Language Acquisition

Jasmine D. Wagner
Psych W3450
Professor Terrace
March 29^th, 2000

Noam Chomsky

And His Psycholinguistic Theory

Noam Chomsky: Brief Biography

Born Dec. 7, 1928 in Philadelphia, PA

Undergraduate and Graduate education at the University of Pennsylvania.

1951-1955 Junior Fellow of Harvard University Society of Fellows. Completed his doctoral dissertation on "Transformational Analysis," major points of which later appeared in his 1957 publication Syntactic Structures.

1955 received his Ph. D. in Linguistics, began work as staff member at MIT

Behaviorist Linguistic Theory

Left-to-Right Probabilistic Model, or the "Markov Process"

Language functions as a series of sequential acts, where each word acts as a stimulus for the next response

Certain words have a higher probability of occurring in sequence, making some utterances more likely than others. Sequences that are grammatical are more likely to be spoken by adults, and more likely to be repeated and learned by children.

The cat
S R

The cat ate
S R

The cat ate the
S R

The cat ate the mouse.
S R

Criticism of Behaviorist Linguistic Theory

Karl Lashley’s paper: "The Problem of Serial Order in Behavior" 1951

Asserted that the temporal sequence of words must be determined by an underlying structure of greater depth than can be found in a series of verbal responses.

In other words:

The associative chain model of language is no good because we just don’t use words that way.

Reasons why:

1. Words themselves have no intrinsic meaning.

Example: The word pronounced "rite" has 4 spellings, many meanings, and can function as a noun, adjective, adverb, and verb.

"The mill-wright on my right thinks it right that some conventional rite should symbolize the right of every man to write as he pleases."

2. People make verbal and written Errors of Anticipation.

Slips of the tongue

Typos

Example:

Intended spoken or written phrase: "I like milk."

Phrase created: "I mike."

3. Selection or understanding of words placed early on in a sentence is determined by words that come later on.

Example:

"Rapid righting with his uninjured hand

saved from loss the contents of the capsized canoe."

- The meaning of the term righting is not determined until several seconds after the word is heard.

4. People easily understand embedded and conjoining sentences.

Example:

"John, who likes Angela, who likes Bill, likes Mary too."

5. People the various meanings of ambiguous sentences simultaneously.

"They are visiting firemen."

Chomsky’s Arguments Against Probabilistic Linguistic Models

Anticipated and supported by Lashley’s arguments

Presented in Chomsky’s 1957 book

Syntactic Structures

Chomsky argues that the transitional probabilities between words in a string (i.e., the probability with which one word may follow another) have no relation to whether that string of words is grammatical.

His reasons why:

"Colorless green ideas sleep furiously."

The probabilities between words pairs are very low (who has ever heard of "colorless green" and "ideas sleep" ?) but most would agree that the sentence is grammatically correct.

However! If the sentence is read backwards as:

ﾒFuriously sleep ideas green colorless."

The transitional probabilities between words remain the same, but the sentence is no longer grammatically correct.

There is no guarantee that a grammatically correct sentence will be generated if each word is chosen by its probability of following the immediately proceeding word.

Example:

"Goes down here is not large feet are the happy days."

The Left-to-Right model cannot generate the infinite number of grammatical English sentences because it cannot combine and embed.

1. If S1, then S2.

2. Either S1, or S2.

3. The man who said S is here.

People can understand simultaneously the various meanings of ambiguous sentences.

ﾒThey are flying planes."

N VP N

"They are flying planes."

N V NP

Phrase Structure Grammar

Criticisms of the Left-to-Right Probabilistic Model of Language, or the "Markov Process" led to the development of a hierarchical theory of grammar:

Phrase Structure Grammar

Elements of language are grouped into categories

Articles (T), nouns (N), verbs (V) are elements

Noun phrases (NP) and verb phrases (VP)

Theories allow for sequences of categories and elements to combine to form sentences

Examples of a few rules:

S -> NP + VP

NP -> T + N

VP -> V + NP

T -> the, a

Successful because the rules can be applied over and over again: Phrase structure grammar is "recursive."

A person who has learned a set of small, finite rules can hypothetically deal with an infinite set of input and output that function by these rules.

