- Hidden generalizations: Phonological opacity in optimality theory
This is the first volume in the series ‘Advances in optimality theory’, edited by Ellen Woolford and Armin Mester. On the one hand, it can be considered a very good start for the series to have one of the eminent scholars in optimality theory (OT) contributing the first volume. Furthermore, McCarthy promises to tackle one of OT’s big problems: phonological opacity. On the other hand, this could also be considered a particularly difficult start since at first glance, the proposal made in the book looks very much like OT filing for bankruptcy.
The main point of M’s proposal is to modify OT’s central tenet of parallelism by using parallel evaluation in a serial fashion. While acknowledging the scale of this move, M points out that the framework nevertheless has assets that make it superior to other conceptions of grammar like rule-based phonology: the advantages of the new approach include its elegant way of explaining conspiracies, markedness, typological variation, language change, and acquisition.
The book is very pedagogical. With the instructions of how to use the book at the end of Ch.1, M takes the reader by the hand and approaches the topic step by step. Besides this nice gesture, [End Page 969] Ch.1, ‘Overview of the issues and the results’, explains what phonological opacity is and why it is a problem for OT, and briefly outlines the solution to the problem that is proposed in Ch. 3.
In Ch. 2, ‘Opacity, derivations, and optimality theory’, M recapitulates the history of analyses of opacity: first, derivational theories as in SPE (Chomsky & Halle 1968) and lexical phonology, and then OT. Before going into the various proposals, he sketches the basics of the theory, that is, the concept of various levels of representation and of linking them via derivation. This outline serves as a preface to the discussion of derivational analyses. M further lays out the basic assumptions of ‘classic’ OT as a prelude to the discussion of approaches to opacity within OT, pointing out which aspects of opacity pose problems for parallelism (i.e. parallel evaluation of candidates).
M concludes in this overview that the crucial element shared by more successful approaches to opacity is the reference to a third representation besides the underlying and the surface forms. In serial rule-based phonology, these are the representations between the applications of rules; in lexical phonology and in stratal OT, which assume the three levels of evaluation, the output of one level serves as the input to the next. Other OT-internal proposals make reference to some failed candidate (sympathy theory, targeted constraints) or an abstract notion of a fully faithful candidate (comparative markedness). M criticizes those OT approaches without a third level of representation for losing sight of the trademark feature of opacity, which is that the opacified process has to be independently active in the language. This results in overgeneration.
In Ch. 3, ‘Candidate chains and phonological opacity’, Mintroduces his proposal, starting with candidate chains (CCs) and well-formedness conditions on them. He then establishes the new family of precedence constraints (Prec-constraints). I discuss this in more detail shortly. The introduction of the new theory is followed by a short discussion of learnability/acquisition. An important question is whether Prec-constraints are universal or whether a learner has to construct them when faced with opaque data. It would be instructive here to have a look at spontaneous opacity, as it emerges in child language. M does not discuss this issue, however, even though he wonders whether there are ‘special circumstances, invisible to learners, where even unlearnable and therefore unlearned Prec-constraints are visibly active’ (119; see Tihonova 2009 for an insightful analysis of spontaneous opacity in child language within OT-CC). At the end of the chapter M comes back to previous approaches and compares the new theory with the earlier attempts one by one.
The last chapter before the postscript, ‘Two case studies...