In the early days of this blog I reviewed a book by Nicholas Ostler, Empires of the Word , in which I quoted its “critical thesis” that a language does not grow through the assertion of power, but through the creation of a larger human community [p. 556]. (See: Words Rubbing Together) The remark seemed important and I’ve cited frequently, but as I have explored its ramifications a variety of mysteries have opened up. For one thing, how does “a larger human community” create something without asserting power? I finally faced up to that question when a couple of voices came together. One was a recent comment posted on this blog (here):
let's [think] back to a time when human societies had no sophisticated hierarchical structure or when there was no elite or domineering group telling the others how or what 'language' they had to speak …
The other was a study by an Italian team found in the June 10 issue of the Publications of the National Academy of Sciences (PNAS) that shows how individuals without a sophisticated hierarchy can agree on a way to speak about the world. The Italians not only offer a rebuttal to the commenter’s thoughtful remarks, but seem to show how a community can have effects without resorting to power. It would all seem quite mystical if the study had not used those anti-mystical tools, computers.
The PNAS paper is “Cultural route to the emergence of linguistic categories” (abstract here) by a team of physicists, Andrea Puglisi, Andrea Baronchelli, and Vittorio Loreto. The authors did find a hierarchy that coordinates the speaking population, but the top level is not composed of powerful individuals. Instead it is the language itself, called in this study “the linguistic level.”
The linguistic level emerges as totally self-organized and is the product of the (cultural) negotiating process among the individuals. [p 7938]
The tedium of the academic jargon obscures the importance of its point, a direct assault on the commonsense assumption that the regularities of language reside in the heads of the individual speakers. This assumption has unobtrusively framed the debate over whether the rules of syntax are learned, innate, or some combination of the two. But suppose the rules reside on a self-organized “linguistic level” instead. Although the idea might sound absurd, the absurd consequences of the resides-in-the-head viewpoint were noted long ago by Lewis Carroll:
`But "glory" doesn't mean "a nice knock-down argument,"' Alice objected.
When I use a word,' Humpty Dumpty said in rather a scornful tone, `it means just what I choose it to mean — neither more nor less.'
`The question is,' said Alice, `whether you can make words mean so many different things.'
`The question is,' said Humpty Dumpty, `which is to be master - - that's all.'
—Alice in Wonderland
We take Humpty Dumpty’s point, but Alice is on target when, at the chapter’s end, she says that conversation with Humpty Dumpty is unsatisfactory. The in-the-head theory of language produces unsatisfactory exchanges because speakers feel no obligation to shape their words and sentences according to a common standard. Few speakers are as liberated about it as Humpty Dumpty, but it is common to read arguments—Pinker’s The Language Instinct contains an example—that people who try to “correct” language are ridiculous, that language evolves, and each person’s language is correct for that person. At best, this attitude leads to mild confusions about what someone means; at worst you get private lexicons that make communication impossible.
Does this tangle have a solution? The PNAS report says yes, create a linguistic level through “the negotiating process among individuals.” They discovered this higher level by running an experiment to test whether memory and feedback was enough to generate a small set of names for the colors of the rainbow. I will spare readers the intricacies of the computer simulation, but it involved “agents” who begin with no color names, but after a series of one-on-one interactions end up with a small set of agreed-upon names useful for identifying the millions of different hues in the spectrum.
When the authors examined the rules each of the agents used for naming colors they found that the agents were not in perfect agreement with one another. They asked, “Given such a strong misalignment among individuals, why is communication so effective?” The answer comes by looking at what happens when two agents are confronted with a color that they name differently. For example, a shade near the blue/green boundary may be called blue by one agent and green by the other. When they interact, both are unable to simply accept the other’s usage but they remember the confusion and are uncertain about the color’s name. Later, one of the uncertain agents may encounter the misalignment color again, in an interaction with a different agent. The second instance will either show that this different agent agrees or disagrees with the uncertain agent’s usage. If the new agent agrees, the uncertainty ends. The agent continues with the old usage. If the new agent disagrees, however, the agent changes its usage to agree with the previous speakers. The PNAS authors call this process “word contagion,” presumably because the word has passed to a new speaker as a result of contact with others.
The critical feature of word contagion is that speakers must be willing to learn from one another. The PNAS team built that bias into their simulation. Humans cannot take that tendency for granted, but, according to this paper, languages only become stable and mutually intelligible when speakers are willing to adjust their speech to one another. Thus, language is like a market or a species; it sounds like an abstraction but it produces real effects because individuals take it seriously when they interact.
Going back to this post’s original question (how does “a larger human community” create something without asserting power?) we find the answer is pretty simple. The members of the community want to create it. So the commenter had it wrong. Language is the result of a community, not an “elite or domineering group telling the others how or what” to say. A great deal of sociology might be deducible from that premise, but with this blog’s focus on speech origins, I will just note one conclusion: membership in a community is older than speech itself. Speech did not produce the community. It was the other way around.