In Syntactic Structures, Chomsky tries to construct a "formalized theory of linguistic structure" and places
emphasis on "rigorous formulations" and "precisely constructed models".
Justification of grammars
Chomsky writes that his "fundamental concern" is "the problem of justification of grammars". He defines "a
grammar of the language L" as "essentially a theory of L", as well as "a device that generates all of the
grammatical sequences of L and none of the ungrammatical ones". Talking about the goals of linguistic
theory, he draws parallels to theories in physical sciences. He compares a finite corpus of utterances of a
particular language to "observations", grammatical rules to "laws" which are stated in terms of "hypothetical
constructs" such as phonemes, phrases, etc.[14] According to Chomsky, the criteria for the "justification of
grammars" are "external conditions of adequacy", "condition of generality" and "simplicity". To choose
which is the best grammar for a given corpus of a given language, Chomsky shows his preference for the
"evaluation procedure" (which chooses the best possible grammar for a language against the aforementioned
criteria) over the "discovery procedure" (a procedure employed in structural linguistics which is supposed to
automatically produce the correct grammar of a language from a corpus) or the "decision procedure" (a
procedure which is supposed to automatically choose the best grammar for a language from a set of
competing grammars).
Grammaticality
According to Chomsky, "the fundamental aim in the linguistic analysis of a language L is to separate the
grammatical sequences which are the sentences of L from the ungrammatical sequences which are not
sentences of L and to study the structure of the grammatical sequences."[16] By "grammatical" Chomsky
means "acceptable to a native speaker".[16] Analyzing further about the basis of grammaticality, Chomsky
shows three ways that do not determine whether a sentence is grammatical or not: its inclusion in a corpus,
its being meaningful, and its being statistically probable. To illustrate his point, Chomsky presents a
nonsensical sentence "Colorless green ideas sleep furiously"[17] and says that even though the sentence is
grammatical, it is not included in any known corpus at the time and is neither meaningful nor statistically
probable.
Chomsky concludes that "grammar is autonomous and independent of meaning, and that probabilistic
models give no particular insight into some of the basic problems of syntactic structure." [18]
Grammar models
Assuming that a set of "grammatical" sentences of a language has been given, Chomsky then tries to figure
out what sort of device or model gives an adequate account of this set of utterances. To this end, he first
discusses finite state grammar, a communication theoretic model based on a conception of language as a
Markov process. Then he discusses phrase structure grammar, a model based on immediate constituent
analysis. He shows that both these models are inadequate for the purpose of linguistic description and as a
solution, proposes his own formal theory of syntax called transformational generative grammar (TGG), "a
more powerful model combining phrase structure and grammatical transformations that might remedy these
inadequacies."
