Adam himself, the ultimate ancestor, in a single original act of baptism. And perhaps our job, as philosophers, is to re-create that original mother tongue or universal character, and rescue humankind from its long captivity in Babel.

Perhaps those things are true, but it sounds like a fairy tale. Our lan­guages are part of our inherited culture, and it now seems wrong to think of our own cultural conventions as the decayed remnants of an ancient dialogue between the first man and God. As philosophers work­ing in a tradition that has often been preoccupied with language, we ought to be able to come up with a better story. So it should come as no surprise that a number of philosophers, ancient and modern, have tried to account for the origins of language in a more convincing way.

Plato made one of the earliest attempts in Cratylus. His way of posing this question has been influential—the quotation from W. V. Quine that serves as the epigraph to this book is one echo of it, as we’ll soon see. Aside from the suggestion made by Hermogenes, in the dialogue, that our language must be a system of arbitrary conventions, however, most of his tentative proposals concerning its possible answers seem to have struck modern philosophers as implausible.

But implausibility isn’t always a defect. Sometimes a very alien way of looking at a problem offers us exactly the new perspective we need. Plato was not naive or foolish. In Cratylus, starting with very different assumptions, he made some suggestions about where the meanings of words come from that we now find surprising.Are they all just bad ideas, in spite of their source? Or could some of them be hidden treasures, insights that look unpromising simply because they’re so unfamiliar?

Here I want to revive in a more contemporary form what seems superficially to be one of the least plausible of those ideas. Socrates is depicted as believing that the words of our existing languages were cre­ated by people he calls nomothetes, lawgivers or legislators. Although Quine called this idea childish, to Plato it apparently seemed obvious. Since neither of them actually knew where the meanings of words came from, this is purely a difference of intuition between two great philoso­phers, one ancient and one modern. That alone should make us curious.

This Socratic story, too, initially sounds like a fairy tale about heroic founders, like the story about Adam in Genesis. Reading the dialogue that way, however, is a mistake. It’s putting the quaint medieval thing we think Plato must have been saying in place of what the text actu­ally says. Although Socrates does talk later about the hypothetical “first legislators” who might have created the most basic elements of human languages before the beginning of recorded history, when he first introduces this idea he’s obviously talking about a process of invention that was supposed to be still going on in his own society at the time of writing. Laws, after all, were still being made in Athens, so the legislators were still around.

“Can you at least say who gives us the names which we use?” Socrates asks Cratylus.“Does not the law seem to you to give us them?” (Cratylus 388d). These questions aren’t about what might have happened long ago, any more than those following them are: “How does the legislator make names?” “To what does he look?” (Cratylus 389a). When he first starts talking about the nomothete, Socrates isn’t telling a fairy tale. He’s mak­ing an observation about the way things work in his own world, in the present. The use of language, he argues in this part of the dialogue, is a technology, like weaving and shipbuilding: “A name is an instrument of teaching and distinguishing natures, as the shuttle is of distinguish­ing the threads of the web” (Cratylus 388c). Lawgivers manufacture its tools, just as the maker of looms makes tools for the weaver. Just as it’s the weaver and not the loom maker who’s the proper judge of looms, he tells us that it’s the teacher and not the lawgiver who’s the proper judge of words.

I don’t want to defend the more fabulous hypotheses about unob­served events in the prehistoric past that Plato gives us later on in the dialogue. What I do want to revive here is the idea that our existing lan­guages are partly the product of ongoing human invention and human judgment, that particular individuals did, and still do, play a role in deciding what our language will be like that is something like the role of Quine’s imagined “syndics.”

How could anyone possibly believe that? The idea that human lan­guages are invented by particular human individuals, “legislators,” appears so ridiculous to us now that it’s hard to pause before rejecting it even long enough to ask what the original claim was. But what was Plato really saying? Why would he have blamed the legislators, in particular?

The lexicographer wasn’t yet available as a suspect. The first recog­nizable precursor to the modern monolingual dictionary seems to have been compiled several decades later, in Alexandria, by Philitas of Cos (Sbardella 2007). Like Cratylus itself, it was largely focused on clarifying the meanings of obsolete or obscure words used by Homer, though the dictionary also included technical terms and words from local dialects. So like Cratylus, it was a response to the rapid evolution of a natural human language, to the highly visible differences between Homeric and Attic Greek. This connection to the evolution of the Greek lexicon explains the otherwise somewhat puzzling contents of Plato’s dialogue, in which a subtle philosophical inquiry into the origins and nature of language is interwoven with abstruse speculations about the sources of various Homeric and modern words.Aristotle’s philosophical lexicon in the Metaphysics, probably written decades later than Cratylus, seems like a bold technical innovation, given that the members of the philosophical generation that preceded Socrates were still writing phi­losophy in verse.

