For years, many linguists—myself included—have been eagerly awaiting the publication of what we call “the wals,” and last year the great day finally came when we could hold in our hands—or, given its heft, more likely on our laps—what non-linguists will see as a big, pretty book of unclear utility called The World Atlas of Language Structures.
The layman will page through the wals and find 142 world maps festooned with multicolored dots standing for languages. Each map addresses a particular aspect of grammar, prefaced by an essay on the variations on that feature that the world’s 6,000-plus languages display. For example, many of us have wrestled with the annoying genders in European languages, such as French’s masculine le bateau (“the boat”) and feminine la maison (“the house”). There are, in fact, many languages with multiple genders to learn (Swahili features six main ones), while just as many other languages have no gender at all (the language I am writing in is one example). The “Number of Genders” map, surveying 256 languages across the globe, shows that languages with way too many genders are concentrated in Africa, while languages with none are most common in Southeast Asia and in the Americas.
The layman is to be pardoned for wondering just why one might care about such matters. To begin with, this massive project reminds us that many human languages are counterintuitively different from English, Spanish, French, or German—the languages we learn most often, and which we are disposed to regard as normative—and different as well from ones farther afield (Arabic and Chinese, for instance).
Take the “Order of Subject, Object and Verb” map: it shows that the word order we Anglophones feel as “normal,” the subject-verb-object sequence of The boy fed the dog, is less common worldwide than subject-object-verb, such that in Japanese it would be The boy the dog fed. And then a handful of languages put things exactly in the reverse of English: in the South American language Hixkaryana, spoken by a few hundred people in the jungles of Brazil, The jaguar grabbed the man is rendered as “The man grabbed the jaguar.” Just why most of the few languages with this order are spoken by small groups in South America is as yet unknown, but there it is.
Still, what I have described so far may leave with you the impression of a coffee-table book, full of neat but inconsequential factoids. Is The World Atlas of Language Structures a linguist’s version of one of those bulky art books or the National Geographic World Atlas, lovely on the end table but almost never opened?
In fact, no. For example, many of these maps have crucial implications for the ideas of none other than Noam Chomsky. I refer not to his political activism but to his day job in linguistics. Forty years ago Chomsky founded a still thriving school of linguists hoping to find imprinted in our brains a “Universal Grammar.” The idea is that we are born with a neurological endowment for speaking and processing language, much of it consisting of “switches” that can be set “on” or “off.” Which are “on” and which are “off” presumably makes the difference between, say, Navajo and French or Japanese and Hebrew. So, some languages use a subject pronoun with a verb (he talks), while others can drop the pronoun, like Spanish habla for he talks, where the pronoun él is optional and usually absent. Another switch would make the difference between The jaguar grabbed the man and Hixkaryana’s The man grabbed the jaguar.
However, linguists outside the Chomskyan orbit have long noticed that in some ways, this “Universal Grammar” looks suspiciously like English and other languages of Western Europe, while excluding other languages—that is, those indigenously spoken by most of the world’s human beings. Take relative clauses, like who ate the cake in The man who ate the cake went home. Most versions of Universal Grammar assume that our brains are configured to produce relative clauses more or less like the ones in English. But in the wals, the “Relativization Strategies” chapter shows that the only people in the world who would put it as The man who ate the cake went home are Europeans, except for speakers of one tiny Native American language in New Mexico. Everywhere else, people render such concepts in other ways. The Turk would say “the cake having-eaten man went home.” The Amazonian Pirahã tribesman would say “The man went home, the man ate the cake.”
To stipulate The man who ate the cake as the “default” type of relative clause implies that Europeans, natives of a peninsula of Eurasia, are somehow “default” humans. The wals gives us a vast amount of data reminding us of what “language” writ large really is, not how language happened to come out on a damp little island across a channel from France.
I haven’t even mentioned the accompanying cd-rom, which has so much extra data that the book is more a companion to the cd-rom rather than vice versa. One of the cd-rom’s handiest built-ins allows you to search for two grammatical features at a time, finding which languages, for example, have three genders and also place the verb after the object.
Correlations like this—although probably not that particular one—can help chart how languages develop over time. Until recently, linguists had to rely on memory and shelf searches to do this kind of work. But since no one can be familiar with 6,000 languages, the results could only be so reliable.
For example, in my own research ten years ago, I needed to know which languages lacked: 1) prefixes and suffixes to mark things like case and tense and 2) tones. I actually knew of several already—creole languages such as Haitian Creole and Papiamentu, which were created only a few centuries ago by African slaves who learned European languages incompletely and then filled out what they had learned into a new language entirely. But what interested me was that creoles seemed to be the only languages that had neither conjugations nor tones nor various other bells and whistles that make most languages so hard for foreigners to learn.
Position of Pronominal Possessive Affixes.
Coding of Evidentiality?
Periphrastic Causative Constructions.
