The Derivational Residue in Phonological Optimality Theory (review)

The theme of the ten articles in this collection is the status of phonological derivation in optimality theory (OT). Although stated nowhere in the collection, it appears as though all of the papers (with the possible exception of Hayes’s) were presented in some form at the Tilburg University Conference on Derivational Residue in Phonology, 5–7 October 1995. In the introduction (1–27), the editors critically evaluate each of the contributions while at the same time dividing them into two main subsets: those that deal with the question of cyclicity and those that deal with phenomena that require rule-ordering in a derivational account.

Four of the papers in the collection deal with rule-ordering phenomena, albeit using different strategies. In ‘Head dependence in stress-epenthesis interaction’ (29–50), John Alderete proposes the constraint Head-Dep (epenthetic segments can’t be stressed) to account for the different ways in which languages treat epenthetic vowels with respect to stress assignment. By taking advantage of the OT concept of constraint ranking, Alderete accounts for partial metrical activity of epenthetic vowels (epenthetic vowels stressed only under certain conditions) in languages such as Yimas. Derivational theories, it is argued, can only account for such patterns by positing two rules of epenthesis, one that precedes stress assignment and one that follows it. Matthew Y. Chen’s ‘Directionality constraints in derivation?’ (105–27) attempts to account for tone sandhi phenomena in the Chinese dialect of Tianjin by proposing constraints that are ‘processual in character’ and extrinsically ordered. Whereas Chen attempts to make a derivational account workable within some form of OT, in ‘Unrecoverable origins’ (51–80) Mary Bradshaw uses tonal alternations in Suma to argue that no current incarnation of OT can account for counterfeeding rule ordering. Similar to Bradshaw, R. Ruth Roberts-Kohno finds in ‘Derivationalism in Kikamba vowel hiatus phenomena’ (269–94) that a multistratal approach to OT could account for the interaction of vowels and empty root nodes in Kikamba. Such an approach, however, would involve positing levels that are not independently justified, whereas a derivational approach can deal with the phenomenon through extrinsic rule-ordering and by positing empty root nodes (which, it is claimed, are independently justified).

Five papers in the collection deal with the question of cyclicity in optimality theory. Although three types of cyclic phenomena are pointed out in the introduction (17), the papers deal with only one type: the application of a rule to a morphological environment which is a substring of a word. San Duanmu’s ‘Alignment and the cycle are different’ (129–52) proposes a cyclic model of OT in order to deal with stress in Shanghai Chinese. René Kager, on the other hand, argues against cyclicity in ‘Surface opacity of metrical structure in Optimality Theory’ (207–45) by proposing in its place paradigm uniformity effects which are achieved via base-output correspondence (a morphologically more complex word maintains identity with its less complex ‘base’). Evidence for this proposal is offered through analysis of stress (and its opaque interaction with epenthesis and syncope) in several dialects of Arabic. The interaction of vowel length and stress in Kashaya is the topic of ‘Uniformity in extended paradigms’ (81–104) by Eugene Buckley, who argues in favor of constraint domains and paradigmatic effects defined under certain circumstances by a suffixed form. In ‘Phonological restructuring in Yidiɲ and its theoretical consequences’ (175–205), Bruce Hayes suggests that phonological systems which have undergone severe restructuring (presumably as the result of e.g. pidginization, creolization, or language loss) should be more closely examined. The thrust of Hayes’s proposal is that when children are faced with a contradictory set of surface data, as apparently were some of the Yidiɲ informants that Hayes’s analysis is based on, they develop ‘anticorrespondence’ constraints, constraints...