It cannot be that we have learned each instance individually, because the possibilities, at least for sentences, are infinite. But the brain that accomplishes all of this is not infinite. Alhough it contains a very large number of neurons that enable us to "do" language, that number is finite. The elements used in doing language must be finite—because the resources of the finite brain suffice to learn them. The answer is that there is a finite set of rules that we have learned, rules that enable us to put together the sounds, words, and sentences of our language and to recognize when they are not being followed.
But each answer to a question leads to more questions. If we know the rules, how did we learn them? We may have tried out the sequence nnnlll when we were babies. No one told us that English does not contain this combination of sounds at the beginning of a word (we did hear it inside of words [only, unless]), but we know it's not English to have it at the beginning. No one explained that un- must go at the beginning of a word and nowhere else, yet if you were told to put it on a new adjective, one you'd never heard before—say, winky—you'd immediately choose unwinky, not winkyun. Likewise, you know the rules that allow you to produce a sentence about Amy's feeling for Stan without scrambling it.
What sort of rules are these? They are not the sort you learn from teachers, who tell you how you are supposed—and not supposed—to say what you wish to express. Rules of that sort, called prescriptive rules, take the form "Don't split infinitives," "Say 'yes,' not 'yeah,'" "Say 'I don't want any,' not 'I don't want none,'" "A pronoun must agree in gender and number with the noun to which it refers," and the like. These are prescriptions for speaking the language in the manner educated people consider "correct." Prescriptive rules may need to be taught. In the case of learning a foreign language in school, they certainly need to be taught.
But the rules linguists focus on when they are concerned with a speaker's implicit knowledge of linguistic systems are the rules inherent in the language or dialect. They describe these rules—they do not prescribe them. For this reason the endeavor is known as descriptive linguistics. Research has demonstrated the vast number and complexity of these rules in all of the many languages that have been studied, yet these are the rules children absorb unconsciously, as their language is spoken around them. They are not taught—indeed they cannot be taught, for most speakers are generally unaware that they know them and could not necessarily articulate them if they wanted to. (What are often taught are the exceptions to the rules, for example: The way to make sheep plural is not to add s but to add nothing; the way to speak of swim in the past is not swimmed but swam.)
Our ability to use our language, what we know about it even if we do not know that we know it, is referred to by linguists as our linguistic competence. This contains the rules pertaining to every component of the language: the phonology (the rules pertaining to the sound system), the morphology (the rules governing word structure), the syntax (the rules governing the structure of sentences), and semantics (the rules concerning meaning). Taken together, all of these comprise the grammar of the language, and linguistic competence is the (largely unconscious) knowledge of that grammar. (Note that this is not the meaning of "grammar" as we commonly use it, which is to refer specifically to what they teach in school—the rules of syntax. A linguist may use the term "grammar" in a number of ways. One of these ways is as I have done here: to refer to the entire set of rules of the language that an individual has internalized. But the term may also refer to the linguist's hypothesis about what this consists of.)
The rules that underlie our linguistic competence can be deduced or inferred by studying the patterns observable in a language as it is actually spoken by a given population. In the little demonstration that follows you will see that, though you are generally unaware of knowing these rules, you in fact know them well and, with some guidance, can bring them to consciousness. A favorite example of linguists for demonstrating phonological rules most people are unaware of knowing is one that concerns the pronunciation of the plural marker—the sound added to nouns to make them plural. This sound, as you all know, is spelled s. The following are two short lists of nouns to which may be added this plural marker:
Try adding the plural as you say each of the words in column A, listening carefully to the sound you make in doing so. Then do the same for the words in column B. If you are indeed listening carefully, you will surely hear the difference. To the words in the first column, you have added the sound s, as you expected to. But to the words in the second column, you have added the sound z. Now if you examine the words in column A and those in column B, you will notice that all of them end in a consonant. What you may not observe right away is that there is a small but significant difference between final consonants in the two sets. The final consonants in the column A words are articulated without the voice, whereas those in the column B words are articulated with the voice. Putting a finger on your larynx (your "Adam's apple") as you say the words will enable you to feel the vibration of your vocal cords; it is this vibration that makes voice happen. You will notice that when you finish saying the vowel sounds in the column A words, the vibration stops. The final sound in each word is not accompanied by vibration. Notice the way you say the plural of each of these words; there is no vibration on the final s sound either.
If you now try the same experiment with the words in column B, you will notice that the consonant at the end of each is accompanied by vibration and voice. And so is the plural sound you attach to them; that's why it sounds like z. You quite automatically put the voiced sound on the words that end with a voiced sound, and you put the sound without voice on the end of the words that end with an unvoiced sound. You can test this by adding the plural to, for example, talp and torb, which you have no doubt never come across before. Because they are not words of English, you have not heard anyone say them, either in the singular or in the plural. Yet if you treat these nonsense words—and any others you make up—as real ones, and make them plural, you will unfailingly attach the version that follows the rule.
And this rule you have just discovered can be simply stated: When forming the plural of nouns, add the voiceless version of the plural marker to words ending in voiceless consonants, and add the voiced version to words ending in voiced consonants. No one taught you this rule, and you were in all probability unaware that you knew it. You have acquired many such rules, which function in all the aspects of your linguistic competence. Although the rules differ from one language to another (the one we have just examined is a rule specifically of English), all languages have such rules and all speakers learn them as you did this one: effortlessly and usually without awareness.
Additional aspects of our linguistic competence concern certain other abilities that we also tend to take for granted, that is, until it is pointed out to us that they are really pretty impressive. For example, we are immediately able to know when an utterance is "all right"—that it accords with our notion of obeying the rules of the language—and when it is not. In this sense it is all right to say, for instance,