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



I had almost no feedback after Lesson Five; that worries me. If it was because you were all swamped with end-of-the-year tasks....exams or taxes, for example....that's okay. If it was because you were all taking well-earned vacations on tropical islands, that's okay too. But if it was because you perceived Lesson Five as defective, that's most definitely not okay. If it's a quality control problem, you have to tell me; otherwise, I can't possibly fix it.


We've now thoroughly discussed two things: what a rule is in linguistics, and why such rules are needed. We've looked briefly at some rules from English phonology (that's the sound system) and from English syntax (the word linguists use for most of what is traditionally called "grammar" in our schools). We've touched very briefly on some other kinds of rules -- for example, the semantic rule that requires all English sequences such as "lift a finger" and "turn a hair" to contain a negative element. We've noticed that the separations of these rules into such categories as phonology and syntax and semantics is in many ways artificial, because the categories interact and overlap.

What I want to do for the next few lessons is expand our knowledge about linguistics rules of all kinds, starting with a discussion of...

The Grammatical Processes of Terran

Although when you study languages you learn scores of rules (often with scores of "exceptions" and "irregularities"), all human languages have only four kinds of rules: deletion rules; insertion rules; substitution rules; and movement rules. (When you read somewhere that there are only three kinds of rules, it's because the writer is defining deletion as the insertion of zero. That's legitimate, but it creates confusion.) If ever you are analyzing a language and you come across any other kind of grammar rule, you can safely assume that the language in question is extraterrestrial. Students always ask me, "But what other kind of rule could there be? I can't think of anything else!", and I tell them that that's precisely the point. For the human brain, languages are specified as having no more than four kinds of rules, and the four kinds are deletion, insertion, substitution, and movement. This is true for all parts of the grammar -- phonology, morphology, syntax, etc. -- separately and in combination. (You might want to try imagining an ET language that lacks one or more of the Terran processes, and see what you come up with.)

Let's assume that -- however linguists want to express it formally -- there is some point in the life of a human thought at which the speaker is aware of the thought about to be spoken but hasn't yet given it a surface shape. We'll call that stage "deep structure" for convenience, although the term is out of fashion in contemporary lingustics. The beginning and ending of a deep structure will always be marked with a pound symbol. Here's an example of each kind of rule in action.

1. Deletion
a. #you eat your breakfast#
b. Eat your breakfast.
2. Insertion
a. #is raining#
b. It is raining.
3. Substitution
a. #John shaved John#
b. John shaved himself.

Note: In structures like these, linguists use a little "i" subscript for each instance of "John" to formalize the fact that they both refer to the same identical individual. If one "John" has a subscript "i" and the other has a subscript "j" it indicates that two different men named John were involved -- John Smith shaved John Jones, for example -- and the rule then would not apply.

4. Movement
a. #I hate eggplant#
b. Eggplant I hate.

Deletion Rules

When a deep structure undergoes a rule (or rules) and the end result is a sequence that can be spoken, the rule has transformed the sequence. Because no transformational rule can ever be allowed to change meaning, deletion rules can only be of two kinds.

5a. #if the Venusians say they will explain their warp drive, they will explain their warp drive#
5b. "If the Venusians say they will explain their warp drive, they will."

Examples 5a and 5b demonstrate the kind of deletion called "deletion under identity," where we understand a chunk of language -- even though it's not there on the surface -- because it's identical to a chunk that is present on the surface.

You could construct a deep structure like "If the Venusians say they will explain their warp drive, they will only be lying" and delete "only be lying"; you have a perfect moral right to do so, and I'll defend that right. However, when you said the resulting sequence aloud -- "If the Venusians say they will explain their warp drive, they will" -- you would have no right to expect that anyone would understand your sentence as ending with "only be lying." The deletion rule you're using allows the deletion of everything that follows the second "will," if and only if everything from "will" on is identical in both parts of the sentence.

6.a. #you beam me up, Scotty#
6.b. "Beam me up, Scotty!"

Examples 6a and 6b demonstrate the kind of deletion called "deletion of a constant." In rules of this kind, the item to be deleted is named in the rule and can only be that item. Examples 6a and 6b show the operation of Imperative Deletion, which allows only the deletion of the word "you."

Terran communication would be impossible if deletion rules weren't restricted -- constrained -- in this way. Suppose I could intend for you to understand my meaning as "I'll be at your place by five o'clock," and I could just delete any part of that sentence I wanted to -- perhaps saying "I'll be at your" or "I'll at place five." There would be no way for you, the listener, to have any idea what I wanted to tell you; my deletion would have changed the meaning of the sequence and the original elements of meaning could not be recovered. That's not allowed.

Insertion Rules

English -- unlike many other languages -- requires the surface subject position of sentences to be filled, even when there's no logical reason for that to happen. In almost all weather sentences, "it" must be inserted, even though there is no actual "it" that rains, snows, thunders, sleets, and so on. (The surface subject position is the position in the sentence immediately to the left of the predicate.) So....

7.a. #is raining cats and dogs#
7.b. "It is raining cats and dogs." (Usually "It's raining cats and dogs.")

This also happens in some other situations in which the surface subject position in the sentence is vacant. For example, the "it" in sentences like 8a and 8b doesn't refer to any actual (or even any hypothetical) "it."

8.a. "It's clear that linguistics is a useful subject."
8.b. "It's obvious that linguistics is a useful subject."

Traditionally, we assume that you get to 8a and 8b from "That linguistics is a useful subject is clear" and "That linguistics is a useful subject is obvious." Those sentences mean the same thing as 8a and 8b, but the "that [X]" chunk has been moved to the end, leaving the surface subject position empty; when that happens, English inserts "it" to fill that position.

(Unless, of course, the sequence of language is an advertisement or commercial -- in which case native speakers of English seem perfectly willing to accept sequences like "Kills germs" and "Makes your garden grow" and so on.)

In Lesson Seven -- unless I get messages from you telling me that more discussion of deletion and insertion is required -- we'll move on to discuss substitution rules; in Lesson Eight, we'll tackle movement rules. And then we're going to try using our knowledge of these four grammatical processes to analyze some sequences from science fictional languages.

In the meantime, if you'd like to try your hand at solving some linguistics problems from a variety of languages, you can find more than thirty such problems -- designed for high school students who are native speakers of English -- at the US Linguistics Olympus Website.

As always, I'd welcome your comments, suggestions, criticisms, questions, and other input, with a note telling me if you prefer to stay anonymous. Please e-mail me directly at

— Suzette Haden Elgin

Updated by webspinner