Index

In this talk, I have two goals. First, I want to present some of the more remarkable discoveries about language that linguists have made over the past half century. But I also want to point out one reason why these discoveries are important. These discoveries bear on some of the most important language-related issues that face us as a society, including an issue that has aroused considerable recent interest: how to teach children to read. In particular, modern linguistics is directly relevant to the current debate raging over "Whole Language" reading instruction. I will come to that issue at the end.

Let me begin with a few simple but remarkable facts about language. Every society on the planet has language. Furthermore, every normally developing child in every society acquires the language of that society. As linguists put it, language is a universal property of human beings.

Now some less obvious, but equally remarkable facts about language — facts that linguists have discovered over the past century or so of research. All languages, no matter where they are spoken, or by whom, share many very specific structural properties. Languages are constructed in fundamentally the same way. Even the differences among languages (and there are, of course, differences) seem to reflect a limited menu of options from which languages pick and choose. For example, linguists often find the most widely divergent societies sharing very particular grammatical rules.

One outstanding example concerns the position of the verb in German. The German verb can occur in three different places: the end of the sentence, second position and first position, depending on the presence or absence of other "auxiliary" verbs and on the kind of sentence in which the verb is found. Remarkably, exactly the same set of rules governs the placement of the verb in a host of other — completely unrelated — languages. Karitiana (an language of the Brazilian rainforest), Vata (a language of West Africa) and Kashmiri (a language of the Indian subcontinent) all share essentially the same set of rules for placement of the verb. These languages are spoken in radically different societies, in different hemispheres, by people who have no particular contact with each other. So why are they so similar?

There is a conjecture about this which goes back about 30 to 40 years to early work by Noam Chomsky. He reasoned as follows: The universality of language and the similarities among languages indicate that language springs from something that all humans have in common. The obvious candidate is human biology. Our ability to acquire and use language must reflect a special mental ability rooted in some way in our genetic makeup. Chomsky called this ability "Universal Grammar." More recently, Steven Pinker has called it our "Language Instinct" in his excellent book of the same name.

The Language Instinct cannot be responsible for all aspects of language, or else there would be no differences among languages. We would all speak the same language. Obviously, our environment interacts with our innate capacity for language so that children in English-speaking environments speak English, while children in Chinese-speaking environments learn Chinese, and so on.

But the key discovery of modern linguistics is the fact that the influence of the environment is so limited. Basically our linguistic environment helps our language instinct figure out which items from the "language menu" are going to get used — but that's it. Languages not on the menu probably cannot be acquired — or at least cannot be acquired in the natural, easy, spontaneous way in which we learn our native language.

Even deliberate acts of "language creation" must pick and choose from the same menu that all languages pick from, so strongly are we prisoners of human biology. Let me give an example close to home: the development of newspaper headlines in the late nineteenth century.

Somehow or other, a distinct dialect (or style) of English came to be standard in headlines — a dialect not used in other settings. Now contrary to what you might think, this dialect is not just a space-saving version of standard English. (I am relying here on work of the UCLA linguist Tim Stowell.) For example, we all know what the missing word is in the "bad" headline CLINTON HAILS END COLD WAR (it's the word "of"). But leaving that word out somehow ruins the headline. Without "of", it's not good "Headlinese". The words that can be omitted in headlines are largely limited to the definite and indefinite article: MAN ROBS BANK (not THE MAN, not A BANK). As it happens, these are just the words that are omitted in many real languages, for example Russian and Japanese, both of which get along fine without "the" and "a," but do need equivalents of "of."

Even more intriguing to the linguist is the interesting use of tense in "Headlinese". With verbs of action, the present tense can be used for events in the recent past. The headline CLINTON WINS ELECTION tells us what happened yesterday. Verbs that describe states rather than actions are different, however. The headline CLINTON OWNS ROLLS ROYCE could only be understood as present tense. Lo and behold, a large group of West African languages (everyday languages, not newspaper dialects) work exactly the same way (as does Haitian Creole). In these languages (as discussed by the Canadian linguist Rosemarie Dechaine) action verbs in the present tense can — just as in English newspaper headlines — be understood as past tense. But verbs like "own" in the present tense cannot be understood that way.

Why? Certainly not because late nineteenth-century newspaper writers were combing Russian or West African grammars for hints about how to write. The answer is probably quite different. You just can't make up a new language. Every language — even "Headlinese" —is made up of bits and pieces of other languages. And no one can escape this fact. Probably the newspaper writers of the nineteenth century did want their headlines to be shorter and more urgent-sounding. Inadvertently, probably without really knowing what they were doing, they slipped out of English into another possible human language, a language with the tense system of West Africa and the noun system of Japanese and Russian.