Euclid’s Elements, which begins with a list of formal definitions, hadn’t yet been written. Mathematical definitions, in some sense of the word definition, may have been important to Plato, but his writings don’t con­tain many recognizable formal mathematical definitions of a Euclidean kind. In fact, in The Forgotten Revolution (2004:171–202), Lucio Russo argues that they’re a slightly later invention.

One type of definition that would have been familiar to Plato, however, is the legal definition. It’s hard to have laws or a constitution, as ancient Athens did, without explicitly defining lots of terms like citizen, prop­erty, and murder. Formal contracts rely on explicit definitions of their terms or on tacit and customary definitions that can be made explicit. In court, lawyers argue about the proper interpretation of those terms, about whether or not they’re actually in the possible world specified by one of the contract’s clauses. Since the founders of Greek cities often wrote their first constitutions, and great constitutional reformers like Solon explicitly redefined terms like citizen as part of their reforms, it must have seemed reasonable to project this deliberate, rational, explicit, socially central activity all the way back past Homer to unknown original definers existing at some time in the distant past. The overall hypothesis would then be simple continuity, that the present is a good clue to the past, that things probably got the way they are now as a result of the sorts of deliberative processes we still see going on all around us all the time. In this kind of story, the meanings of words are settled in the course of our ongoing efforts to settle the meanings of words.

At the very beginning of this extrapolated process, Plato saw more or less the same paradox that Quine pointed out in the opening epigraph when he spoke about syndics and vicious regress. To Plato, though, the paradox was a genuine puzzle about our prehistory, and not a reason to reject the whole idea of human agency. Here’s his version of the problem:

socrates: Let us return to the point from which we digressed. You were saying, you remember, that he who gave names must have known the things which he named. Are you still of that opinion?

cratylus: I am. socrates: And would you say that the giver of the first names also

had a knowledge of the things which he named? cratylus: I should. socrates: But how could he have learned or discovered things

from names if the primitive names were not yet given? For if we are correct in our view, the only way of learning or discovering things is either to discover names for ourselves or to learn them from others.

cratylus: I think there is a good deal in what you say.

socrates: But if things are only to be known through names, how can we suppose that the givers of names had knowledge, or were legislators, before there were names at all, and therefore before they could have known them? (Cratylus 438a)

Apparently the assumption is that we obtain our most fundamen­tal knowledge—our knowledge of the necessary or analytic truths that Platonists supposed were the sole contents of mathematics or any other genuine science, our knowledge of what’s “true by definition”—at least partly from the meanings of words, from our language itself. The original nomothetes, the very first legislators or, more literally,“rule givers,” who supposedly created the most basic elements of our original languages, must have had another method of knowing those first truths, since they incorporated them in their creations, but what was it? Sensory informa­tion alone, without any logical rules or preexisting categories to use in classifying and interpreting it, seems inadequate.

The worry appears to be more or less the same as the one about a vicious regress that Quine expressed; although if Quine’s “syndics” are analogues of Plato’s nomothetes, Quine was more skeptical of their actual existence. In Plato’s paradox of the first nomothetes and Quine’s paradox of the syndics, both philosophers are in fact pointing to a philo­sophical puzzle we still face in almost the same form.

Anticipating his own later work, Ludwig Wittgenstein ([1922] 1998) asserted in Tractatus Logico-Philosophicus that “all propositions of our colloquial language are actually, just as they are, logically completely in order” (5.5563). When he said this, he was saying something that appears to be true—but why should that be so if our language was handed down to us from people who knew much less about the world than we do? How did ordinary language ever get into such good order in the first place? Who could have arranged it that way without having a language that was already in perfectly good order to use in figuring out how to do it? When did they do that, and what did doing it consist of? Were they doing something different from what we do now? Was there some extra human activity at some point in the past that resulted in words acquiring the meanings they have at present? Why and when did we stop doing whatever it was? Or are we still doing whatever activity has had those consequences over time, even now, without real­izing it? John L. Austin (1956) claimed that ordinary language “embod­ies . . . the inherited experience and acumen of many generations of men” (11). How did their experience and acumen become embodied there? Plato’s willingness to focus on this paradox and, in the end, admit that he had no real solution for it is arguably better than the modern tendency to simply ignore it or push it off into the unobserv­able past.

