Chapter 4: First Phrases: Glimpses of Grammar

1

Chapter 4: First Phrases: Glimpses of Grammar

What Ignites Grammar?

At one sweet and precious moment, a one-year-old child points to sneakers and says "Daddy.” That child has very overtly entered the human species.

Really? Look closely. Suddenly a word captures not an object itself (sneakers) but a relation between a thing and its owner, specifically, a possessive relation (Daddy's sneakers). A new, infinite, realm of communication opens. Reference itself has changed from object and event names to relationships. (nn1)

For each child, of course, a different verbal gem ignites this new realm of reference. One child points at an empty crib and says "doll." Another child points at a chair and says "sweater." Why? The astute parent realizes that it is because the child had left the sweater on that chair earlier, so “sweater-location” is intended. Location (“on chair”) and possession (“Daddy's sneakers”) are the usual relations of a child's first expressions, but others come quickly. A child points to the door and says "coat." (nn2) Hmm, it takes some more parental puzzling, but then it is clear: you put on a coat to go out the door. The coat reflects the "manner" of going outdoors---a relation to be captured ultimately by with or without (“Don't go out without a coat!”) All at once single words have "relational reference" (like manner-of-action) on top of direct reference (like naming) and now the instrument of communication has begun to grow.

But why do we claim that membership in the "human species" rests in this moment? Because animals (as far as we know) do not refer to objects in any creative way at all. As we have noted, a monkey can have a gesture that says "this is my lover" or "stay away from my child," but cannot say "my stomach hurts.” Fixed social relations are just that, fixed in biologically innate gestures, not creative relations which can be freely invented and applied between objects and people. This infinite, expressive quality of human grammar is special, not found naturally and without instruction elsewhere in the animal kingdom.

Are we just guessing about these "relations"? The answer is, yes, a little bit. How can we "prove" that the child meant possessive? What if our child, who just said "Daddy" (for sneakers), had had an Agent relation in mind: "put on your shoes Dad, let's go out and play.” Proof of the possessive intepretation is hard to guarantee, but support can be found with a little imagination. We can informally test the possessive claim with toys or pictures of them, by using ellipsis:

Exploration 4.1: Early Relations
Point to a toy that belongs to a cat (also a toy). And say:
“this is the cat's toy”
[put in some simple pictures here]
Now with a toy dog and another toy next to it, say:
“now point to the dog's”
and see whether the child points to the dog or the toy.
You might vary it with similar examples and sometimes say:
“now point to the dog”
and see if they have the difference and understand the possessive. (See also experiments on possessives and ellipsis in chapters 7 and 8.)

Caption: Understanding possessives

Exploration 4.2: Basic Relations for Older Children
With an older child a more complex variant would be revealing.
Put two dogs and two toys next to each other, and say:
“Here's the cats' toys.
Now show me the dogs'.”
The second form could of course be understood as a plural:
“Now show me the dogs.”

Caption: Understanding complex possessives

Which will a child take and why? If the child prefers the plural they will show that they prefer a reading where one does not have to borrow parts from a previous sentence. The possessive requires one to fill in:

“Now show me the dogs’ (toys)”

If the child prefers the possessive, it suggests that they prefer to use the whole discourse, and obey the thrust of the previous sentence. Current literature suggests that children do not (or even cannot) use discourse, so they should resolutely prefer the plural. I suspect that the opposite is true and this exploration would show it. We will return to this question.

Exploration 4.3: Another Game
Wander around the house and point to things. Point to something that belongs to Dad and say "Dad's," then to something which is Mom's and say "Mom's." Then, crucially, point to something the child knows is Mom's and say "Dad's."
Now see how the child responds. If the child gives it to Dad, then "Dad's" was understood as “(give) to Dad.” If the child somehow frowns or says "no," then we have evidence that the possessive guess was right. If he says “no that a hat,” then he understood “Dad” and not “Dad’s.”

Caption: Elliptical possessives

New Relations: Agent, Instrument, Object

Within days new relations tumble out of the child's mind. One child who wants a parent to play a record, first points at the record and says "Mom,” a day later he says "acord,” and a day later he says "round." The same activity is seen three ways: by agent (“Mom”), object (“record”) and manner (“round and round”). It is as if, through these new relations, angles on existence are being elevated to consciousness.

How much can we accomplish with these new "relational nouns"? A lot, but grammar has more tricks up its sleeve to express thoughts efficiently. First the implied relations grow into familiar prepositions: by, with, of, at, for, that capture exactly the core relations that grammar uses: agent (by), instrument (with), possessive (of), locative (at), goal (for). This is just one step in the expression of mental operations through grammar.

These grammatical relations do not cover everything. For instance they do not cover actual human relationships. The word détente may reflect the relationship of the Soviet Union to the U.S. ---or even the feeling of two-year-old Johnny toward his three-year-old sister Jennie, but "détente" is not captured directly in any of grammar's prepositions. Prepositions only go so far. In fact one of the important thrusts of modern grammar is that the explicit (and efficient) relations are really just a handful: agent, object, possessor, experiencer, goal, instrument, manner. Both other words and other mental operations are needed to convey complex relations.

Have we missed something of the human essence here? A child's personality is also richly on display. The child whose first word is "kiss" has something different on her mind than the child who first word says "no.” Personality, of course, gets into everything. Language carries personality right along with meaning.

Now the child is ready to move on. What new power is held in the simple act of combining two words? The primary set of relations (those prepositional relations) flower into endless new branches which we will carefully explore. Each branch of language articulates a form of mental computation, used in language and elsewhere. Each step is a further display of what it means to be human. Of course, a child (or you) are still human even if nothing is said. Humanity is just not as vividly on display.

