Lexical Phonology is a theory about the organization of grammar. In particular, it deals with the relationship among phonology, morphology, and the lexicon. Its basic claim is that all morphological processes, and many phonological ones, are carried out in the lexicon. On this view, phonological rules fall into two classes:
(1) Lexical Rules: which may interact with morphological rules.
(2) Postlexical Rules, which may not interact with morphological rules.
This organizing principle is expressed by placing all lexical phonological rules in the lexicon, and all postlexical rules in a separate phonological component that is ordered after the rules of syntax. By definition, any phonological rule that applies in constituents larger than words must be postlexical, since such constituents are created by the syntax.
III. Phonology and Morphology:
With regard to morphology, Lexical Phonology is a strong version of the so-called Lexicalist Hypothesis. It's because it assumes that all word formation, including inflection, is carried out in the lexicon. This position follows from the fact that lexical phonological rules may have to apply both to derived words and to inflected forms of words Given the assumption that morphology and part of phonology are carried out in the lexicon, we expect some interaction between morphological and phonological rules. The classic view (Booij 1981, Kiparsky 1982) is that morphology and phonology apply in tandem. After every word-formation rule, lexical phonological rules re-apply. The effect of this is tha lexical phonological rules apply cyclically. This mode of application is illustrated by the following abstract example, in which the phonologival effects are represented as prime signs:
(1) Cycle 1: Root (taken from the lexicon)
Cycle 2: [Root]Y] word formation rule 1
[Root]Y]¢ phonological rule A
[Root]Y] ¢¢ phonological rule B
Cycle3: [[Root]Y] ¢¢Z] word formation rule 2
[[Root]Y] ¢¢ Z] ¢ phonological rule A
[[Root]Y] ¢¢Z] ¢¢ phonological rule B
The interspersing of word-formation rules and cyclic phonological rules predicts that we may find word-formation rules whose application is conditioned by the prior application of a phonological rule. An example is the derivation of the Dutch word ambassadrice "female ambassador" from its neutral counterpart ambassadeur "ambassador", derived in turn from ambassade "embassy". The relevant word-formation rule replaces the suffix -eur with -rice when the stem ends in d or t. The derivation is as follows:
(3) Cycle 1 : [ambassad∂]
Cycle 2 : [[ambassad∂]eur] -eur suffixation
Cycle 3 : [[[ambassad] ]rice] -rice suffixation
The crucial point here is that the phonological rule of prevocalic schewa-deletion makes the stem ambassad end in an alveolar obstruent, as required for suffixation of -rice.
The classic version of Lexical Phonology, as outlined above, also implies that phonological rules may apply before morphological rules. This seems to be correct, since word formation rules may require information about the (rule-derived) stress patterns of their input words. For instance, the choice between the two denominal adjectival suffixes -ief and -isch in Dutch depends on whether the last syllable of the input word bears main stress: if so, -isch has to be selected-- otherwise, -ief.
A weaker claim about the interaction between phonology and morphology is that this interaction is only indirect. That is , all word-formation rules apply first, and the cyclic phonological rules than apply.
The common core of the "direct" and the "indirect" view of phonology/morphology interaction is that cycles are the organizing priciple for the application of lexical phonological rules. A further property of lexical rules is that their application is governed by the S[TRICT] C[YCLE] C[ONDITION], one of the most important principles of Lexical Phonology. The essence of the SCC is that feature-changing (applications of) cyclic phonological rules apply only in derived environments. The environment may be derived either morphologically or phonologically. In the former case, a new structure is derived from the operation of a morphological rule. In the latter, it results from a phonological rule. Consider /servis+e/, the Polish word serwis "service." In contrast to the first sequence /se/, the second /se/ in this form is derived morphologically: -e is the ending. Polish has a cyclic palatalization rule that palatalizes coronal consonants before front vowels. The SCC now correctly predicts that only the second /s/ palatalizes: [servi'se].
The SCC places a strong restriction on the abstractness of underlying forms. For example, we can exclude in a principled way the proposal that, in Polish, the underlying form of Chile [ˇc ¢ile] can be /kile/. This is because Velar Palatalization, the rule that effects the change k >ˇc , is cyclic , and hence cannot apply in non-derived environments. Consequently, an underlying /k/ of /kile/ would surface as a [k]. Therefore we must assume the underlying from /ˇc ile/. Thus the abstractness of underlying forms is severely restricted.
The SCC does not imply that cyclic rules cannot apply to non-derived environmnets. However, in such environment, they can only add information. For instance, since syllabification rules and stress rules only add prosodic information, they can apply on the first cycle.
Segmental rules can also apply on the first cycle, if they fill in certain feature specifications. Nasal Assimilation in English specifies that the underlyingly underspecified nasal consonant N has the same place of articulation as the following obstruent, as in damp, end, or thank. Note also that the structure-adding applications of rules do not create derived environments. Otherwise, all cyclic rules would incorrectly apply on the first cycle, because the syllabification rules already apply on that cycle.
Bright, William(ed). 1992. International Encyclopedia of Linguistics. New York:
Kaisse, Ellen M & P. A. Shaw. 1985. On the Theory of Lexical Phonology.
Phonology Year Book 2: 1-30.