Let's return to why this should be. I said that the common properties of human languages (including the menu from which they are assembled) stem from something biological — something that is rooted in our very nature as human beings. I described Chomsky's conjecture that this fact could also help explain how we could acquire language. I called this idea a "conjecture" because when Chomsky first started thinking about this in the '60s and early '70s, our knowledge about how children actually acquire language was quite primitive. One of the most exciting developments of the last 10 years has been an astounding series of advances in our knowledge about actual language acquisition by children. In fact, by now, "Chomsky's conjecture" is as close to an established fact as one can have in this field.

For example, a large body of research on newborn infants has established that specific details of language structure are in place in the brains of infants — even, in some cases, at birth.

In many cases, linguistic knowledge is present despite the absence of any language experience that could account for the knowledge. Children know things that they have never been taught and have never heard. For example, infants display knowledge of specific linguistic contrasts among sounds that are not part of their mother's native language. As they grow older, they seem to lose access to this knowledge — as if this aspect of language acquisition involved "unlearning" innate knowledge.

One of the more spectacular pieces of evidence comes from a recent paper by Susan Goldin-Meadow and Jana Iverson of the University of Chicago. Spoken language is not just verbal, but is regularly accompanied by gesture. (If you don't believe me, pay attention to absolutely anyone when they talk informally. Their hands will be going the whole time.) Gesturing has a grammar; it follows rules. Linguistic gesture seems to be a lot like intonation, the rise and fall of pitch that is also a part of everyday speech. It doesn't convey the main message when you talk, but it does play a role in communication. As with the rest of language, there are universal aspects of gesturing, and there are aspects that are language particular. Goldin-Meadow and Iverson studied the gesture of blind children. These children, since they cannot see the gestures of others, have no way to learn from others how people gesture when they talk. The researchers studied blind children who were asked to recount stories, give directions, and perform various other linguistic tasks. They videotaped the gestures and compared them with the way sighted people gesture. While there were interesting differences, the similarities were overwhelming. Blind children gesture when they talk in basically the same way we all do. Why? Obviously the answer is not learning from experience. The answer has to be: Gesturing is part of our language instinct and, as such, is something we do not have to be taught. We are born knowing about it.

Even in areas where experience obviously does play a role, there is evidence that we as humans are designed to make special use of this experience. Language acquisition, even when it obviously does involve learning, is so remarkably rapid and precise. At birth, for example, a child can already tell the mother's language from any other language, so some kind of incredibly rapid language acquisition must have been going on in the womb.

I think we cannot escape the conclusion that language is indeed a human instinct. Much of it we know without ever having to learn it, and those aspects that are learned are learned quickly, automatically — even before we are born.

Now I hope this was interesting, but even if it wasn't, I'd like to argue that there is a reason why newspaper editors should pay attention to this.

A large literature exists that denies everything I have said. This literature —not the hundred years of linguistics that I have just described — drives many of our society's most important decisions about language, including decisions about how to teach children to read. This literature claims that language is not an instinct. Language, in this view, is a tool, something created by people in order to communicate — just as a bicycle was created in order to get places, and a hammer was created for driving nails.

The appeal of this view is easy to see. It is natural to think that language exists because people want to communicate. Communication, after all, is one of the main uses to which we put language. (Others include organizing our own thoughts, as when we talk to ourselves, and giving vent to our emotions.) Reinforcing this alternative view is the fact that some aspects of language really do serve the purposes of communication. For example, as a language user, you can tinker with some aspects of language. You can make up new words and new expressions — "outsource," "downsize," "grow the economy." But as we've just seen, many of the most interesting aspects of language do not fit this folk view of language-as-tool. Our innate knowledge of speech sounds, for example, was nobody's invention, and may even predate language in our evolutionary history.

What happens when the view of language as tool comes to dominate public discourse? One thing that happens is a blurring of the boundaries between language and other communicative systems. In the area of education, this blurring of the boundaries permitted the rise of a dominant orthodoxy called "Whole Language," which has been much in the news over the past year.

Whole Language starts with the idea that the "really important" properties of language are its uses as a vehicle for communication. Consequently, the fact that all human beings possess language is attributed to nothing more than our intense need to communicate with each other. (If you're thinking that this view does not account for any of the evidence I just discussed, you're right.) Now the argument turns to reading and writing. Why should kids have difficulty learning to read when they don't have any difficulty learning to talk?

