Barbosa, P, D. Fox, P. Hagstrom, M. McGinnis and D. Pesetsky (eds) (1998) Is the Best Good Enough. MIT Press and MITWPL: Cambridge MA.
Reviewed by Melissa Svendsen, University of Victoria.
Synopsis This volume contains sixteen articles generally focused on the application of Optimality Theory to syntax. Optimality Theory originated in phonology, with Prince and Smolensky's "Optimality Theory: Constraint Interaction in Generative Grammar" (1993), and has subsequently been extended to morphology, syntax and other areas of generative linguistics. The central premise of Optimality Theory is that universal grammar consists of a set of violable constraints, and that the grammars of individual languages consist of different rankings of those constraints. The articles in this volume explore the implications of this model for syntax and related subjects. Most, though not all, of the articles assume the basic tenets of Optimality Theory, which are that constraints are ranked and violable.
In "Whot?" Peter Ackema and Ad Neeleman give an Optimality Theoretic account of variation in wh-questions in English, Czech and Chinese, using constraints requiring that question features and scope be marked and that elements not move. In English, the constraint that question features must be marked outranks the constraint against movement, which in turn outranks the constraint requiring that scope be marked; in Czech, the constraints requiring that question features and scope be marked both rank above the constraint against movement; and in Chinese the constraint against movement is ranked above the other two.
In "Optimality and Inversion in Spanish," Eric Bakovic accounts for variation among dialects of Spanish with respect to subject/verb inversion in matrix and subordinate wh-clauses. He does this by interleaving two constraints -- one requiring that syntactic operators c-command their extended projections and another against traces -- with a universal subhierarchy of constraints that favour syntactic operators being in specifier position. The constraints in this subhierarchy are ranked according to the degree to which the operator in question is 'argumental'.
In "Morphology Completes with Syntax: Explaining Typological Variation in Weak Crossover Effects," Joan Bresnan, working within the framework of Lexical-Functional Grammar, suggests that cross-linguistic variation in the 'weak crossover effects' results from competition between constraints on parallel structures. In her analysis, a constraint on linear order operates on categorial structure (roughly a phase marker), and a constraint on syntactic rank operates on f-structure (roughly a predicate structure). Since the categorial and predicate structures of a sentence may not be exactly parallel, the constraints operating on them may conflict.
In "Anaphora and Soft Constraints," Luigi Burzio argues that three interacting hierarchies of violable constraints account for the distribution of anaphors better than does standard Binding Theory. Like Bakovic, Burzio argues for the interleaving of constraint hierarchies which themselves have fixed internal structure. In other words, any constraint from Hierarchy A may outrank any constraint from Hierarchy B, but Constraint 1 from Hierarchy A always outranks Constraint 2 of Hierarchy A.
In "Some Observations on Economy in Generative Grammar," Noam Chomsky gives a thumbnail sketch of "considerations of economy" in generative syntax as a whole, and in the Minimalist Program in particular.
In "Locality in Variable Binding," Danny Fox argues that if two potential binders of a given variable yield the same interpretation, the existence of the local binder blocks the binding of that variable by the more distant binder. He goes on to suggest that while some inviolable conditions, such as c-command as a prerequisite for binding, can exclude representations from the candidate set, others, such as Binding Conditions and Parallelism, apply to the output of the Optimality Theoretic selection. Thus he seems to lean toward confining Optimality Theoretic processes to certain components of the grammar.
In "Optimality Theory and Human Sentence Processing," Edward Gibson and Kevin Broihier argue that the 'winner-takes-all' approach of standard Optimality Theory (in which the violation of any number of lower ranked constraints is preferable to the violation of just one highly ranked constraint) does not account for the facts of sentence parsing. Instead, they argue for a 'cumulative constraint weighting system', in which the violation of a highly ranked constraint may be preferred to multiple violations of lower ranked constraints.
In "Optimal Subjects and Subject Universals" Jane Grimshaw and Vieri Samek-Lodovici account for pro-drop (the absence of a pronominal) and subject-verb inversion in Italian by proposing that Italian has a high ranking for two constraints -- one against overtly realizing an argument that is coreferent with the topic and another requiring that the left edge of a focused constituent be aligned with the right edge of a maximal projection.
In "Semantic and Pragmatic Context-Dependence: The Case of Reciprocals" Yookyang Kim and Stanley Peters argue that there are eight possible interpretations of reciprocal elements (each other and one another), which form a candidate set from which the grammar selects the strongest one that is consistent with the context.
In "When is Less More?" Geraldine Legendre, Paul Smolensky, and Colin Wilson give an Optimality Theoretic formalization of the Economy Principle 'Shortest Move' ('Shortest Link' in their system) using the constraint MinLink, which is itself a universal subhierarchy of constraints against a chain link crossing a maximal projection which is not L-marked. Using MinLink and a small set of additional constraints (most of which are independently motivated), they account for cross-linguistic variation in a variety of extraction phenomena, such as wh-island and superiority effects.
In "Reference Set, Minimal Link Condition and Parameterization," Masanori Nakamura argues that wh-extraction in Tagalog involves competition among forms differing in both syntax and morphology, with the construction with the shortest wh-chain blocking competing constructions involving longer wh-chains. (This paper is solidly in the Minimalist Program and, like Chomsky's contribution, can at most be said to have an Optimality Theoretic 'flavour'.)
