Grammar begins with understanding basic rules and their exceptions. The Ancient Indian Pāṇinian tradition is distinguished by viewing this phenomenon as a process of interlocking domains wherein the specific (the «exception») is excluded from the general (the «rule»). This process can be so intense that, at the end of the description, there is no more domain for the general to apply. This is a very productive method that has been applied by modern linguists to general theories of syntax and phonology.

Paul Kiparsky (1973) developed a version of this system into what he called the ‘Elsewhere Condition’ (EC) where a more specific rule blocks a more general one. His theory has since been long refined and contested, and applied to many languages and branches of linguistics. Chomsky and Halle’s foundational Sound Pattern of English (1968) employed, perhaps from an earlier version of Kiparsky, a similar approach to phonology.

Evidence for this approach goes, therefore, far and wide, but it also betrays an interesting lack of uniformity. The EC applies in circumstances where lexical variation and morphology interact (as in the example of English comparatives) and in those where suppletive verbal forms interact with semantics. Therefore there are reasons to believe this principle is general to language in all its forms and may even belong to rule formation when it applies to phenomena outside language. Our paper will expand the list of examples by returning to Pāṇini’s grammar and modern languages.

Prince (1993), in attempting to overturn the EC, has ironically argued that the principle may not merely be a means of describing rules but of the process by which language is generated, as Pāṇini does it. We will take this argument more seriously.

The EC and other disjunctive forms of rule ordering in linguistics very closely resemble patterns of behaviour observed in early theories of cognition. Classic work in Gestalt theory related to the recognition of subject and ground likewise rely on vision using set patterns of cognition to take one of two inexorably defined paths. Most productive for this comparison is the theory of affordances developed by J. J. Gibson (1979). We will return to the roots of the American psychologist’s arguments made in the early 1960s and apply them to our case with the aim of explaining why the EC operates in so many heterogeneous phenomena.
Doubts arise, however, when the comparison is extended to an attempt to account for the mechanism underlying such phenomena. Few would accept, for example, that a speaker uses the EC in generating his own (correct) grammatical forms. We will attempt to find examples to illustrate cases where the comparison of these theories is productive and those where it is not.