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


It Depends on What the Meaning of "Rule" Is....

I want to begin this lesson with two quotations. The first, by Peter Norquest, is from a 6/16/01 review of April McMahon's book, Change, Chance and Optimality, on Linguist List; Optimality Theory (OT) is one of the currently fashionable theoretical models in linguistics. Norquest writes:

"OT stands in sharp contrast with rule-based theories due to the fact that it does away with rules altogether and relies solely on a single formal object, that of constraints."

The second quotation comes from "Optimality: From Neural Networks to Universal Grammar," by Alan Prince and Paul Smolensky, pp. 1604-1610, Science for 3/14/97, on page 1604:

"In an English sentence like it rains, a constraint requiring all words to contribute to meaning (unlike the elment it in this usage), conflicts with a structural constraint requiring all sentences to have subjects; and the latter controls the outcome."

Our Real World Linguistics rule says that when the deep structure of an English sentence has no subject nominal in it you have to insert a meaningless "it" to fill that gap. OT says that English has a constraint barring all sentences that have no subject nominal. This grammatical fact about English can be written as either a constraint or a rule. When you write it as a rule you don't have to include that constraint; when you write it as a constraint you don't have to include that rule. To say "When you send a child out into the rain, put a raincoat on the child" is a rule; to say "No child is allowed out in the rain without a raincoat" is a constraint; the effect on the child in the real world is identical.

It's certainly possible to take the position that a rule actively does something, while a constraint passively sees to it that something isn't done. The constraint approach is considered more elegant. But I see no Real World justification for claiming that either rules or constraints don't exist, especially in an introductory course, and I plan to go on using both terms.

The Grammatical Processes of Terran, Continued:
Substitution Rules

We've now looked at some English insertion rules and deletion rules; in this lesson we're going to consider some substitution rules. Linguists indicate that two or more items refer to the same entity (that is, that they are coreferential) by putting a tiny subscript "i" at the end of each one. Since I can't do that conveniently in this format, I'm going to use a different notation: Two or more items that are coreferential will both be in boldface type in my examples. For instance....

1a. #when the captain announced our destination, the Venusian delegate was shaving the Venusian delegate#
1b."When the captain announced our destination, the Venusian delegate was shaving himself."

This is the reflexive substitution rule in action; it notices that both instances of "the Venusian delegate" in example 1a are in boldface type, which means that they refer to the same individual; it notices that one Venusian delegate is the subject and the other Venusian delegate is the Object of "was shaving"; and it therefore substitutes "himself" for the second one, giving us 1b. If the two hadn't been in boldface type, that would have meant that there were two Venusian delegates, one of whom was shaving the other one, and the rule wouldn't apply.

2a. #when the officers announced our destination, the Venusian delegates were shaving the officers#
2b. "When the officers announced our destination, the Venusian delegates were shaving them."

This is the plain vanilla pronominalization rule we all learned (sort of) in grade school. [You'll probably remember a grammar incantation along the lines of "a pronoun is a word that replaces the noun which is its antecedent and agrees with it in both number and gender."] The rule notices that there are two instances of "the officers" and that they refer to the same persons; it substitutes the pronoun "them" for the second one. (You could propose that an ET language always substitutes a pronoun for the first of a coreferential pair instead of the second, and that might say something about the brain of the ET speakers of that language -- but it would still be only a substitution rule.)

I Can't Find The Small Blue Squishy It!

That grammar incantation back there ("a pronoun is a word" etc.), as it's all too often taught to kids, doesn't work. Suppose we start with example 3a...

3a. #I've been looking for the small blue squishy creature, but I can't find the small blue squishy creature#

Apply the grammar incantation, using a pronoun to replace "the noun that is its antecedent," and you'll get the following sentence, which speakers of English will reject as unacceptable (indicated by the asterisk at the beginning of the sentence):

3b. *"I've been looking for the small blue squishy creature, but I can't find the small blue squishy it."

It would be nice if all youngsters were taught that pronouns replace the whole noun phrase -- that is, the noun plus all the items that go with it -- instead of just the noun. If they were, the rule would work and the result would be this:

3c. "I've been looking for the small blue squishy creature, but I can't find it."
[Note: We need a note here about terms, to avoid confusion.

You'll notice that I use both "noun phrase" and "nominal." A noun phrase of English can include various modifying items -- for example, "a" or "the" or "this," adjectives like "blue" and "squishy," whole embedded sentences such as "who is known to have seen a dragon" in "the woman who is known to have seen a dragon." A nominal is a noun phrase plus its case marker. I'd rather not worry you with all that right now, but I know that if I don't mention it I'll get grumpy e-mails. Feel free to let your mental grammar -- which always knows which items belong together as one grammatical chunk -- take care of it.]

Fuzzy Coreference

Theoretically, substitution rules can only be used when the two or more items involved are exactly the same item, but Terrans (at least, English-speaking Terrans) are flexible about these things; often, they'll accept what's called "fuzzy" coreference. For example:

4a. #after the firelizard's tail was cut off, the firelizard's tail grew back#
4b. "After the firelizard's tail was cut off, it grew back."

That's fuzzy. Suppose it's true that you can cut off a firelizard's tail and know that the firelizard will grow another one; the new tail is not the same tail that you cut off. But the Terran mind says, "Heck, that's close enough," and applies the rule, and nobody refuses to accept sentences like 4b on standardized tests. Stretch the fuzz just a little farther, however, and you'll lose points. For example:

5. "Norm is an orphan, and he misses both of them."

Speakers of English are fully capable of understanding that sentence. They understand that "them" is being substituted for "his parents"; in a conversation, they'd let the whole thing go by. But they won't accept the sentence on a test.

Enough for now; next time, we'll do movement rules.


I've said that the Terran brain -- even when it's the brain of a science fiction writer -- can't come up with any grammatical processes other than these four that we're considering at the moment: deletion, insertion, substitution, movement. I've suggested that anyone who can come up with additional grammatical processes has to be an ET. A number of you have responded with "Oh, yeah? Just give me a little time!" That's wonderful; I'm looking forward to getting your proposals for a fifth grammatical process... and a sixth.... and a seventh. Please do send them; we'll discuss them.

As always, I'd welcome your comments, suggestions, criticisms, questions, and other input, with a note telling me if you prefer to stay anonymous; e-mail me directly at I'd also be grateful if you would spread the word that the course exists, so that this discussion we're having can include more people; please tell your friends and your listmates. Thank you for your help.

— Suzette Haden Elgin

Updated by webspinner