Most, perhaps all, formal linguists subscribe to the view that the mind is modular and that the human faculty of language is one such module: a cognitive unit that is domain-specific (it can only deal with linguistic material), encapsulated (it is insensitive to what happens elsewhere in the brain: grammaticality judgements do not change when we are depressed) and obligatory (a person cannot switch it off at will, e.g. one cannot decide not to understand what another person is saying).
Modularity is so uncontroversial that Cairns & Raimy do not mention any references arguing in favour of it in their introduction to Contemporary views on architecture and representations in phonology (henceforth Contemporary views). They merely stipulate (p. 1) that ‘the modular approach seems natural simply because phonology is a component of human cognition, ultimately a biological object; all biological entities more complex than viruses are arguably best understood in a modular framework’. Furthermore, they give a methodological argument for modularity: it ‘enables us to break the larger questions of phonology into smaller ones’. The authors thus deduce the necessity of modularity within phonology from the plausibility of modularity in biological systems; this is a conclusion which some might find too hasty.
Cairns & Raimy present their book in this light. The chapters in the volume, which were presented in an earlier form at the City University of New York (CUNY) Phonology Forum Symposium on Architecture and Representation in Phonology in February 2004, can be seen as contributions to an ongoing discussion on the question of the nature of the modular organisation of phonology: what modules are there, and how do they interact? Cairns & Raimy themselves give a few interesting examples of the success of modular thinking.
One of these (discussed also in an independent chapter by Cairns in Contemporary views) is the idea that metrical structure and syllable structure are present on different planes of phonological representation rather than being integrated into one prosodic hierarchy. Cairns adopts the principle of Syllable Integrity (Cairns & Raimy refer to Everett Reference Everett1998, although the concept can already be found in e.g. Prince Reference Prince1980 and Kager Reference Kager1993).
(1)
Cairns & Raimy argue that this principle is too strong, since a language like Southern Paiute violates Syllable Integrity. They therefore prefer an approach in which feet are built independently of syllable structure. The extent to which this architecture is truly ‘modular’ is not entirely clear, however. There are languages which respect Syllable Integrity (its original motivation), therefore there must be a place in the theory where it is operative, and where both metrical structure and syllable structure are visible. A well-known problem for modular theories is: what is the status of this interface? Is it yet another module? And how much is the modular set-up of a theory weakened by allowing such interface devices?
A lack of explicitness about what exactly constitutes a module, or how we distinguish between the interaction between two modules or their integration – in other words: how do we decide whether we have two modules which have somewhat similar internal structures and a strong interface between them or whether we are dealing with one integrated module? – makes the book less of an integrated whole than the reader would expect after having read the introduction. Most of the papers are very interesting and offer important contributions to several debates within phonology in their own right, but they do not shed a great deal of light on the modular architecture of phonology. It is not my purpose in this review to give a summary or an evaluation of every individual paper; what follows should therefore not be read as reviews of the separate papers, as the authors do not always intend to discuss modularity (and may not have been aware that this was an organising principle laid out in the introduction). Nonetheless, I believe it is very useful to take this opportunity to explore what we can learn about the current role of modularity in phonological theory from the individual chapters.
One example of how modular organisation might be more fully engaged can be made by considering the excellent article by the late G. N. Clements, ‘The role of features in phonological inventories’. In this chapter, Clements provides a clear and elegant summary of the ideas that he developed in the last years of his life on the structure of phonological inventories. Clements describes a number of such principles, such as Feature Economy (‘the tendency to maximise feature combinations’), Markedness, Robustness and Phonological Enhancement, and he shows how these principles make statistical predictions on the structure of inventories, which he verifies on the basis of the UPSID database (Maddieson & Precoda Reference Maddieson and Precoda1990).
The volume contains two responses to Clements' chapter, by Morris Halle and by Bert Vaux, pointing out problems both with some individual features and with the empirical foundations of the UPSID database. However, it seems to me that Clements' main findings are uncontested by these commentators – while there are many problems with the database, these do not seem to lead to a bias in the direction of Clements' findings, which therefore still stand.
Clements' article is thus a useful and research-provoking contribution. But it does not really attempt to shed light on the issue of modularity. On the contrary, it is not entirely clear where principles like Feature Economy are be located, given that they seem to fit better within a structuralist conception of grammar than in a generative one – they are not principles describing the derivation of individual forms, but rather laws describing tendencies within overall systems. Furthermore, principles of Markedness, Robustness (‘there is a universal hierarchy of features such that, in constituting their inventories, languages draw upon higher-ranked features in the hierarchy before drawing upon lower-ranked features’; pp. 42–43) and Phonological Enhancement all refer to principles which are probably grounded in phonetics. The way in which such grounding works is of course an important topic for anybody who is interested in modularity; but it is not answered in Clements' chapter, or in his other contribution to this volume, a response to the chapter by Vaux & Wolfe.
This particular question, however, is taken up in two other papers in this volume, ‘Phonetic explanations for recurrent sound patterns: diachronic or synchronic?’, by Juliette Blevins, and ‘Phonetic influence on phonological operations’, by Thomas Purnell (both of which are situated in the section of the book titled ‘Interactions’). Both authors argue in favour of a modular view – without actually mentioning the concept – of the relation between phonology and phonetics: the two have to be teased apart, and approaches such as that of the authors in Hayes et al. (Reference Hayes, Kirchner and Steriade2004) are therefore argued to be on the wrong track. Blevins and Purnell, although disagreeing on many issues, both point out that there is some evidence for ‘phonetic knowledge’, which implies that phonetics also has some place in the modular architecture of the mind.
