Optimality and Cumulativity in the Phonology of English

Review of Michael Hammond. The Phonology of English. A Prosodic Optimality-Theoretic Approach. Oxford: Oxford University Press, 1999. The Phonology of the World's Languages series.

Marc van Oostendorp

Review for GLOT International, Spring 2001

0. Introduction

Hammond's Phonology of English (PoE) has three goals: (a) introducing students to Optimality Theory; (b) introducing students to distributional regularities of English, i.e. mainly syllable structure and stress; (c) "making a number of novel theoretical proposals within Optimality Theory." (PoE: viii) It is very ambitious to try to combine these goals in one book, and even though some parts of PoE are really impressive and very helpful to the student of English phonology – for instance, the book features many tables with possible and impossible consonant clusters – I am sorry to say that in my view it has not entirely succeeded in any of these three goals. In this review I will discuss each three of them in turn and explain why I think the author has been unsuccessfull in his aims and goals.

The book series The Phonology of the World' s Languages, edited by Jacques Durand of the Université de Toulouse-le-Mirail and published by Oxford University Press, contains many very important in depth descriptions of phonological systems of single languages. Until the present day, interesting and important volumes appeared on Armenian, Dutch, German, Hungarian, Kimatuumbi, Norwegian, Portuguese, and Slovak – all of them written by outstanding scholars in the field. The present volume deals English, probably the most widely studied language in the phonological literature. It has been written by the well-known American phonologist Michael Hammond, and it presents the first volume in the series that has been framed exclusively in Optimality Theory. All other books take a derivational, rule-based point of view, which seems simpler from the point of view of language description.

PoE differs from other volumes in the series also in other respects. It does not give an overview of all the relevant facts of the language, for instance. Many well-known phenomena are completely ignored, such as the effects of the Vowel Shift and velar softening. Rather than this, the aims of the book are to introduce the reader to Optimality Theory and, at the same time, to describe "distributional regularities in monomorphemic English words" (PoE:vii). The argument for the latter restriction is partly theory-internal to Optimality Theory – "the theory lends itself to this because of its focus on surface regularities" (ibid.)" and partly based on a particular view on the empirical basis of generative phonology: "[t]he data lend themselves to this because while the corpus is clear, decisions about which forms alternate with each other or are even related morphologically to each other is far more subjective." (ibid.)

PoE is organized in the following way. Chapter 1 contains an introduction to Optimality Theory and to some of the basic facts of English (e.g. the phoneme inventory). Chapters 2, 3 and 4 deal with syllable structure; chapter 2 gives a basic introduction to the kind of evidence phonologists use in favour of distinguishing the syllable as a constituent and a general overview of moraic theory, while chapter 3 deals with the phonology of syllable margins and chapter 4 with syllable peaks. Chapters 5-8 give an overview of stress theory. Chapter 5 introduces the notion of stress and evidence for the foot as a constituent; chapter 6 discusses the relation between syllable structure and stress in English; chapter 7 discusses the position of the "rightmost" stress foot, and chapter 8 with the other stress positions. Chapter 9 is an afterword with a summary of the results and some remaining theoretical and empirical problems. Every chapter ends with a short section on "further reading" . The book contains a bibliography (see below), a subject index and an index on all the English words that are discussed in the book.

1. PoE as an introduction to Optimality Theory

Other than most books in the Phonology of the World's Languages series, PoE is intended for an audience of beginning students. Some chapters (1, 2, 5, 6) are written exclusively or primarily for this group. They demonstrate the structure of the argumentation that is most common in modern generative phonology and explain the basics of Optimality Theory. This is first and foremost a theory of language typology. One of the attractions of the theory for the beginning student is that superficially quite complicated patterns of data in a number of different languages can be described by simply ranking a small set of constraints in various ways. Optimality Theorists may differ in their assumptions about the form of the constraints and about the structure of the representations their theory deals with – in short, they can differ in principle about the explanation of any universal aspect of human language. Yet they should all agree about the way in which languages differ from one another, since constraint ranking should always be involved in this.

It is much harder to find evidence for constraint ranking for a student of only one language system; so-called 'Emergence of the Unmarked' effects could provide such evidence, but it is not always very easy to find such effects. They do not occur in PoE, as far as I am able to see. Now all of this does not need to be a problem if the language one restricts oneself to is precisely English. Because of its wide geographical dispersion and other sociocultural factors, there is an enormous amount of variation in the language. In the words of Harris' (1994:xii) introduction to his own excellent textbook of English phonology (ominously missing from PoE's bibliography): "In fact, a good case can be made for saying that in English we find a hefty sample of the phonological phenomena that are possible in any natural language." Studying the microvariation between dialects and sociolects even has certain advantages over studying macrovariation between distant languages, since it allows one to isolate phonological differences in systems that are otherwise very similar.