To a linguist, the most likely answer to this question is obvious. Speaking is the result of innate capacities — our "Language Instinct." Written language is a technology that makes use of this instinct, by providing a notation for the structures of speech. But there is little evidence that written language itself arises from any instinct. Except, perhaps, for newspaper editors, we have no "writing instinct," no "reading instinct." Consequently, from the standpoint of linguistics, it is likely that most children will need to be taught to read in a way entirely alien to language acquisition. Learning to read, unlike learning to speak, is expected to look like the acquisition of technological skill — the ability to mentally translate marks on a page into speech. This view dovetails precisely with what cognitive psychologists have learned about reading, and is beginning to find support in the best recent studies of actual classrooms.

To a Whole Language educator, who disregards most of the lessons of linguistics, the answers are quite different. Whole Language starts from the premise that there is no special Language Instinct. Consequently, learning to read and learning to write should be a lot like learning to talk. From a Whole Language perspective, all communication systems are equal, and should be learnable in much the same way. Therefore, Whole Language finds little value in systematicity or explicitness in reading instruction. Instead, the entire focus is on getting kids enthusiastic about reading, making them "feel like" readers, and showing them at most an assemblage of "strategies" that might clue them into the reading process. Whole Language gambles everything on the hypothesis that children, placed in the right environment, will learn to read in the same way they learned to speak.

Consider the following, from a recent brochure of the National Council of Teachers of English, aimed at parents: "We used to think . . . that in order for children to learn to read and write whole texts, they had to learn the smallest parts of language first. Because of this 'part-to-whole' belief, it made sense to first teach kids letter names, then 'sounds' the letters make, then introduce them to simple words and short sentences. We spent a good deal of time working on what we thought were prerequisite skills to reading and writing. Though many children learn to read and write under these conditions, many did not. In fact, such instruction made it difficult for most children to understand the joy and benefit of reading and writing — to make them lifetime readers and writers." Now comes the clincher. "We now know that learning to read and learning to write are a lot like learning to talk. We would think it funny if parents hovered over their newborn's crib chanting the sounds of language one at a time. Parents are not trying to teach language, but rather trying to communicate with their child."

Reading this, I ask: Where is the response to any of the evidence from linguistics? In fact there is none. Yet linguistics is very relevant to the discussion. For example, we know why parents don't need to "chant the sounds of language" to their infants. It is because their children were born knowing these sounds. There is no evidence that anything comparable is true of letters of the alphabet. Children are born knowing the sounds of the world's languages, but they are not born knowing even which side of the book is up!

Unfortunately, I have yet to find any Whole Language publication that even addresses this fundamental disparity between what we know about language and the wishful thinking that lies at the foundation of the Whole Language movement. If this disparity were purely an intellectual curiosity, there might be no reason for others to pay attention to the issue. But in fact, difficulties faced by many schoolchildren learning to read are widely attributed to the growing influence of Whole Language on schools throughout the English-speaking world. If these conclusions are true, it is important to learn why they might be true. Linguistics provides a piece of the answer.

Once the news gets out, what do we do next? Well, linguists are not the ones to answer that question. Linguistics is relevant to the conceptual foundations of the debate, but linguists must pass the baton to others when it comes time to make practical decisions. Nonetheless, from what I can tell, the lessons of linguistics provide exactly the right baton. The view of language provided by linguists fits exactly with the view of reading provided by cognitive psychologists, which in turn accords quite beautifully with the classroom results emerging from the most recent educational studies.

So how could Whole Language happen in the first place? Who's at fault? Unfortunately, linguists have to shoulder their share of the blame. Despite the obvious relevance of linguistics to public issues like reading, linguists have rarely made their results accessible to the public, or taken the time to enter these debates. With the publication of Pinker's book and others, this state of affairs is beginning to change.

Right now, there is considerable turmoil about reading instruction in the press and in government, with state legislatures passing laws that mandate changes in teacher education — laws that often mandate an alternative to Whole Language called "phonics." Phonics —systematic instruction in the relation between written language and the sounds of spoken language — can be a rational alternative to the problems of Whole Language. But I have some worries. Can laws favoring one form of instruction substitute for the scientific knowledge that should be the real motivation for this choice? I suspect not. I suspect that a truly productive revolution in teaching will require the closing of the gap between scientists' knowledge and popular knowledge about language and reading. This will take time and, of course, efforts by journalists as well as scientists and educators.

Title page

Introduction: what editors discovered by meeting with language experts

Structure, goals and comprehensibility revisited

Hierarchy of abstraction key concept in powerful writing

The question: Why is writing clear? Why is it concise or direct?

First mention drives how people read and comprehend language

The ivory tower of Babel and the myth of a general literacy