Geoffrey K. Pullum
University of California, Santa Cruz
My aim in this paper is to discuss an intensely complex cluster of interlinked concepts involving distinctions between (i) descriptive and prescriptive grammar, (ii) constitutive and regulative rules, (iii) conservative and liberal attitudes, and (iv) standard and non-standard dialects. I cannot hope to be comprehensive, but I will try to be clear.
Sometimes it seems to me that the issues we are dealing with here have been mired in the same controversies for a good forty years. Other times it seems more like a hundred and forty. There are few signs of any knowledge about grammar dating from after 1900 having become known to a broad cross-section of the general public or having had an impact on education. Perhaps it is time to attempt to understand the situation better, rather than simply to deplore it.
Correctness conditions
I begin by taking it for granted that there are conditions we might call correctness conditions for natural languages. (Whether they are standard languages, non-standard dialects, or undescribed tribal languages of preliterate peoples does not matter: all have correctness conditions.) And I will also assume that it is possible in principle to be perfectly explicit about such conditions. In terms of the distinction drawn familiar thirty-five years ago by John Searle,1 They are constitutive, not regulative. They do not regulate the use of the language, in the sense that one could use it either in ways that comply or in ways that don’t; they constitute the language, in the sense that not respecting them amounts to not using it at all but doing something else instead.
Modern descriptive linguists try to figure out from the available evidence the principles that constitute the language being described, and to give explicit, and potentially falsifiable, formulations of them.
Linguists often make little effort to distinguish the conditions themselves (the