Language is all around us. Language allows us to share complicated thoughts, negotiate agreements, and make communal plans. Our learning, our courting, our fighting — all are mediated by language.

You can think of language as a technology — humans manipulate their bodies to produce sounds, gestures, and appearances that encode messages using a shared system.

How then does the technology of language work? Answering this question is surprisingly hard; our language skills are automatic and therefore hard to reflect upon. Nevertheless, throughout the centuries, scholars have devised ways to study human language, although there is still much more research to be done and many mysteries to explore. The field of scholarship that tries to answer the question "How does language work?" is called linguistics, and the scholars who study it are called linguists.[1]

How do linguists learn about language? edit

Linguistics is a science.[2] This means that linguists answer questions about language by observing the behavior of language users.

Astronomy has its enormous telescopes, particle physics has its supercolliders, biology and chemistry have intricate and expensive apparatuses, all for learning about their particular facets of the world. Modern linguists go straight to the source by observing language users in action. One of the charms of linguistics is that the data is all around you; you need nothing more than a patient ear and an inquiring mind to do original linguistic research of your own. But you need not start from scratch — generations of linguists before you have laid a fairly stable groundwork for you to build on. Throughout the history of linguistics, the primary source of data for linguists have been the speech, writing, and intuitions of language users around them.

This is not the only way one could imagine learning about language. For example, one could study respected authorities. But this approach raises an obvious question: how did the respected authorities learn what they knew? If each language were invented by an ancient sage, who determined once and for all how that language worked, the authoritative approach would have great appeal. We would go to the writings of the Founding Sage of Danish, for example, and to the writings of the sage's immediate disciples, to find out the Original Intent, much as American judges refer to the Constitution. But, as far as we can tell, this is not how most languages come to be. We have ancient authorities in plenty, but in most cases these authorities were merely trying to codify the practices of the people who seemed to them most skillful in the use of language. In other words, these authorities were themselves scientific linguists of a sort: they observed language users and tried to describe their behavior.

Describing and prescribing edit

In literate cultures, it is common to have a tradition of language instruction. In formal classes, students are taught how to read and write. Furthermore, the teacher tells the students rules of proper usage. This is what is referred to as a prescriptive tradition, in which students are told what to do. It is similar to being taught the proper way to do arithmetic or knit a sweater. Formal language instruction is usually normative, which means that it involves a sense of "should and shouldn't", a notion of right and wrong behavior.

By contrast, linguists follow a descriptive tradition, in which the object is to observe what people really do, and form theories to explain observed behavior. Any specific use of language is only considered right or wrong on the basis of whether it appears in ordinary, natural speech.

As a member of a literate culture, you have probably been exposed to a certain amount of your culture's traditional language instruction. When you first take up the study of linguistics, you will probably experience some discomfort as you observe language behaviors that you have been taught are wrong. It will be hard to suppress an almost instinctive reaction: "This behavior is incorrect. My observation is no good; the person I am observing is an unreliable source of information."

It is important to remember that traditional language instruction and scientific linguistics have completely different goals and methods. Traditional language instruction is intended to train students to use a standard language. Language standards exist largely to make sure formal communication is possible between distant regions, between generations, between centuries, between social classes. Modern civilization arguably depends on such formal communication. Its rules must be constant over wide areas, over long spans of time, across different social and economic groups. This leads to an interesting contradiction:

  • The formal rules of a standard language are almost arbitrary. It doesn't matter in detail what they are, so long as everyone agrees to them and more or less follows them when formal communication is needed.
  • Traditional language teachers need to imbue these mostly-arbitrary rules with a sense of rightness, in order to enlist the students' moral sense in the cause of preserving the stability of the standard language.

The natural result is that students emerge from traditional language instruction with a strong sense that certain language behaviors are simply wrong. Most members of a literate culture have this moral sense about language, and find it hard to suppress. Yet to do objective science, to find out how language really works, it is necessary to adopt a detached viewpoint and to treat all language users as valid. The first principle of linguistics is: Respect people's language behavior, and describe it objectively.

