by Suzette Haden Elgin, Ph.D.
(linguistics; UCSD)



So far in this course we've focused primarily on phonology — the sound system of language. In this lesson we're going to shift briefly to syntax, the part of language traditionally called "grammar." Before we do that, however, I have an urgent query to deal with.

Several people have sent me messages saying something like this: "Chomsky says languages don't have any rules, so why do you keep talking about them?" The source for their queries was Noam Chomsky's latest book, New Horizons in the Study of Language and Mind. On page 7 of which he says theoretical questions suggest that we should...

challenge the traditional assumption, carried over to early generative grammar, that a language is a complex system of rules, each specific to particular languages and particular grammatical constructions: rules for forming relative clauses in Hindi, verb phrases in Swahili, passives in Japanese, and so on.

Out of context, this quotation would lead you to believe that learning about "rules" is a waste of your time. Let's consider it in context instead. Chomsky has proposed that there is a core of grammar which is part of the mental equipment of every human infant, composed of principles and parameters. He claims that what makes human languages different is the different ways they set the switches for those principles and parameters: Every Terran language has Switch X, but in some languages Switch X is set for "On" while in others it's set for "Off." He says that if we had the necessary information about all those settings we could look at a given arrangement of switches and know immediately that it was English or Chinese or whatever — which would perhaps make grammar rules irrelevant. However, we don't have that information. Not yet. Until we do, rules continue to be what we have to work with.

[You can read more about this in the Chomsky book, or in the "Language and the Brain" chapter of a book of mine called The Language Imperative. My book is easier to read; Chomsky's book is much more scholarly and technical, but not impossibly so; I strongly recommend it.]

It's impossible to understand how linguists go about analyzing and describing languages unless you understand the concept of language rules. It may be possible to do linguistics without rules; I don't know about that, and am not prepared to argue with Noam Chomsky in public. However, I'm sure it's not possible to learn linguistics without rules.

Moving right along....

To investigate some phenomenon in a language, you'll remember, you gather data — examples of the phenomenon — look for patterns in the data, propose hypotheses to account for the patterns, test the hypotheses, and repeat until you've gone as far as you can go. We've been looking at phenomena from the English sound system; now I'd like you to look at a set of data from English syntax. Items with an asterisk [*] at the beginning aren't acceptable to native speakers of English.

1a. Every morning, I go feed the dragon.

1b. Every morning, we go feed the dragon.

1c. Every morning, they go feed the dragon.

1d. Every morning, Tracy and Haj go feed the dragon.

1e. *Every morning, he goes feed the dragon.

1f. *Every morning, she goes feed the dragon.

1g. *Every morning, it goes feed the dragon.

1h. *Every morning, Tracy goes feed the dragon.

If you look at these data carefully, the solution looks obvious. "Go feed" is okay with first person singular and plural ("I" and "we"), with second person singular and plural ("you"), and with third person plural ("they," "Tracy and Haj"). "Go feed" is not okay when the subject is third person singular ("he," "she," "it," "Tracy"). This would lead us to propose the hypothesis that English has a rule like the following:

Rule 1. The sequence "go + VERB" is not allowed with a third person singular subject.

Now we need to test the hypothesis. We could check to see if the same thing happens in all "go + VERB" sequences, not just "go" followed by the verb "feed." We'd find that it does, which gets us no farther; we need to try checking something else. For example, what happens if we use "go + VERB" in questions? Look at the next set of data, please.

2a. When do I/you/we/they/Tracy and Haj go feed the dragon? Every morning.

2b. When does he/she/it/Tracy go feed the dragon?

And look at these data.....

3a. Can I/you/we/they/Tracy and Haj go feed the dragon?

3b. Can he/she/it/Tracy go feed the dragon?

Notice that 2b and 3b have third person singular subjects, but they're acceptable. Does that mean we should rewrite Rule 1 by adding "except in questions" at the end? Maybe. That would be a refinement of our hypothesis. But before we decide to do that, there are other kinds of data we haven't looked at. For example, what happens when you use "go + VERB" with the past tense? As in examples like these....

