Creation myths of generative grammar, and the mathematics underlying Syntactic Structures

Syntactic Structures (SyS) is perhaps the most influential book ever published on syntactic theory. It locked syntacticians into a half-century of work dominated by the generative conception of grammar. Its empirical analyses of English are as widely admired as its theoretical rigor.

Without in any way denying the lasting impact, brilliant clarity, or persuasive force of SyS, I examine here a few cases of exaggeration or misapprehension, and challenge some creation myths about generative grammar that seem to have emerged from inaccurate recollections.

• Linguists and cognitive scientists speak of SyS as having launched a cognitive revolution, and journalists tout its revolutionary theories of the mind and of universal grammar. None of this is accurate.
• SyS has been treated as if it sprang almost out of nowhere, but the relevance of the work of Emil Post (not referred to in either SyS or the precursor typescript The Logical Structure of Linguistic Theory) has been greatly underestimated in this connection.
• SyS contains no proof that natural languages are not finite-state, and no sound argument against stochastic modeling of grammaticalness.
• SyS has been credited with launching a cognitive revolution, and of promulgating theories of the mind, or of universal grammar. None of these things are true.
• The descriptive proposals advanced in the course of its arguments for transformations were not fully original, and had many failings, both formal and empirical.

What is perhaps most of interest in the context of this conference is that its claim to be based on formal methods does not square with its loose and confusing use of symbols. It shows little sign of the care found in the best successful formalizations. Its supposedly formalized restatements of transformations are sometimes less explicit than the intuitive presentations of their content earlier in the text.