In this book we will adopt this objective stance: language behaviors are not intrinsically right or wrong, and we seek to describe what they are, not to prescribe what they should be.

Hidden knowledge: how linguistic inquiry works edit

Linguists often say that they study the knowledge that a native speaker must have in order to use their language — not formal, school-learned knowledge, but a more subtle kind of knowledge, a knowledge so deeply-ingrained that language users often do not know they know it.

We will illustrate the type of knowledge we mean with a "consciousness-raising" exercise. We will show you some things about English that you must already know, but almost certainly don't 'know you know'.

Case study 1: English plurals edit

A fork
  1. Suppose you have one fork, and I hand you another one. Now you have two forks.
  2. If you have a spoon and find another, you have two spoons.
  3. If your garden has a rosebush, you might plant a second to have two rosebushes.

(We will use boldface for language examples, that is, things that people might actually say.)

In order to speak English, you have to know how to make the plural, or multiple form, of most nouns you hear. You probably do this effortlessly. If you ask most people how to do it, they will say "Oh, you just add s."

But listen carefully.

  1. To form the plural of fork, you add a hissing sound, the first sound in the word sap.
  2. To pluralize spoon, you add a buzzing sound, the first sound in the word zap.
  3. To pluralize rosebush, you add an entire extra syllable, sounding similar to the word is.

You use these three different plural endings every day, effortlessly, without conscious reflection. You always use the right one. It is even amusing to try to use the wrong plural ending. You can say *forkiz, or *spoonce, but you never do. (We use an asterisk to draw your attention to the fact that few people would ever say these things.) You must know the rules governing the use of these different plural endings somewhere inside you, but in all likelihood you never knew you knew until this moment. You must have some way to select the correct ending to use with each word, otherwise you would occasionally say things like *rosebushss. But unless you have thought about this before, it is almost certain that even now that you have been exposed to the concept, you still have no idea how you manage to select the appropriate plural suffix every time. Here is something that you definitely know, but you cannot state it out loud. It is unconscious knowledge.

Can you analyze your own behavior and figure out how you decide whether to use -s, -z, or -iz? Take a few minutes and try. Write out a dozen or so common English nouns and classify them according to what plural ending you would use. Do you see any patterns? (Watch out for completely irregular nouns like foot/feet; for now we are only concerned with "S-plurals".)

One theory you might come up with is that the correct plural suffix must simply be memorized for each noun. This is a perfectly reasonable theory. Perhaps forks sounds better to us than *forkiz simply because the former is the only plural we have ever heard. However, this is not the case, because for new words we have never encountered we can still pick out a plural ending that sounds right. For example:

  1. If you have a zug, and you find another, you now have two ...
  2. Mike just finished making his third bidge, so he has made three ...
  3. I inherited a blick from each of my grandparents, which is why I have four ...

(Jean Berko-Gleason first studied examples similar to these as set in her wug test).

Complete these sentences with the appropriate plurals. Then have five English-speaking friends do it, but don't let them collude: force them to form their own plurals. You and your five friends will all agree: the first example gets a buzzing plural -z, the second gets a whole syllable -iz, and the third gets a hiss -s. And none of you have ever heard those words before, nor do you have any clue what they mean. If plural endings were simply memorized, you and your friends would have had to guess the endings, and you would likely have made different guesses.

The second principle of linguistics is, Language knowledge is often unconscious, but careful inquiry can reveal it.

The idea of deep structure and the general outline of linguistic theory edit

What have linguists learned about how language works? What is the overall shape of modern linguistic theory?

Linguists espouse a variety of theories about language; differences between these theories are sometimes quite striking even to laypeople and sometimes so subtle that only well-read linguists can understand the distinctions being made. Arguments between linguists who support different theories can be quite heated. But underneath the noisy debate about details there is widespread consensus, which has been coalescing since the 1950s. This consensus sees, in every corner of human linguistic ability, at least two layers: a surface structure consisting of the sounds we actually speak and hear, and the marks we write and read, and a deep structure which exists in the minds of speakers. The deep and surface structures are often strikingly different, and are connected by rules which tell how to move between the two kinds of structure during language use. These rules are part of every language-user's unconscious knowledge.

