Transformational Grammar. It is noticeable that the
structuralistic linguists of the thirties, forties, and fifties
have in the main focussed their attention on the exter-
nal side of language, on phonology and morphology
rather than on syntax. In this they followed in the
footsteps of their nineteenth-century predecessors. The
typical “historical grammar” of the last century was
two-thirds phonology and one-third morphology. It is
true that B. Delbrück wrote a Comparative Syntax of
the main Indo-European languages, published 1893-
1900, but in the main, syntax was neglected: one had
no methodological tool, comparable to the “sound
law,” to treat it in a scientific way. The theoretical
framework remained the traditional one. Many lin-
guists expressed their dissatisfaction with it, but few
had anything to put in its stead. An exception is the
Dane, Otto Jespersen, who introduced, among other
things, the distinction between nexus and junction
expressions, which roughly equals that made by Aris-
totle between predicative and non-predicative expres-
sions. Jespersen also elaborated a theory of rank, de-
signed to explicate the idea of syntactic rules.
Structuralists in general felt that syntax—like
semantics—would have to wait until a sure foundation
for grammar had been built on phonology and mor-
phology. That seemed natural, as “immediate constit-
uent analysis” (Bloomfield's version of traditional par-
sing theory) had to be based on the morphemes as
ultimate elements.
A dramatic change, however, took place in the fifties.
Noam Chomsky, professor of linguistics at Massachu-
setts Institute of Technology, published in 1957 a little
book called Syntactic Structures, in which he outlined
a new approach not only to syntactic analysis, but to
grammar and linguistics in general. It was followed
in 1964 by a far more extensive treatment, Aspects of
the Theory of Syntax. The new approach, generally
called “transformational grammar,” has had a tremen-
dous impact, and has led to almost feverish activity
in practically all linguistic fields during the late fifties
and sixties. The number of publications influenced by
Chomsky's ideas runs into thousands, and they appear
in all parts of the world. It seems evident that linguis-
tics is undergoing a change of orientation which is quite
as spectacular as the one that led to the establishment
of historical linguistics in the early nineteenth century.
The American structuralists in the Bloomfield tradi-
tion had carried to an extreme the empiricism and
positivism that had prevailed in linguistics for over a
century. Hence their neglect of the content side of
language. Hence also their insistence on the observa-
tion and classification of the material—the text—as the
chief object of linguistic analysis.
Chomsky made a clean break with these views. A
theory of language, as of anything else, cannot be
produced mechanically from the material, but has to
be invented. The classification of the “surface” proper-
ties of the text is not likely to yield the most fruitful
basic concepts of linguistic theory. Instead, the basic
concepts may very well be highly abstract con-
structions, connected with the observable reality only
in an extremely complicated fashion. The basic con-
cepts of, say, nuclear physics, illustrate this point. In
principle transformational theory aims at constructing
a theory which stands to language as physical theory
stands to the world of matter. One of its basic assump-
tions is that linguistic expressions—sentences—have
not only a surface structure, but also an underlying
deep structure which is not immediately available for
inspection. Roughly speaking the deep structure of a
sentence represents its content, the surface structure
its form.
Chomsky has himself pointed out that the notion of
deep structure may be said to be implicit in the seven-
teenth-century idea of general grammar, an idea that
had been almost totally eclipsed for a century and a
half. But there is at least one important difference
between the Port-Royal grammarians and Chomsky.
The seventeenth-century grammarians left it to the
intuition of the intelligent reader to establish the con-
nection between the general grammar and the actual
grammars of the particular languages. But the modern
transformationist aims at stating explicitly (i.e., for-
mally, mathematically) how the deep structures of a
language can be transformed into surface structures.
Starting from a precise, postulated set of primitive
terms and operations it should be possible to generate
fully any grammatical sentence in the language. Such
a grammar is called a “generative grammar,” a term
which is also sometimes used for the new approach
as a whole.
Transformational grammar clearly bears the imprint
of its own age. Like most linguistics in the twentieth
century it is synchronic rather than diachronic. Its
employment of postulates and abstract models recalls
the contemporary models of the advanced natural
sciences. Its insistence on strict formalization is natural
in an era where the hard work of comparing the theory
with the facts can be handed over to the electronic
computer.
But at the same time, transformational grammar is
also the product of more than two thousand years of
thinking about language. The distinction made be-
tween deep structure and surface structure may be said
to complete the work that the Port-Royal grammarians
had started, namely, to utilize to the full the contri-
bution of Aristotle and of the philosophers to language
study. And it is clear that the formalization of trans-
formational grammar could not have been attempted
without the strict methods of analysis developed by
structural linguists. They, in their turn, owed much of
their acumen in these matters to the Junggrammatiker
and, ultimately, to the Sanskrit grammarians. Linguists,
like all others, develop their science by standing on
the shoulders of their predecessors.
