Since the advent of Optimality Theory (OT) nearly twenty years ago, the field of phonology has witnessed a general shift in focus away from issues related to the architecture of the grammar and about the nature of representations to questions about how to accommodate complex phonological interactions in an inherently non-derivational paradigm and the role of functional, psycholinguistic, and phonetic factors in phonology. Although the paradigm shift has both directly and indirectly led to many important advances in our understanding of phonology and its relationship with other components of grammar, it also had the undesirable effect of fostering neglect of issues that belong to a fruitful research program which has advanced the field for over forty years. The present volume, edited by Eric Raimy and Charles E. Cairns and aptly titled Contemporary views on architecture and representations in phonology, serves the useful function of returning our attention to these questions of architecture and representations, many of which were quietly ‘swept under the rug’ without adequate resolution, despite being not directly related to the debate about the efficacy of OT as compared to its derivational predecessors. In fact, one of the virtues of this book is its ability to extricate itself almost completely from the OT versus derivations ‘quagmire’ while contributing to our understanding of a diverse array of phenomena, including features, syllable structure, metrical stress theory, and reduplication, and providing overarching and topical discussion of the role of substantive biases in synchronic phonology and the modular nature of phonology.

The contributions in this volume are based on conference presentations at the 2004 City University of New York (CUNY) Phonology Symposium on Architecture and Representation in Phonology. Following an introductory chapter by the editors (‘Architecture and representations in phonology’), the book is divided into five sections, each consisting of a set of thematically related chapters, many of which are written in direct response to another chapter within the same section. The first three sections of the book grapple with three different phenomena: ‘Phonological features’, ‘The syllable’, and ‘Metrical structure’, while the last two sections deal with broad issues related to the ‘Architecture’ of the grammar and the proper formalization of substantive biases in phonological theory (‘Interactions’).

Chapter 2 of the volume, ‘The role of features in phonological inventories’, written by G. N. Clements, attempts to predict typological generalizations about phoneme inventories on the basis of a traditional (The Sound Patterns of English-style) set of articulatory-based phonological features. Clements's feature-mediated approach stands in contrast to other accounts that attribute cross-linguistic patterns directly to phonetic notions of articulatory ease and perceptual distinctness (e.g. Lindblom & Maddieson Reference Lindblom, Maddieson, Hyman and Li1998). Drawing on the UPSID (UCLA Phonological Segment Inventory Database, Maddieson & Precoda Reference Maddieson and Precoda1989), Clements shows that distributional skewings in phoneme inventories reflect the interaction between five principles: feature bounding, feature economy, marked feature avoidance, phonological enhancement, and robustness. The first of these principles constrains the number of contrasts a language may employ by projecting those contrasts through a discrete set of phonological features. The other four principles work, in some cases antagonistically to one another, to further limit the set of contrasts adopted by a language. A virtue of Clements's theory is that its reliance on a discrete and relatively small set of features potentially makes its predictive power easier to assess, relative to an account that appeals to more finely graded phonetic continua, which can be notoriously difficult to quantify, particularly in the articulatory domain. Ultimately, though, the core principle that guides more directly phonetically-driven approaches to segmental typology, namely the competition between minimizing articulatory effort and maximizing perceptual distinctiveness, surfaces in Clements's account in the form of competition between the principle of economy, which constrains the number of feature contrasts exploited by an inventory, and the notion of robustness, which acts to maximize the use of the articulatory and acoustic space.

Two shorter chapters by Bert Vaux (‘The role of features in a symbolic theory of phonology’) and Morris Halle (‘Two comments on “The role of features in phonological inventories”’) respond to Clements's contribution by voicing their support for his featural account of inventory construction. Vaux points out some concerns about the primary data from UPSID on which Clements's work is based and about UPSID's interpretation of the data. Such problems are inevitable, however, in a large database like UPSID and likely to be insignificant given the sheer magnitude of cases included in the survey. Halle, who is also overwhelmingly positive in his assessment of Clements's thesis, takes issue only with the appeal to a feature of [voice] rather than one which refers to the stiffness of the vocal folds (Halle & Stevens Reference Halle and Stevens1971).

The second section of the book focuses on the proper representation of syllables. This is a traditional subject of contention in generative phonology, although less so in recent years due to the widespread adoption of flat-structured moraic models of the syllable (Hayes Reference Hayes1989) and in light of evidence suggesting that the range of phenomena probative in diagnosing syllables is more circumscribed than once thought (Steriade Reference Steriade1999). In a lucid and thoughtful chapter (‘The appendix’), which gives an excellent overview of the state of the art of syllable theory, Bert Vaux & Andrew Wolfe present compelling (phonetic, phonological, morphological, and psycholinguistic) evidence in favor of an approach that assumes syllable appendices linked to constituents above the syllable, where this suprasyllabic constituent varies across languages and even potentially within languages. As the authors discuss, it is most difficult to tease apart the merits of an analysis that assumes stray unparsed segments and one that posits a direct linkage between extrasyllabic elements and a higher prosodic constituent. A crucial piece of evidence they bring to bear on this issue is the observation that even extrasyllabic elements are often restricted (e.g. to /s/ in many languages) and that this restriction should follow from a licensing constraint mediated by a prosodic constituent, parallel to constraints operative over core syllables.

