Do linguistic rules apply “across the board”?

Most linguists take it as self-evident that linguistic generalizations are, or should be, formulated in terms of abstract variables and that they apply “across the board”, that is to say, to all items belonging to the class defined by the variable (unless they are pre-empted by more specific knowledge). Of course, there are exceptions, especially in morphology – but, it is often assumed, behind every exception there is a default rule that applies “elsewhere”.

In this talk, I argue that while such “big mean rules” exist, they may be a special case: an exception rather than the norm, at least if we consider language as a cognitive phenomenon. In other words, even when highly general patterns arguably exist at the community level, they are not necessarily represented in individual speakers’ minds. Furthermore, in some cases, languages lack a default pattern even at the community level: that is to say, a language might not offer a standard solution for some types of communicative problems. 

I also reflect on why modern linguists have tended to overemphasize general rules and principles. This is partly due to the failure to distinguish between community grammars and mental grammars; partly to theoretical commitments which affect judgments; and in some cases, to more or less deliberate attempts at mystification.