One of the fundamental observations of linguistic theory is that linguistic elements do not interact over arbitrarily great distances. Instead, relations obtain locally. This conviction would seem trivial if all phonological interaction were obviously confined to adjacent segments, or all syntactic relations to adjacent words. In reality, the assumption of locality requires work for those assuming it; in phonology, it has for this reason been one of the important forces driving the elaboration of metrical and autosegmental representations.
The first goal of this paper is to present an argument for local spreading that is strict in two ways. First, spreading respects strict segmental adjacency. An essential result of this view is that segments are either blockers or participants in spreading; there is no transparency or skipping. Second, segmentally strict locality is inviolable; in Optimality Theoretic terms, Gen does not produce structures that disobey it. As we will see, these assumptions are fundamental to an account of certain asymmetries in long-distance feature spreading. In particular, strict locality (along with other basic assumptions) explains why vocalic place features spread long-distance, while consonantal place features do not. More generally, strict locality is a likely key to resolving a basic problem: some features spread long-distance and others do not, a fact that follows from nothing in the theory otherwise.