Phrase structure grammar can account for embedded and ambiguous sentences in a way that previous linguistic theory couldn’t.

HOWEVER

It bothered Chomsky that Phrase Structure Grammar dealt only with syntax of a sentence, not with the underlying semantics of the expression.

Wanted to demonstrate the presence of a structure beneath simple strings of words.

This led Chomsky to expand his theory of:

Transformational Grammar

A Transformation is an operation which converts one phrase structure into another using simple operations such as:

Substitution (pronoun for noun or a question word for a noun)
Displacement (moving a word from the end to the beginning of a sentence)
Permutation (reversing the order of two words, such as a subject and its auxiliary)

Chomsky used transformations and logic to give proof of the existence of a "Deep Structure" of language.

Charlie has asked Zack to buy avocados.

Charlie has asked Zack to buy what?

Substitution

What Charlie has asked Zack to buy?

Displacement

What has Charlie asked Zack to buy?

Permutation of subject and auxiliary

The Declarative statement

Charlie has asked Zack to buy avocados.

Has been transformed into the Interrogative

What has Charlie asked Zack to buy?

However, the transformation of

The boy hit the ball.

Into

What did the boy hit?

Is not as simple.

The boy hit the ball.

The boy hit what?

Substitution

What the boy hit?

Displacement

This last statement does not quite correspond to the desired "What did the boy hit?" even though it sounds quite like something uttered by many 2-year olds at a certain stage of linguistic development!

Where did the additional word "did" come from?

It appears as though we are missing an auxiliary to reverse the subject with!

In this case, we have to introduce a "dummy auxiliary," the verb do, in its past tense, did, in order to complete the correct grammatical transformation of the statement into a question.

We can then switch the auxiliary and its subject and result in the correct statement:

ﾒWhat did the boy hit?"

Chomsky reasoned:

Although the sentence "The boy hit the ball" has no apparent auxiliary, it must have some abstract auxiliary in its underlying structure.

This is based on the fact that the auxiliary appears in both its interrogative and negative forms.

What did the boy hit?

The boy didn’t hit the ball.

Not everything we know about a sentence is revealed by the string of words that we utter.

This is the distinction between the "Deep Structure" and the "Surface Structure" of a sentence.

Deep v. Surface Structure

The "Deep Structure" of a sentence contains what we know, while its "Surface Structure" is the grid, the pattern of words we use to express this knowledge.

There can be many different Surface Structures for one Deep Structure.

Deep Structure:

Cold coffee

Surface Structures:

ﾒI wonder if you could heat this coffee."
"Can you heat the coffee?"
"This coffee sure is cold!"
"Don’t you know I always expect heated coffee?"
"Cold again!"

Deep and Surface Structures are connected by a set of grammatical transformations.

When a child is learning a language, he is not merely repeating and memorizing a set of utterances (Surface Structures, like "cold coffee!"), but learning a set of rules by which he can translate an internal idea (a Deep Structure) into an infinite set of case-specific statements (Surface Structures) in order to express his mental states.

Children are predisposed to this method of language acquisition due to an innate mechanism, the LAD (Language Acquisition Device).

The LAD allows children to quickly hypothesize, test, and organize the grammatical rules of their language so they can begin to communicate as early as possible.

ﾒThe child is perhaps like a pianist waiting to sight-read a piece of music. The pianist will know in advance that the piece will have a rhythmic beat, but will not know whether it is in two, three or four time until she sees it. She will know that the notes are within a certain range - but she will not know in what order or combinations they come." (Aitchison, pg. 95)

Quite different from what came before!

Different from Behaviorist theory described earlier, especially in Skinner’s 1957 Verbal Behavior (MUCH criticized by Chomsky) wherein…

The same way a flashing light "controls" a rat’s "bar-pressing behavior" ...

a hungry tummy "controls" one’s "pass the bread!" behavior.

Not to mention that human infants are seldom rewarded for "good grammar."

In most cases (as studied by Brown, Cazden, and Bellugi in 1968) an infant’s incorrect grammar goes unnoticed. An infant will rewarded by its mother when it makes a novel utterance, or when it makes an informationally correct statement - even if the utterance is grammatically wrong.

And when the child is rewarded...

She is not rewarded with pellets.

Like in Skinner’s conditioning studies.