The use of traditional trees need notbe proscribed altogether. However, wemust understand thattherelationbetween
Det andtheis not the same as that between NP and Det. NPdominatesDet—a syntactic relation; but Detcorrespondsto
morphophonologicalthe—an interface relation. Accordingly, when it is convenientto use traditional trees, I'llemploya
dotted line for the interface relations, as in (22a). This will abbreviate the more proper notation in (22b).
Where do traditional phonological derivations (àlaChomsky and Halle 1968)fit into this picture? Their input
(“phonological underlying form”) is given by the interface with syntactic structure, here the sequence of
morphophonological elements. The output of phonological derivations is phonetic form, the linguistic structure that
interfaces with speech perception and speech production. This consists of the syllabic,metrical, and intonational tiers.
As seen in (13), these two do not entirely match in bracketing or segmental content (and the disparity is much greater
when we get away from simple English examples). It is this gap (or interface) that traditional phonological derivations
fill, converting the morphophonological structure step by step into the phonetic form.
Optimality Theory provides an alternative approach to this interface, with the morphophonological sequence again
serving as input and theother tiersas output. The difference betweenthetwoapproaches (and itis radical in practice!)
is that traditional phonology bridges the interface with a sequence of derivational rules, whereas Optimality Theory
does itwitha setof simultaneouslyappliedviolableconstraints. (SeeJackendoff1997a, section 6.2 for somewhat more
extended remarks on the role of traditional phonological derivations.)