Things that I Think Other People Are Wrong About: Grammar

Right now, I feel like writing about grammar. So, without further ado…

The Mainstream Approach: Constituency Grammar

The most common theory of grammar going around these days is constituency grammar, formulated by illustrious linguist and MIT professor emeritus Noam Chomsky.

hate constituency grammar.

So, instead of constituency, I’m going to advocate for another theory that I like better, namely, dependency grammar. Dependency grammar apparently goes all the way back to ancient India, if I remember right. It was either revived or independently discovered again by a guy whose name escapes me, but the voice in the back of my head says he was French.

Now, here’s why I like dependency grammar better than constituency grammar. Constituency grammar analyzes sentences by breaking them up into their constituent parts. So, a sentence is analyzed into a noun phrase + a verb phrase, where the noun phrase is the subject and the verb phrase is the predicate. The noun phrase in turn is analyzed into an article + noun, or article + adjective + noun, or article + noun + prepositional phrase, etc. The verb phrase is also analyzed into just a verb, or verb + noun phrase (where the noun phrase will be the direct object or complement), or auxiliary verb + infinitive + noun phrase, or auxiliary verb + infinitive + noun phrase + adverb, etc. Again, the noun phrase within the verb phrase can in turn be analyzed into its own parts. In constituency grammar, the grammar of a language is defined as the list of all the possible rules for what sequences of grammatical elements can be substituted for what other sequences. Notice that the emphasis in constituency grammar is on the parts of the sentence; the relationships between parts are derived from the structure of the sentence.

To me, constituency grammar just seems like an overly complicated, convoluted mess. First of all, there’s an infinite (or practically infinite) variety of possible sentences, which means that there’s going to be an infinite number of possible combinations of constituents that we’re going to have to account for. It’s always a huge hassle to come up with exhaustive lists of anything, let alone sufficient grammar rules to account for all possible sentences. I’m not claiming it’s impossible, but it’s such a hassle. It’s like if someone were to ask you to add up all the numbers from 1 to 100. If someone were to ask you to do that, you would probably think, “There has to be a better way to do this than just adding it all up by hand. Please, let there be a better way to do this than just adding it all up by hand.” And of course, there is, though it’s not exactly obvious at first glance. To me, constituency grammar is analogous to the inefficient face-value approach―whereas dependency is the less obvious but more elegant and simple solution. Second, constituency grammar, though claiming to be a universal grammar capable of accounting for all languages, seems to me to be biased towards English (and similar languages―specifically, analytic, subject-prominent languages). In constituency grammar, the two primary components of the sentence are generally said to be a noun phrase, corresponding to the subject, and a verb phrase, corresponding to the predicate. The subject is considered a separate element on par with the verb phrase, while the direct object is considered a subordinate element of the verb phrase. Why does the subject get special treatment? Isn’t it just as closely related to the verb as the object? I mean, come on, the object receives the action just as much as the subject gives it. I can only think of two reasons to set the subject apart like that: 1. “The subject is what the sentence is about”; and 2. “You can have a sentence without a direct object, but you can’t have a sentence without a subject.” But these objections only hold in English (and other languages like it). In English, the subject is implicitly understood to be what the sentence is about because English is what’s called a subject-prominent language. In topic-prominent languages, it’s possible, and perfectly normal, to have sentences that are “about” something other than the subject, be it the direct object, or the location where the action described takes place, or the time when it takes place, or something else entirely. And as for 2., of the four languages I have some competency in, English is the only one that makes it mandatory to specify a subject; mandatory subjects are by no means universal. Furthermore, as I said above, the emphasis in constituency grammar is on the parts of the sentence; the relationships between the words are secondary and derive from the way the parts go together. In languages like English, where the relationships between parts are indicated by the relative positions of the parts, this is well and good. But in languages with freer word order, it’s not always easy to separate the sentence into a nice, tidy verb phrase and a nice, tidy noun phrase. Just ask any high school student in AP Latin.