4a. *I/you/we/they/Tracy and Haj went feed the dragon yesterday.

4b. *He/she/it/Tracy went feed the dragon yesterday.

4c. Did I/you/we/they/Tracy and Haj go feed the dragon yesterday?

4d. Did he/she/it/Tracy go feed the dragon yesterday?

Clearly, Rule 1 can't be right. We could just add more stuff to it, of course. We could try rewriting it to say, "The sequence 'go + VERB' is not allowed with a third person singular subject — except in questions, and is not allowed at all in the past tense — except in questions." However, science likes economy. The fancier and more complicated a rule gets, the less likely it is to be valid. Linguists know from centuries of experience that a rule like our rewritten one is beginning to have a suspicious smell; they would want to look at many other sets of "go + VERB" examples before accepting it.

If we were in a classroom, I'd be standing at the blackboard and you would be coming up with those examples so that I could write them on the board for examination. We'd try "go + VERB" with future tense ("We/he will go feed the dragon shortly"), we'd try it with commands ("Go feed the dragon this minute!")....and eventually one of you would suggest trying this:

5a. *Now I'm going feed the dragon.

5b. *Now we're/you're/they're going feed the dragon.

5c. *Now Tracy and Haj are going feed the dragon.

5d. *Now he's/she's/it's/Tracy's going feed the dragon.

5e. *Now he is/she is/it is/Tracy is going feed the dragon.

All right? We could tinker with that a little, and we'd find that the same thing happens with "was/were going" and "will be/would be/should be/could be/must be going," and so on. And we'd find that if we construct similar examples in question form — like "*Is he going feed the dragon now?" -- they'd all be unacceptable. Now our rule would have to be rewritten like this:

Rule 1, Revised: The sequence "go + VERB" is not allowed with a third person singular subject, (except in questions), and is not allowed with past tense (except in questions), and is not allowed at all with progressive aspect.

["Aspect" has to do with whether the act or state specified by the verb is finished or not. When you say "I'm feeding the dragon," you haven't finished doing that yet; that's called progressive aspect. Politely ignore people who tell you that "progressive" is a tense, please.]

At this point, the smell of our rule is hopelessly fishy. You would suspect that the more data we looked at, the more different kinds of "go + VERB" patterns we looked at, the more "exceptions" we'd have to add to our hypothetical rule — and you'd be right. You'll remember that linguists consider even one exception to be evidence of failure. Either linguistics is not a science, or we're missing something.


If you want to work out the solution now for yourself, don't read any farther; you can come back and check your solution against mine later. Otherwise....

This space intentionally left blank.

Suppose we look at all the data we've already examined, with an open mind, instead of digging this same dry well deeper and deeper. We want to know what all the unacceptable sentences have in common that makes them different from all the acceptable ones. Eventually, it will occur to us that in all the unacceptable sentences — but none of the acceptable ones — "go" has an ending. A suffix. Sometimes it's "-es" as in "*Haj goes feed the dragon every morning"; sometimes it's "-ing," as in "*I'm going feed the dragon right away." Sometimes it's "-ed," as in "*Yesterday we went feed the dragon"; the way we spell "go+ed" in English is "went," and the two things are equivalent. This tells us that all those factors we've been considering were red herrings, and that we need to propose an entirely new hypothesis for the English "go + VERB" construction, like this:

Rule 2: No ending may be added to "go" in a "go + VERB" construction.

There! We'd still have to test this rule, and I encourage you to do so; it's still a hypothesis. However, we can say that it handles all of the data we've looked at so far, without the string of exceptions that we had to tack on to make Rule 1 work. It could be written in a different way, yes. It could even be wrong, although that would surprise me. But it's unquestionably a far better hypothesis.

There. As always, I'd welcome your comments, suggestions, criticisms, questions, and other input, with a note if you prefer to stay anonymous. Please e-mail me directly at Final note: To see vast quantities of interesting and useful material about linguistics and languages — plus an Ask-A-Linguist page where you can e-mail questions and have them answered by a panel of linguists, go to The Linguist List.

— Suzette Haden Elgin

Updated by webspinner