in much of the 20th century has been driven by the analysis of a linguist: Noam Chomsky

the battlefield: syntax
typical analysis: out-of context sentences

Chomsky: how can we produce and comprehend an unlimited number of acceptable and only acceptable word strings
that is: what type of machinery could do this (called a descriptive grammar)

1956: B. F. Skinner: Verbal Behavior syntx can be understood as a phenomenon of chaining (in much the same way that chaining occurs when an animal learns a sequence of behaviours)

Chomsky correctly saw this as an example of a finite state descriptive grammar

device: W1-> W2-> W3 etc
that is, there is a left to right system in which word is dependent on the words that preceded it,


chomsky: is interested in competence models: the ideal conceptualization of conditions that must be met (not effected by performance factors, such as attentional limitations, memory etc)

his data: intuitions that the average person could tell you about their language

example: is this word string grammatical (an acceptable word order in your language)

consider: "colourless green ideas sleep furiously"
is this acceptable

-note: grammaticality but not semanticity

modularity in which syntax takes priority

using this type of logic he argues that, indeed, a skinner-like model is a realistic descriptive grammar (that is it will produce acceptable and only acceptable word strings) BUT fails to capture all (or in fact, many) the possible word strings that are produced everyday


example: the problem with embedding
Charles loves Camilla
who does charles love?

Charles even when though married loves Camilla
who does charles love?


Chomsky (round 1); phase structure grammar

argues that the grammar can't be a left-right system but rather a hierarchical system
can be explained as re-write rules (not we have here a version of production rules, and a production system)

S-> VP NP
NP->(det) A*N
VP->V (NP)
VP->V S

(See pp. 367-372)


Chomsky (round 2; transformation-generative grammar)

Problem: if we use phrase structure as a roadmap to understanding a sentence, we are faced with several problems

1. Phrase structure ambiguity
-one sentence can be given more than one phrase structure...so which meaning is being intended

2. Items with different phrase structures can mean the same thing...but how do we know this? 3. Reflexive form
    why is acceptable to say
    I love you    or    You love me

but not
    I love me*     You love you*


Suggests there are two levels of analysis

at a d-structure level: the thematic relations are represented (ie something like propositions in which who did what to whom is represented)

at a s-structure level: represents the sentence as output

a consequence; one needs a set of rules that permits the translation of d-level meaning to s-level forms (changes word order without changing meaning)


chomsky and the innateness hypothesis

1. Children don't seem to get feedback on grammaticality.. .so how do children learn to produce only grammatical sentences

2. Children pick up grammar from impoverished input


3. Within a language there are correlational characteristics that have counterparts in other language
-example in English VP; verb ALWAYS comes before its' object
we say "take the cheese" and never "the cheese take"

but in some other languages one always gets the opposite

principles and parameters approach
-there is a universal grammar which consists of s et of "switches", each of which can be set to one of a number of parameters (values)
-language learning, by this account, consists of--via exposure to a particular language-- setting the appropriate switches of that language


Psych 235 Home Page