Two Words to Many Words

The child's grammar is now on the verge of explosion. Two-word sentences follow that make those relations explicit:

"baby highchair" = baby is in the highchair

"Mommy eggnog" = M had her eggnog

"sat wall" = he sat on the wall

"throw Daddy" = throw it to Daddy

"pick glove" = pick the glove up

agent-action "baby eat"

verb-object "pick glove"

agent-loc "baby highchair"

possessive-obj "mommy sock"

Agent-object "mommy eggnog"

verb-Goal "throw daddy"

Is this an "anything goes" situation? One surely cannot be confident of these judgements. Does the child really distinguish between goal (as in a dative: “throw daddy a ball” => Daddy = goal) and the object (ball). It is difficult to be certain. It is possible that a more abstract notion of general Object (or complement) defines what the child chooses, and no distinction between direct and indirect object is part of the grammar (even though the interpretations are different--the child does not heave Daddy when he says “throw Daddy.”)

What we can assert with some confidence is that a range of common relations are excluded and almost never occur:

Excluded:

identity *"mommy lady” (= mommy is a lady)

conjunction: *”fork knife” (fork and knife)

two clause: *"said ate” (=I said I ate)

Why are these forms absent? There may be multiple reasons. First it is possible that some part of the child's grammar must mature before they are available. In particular "I said I ate" requires us to put one sentence inside another. We also fail to find identity as in: "Mommy girl.” That comes when the child grasps the meaning of is (“mommy is a girl”) as a kind of equivalence which in turn may follow is as introducing a property (“Mommy (is) tall”).

The omission of conjunction, however, points at exactly how the structure is built. Now we must start to reason very carefully. It looks as if the child just "combines" two words. The more we contemplate things, the more surprising they become. We have seen that not all meanings are used. Why wouldn't the simplest connection be "and"? Why doesn't a child start out with lots of expressions that capture things conjoined in everyday life?

MommyDaddy

hatcoat

shoessocks

meatrice (=meat + rice)

Where should we locate this bias against conjunction? It is unlikely that the restriction is cognitive. The child certainly can put two things together elsewhere in life. Another possibility is that words are not just added: their combination invokes a syntactic structure which in turn invites certain meanings but not others. We are moving toward seeing a fundamental asymmetry at the core of how grammatical structure is built by the human mind.

If we just put two words together, then we would be making a list like:

house, doll

coffee, cup

The mode of combining that produces:

dollhouse

coffeecup

is fundamentally different. It is a modifier relation where one element is the "main" element: a dollhouse is a house and a coffeecup is a cup. The choice is clear in a contrast like: boathouse/houseboat. This is the most basic syntactic act which distinguishes human language. It combines two things and decides who controls or dominates the combination. If we take an adjective and a noun, the relation seems obvious:

“red, house” = red-adjective + house-noun

The combination is still a noun. We can represent it by saying the whole thing is like one pure noun so we put noun brackets around it and label it: NOUN[red house]. Another way to illustrate the parts it is to combine them in one phrase with a label on top. We could make the label the thing itself to illustrate what controls:

house

/ \

red house

or the syntactic category which has become customary.

Noun

/ \

Adjective Noun

| |

red house

This "symbolic tree" just shows how we "combine and decide which controls,” Link and Label. In theoretical terms, the operation of link-and-label is called Merge. (nn3) Linking is not just "linking,” though. This is a distinctive human operation that says something more precise than "combine.” The label captures the nature of the link.

Each phrase above involves some decision:

"sit wall" = verb => about sitting, not a kind of wall

"pick glove" = an action, not a glove category

"mommy eggnog" = possession or hidden action, not a kind of eggnog

Again, among the excluded structures is what one might think of as the easiest thing, a list, a conjunction. But we do not usually find:

knife fork

Mommy Daddy

drink food

That is, children do not choose conjunction as the first kind of two-word utterances. (We exclude one word sequences with pauses which may occur). Cognitively it looks like the simplest form, but structurally it is not. Here is why:

If I say:

“Mommy Daddy”

is it basically Mommy or basically Daddy? We could not assign a higher structure and decide what controls:

Mommy

/ \

Mommy Daddy

Deciding that it is basically Mommy or basically Daddy is a choice most families would choose not to make. Now we can see that our structural claim about choosing what is on top actually predicts that conjunction will not occur. Conjunction allows neither form to be the primary one. Instead it is a hidden democratic "and." Thus we do not have:

*"Mommy Daddy"

because it would really mean "Mommy and Daddy" where and-ness is the dominating feature.

(and)

/ \

Mommy (and)

/ \

(and) Daddy

But again if I say houseboat we know it is basically a boat, and if I say boathouse we know it is basically a house.

boat

/ \

house boat

Since neither "Mommy" or "Daddy" is and-ness, the core concept is unexpressed, so now it should not seem so surprising that a child does not start with "Mommydaddy.”

In technical terms, a two-word utterance involves a form of dominance or subordination. One word subordinates another. This seems so natural to us that it is easy to miss how profound an act of deployment of mental information is involved in just building what we call a Merge structure.

Exploration 4.4: Domination or Coordination
We can go right after this claim. Which word does the child take as dominant? Here are some pairs that can pull it out:
[Insert sketch of boathouse and houseboat]
Picture of: boathouse
Picture of: houseboat
Two pictures in one: boat, house
which of those three will the child pick when we say:
"boathouse.”
One could create others (dollhouse, housedoll, or doll, house). Or let us create a new one from slippers, a shoe-tree, and a picture of shoes
and a house:
home-shoe
shoe-home,
home, shoe.

Caption: Compounds