In response to Vaux & Wolfe's article, Eric Raimy (‘A case of appendicitis’) presents evidence in favor of a more flexible theory of the syllable, allowing for unparsed segments, (e.g. in Nxaʔamxcín) and even degenerate syllables (e.g. in Thao). Charles E. Cairns argues for yet another possibility in ‘Phonological representations and the Vaux–Wolfe proposal’: a 3-D theory of phonological representations (Halle & Vergnaud Reference Halle and Vergnaud1980) that does not rely on the Prosodic Hierarchy above the syllable. A feature of Cairns's proposal is that it expands the richness of onset clusters cross-linguistically by assigning many segments that other models would relegate to an appendix to the onset itself. In ‘Does sonority have a phonetic basis?’, G. N. Clements builds further on Vaux & Wolfe's proposals by suggesting that syllabification is ultimately governed by phonetic sonority, in particular, by ‘perceived resonance’. Indeed, the quantification of sonority has been the subject of extensive research, most recently by Parker (Reference Parker2002, Reference Parker2008), who shows that an intensity-based linear function can predict most features of standard sonority scales.

The third section of the book is devoted to metrical structure. The lead article by William J. Idsardi, ‘Calculating metrical structure’, presents a revised version of his simplified bracketed grid theory (SBG), applies the updated theory to a few case studies, and shows how it can be computationally implemented using finite state automata (FSA). The changes introduced to the theory have the virtue of reducing metrical operations to two basic types: bottom–up projection of grid lines and grouping of grid marks into constituents. One of the interesting features of SBG is its directional parameterization of bracket assignment, which allows for the distinction of left and right brackets in addition to no brackets. This directionality parameter opens up the possibility of distinguishing different types of feet, which could potentially be a useful device for accounting for complex foot structures and instances of metrical incoherence between different phenomena (Vaysman Reference Vaysman2009). As Idsardi shows, building on work by Karttunen (Reference Karttunen, Karttunen and Oflazer1998), the straightforward computational implementation in the form of FSAs is a further selling point for his theory vis-à-vis standard OT implementations of metrical stress theory that allow the counting of violations.

In ‘Long-distance dependencies and other formal issues in phonology’, Charles Reiss adopts a less sanguine view of the prospects of translating rule-based analyses into FSAs, pointing out that long-distance dependencies and natural classes are not readily implemented in an FSA. He also suggests a slight revision to Idsardi's metrical theory: replacing directional foot marks with non-directional boundary marks. As Reiss observes, this move leads to a more parsimonious inventory of representations. What is unclear from either Idsardi's or Reiss' chapter is the degree to which a theory with directional foot marks offers empirical coverage not provided by an analysis with non-directional boundaries.

B. Elan Dresher's chapter employs SBG to provide an analysis of ‘Stress assignment in Tiberian Hebrew’. The Tiberian Hebrew stress system features various processes that suggest different footing algorithms as well as contextually governed variation in stress patterns. As a detailed case study, Dresher's chapter, together with the chapter by Halle & Nevins on Slavic (on which see below), brings an additional dimension of interest to the volume, which otherwise is dominated by discussion of broad theoretical issues supported by shorter data presentations.

The fourth section of the book centers on an article by Andrea Calabrese (‘Markedness theory versus phonological idiosyncracies in a realistic model of language’), in which he proposes a model of phonology designed to reconcile the conflict between substantive notions of naturalness, which offer an explanation for many observed patterns, and unnatural language-specific rules that inevitably creep into the grammar but defy phonetic explanation. The source of naturalness in Calabrese's theory is the ‘markedness module’, a component of Universal Grammar that consists of a series of filters and markedness statements triggering repair operations in the course of the derivation. Crucially, the markedness module contains not only information about markedness but may also contain unnatural rules on a language-specific basis, such as the raising of /o/ before word-final voiced consonants other than nasals in Polish. Calabrese gives a detailed account of his theory, which is built on a lengthy research program extending back to his dissertation (Calabrese Reference Calabrese1988), in which he introduced the notion of filters. After exemplifying his theory through case studies of French, Bulgarian, and Icelandic, Calabrese conjectures on the nature of historical change.

