Lexical Phonology and the History of English (review)

Despite the current optimism about purely constraint-based theories, rule-based phonology lives on. This book is an example, but it is not an argument against pure constraint theories. Rather, McMahon for the most part assumes a version of lexical phonology and asks what light historical change and dialectal differences can throw on that theory.

The data are from English, primarily American and British dialects; Scottish and nonrhotic dialects are singled out for the light they shed on theoretical issues.

Ch. 1, ‘The role of history’, is about history in two senses: what diachronic developments can tell us about synchronic theory (M feels diachronic evidence has been undervalued) and the history of generative phonology. The history of lexical phonology is further explored in Ch. 2, ‘Constraining the model: Current controversies in lexical phonology’. As one might guess from that title, there are no arguments for a derivational theory as opposed to a constraint-based theory, nor in favor of lexical phonology over other rule-based approaches. Rather, the argumentation concerns analyses within lexical phonology and, within that theory, the role of constraints in blocking excessive abstractness, in particular, the elsewhere condition, the strict cyclicity condition, structure preservation, and the derived environment constraint. (This last constraint blocks stratum one rules from applying in nonderived environments.) Analyses by Morris Halle and K. P. Mohanan serve as foils. But M reintroduces a certain kind of abstractness by claiming that certain bracketing configurations can block phonological rules. It is not clear how well her claims would translate into a tree-based view of the same structures.

Ch. 3, ‘Applying the constraints: The Modern English vowel shift’, illustrates M’s claims with a concrete analysis. The derived environment constraint prevents many abstract analyses, but there is a price: ‘The strict imposition of constraints inevitably prohibits a maximally simple phonology, if simplicity means a minimal number of rules used maximally … the optimal phonology will no longer be the one with the most simple and elegant analysis’ (138). M is not afraid to split vowel shift into two synchronic rules, for tense and lax vowels respectively.

Ch. 4, ‘Synchrony, diachrony and lexical phonology: The Scottish vowel length rule’, is a detailed historical analysis of how Scottish dialects of English underwent rule inversion, winding up with rather different underlying forms from other English dialects. Certain neutralization rules turn out to be exceptions to a pure version of the derived environment constraint.

Ch. 5, ‘Dialect differentiation in lexical phonology: The unwelcome effects of underspecification’, underscores the fact that different dialects may have strikingly different underlying representations of lexemes and different inventories of phonemes—a point which might seem obvious but for the fact that some phonologists seem to have implicitly assumed otherwise. Likewise for the oft implicit ‘preference for computation over storage’ (218), in the form of underspecified underlying representations; M argues that underspecification allows sidestepping otherwise motivated constraints and should therefore be disallowed.

The final chapter, ‘English /r/’, examines several nonrhotic dialects. M argues for an r-insertion analysis (as opposed to a deletion analysis or both insertion and deletion). Because the process was originally one of deletion (rule inversion, again), the environment after which r-insertion takes place is now an arbitrary set of vowels. This is an important point, deserving greater discussion than M gives it (and than I can give it here). For if phonological rules can have synchronically arbitrary sets of segments in their environments, then the argument used by early generative phonologists against the curly braces of their structuralist predecessors is lost. Moreover, the segment which is inserted is itself arbitrary (/r/ is not an unmarked vowel or consonant of English), again as a result of rule inversion, again undercutting that mode of argumentation which assumes a phonological analysis must be maximally natural. M quotes Wolfgang Wurzel as saying, ‘given a purely synchronic approach, many grammatical facts defy explanation which, from a historical point of view, are quite explainable’ (283). This...