Introducing… Dependency: THE Syntactic Relationship

Now, let’s take a look at dependency grammar. The beauty of dependency grammar is that it reduces all relationships between words to one kind of relationship, called (you guessed it) dependency. Dependency is an asymmetrical binary relationship―that is, it’s a relationship between two things, one of which is subordinate to the other. The subordinate word is called (you guessed it) the dependent, and the dominant word is called the head. In all dependencies, the head is the “main” concept, while the dependent in some way clarifies or modifies the meaning of the head. There are two kinds of dependencies, defined by how their dependents relate to their heads. In some dependencies, the dependent just adds extra information to the head; it is not necessary to the coherence of the sentence. In these cases, the dependent is a modifier. In the other kind of dependency, the dependent provides information that is vital to understanding the head; without the dependent, the sentence will not be coherent. In these cases, the dependent is an argument of the head. In all dependencies, the whole phrase acts as if it is “concentrated” at the head; if a word is going to act on the phrase as a whole, it acts on the head. Because of this, it is possible to reduce any sentence to a hierarchy of dependencies: once you figure out the dependencies within one phrase, the phrase as a whole will fit nicely with the rest of the sentence by having its head depend on something else. You could say that the head is the interface between a phrase and the rest of the sentence. The word at the top of the hierarchy―the word that doesn’t depend on any others―is the main verb of the main clause, the one word that directly says that a certain state holds in reality. (But more on that in another article.)

And… yeah. That’s about it, I think. None of those innumerable “transformation rules” that constituency depends on, no pun intended.

Below is a sentence diagrammed under constituency grammar (left) and dependency grammar (right). Notice how much simpler the dependency diagram is.

(Legend: S = sentence; NP = noun phrase; VP = verb phrase; D = determiner (article, or other word that indicates number or definiteness); PP = prepositional phrase; N = noun; V = verb; P = preposition; A = adjective)

I realize that that discussion above was pretty abstract, so now let’s get more into specifics, using this diagram as an example. As you can see, like I said earlier, the main verb “is” is at the top of this diagram because it carries the overall function of the sentence, namely stating that a certain state holds in reality. Its two dependents are the subject “house” and its complement, “red.” These words are dependents because they clarify how exactly the action/state of the main verb occurs, in this case what particular thing is existing and how that thing exists. Since we can’t conceive of existence without some particular thing existing, these words are necessary to the meaning of the sentence and are therefore arguments, as opposed to modifiers, of the verb. The noun “house” in turn has two dependents, “the” and “at.” “The” indicates that this is a definite noun, i.e. a particular house that the listener is already aware of as opposed to some other house that the listener is hearing about for the first time. This is a dependent because it clarifies what particular house we’re talking about. The other word directly depending on “house” is “at.” This one is a bit harder to see because you can’t really think about “at” without thinking “at what?” But a little thought shows that this treatment of the word is correct. “At” is a preposition. A prepositional phrase provides new information about a noun: “the house at the end of the street” as opposed to some other house at some other place. Prepositions provide this information by specifying a thing’s relationship with some other thing. In this case, the preposition “at” is relating “house” and “end.” It is appealing to the relation of “at-ness,” i.e. proximity, while the word “end” clarifies what location the house is at. Since “at” specifies a relationship and “end” specifies what that is a relationship with, “end” depends on “at.” Since we couldn’t conceive of “at-ness” without a particular location to be at, “end” is an argument, not a modifier. “End” in turn has its own determiner and prepositional phrase depending on it, but those are parallel to the dependents of “house,” so I don’t think there’s any need to go into detail there. In this case, the adjective in the sentence is a complement of the main verb, so it isn’t depending on a noun. But if it were in front of a noun, like if the sentence said “the ugly house at the end of the street,” it would depend on its noun. So, there would be another “A” underneath the N corresponding to “house.” Since adjectives don’t need arguments, there would be nothing underneath the “A,” unless there happened to be an adverb modifying “ugly” (“the incredibly ugly house…”). Adverbs function just like adjectives, except that they depend on verbs, adjectives, and other adverbs instead of on nouns.

So, there you have it. I know I tend to rush through things, so sorry if some of it wasn’t clear; I’ll probably come back and update if I think some of it isn’t explained well.


One thought on “Things that I Think Other People Are Wrong About: Grammar

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s