In a short commentary (‘Comments on diachrony in Calabrese's “Markedness theory versus phonological idiosyncracies in a realistic model of language”’), Ellen M. Kaisse expresses reservations about Calabrese's suggestion that phonological change spreads on a word-by-word basis. She provides a concise but clear summary of the differences between lexical diffusion and the regular Neogrammarian mechanisms of sound change, arguing that diachronic shifts primarily belong to the latter category and characteristically begin as gradient phonetic shifts before being regularized as discrete phonological rules. In ‘Nuancing markedness: A place for contrast’, Keren Rice expounds on Calabrese's incorporation of markedness in his theory, showing that, crucially, markedness is sensitive to contrasts. Only when contrasts are at stake in a particular language do universal statements of markedness come into play. For example, the output of place neutralization in final stops is predominantly coronal, as a theory positing coronal as the unmarked place would predict, but it may also be labial or velar. However, the markedness of coronal place becomes more apparent in place assimilation processes which occur in languages with contrasts in preconsonantal place of articulation. The lesson gleaned from Rice's discussion is that ‘idiosyncracy … yields to systematicity when the critical role of contrast is understood’ (318).

The first two chapters in the final section of the book focus on the role of phonetic factors in a synchronic grammar. Juliette Blevins's contribution, ‘Phonetic explanations for recurrent sound patterns: Diachronic or synchronic?’, is an excellent introduction to the topic. She presents evidence suggesting that phonetic explanations in many cases are more compelling when viewed as historical changes rather than synchronic processes. Her conclusion is that there is little to be gained by redundantly encoding in the grammar phonetic explanations of questionable synchronic relevance, if they must be independently appealed to in explaining language change. In ‘Phonetic influence on phonological operations’, Thomas Purnell also adopts the stance that phonetics and phonology are clearly separate modules of language, arguing that formalizing phonetic knowledge in the synchronic grammar is problematic because of the variability and gradience of phonetic data, the indeterminateness of phonetically-informed predictions, and the availability, in many cases, of alternative, more traditional phonological accounts. A positive message that emerges from Purnell's chapter is its appeal for more experimental evidence which directly tests the psychological reality of phonetic knowledge.

The chapter by Morris Halle & Andrew Nevins, ‘Rule application in phonology’, defends a traditional derivational model of phonology, featuring ordered rules and a distinction between cyclic and postcyclic rules. Their defense of this traditional model rests on a presentation of complex and often opaque rule interactions in Slavic, which would be extremely difficult to handle in a one-step constraint-based framework. The last paragraph of their chapter gets to the crux of the perennial debate about abstraction in phonology, observing that analyses excluding abstract yer vowels on the basis of their questionable psychological plausibility are ‘forced to employ considerably more complex and less perspicuous solutions than the handful of rules’ that Halle & Nevins posit (378).

The final chapter in the book by Eric Raimy, ‘Deriving reduplicative templates in a modular fashion’, is a cogent defense of his Precedence-Based Phonology model (PBP) of reduplication (Raimy Reference Raimy2000) against criticisms leveled by proponents of Prosodic Morphology (PM) theories of reduplication. Raimy argues that various independently motivated biases (or, in Raimy's terms, ‘organic constraints’) in the domains of cognition, acquisition, and grammatical architecture not only constrain overgeneration of patterns in his PBP account of reduplication but, in some cases, are also assumed by the PM model (albeit in different guises). For example, many unattested but logically possible reduplication patterns requiring reference to three predicates are precluded by a cognitive bias or constraint in favor of a binary ceiling on counting, which is not unique to reduplication. This computational constraint plays an important, if indirect, role in the PM model as well, by ensuring that the prosodic categories referenced by the PM model are maximally binary. Raimy adopts the sensible view that ‘U[niversal]G[rammar]'s role is to constrain the hypothesis space within which statistical learning occurs’ (396). His use of basic organic principles has a general appeal that extends beyond his reduplication framework. For example, in work on syllable weight (Gordon Reference Gordon, Hayes, Kirchner and Steriade2004), I suggest that weight distinctions are constrained by a cognitively imposed bias against representations referencing more than two phonological predicates. Similarly, the criticism that a direct-phonetics model of OT overgenerates (see Zhang Reference Zhang2002) is mitigated by the fact that phonetic differences must be of a certain magnitude to be perceptible and thereby accessible to the grammar.

All in all, this book is a treasure trove of treatises on a diverse array of topics approached from a variety of angles. The contributions are thought-provoking and raise important questions about the nature of phonology and its interaction with other components of the grammar, mechanisms of language change, and basic physical and cognitive forces external to the grammar. Although the viewpoints expressed in the book are often incongruent, this dissonance is perhaps the greatest virtue of the book since it provides ample fodder for further research. As the editors point out, ‘lack of unanimity is positive because of the question that it raises. Understanding progresses when the field comes up with better questions’ (15).