In "On the Nature of Inputs and Outputs: A Case Study of Negation," Mark Newson accounts for differences in the syntax of negation in English and Hungarian. He suggests that in English a constraint requiring that all elements be licenced is highly ranked, while in Hungarian a constraint requiring that all heads be overt is highly ranked.
In "Some Optimality Principles of Sentence Pronunciation," David Pesetsky argues that the laws governing movement are separate from the laws governing the pronunciation of moved elements, and that the latter are best characterized as Optimality Theoretic constraints.
In "Constraints on Local Economy," Geoffrey Poole, working within the Minimalist Program, argues that the Minimalist principle 'Procrastinate' can not account for economy in derivations. Instead, he proposes that the relevant principle is one whereby an element moves or is inserted into a position in which all its formal features are checked.
In "The Logical Problem of Language Acquisition in Optimality Theory," Douglas Pulleyblank and William J. Yurkel show that previously proposed learning mechanisms can not adequately describe how language learners acquire constraint rankings. They then propose a 'genetic algorithm,' in which the 'fittest' constraint sequences recombine to form new sequences, which are then evaluated with respect to incoming data, yielding a new determination of which constraints are 'fittest'.
In "Error-Driven Learning in Optimality Theory via the efficient Computation of Optimal Forms," Bruce Tesar proposes that every example of an optimal form encountered by a learner contains implicit information about non-optimal forms. If the learning mechanism is currently entertaining a hypothetical constraint ranking, then it can compare the sub-optimal (not-encountered) form that result from that constraint ranking with the optimal (encountered) form, and adjust the hypothetical constraint ranking accordingly.
Critical Evaluation The excellent introduction gives a precise and elegant explanation of the contrast between what the author's call the 'Standard Scenario' -- the dominant paradigm in generative linguistics until very recently -- and what they call the 'Optimality Scenario,' in which rules and constraints are ranked and violable. This is followed be a brief survey of historical antecedents of Optimality Theory within Generative Linguistics, and then by an overview of the articles in the volume.
The overview, which highlights similarities and differences among the approaches taken by the various contributors, is very important to seeing the volume as an integrated whole. It is also a handy reference source, especially since most of the articles lack abstracts, and some of them also lack coherent summaries.
The quality of the papers themselves in uneven, as is the editing. Many present ideas that are both interesting and original, but in ways that are sometimes difficult for the reader to follow. In some, individual constraints are used without adequate introduction. In others, there is a paucity of data, as if contributors get carried away by the theoretical interest of their work, and lose sight of the empirical basis of the discussion.
Simply keeping track of the constraints can be a challenge for the reader. T.S. Eliot said that "the naming of cats is a difficult matter" and the same may be said about the naming of constraints. In Optimality Theory(or at least in the subset of work on syntax with which I am familiar) constraints seem to be named without a great deal of thought for the reader who must endeavour to keep them all straight. Particularly problematic is the fact that sometimes what is essentially the same constraint turns up under several different names. Occasionally this reflects a significant theoretical difference (as in the difference between Grimshaw's Stay and Legendre et al's *t, which state that an element does not move and that traces are disallowed) but this is not always the case. (An example is Newson's Move, which appears to be a combination of Stay and *t). In an anthology of this kind, the contributors might have been encouraged to aim for greater consistency in the naming of constraints.
Where this is not desirable for theoretical reasons, a concordance of constraints would be useful, particularly since one of the main goals of Optimality Theory is to posit a set of relatively simple universal constraints which account for a wide variety of complex phenomena. Although the field of Optimality Theory and syntax is still very young, it is not too soon to start keeping track of which constraints account for more than one phenomena, particularly since Optimality Theory must be defended against the charge that its practitioners simply cut constraints out of whole cloth as the need arises. (While Optimality Theory is very much established in phonology, in syntax it is still very much the new kid on the block.)
This book is part of an ongoing dialogue about syntax and Optimality Theory, and as such it is not self-contained. Contributors build upon or refine analysis presented in previous papers (many of which are available on the Rutgers Optimality Archive). The subject matter ranges widely, both in terms of the types of data considered and in terms of the theories within which they are discussed. Many papers assume considerable prior knowledge on the part of the reader. Readers will find it helpful to keep on hand a linguistic dictionary or a general linguistics text with a good glossary.
Nonetheless, any reader with a serious interest in the application of Optimality Theory outside of phonology will find reading this book very useful, and anyone doing research involving Optimality Theoretic approaches to syntax will find reading it a necessity. This book is a valuable contribution to the field.
Bibliography Prince, Alan S., and Paul Smolensky. "Optimality Theory: Constraint Interaction in Generative Grammar," RuCCs Technical Report #2, Rutgers University Center for Cognitive Science, Piscataway, N.J. (To appear; Cambridge MA. MIT Press)
Melissa Svendsen is a graduate student at the University of Victoria. She is currently writing her master's thesis on an Optimality Theoretic account of cross-linguistic variation in wh-movement, focusing in particular on optional wh-movement in Babine Witsuwit'en (an Athabaskan language).
|