The idea of deep structure is an unintuitive one. It is natural to be skeptical about it. Why do linguists believe that language structures inside the mind are so different from what we speak and hear? We will use another case study to give an example of the evidence.

Case Study 2: The English auxiliary wanna edit

(1a) Rachel doesn't want to do her linguistics homework.

(1b) Rachel doesn't wanna do her linguistics homework.

In many varieties of English, the two words want to can often be contracted into wanna. English users are more likely to do this in speech than in writing, and are more likely to do it in relaxed, informal contexts. (Linguists use the word registers to describe the different behaviors language-users adopt depending on context.) The pronunciation of wanna lacks the clear t sound of want to. English users evidently must know both variants.

Can want to always be contracted? Consider the following examples.

(2a) Who do you want to look over the application?

(2b) *Who do you wanna look over the application?

Again, we are using an asterisk to call your attention to the fact that the second example is not natural English to most native users. It is, in fact, traditional in linguistics to use an asterisk to mark an example that is somehow unacceptable or unnatural for native users.

As in our first case study, we seem to have found a mysterious piece of unconscious knowledge that English users all share. We do not resist changing (1a) into (1b), but something makes the change from (2a) to (2b) much less comfortable. What could it be? How do English speakers decide when want to may be contracted?

Perhaps the contraction is inhibited by the fact that (2a) is a question. We can test this theory with a similar example.

(3a) Who do you want to invite to the party?

(3b) Who do you wanna invite to the party?

Here, the contraction works fine. So the question-theory cannot be correct. In fact, the similarity between (2a) and (3a) makes (2a)'s resistance to contraction quite puzzling.

What follows is not the answer to the puzzle. Rather, it is a sketch of part of a theory that some linguists use to explain the observed behavior of wanna. This theory was arrived at by considering many, many examples, and consulting many, many native English speakers. It is not in any way authoritative, but it illustrates the point we are trying to make. Consider some possible answers to questions (2a) and (3a), which we repeat for ease of reference:

(2a) Who do you want to look over the application?

(2c) I want Yuri to look over the application.

(3a) Who do you want to invite to the party?

(3c) I want to invite Yuri to the party.

Notice that in sentence (2c), the name Yuri comes between want and to, separating these two words, while in (3c), the words want to are still next to each other. Let us hypothesize that in our minds, questions like (2a) and (3a) have some kind of mark that shows where we expect the answer to be inserted. Linguists sometimes call such a mark a trace, and symbolize it with t. So we might show this "mental form" of our two questions as follows:

(2d) Who do you want t to look over the application?

(3d) Who do you want to invite t to the party?

If such traces really exist in our minds, they would provide a very elegant explanation of when want to can be contracted. The proposed explanation is that we can contract want to only when there is nothing between the two words in the mental form of the sentence. We already knew this to be true when the interrupting material is audible. Of course want to cannot be contracted in (2c), because Yuri is in the way. Our proposal is to extend this explanation to inaudible material, and to say that want to cannot be contracted in (2a) because a trace is in the way.

You might object that we have invented traces precisely to explain when want to cannot contract; that we will simply hypothesize that every uncontractable example has a trace in the middle. This is a fair objection, but remember that we are not putting traces wherever we want, but only where we expect the answer to the question to fit. You are encouraged to try more examples on yourself and your friends.

This step of introducing traces to explain when want to may be contracted is a serious and profound piece of theory-building. We are saying that sentences in the mind are not exactly like their counterparts spoken aloud. They are not mere mental tape-recordings — they can possess aspects, like traces, that cannot be heard. As soon as we take this theoretical step, we open up a new question: 'How is language represented in the mind?'. Linguists use the term deep structure to discuss the way sentences are represented in the mind. In contrast, surface structure means sentences as we hear or read them.

This leads to the third principle of linguistics: Sentences have a deep structure in the mind, that is not directly observable, but may be inferred indirectly from patterns of language behavior. It is this third principle that separates recent linguistic scholarship (since about 1950) from the centuries of work that went before.