To me, it seemed predictable that creoles would be like this. Creoles started out as makeshift lingos with only some hundreds of words, and little “grammar” at all. Eventually slaves turned the lingo into a real language, as structured and nuanced as any. But things like suffixes and tones only develop over time, and creoles haven’t had much.
Suffixes, for instance, start out as separate words. The past marker –ed in walked likely began as the word did—“I walk-did”—and wore down into a mere ending (technically, this would have happened in the ancestor to English and its closest relatives like German). This process takes several centuries at least. Tones take just as long. In Mandarin Chinese, ma can mean “horse,” “mother,” “hemp,” or “licentious” depending on its pitch. This starts with something hard to even perceive: the pitch you utter a vowel on varies depending on what consonants are nearby. Say pet and then peg, and notice that you will tend to say the e with a slightly higher pitch in pet than in peg. Now, imagine if over time the t and the g wore away, as the t in often has for most people. Without the t and the g, the only difference between the two words is that pitch difference: “high” pe and “low” pe. After a while you’ve got a language like Mandarin.
Any language that has been around for a long time cannot help but develop such subtle distinctions. But creoles, unsurprisingly, have little to none of these refinements. They are the world’s “cleanest” and most streamlined languages.
Non-linguists would be surprised to observe the shock and horror that my pointing this out elicited among specialists. Or maybe not so surprised. Certain familiar currents in modern academia led certain parties to propose that I was insulting the intelligence of creole speakers, etc. Then others insisted that even an old language could just happen to end up as streamlined as a creole, and that I was, as it were, stereotyping creole languages.
The wals gives me new ways to respond to such objections. Heretofore, no one has presented me with an old language with a grammar like a creole’s. But no one knows all 6,000 languages of the world, including me, and so the possibility always looms that one, or several, are out there that I just happen not to have come across.
How nice, then, that with the wals’ cd-rom, in just a few seconds I can search for languages that lack 1) prefixes and suffixes and 2) tone. Other than creoles, the cd-rom coughs up a mere seven. And since the cd-rom also includes a handy description of each of the 2,560 languages in its database, I can see whether these seven are truly as structurally uncluttered as creoles.
Three of them are no problem because the scholars who did the suffix map actually addressed only prefixes and suffixes that mark case or tense, whereas these three languages have plenty of other prefixes and suffixes that do different things.
Then the deeply obscure Amazonian language Wari’ only comes out in the tally as having no prefixes or suffixes because it did once have them, but melted them into overloaded little clumps that carry almost a sentence’s worth of meaning by themselves. For example, na– in Wari’ means “he, when referring to a concrete event and in the present or past but not the future.” This is a kind of “mega-affix,” and was the product of eons of the kind of drift that in English turns Did you eat? into Jeet?
I have observed in my work that there are only two places where old languages lack prefixes, suffixes, and tone, the South Seas and Southeast Asia. That is exactly where the remaining three languages the wals came up with are spoken. And what distinguishes them from creoles is that they have words that do the same work as prefixes and suffixes and took just as long to develop. Some languages are like fastidious people who prefer to keep foods clearly separate from one another on their plate: instead of making something into a prefix or suffix, such languages allow this or that bit of speech to remain as a separate word, even if it has the same meaning that a prefix or suffix would. One of the languages in the wals’ database is Khasi of India. It has gender just like French, but marks it with little words before each noun. The cabbage is masculine— ‘u kubi— while butter is for some reason a lady— ka makhan. So Khasi takes us right back to French 101, but with separate words instead of endings.
Now, the wals does not cover every language in the world, and each map only refers to at most 500 or so languages and often just a hundred, since no researcher could thoroughly study 6,000 or even 2,500. But each author’s sample covers the entire world as evenly as possible, and thus it is significant that the work as a whole reveals no old language with a grammar as tidy as a creole’s. This is the kind of finding that linguists with an affection for numbers are especially receptive to, and the wals lends me the tools to reach people like this more easily in the future.
Yet I suspect that many linguists will not be able to resist curling up with this massive volume on rainy days just for the fun facts. For example, a great many languages spoken along the southern margin of the Sahara don’t have a p sound. It’s one thing if a language doesn’t have click sounds or umlauts, but something as meat and potatoes as p? Interesting, then, that cosmetic modification of the lips has been widely practiced among people in just this area, such that p would be the hardest sound to make. I doubt if that has any larger implications for much of anything, but I’m glad to know it anyway. (And just in case—no, as streamlined as they are, creole languages do have p sounds.)
John H. McWhorter is a senior fellow at the Manhattan Institute. He is the author most recently of Winning the Race: Beyond the Crisis in Black America (Gotham Books). Among his other books is Defining Creole (Oxford Univ. Press).
Copyright © 2006 by the author or Christianity Today/Books & Culture magazine. Click here for reprint information on Books & Culture.
