Generative Grammar and Other Very Strange Ideas

The idea that grammars "generate" languages emerged in 1954 in writings by Zellig Harris and Charles Hockett. Ultimately it was adopted so enthusiastically that it became part of the fabric of how linguists conceive of syntax, an almost invisible tacit assumption. As such, it is hard to question. But I will try. I argue that the consequences of assuming generative grammars have not been thought through seriously enough. It is not only an inherently strange idea but also (despite useful applications in mathematics, logic, and computer science) an inappropriate one for characterizing human languages. Some of my arguments relate to the structure and properties of generative grammars themselves; others concern their fit with certain essential properties that human languages seem to share.