Using linguistic evidence cautiously edit

Before we proceed, we must say a few words about the mode of inquiry we are using. As we throw various examples at you, we are either marking them with asterisks — starring them, as linguists say — or we are not. In essence, we are asking you to go along with our judgment about whether or not the examples are natural, native English. We would prefer to be scientific about it; one way of doing this would be to perform a study in which we present our examples to a few hundred native English users, and have our subjects tell us whether they thought the examples were good English or not.

But such studies take a lot of time and effort, and it's easy to make mistakes in experimental technique that would weaken our confidence in the results. It is extremely tempting to use oneself as an experimental subject, and use one's own judgment about whether an example is natural English or not. There are obvious pitfalls to this approach. One may not be as typical a speaker as one believes. Or one's judgment may be unreliable in the highly artificial situation of asking oneself questions about one's own language.

But nevertheless, in some cases, the situation seems clear-cut enough that we can give examples, as we have been, in the reasonable confidence that the reader will agree with our judgments. This is a shortcut, and is not a good substitute for real data. But in reality, a lot of linguistics gets done this way, with scholars using themselves as informal experimental subjects. Doing research in this way incurs a debt, the debt of eventually backing up our claims with real experimental studies. It's fine to get preliminary insights by probing our own intuitions about language. Eventually, though, we must do real science, and we must remember that in any conflict between experimental data and our own intuitions, real data always wins.

Structure of this Book — Layers of Linguistics edit

As you may already have noticed, language is a hugely multifaceted entity. When we learn how to write in school, we are taught that individual letters combine to create words which are ordered into sentences that make up a composition. Spoken language is similar, but the reality of language is much more subtle.

The structure of language may be separated into many different layers. On the surface, utterances are constructed out of sequences of sounds. The study of the production and perception of these sounds is known as phonetics. These sounds pattern differently in different languages. The study of how they group and pattern is known as phonology. These groups of sounds then combine to create words, which is morphology. Words must be ordered in specific ways depending on a language's syntax. The literal meaning assigned to words and sentences is the semantic layer, and the meaning of sentences in context is known as pragmatics. Each of these may be considered a branch of theoretical linguistics, which studies the structure of models of language.

Don't worry if it's not yet clear to you what each of these subfields of linguistics deals with. The first chapters of this book go through these fields layer by layer, building up a clear picture of what linguistics is. We will then explore various topics of inquiry which apply our linguistic knowledge of these layers to solve real-life problems, a pursuit known as applied linguistics. Branches of applied linguistics include: computational linguistics, anthropological linguistics, neurolinguistics, sociolinguistics, psycholinguistics, discourse analysis and language acquisition.

Workbook section edit

Exercise 1: Me and John edit

If you took English classes at school, you may have been warned against using the following sentence:

(4a) Me and John are friends.

You probably were instructed to replace it with the following:

(4b) John and I are friends.

Pronouns such as me, him, her, ... are termed 'objective pronouns' because traditionally they are considered to never appear as the subject[3] of a verb, and prescriptivists rule that as such usage of them in this position is "incorrect". However (4a) is not marked with an asterisk because it is still largely acceptable to native English speakers, and as descriptive linguists we are interested in both forms.

Now note that certain arrangements of pronouns (I, me, John, etc.) in the sentence make it ungrammatical to all English speakers:

(4c) *I and John are friends.

(4d) *Her and John are friends.

(4d) *I and him are friends.

List all possible combinations of two pronouns in the sentence "___ and ___ are friends." that you can think of, and label each sentence which would not be said by native speakers with an asterisk. Then create a theory as to what makes any sentence of this form unacceptable.

Notes edit

  1. In common parlance, a linguist can simply be someone who speaks many languages. This is not the sense we mean, and whenever we say linguist, we are referring to a scholar of linguistics. We will call someone who speaks many languages a polyglot. Some linguists are also polyglots, but it is perfectly possible to be either one without being the other.
  2. Some debate this statement, contending that modern linguistic theory lacks predictive power. We will, however, let this assumption stand for now.
  3. Generally (though not always) whatever is performing the action. Don't worry about the definition of subject for now — it will be defined in a later chapter.