Andrea Calabrese, on the other hand, in ‘Markedness theory versus phonological idiosyncrasies in a realistic model of language’, sketches a very explicitly modular architecture of phonology (the most fully worked out architectural proposal in the volume), in which there is place for a ‘markedness module’, certain components of which ‘represent the sensorimotor system in the linguistic computational system’. This module, therefore, constitutes a direct interface between the linguistic system and the sensorimotor system. (Calabrese's paper is followed by two interesting responses, by Ellen M. Kaisse and by Keren Rice.)
Contemporary views also contains a chapter in which Raimy explains how a view of a phonology which is internally organised in terms of submodules can account for reduplication patterns, when viewed in interaction with other modules, e.g. morphology and language acquisition; this chapter presents perhaps the most elegant argument in favour of modularity in the volume.
In terms of explicitly examining the role of modularity in the grammar, however, many other paper address this issue much less directly. Before proceeding, I should say that I take this lacuna to be more of a reflection of the state of current phonological theorising at large, or more precisely of the focus of phonologists' interests, rather than a shortcoming of these particular proposals. ‘The appendix’, by Bert Vaux & Andrew Wolfe, provides detailed arguments in favour of analysing extraprosodic consonants as being linked directly to a node higher in the prosodic hierarchy than the syllable node rather than being (a) part of complex margin constituents, (b) onsets in empty-headed syllables or (c) prosodically unaffiliated. Vaux & Wolfe's arguments come from a wide arrange of possible data, such as phonetic realisation, psycholinguistics and intuitions about the number of syllables. Importantly, however, it is not always clear how all of these data are relevant within a modular view of ‘encapsulated’ phonology, i.e. a phonology which communicates with the outside world in only a very limited way. It may not be impossible that all the ‘external’ evidence indeed bears on the issue in question, but Vaux & Wolfe do not discuss how this would work. For example, in presenting a problem for any account which says that extraprosodic segments are always peripheral in the word, the authors point to the s consonants in English words such as midst, Baxter, extra, expect and others, and claim (p. 128) that ‘from a synchronic standpoint there is no reason to believe that an unschooled native speaker would parse them into more than one morpheme’. This presupposes that we take (explicit) parsings by ‘unschooled native speakers’ as somehow indicative of morphological structure, which is not a standard methodology in morphology or syntax. It is actually one of the standard assumptions of modular theory (Fodor Reference Fodor1983) that consciousness has no access to the internal structure of a module (consider, for example, optical illusions, for which no amount of meta-awareness of visual perception can defeat the illusion). The volume contains three responses to Vaux & Wolfe's chapter, one by Cairns, one by Clements and one by Raimy. The last of these makes some of the same points as are raised here.
Turning to a paper within the ‘Metrical structure’ section of the book, William J. Idsardi's paper ‘Calculating metrical structure’ presents an updated view of his simplified bracketed grid (SBG) theory of metrical computation, as well as showing how this theory can be fruitfully translated into finite state automata. Two responses to Idsardi's paper, one by Harry van der Hulst and one by Charles Reiss, point out that such a translation is only of limited use, given that it does not really provide more insight into the linguistic phenomena at hand. Idsardi shows that a translation of an OT analysis of similar metrical facts ends up being much more complex, but as Reiss shows, this does not even tell us anything about the relative complexity of the two analyses per se. (A third response to Idsardi's paper, by B. Elan Dresher, shows how SBG theory can solve a long-standing problem in the analysis of Tiberian Hebrew metrical structure.)
The only chapter in the volume not mentioned so far is Morris Halle & Andrew Nevins' ‘Rule application in phonology’. As the title suggests, the chapter is mostly concerned with some aspects of the way in which phonological rules are applied; the authors' proposals are illustrated with respect to the analysis of verbal forms in three Slavic languages (Russian, Czech and Serbo-Croatian). At a few points, the paper does touch on interface issues, for instance when the authors suggest, towards the end of the paper (p. 378), that ‘theme vowels serve a parsing purpose, functioning as boundary markers between contentful morphemes that assist the hearer in recovering the constituency of the linear phonetic string’. However, the implications of this proposal for modularity or for the way in which the interface between the linguistic system and the outside world should be understood are not further discussed.
The conclusion of this evaluation, based on the articles and responses, is that modularity does not seem as popular a topic of linguistic research as it was approximately twenty years ago. Among current phonological frameworks, Optimality Theory (Prince & Smolensky Reference Prince and Smolensky2004) is a theory that is very uneasy with modularity assumptions, because it assumes that all aspects of phonology (indeed linguistic structure in general) are generated in parallel, and all relevant constraints can freely interact with each other. Determining modularity within such a system is not a trivial task. (In fact, as pointed out by Andrew Nevins (personal communication), factorial ranking of constraints would be greatly simplified if it were known, via modularity interactions, that certain constraints never interact with each other.) Nonetheless, at the same time an opposite trend is developing: the Minimalist Program in syntactic theory (and the related Biolinguistics Program) has developed a very strong interest in so-called ‘third factor’ (extralinguistic) explanations for linguistic principles (Chomsky Reference Chomsky2005), thus also blurring the rather clearly encapsulated architecture of the language faculty once prevalent in approaches embracing ‘the autonomy of syntax’.
For many formal linguists, therefore, while the modular picture of linguistic theory might still be attractive in some way – perhaps as our ultimate ideal of what the perfect theory should look like – it does not guide current research, at least not with respect to submodules within phonology or phonetics. Most of the chapters in the volume confirm this evaluation; while their overall quality and impact is very high, modularity does not seem to be feature prominently in the agenda of many authors, despite the editors' emphasis in their introduction that this is an organising principle of the book. I believe that the lack of attention to these questions (such as the interaction between modules, and where certain principles are located within the grammar) is clearly a weakness of our field as a whole. It is to be hoped that there will be many more volumes with the aims set forth in Contemporary views that can stimulate work on this important topic.
