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



In Lesson One, after defining some terms, we discussed one basic concept: that all human languages are rule-governed. Now I'd like to give you a firmer grasp of what grammar rules are like, how you follow such rules, and what that means for the way you use your language. I think the simplest way to begin is to go through a typical linguistics problem with you so that you can see a rule in operation. I'm going to use a problem from the sound system of English (its phonology), and I'll define any necessary terms as we go along.

The Problem: Why Is That G Silent?

When linguists take up the task of writing a grammar for a language — which means writing the set of rules that determine the language behavior of its native speakers — they take the whole language as the database for that task. They look at all the data, with, as their first goal, the identification of patterns. They describe the patterns that they perceive in the data. They construct hypotheses — proposed grammar rules that will explain the patterns — and then they test them. We're going to look at a very small subset of English data and work our way through this process together. Here's our data:

sign, resignation, malignant, signature, resign, benign, malignancy, malign, significant

In a traditional language arts class, you wouldn't be asked to figure out how these words work or what grammar rules apply to them. You would be taught that "sign, resign, benign, malign" all have a "silent G," and you would be expected to memorize that fact and to prove that you had memorized it by taking tests. We're going to do it differently.

We'll start by looking for patterns in the data, and we'll find that there are at least two ways we can sort it. Suppose we sort the data into words that resemble one another closely. We'll get an arrangement like this:

1. sign, signature significant; resign, resignation; malign, malignant, malignancy; benign.

The question is, why should some of the words in each of those groups have a "silent G" while others have a G that gets pronounced? We know it's not random; it's not that some monarch got up one day and decreed that it would be that way. It can't be that G is silent in words of one syllable — a hypothesis that might have crossed our minds — because G is silent in the two-syllable words "resign, malign, benign." Let's sort the data again, in a different way.

2. sign, resign, malign, benign; signature, significant, resignation, malignant, malignancy

That gives us two subsets, one with all the items that have a silent G, and one with all the items that pronounce the G. We want to find out what causes a word from this class of words to end up in one of those sets instead of in the other.

Note: While we're here, I want to do one other thing. Suppose I were to give you these two hypothetical English words, which I have just made up on the spot: "sebrign" and "olignation." Notice that you know immediately how to pronounce my fictional words and that you know which subset they belong in. This proves that you do know the rule we're looking for.

Now, we have to ask ourselves what "sign/resign/malign/benign" have in common that makes them different from the rest of the words in the set. We already know that they have a silent G, but that doesn't explain anything at all; it describes them, but that's not enough. What else do all four have in common?

If we were making this up, a traditional way to proceed might be to propose the following rule:

All six-letter words with GN in them have a silent G; the word "sign" is irregular and has to be learned as an exception.

For this nine-word set of data, that's a possible rule. However, linguists hate exceptions. When we have to call something "irregular" or "an exception," we consider that a failure; sometimes it's the best we can do, but it always has to be the last resort. What are we missing?


If we were together in a classroom, I'd help you until you found the answer to my question yourself. Since I'm not with you, I can't do that. The answer is that in "sign/resign/malign/benign" — but not in the other five items — the GN comes at the very end of the word. Eureka! We can now propose as a hypothesis this rule:

When an English word ends with GN, the G has to be silent.

We can test that hypothesis with a big dictionary of English words. We can test it by asking native speakers of English to check our results, both with existing words and with words we construct. We'll find that it's valid. That statement is indeed a phonological rule of English — a rule of the English sound system — and it's a rule that was already stored in your internal grammar. It's the rule you followed when you decided how to pronounce "sebrign" and "olignation."

Note: Often, even though linguists have found a rule and it appears to them to be correct, they're unable to say why the rule exists. In this case, however, we do know a thing or two. We know why the silent G isn't dropped from the spelling of words like "sign." English keeps it in the written form of the word, very wisely, because that G tells readers that "sign" and "signature" and "significant" and "signatory" and "insignia" (and many more) are related historically and in their meanings. This is valuable information, and worth preserving; it's helpful to people who are reading written English. It's not, as you have probably been told, evidence that the English spelling system is crazy and should be "reformed."

I'd be pleased to have your comments, criticisms, questions, and any other feedback — for this lesson or any previous one. I'd be glad to hear from you. Please e-mail me directly at

— Suzette Haden Elgin

Updated by webspinner