Without the somewhat fabulous notion of “original legislators,” which Plato used partly to motivate his own version of Quine’s paradox of the syndics, his suggestion presumably would just be that all the way back to the unknown beginnings of human language, people have sometimes disagreed about what words like citizen, murder, and delivery mean, or should be taken to mean, and that by settling these disputes, prestigious or powerful third parties like Solon have, over time, created communal consensus about their correct meanings. The story also seems to include some further process in which teachers, as their users, make judgments about the usefulness of the words that the legislators have given us in this way, about whether a new definition of a word is better than what came before it.

Even if that isn’t true—even if that sort of adjudication is never nec­essary or is so exceptional as to be unimportant, even if people never argue about what was meant by something that was said or what a word really means, or even if those billions of iterated arguments have had no cumulative consequences at all—the idea isn’t obviously absurd or genuinely childish, so our own immediate impulse to reject it is a bit puzzling. Why should Plato’s theory about where meanings come from strike modern people as ludicrous?

The problem, I think, is that in a modern context, Plato’s fable about the origins of meanings quickly runs into another, much more thor­oughly entrenched fable.When the imagination of most modern people is directed to the same unknown origins of the same natural human languages, it produces a very different picture. We naturally gravitate toward some version of behaviorism: a distant ancestor must have habit­ually made a particular sound when he carried out some action, and perhaps from this association, his peers came to regard that as a name for the action, or else an imitation of a sound gradually became a way of referring to the thing that made that sound. Then things continued to happen in more or less the same perfectly and completely accidental and unconscious way until finally we ended up with the very complex human languages we have today.

But this, too, is recognizably just an ancient philosophical myth about the origins of language in the distant and unobservable past, a myth that Brian Skyrms (2010:5) attributes to Democritus. It’s the myth that became a modern folk belief, perhaps because it fits so well with David Hume’s idea that we learn about the world mostly by unconsciously keeping track of recurring associations. That doesn’t mean it’s the truth about the actual evolutionary history of the human animal. The part of the theory that tells us how we get from the caveman’s crude pro­tolanguage of groans and grunts to the full complexity of a modern human language doesn’t seem to have many details. The classical fable that instantly sounds right to us could be just as fabulous as the one that instantly strikes us as wrong.

To my own surprise, I’ve come to believe that there’s an element of truth in the apparently less plausible Platonic story that’s easy to miss, one that seems to be almost completely obscured by the paradox that both Quine and Plato have described. It isn’t that our languages were deliberately invented by particular groups of people, legislators or syn­dics in the formal sense of those words, sitting around particular tables, at particular times in the past. It seems to me that they’re more like our dogs, our wolfhounds and sheepdogs and dachshunds, our retrievers and pointers and greyhounds. We didn’t invent them exactly, but our ancestors did repeatedly make deliberate, more or less rational choices in the process that made them what they are today, choices among a long series of slightly incrementally different variants, unconsciously shaping the dogs into precisely what their human breeders needed them to be.

What most people missed about this activity for most of the time they were engaged in it was that its results accumulate in the powerful way they do. A long series of small, seemingly inconsequential choices among accidentally occurring variations produced big, consequential changes over time. The animals were optimized, mostly inadvertently and without much coordination by their breeders, as specialized tools for particular human activities. The same thing could be true of the more voluntary aspects of our daily use of language. We may have sim­ply overlooked the way their results accumulate.

The case of sheepdogs may offer a particularly good analogy. Many of the choices humans have made about which sheepdogs to keep and breed from have been made in the course of their efforts to manage sheep, so the choices were mostly made in passing, while doing some­thing else, but the existence of that other, more immediate, objective doesn’t mean they still weren’t choices about dogs.

The lecturer in the painting used as the frontispiece for this book, Rembrandt van Rijn’s The Anatomy Lesson of Dr. Nicolaes Tulp, is try­ing to manage his society’s accumulated knowledge of human anatomy, trying to get the members of his audience to improve their cultur­ally transmitted ideas about the newly discovered lymphatic system. (There are other theories about what he might have been talking about, but it makes no difference to the argument.) He’s helping a piece of cul­ture—our new knowledge about the lymphatic system and everything needed to make that new knowledge comprehensible—propagate through a professional community, surgeons. He’s using language to do it, though, choosing his words carefully, and the members of his audience are likely to be picking up the way he uses terms even as they struggle to grasp the anatomical knowledge he’s using them to convey.

