Some recent works in phonology and morphology assume a BLOCKING mechanism t which precludes the application of a general rule whenever a more specific one is operative. The principles which permit such blocking are known as DISJUNCTIVE ORDERING or the ELSEWHERE PRINCIPLE, and the blocking itself can be construed either as a formal constraint on rules (Anderson 1969(Anderson , 1986;;Kiparsky 1973; Matthews 1972, 191 ft.) or as a general cognitive condition (Kiparsky 1982b).
However, Janda and Sandoval (1984) cast some serious doubts on the validity of such a condition "at least for lexically-free rules of morphology" (p. 39). They note a large number of counter-examples to blocking and argue that other mechanisms can explain the facts that appeared to militate in its favor. Anderson (1986) in a well-argued paper shows how one can explain some of the apparent counter-examples discussed by these last two authors, 2 and concludes that "there seem good reasons to anticipate that comparable answers could be developed for the remainder of [the two authors]'s reservations" (p. 15).
In this paper, I would like to analyze some morphological rules of French which are in some ways analogous to the problematic rules discussed by Anderson (1986). Some of them will make more precise the concept of blocking; the others, unfortunately, still appear to constitute counter-examples.
* I would like here to heartily thank Igor Mel'~uk and Arnold Zwicky for many helpful comments on an earlier version of this paper. i According to Anderson (1986, p. 4), this revives a condition originally proposed by P~nini. 2 He actually refers to an unpublished paper read to the 1983 Annual Meeting of the Linguistic Society of America by Maria Dardis and Richard Janda with the same title, and apparently the same content.