Although Dr. Tulp may be choosing his words in passing while trying to do something else, that doesn’t mean he’s not still choosing them, and choosing them carefully. He’s renaming important parts of their professional world, and the surgeons who are familiar with only the old names, or their old senses and referents, appear to be attending closely to this nomothetic act, like anxious shepherds thinking of buy­ing a new breed of dog to herd their sheep.

If that’s a good analogy, if words really are something like sheepdogs, then we need a philosophical theory of where their meanings come from that’s very different from anything we currently have. They aren’t just conventions, though that’s one of the things they appear to be, and they aren’t just tags attached to objects long ago, though that, too, is one of the things they can be. They’re also carefully curated pieces of culture subject to cultural evolution, like every other kind of human culture.

Of course, there’s nothing new about the idea that the meanings of names are passed down to us from previous generations. That’s just Saul Kripke’s (1980) historical-chain-of-transmission theory of reference (Burgess 2013:28–33). There’s nothing new about the idea that speakers may introduce variations in meaning as the name passes down the chain (Evans 1973; Kripke 1980:163). There’s also nothing very novel about the general idea that conventions can evolve. H. Peyton Young’s “The Evo­lution of Conventions” (1993) is a classic of evolutionary game theory. Skyrms (2010) wrote a book about the evolution of signaling conven­tions, in particular.Austin (1956) spoke of words in ordinary language as having “stood up to the long test of survival of the fittest”(8). What may be less familiar is the idea that the evolution of the conventions of our language might involve what Charles Darwin called “artificial selection.”

Daniel Dennett (2009a) has already suggested that domestication might be a good model for the evolution of technical terms, “anchored by systematic definitions fixed by convention and reproduced in the young by deliberate instruction, rehearsal, and memorization” (4), though he expressed some doubts about the model’s applicability to ordinary language. But we play a direct, and at least partly conscious, role in selecting the versions of all our words that will be heard or read by other people and therefore might be reproduced. This suggests to me that domestication is the only model of the evolution of words in general that can possibly fit the facts.

If the meanings of words are part of our cultural inheritance, if we learn them from those around us, and if they are “conventional,” in David Lewis’s ([1969] 2002) specific sense of the word convention (which I will explain in chapter 2, since it’s different from the sense that some­one like Young [1993] might attach to the word), then it seems to fol­low logically that they must be thought of as domesticated, in Darwin’s ([1859] 2009:36–48) sense of the word domesticated, that we have no choice. Something becomes domesticated when we acquire a veto over its reproduction. The farmer chooses which seed to plant and rejects the others. The shepherd picks which dogs to breed from, or keep, and rejects the others. For something to be a convention, in Lewis’s ([1969] 2002) sense, as opposed to, for example, an unconsciously imitated man­nerism, or an absolute practical necessity, it must be possible, in prin­ciple, for us to choose to reject it in favor of some “almost equally good” alternative (68–76). Darwinian domestication occurs when we’re put into a position to reject some candidates for inclusion in a gene pool on the basis of our own human preferences, perhaps in pursuit of purely local and temporary goals, like catching rabbits or herding sheep, and thereby increase the contribution of other alternatives.

Self-interested participation in conventions is another requirement of Lewis’s theory. Otherwise, they become obligations requiring sacri­fice, which is a different phenomenon.We rationally choose among the available conventions regarding the meanings of the words we might use when we speak, choosing some precise sense in which to use them or choosing to use entirely different terms. If we do this in the same self-interested way that we rationally accept or reject greyhounds or racehorses for breeding and they’re subsequently passed on to others for acceptance or rejection in the same way, over and over, how can they not follow the same general evolutionary pattern as other domes­ticated replicators?

If we don’t rationally accept or reject conventions regarding the mean­ings and associations of words, what was Lewis trying to do by writ­ing Convention? Wasn’t he trying to persuade us to adopt an amended and clarified convention regarding the meaning of the word conven­tion, as the book’s title suggests? Was he trying to persuade us to change behavior—our way of interpreting and using that word—about which we have no choice?

Excerpted from “The Domestication of Language” by Daniel Cloud. Copyright (c) 2015 Columbia University Press. 

This appeared in SALON on January 11, 2015.