Unfortunately, the author of PoE has decided to discuss only one variant of English, a variant he refers to alternatingly as "my own dialect of American English" (p. viii) and "the most standard dialect of American English" (PoE: 2). The argument given in PoE for restricting oneself this way, is weak: "It would be impossible to treat the range of material offered here with proper attention to all varieties of English." Of course, nobody would expect an introductory book to discuss all varieties of English in detail; but that does not mean that it is equally impossible to sometimes compare a few other varieties with respect to a certain phenomenon.

PoE's choice implies that concrete language variation does not play a role in the book at all. Chapter 5 includes an extensive discussion of a typology of stress systems derivable from the constraints Parse-s , WSP, RL/LR, NonFinality and FtBin (PoE: 172-190), but this discussion is very abstract. It is based on patterns such as LHLL or on made-up words, such as patakasa. PoE demonstrates that 8 possible systems can be derived from these constraints, and shows that English conforms to one of them. Unfortunately, he does not show that the other rankings describe real languages as well, and does not even mention the names of those other languages, even though this would have made the whole discussion much more attractive to the beginning student.

Another unfortunate consequence of demonstrating Optimality Theory on the basis of a meticulous analysis of phonotactics and metrical structure in one variant of English, is that the analysis sometimes lacks elegance by necessity. For instance, the fact that English does not have syllables with a short, tense vowel in the rhyme, followed by a [zÿ], is explained by assuming that this consonant has two moras when it occurs as a coda consonant, and by the following constraint (PoE: 137):

(1)?Pre-[zÿ] tensing}

Vowels are tense (short) before coda [zÿ].

Of course, it is hard to think of a principled explanation for this curious phenomenon, but on the other hand, the part of PoE that deals with phonotactics has many such ad hoc solutions because apparently there still are a lot of things that we do not understand well enough to fit into a constraint-based analysis. To me, this does not seem a very attractive Invitation to phonological theory for the interested beginning student.

A final reason why I would be hesitant to recommend PoE as a beginner's guide to Optimality Theory is that it presents quite a number of tentative and confusing suggestions. For instance, on page 27 it is claimed that "all constraints in all languages must be instances of GA/NGA [i.e. Generalized Alignment in the sense of McCarthy and Prince 1993, or Negative Generalized Alignment, i.e. Alignment constraints prefixed by a negation marker] or GC [Generalized Correspondence in the sense of McCarthy and Prince 1995]. The analysis of English to be presented in the following chapters suggests that this may be true." The student who has just read this, can easily get confused by statements such as the following (PoE: 48, about restrictions on English word onsets): "Some of these are readily treated in terms of syllable structure and some are best treated in terms of alignment (GA/NGA) constraints on simple linear order". Since syllable structure constraints are distinguished from alignment constraints in PoE, we should perhaps conclude that syllable structure constraints are correspondence constraints. It is not clear at all, however, how they could be formalised as such.

2. PoE as an introduction to English phonology

Apart from providing an introduction to Optimality Theory, PoE also aims to introduce the reader to topics in English phonology. In itself this is a worthwhile enterprise: it is good, for instance, to see a complex stress system such as the one found in English, analysed in a relatively new framework such as Optimality Theory.

Unfortunately, however, PoE does not satisfy as an introduction to English phonology. I already mentioned the fact that many important issues are simply missing. Next to the Vowel Shift and velar softening, I could mention r-dropping, r-epenthesis, flapping and tapping, trisyllabic and trochaic shortening, all kinds of assimilation, the distribution of clear and dark l, phrasal stress and intonation. I would consider a basic knowledge of each of these phenomena essential for every serious student of English phonology, but one could argue that it is not possible to treat all of them in one book. If that is true, this probably is also the reason why there is no reference to literature on Germanic languages other than English. For instance, PoE contains a discussion about the difficulty of distinguishing the opposition tense/lax from an opposition in length. This is not special to English, however, and a scholarly discussion of this should at least refer to the fact that similar discussions have been conducted on Dutch and German as well (for instance in the volumes on these languages in the same series).

More importantly, PoE does not even refer the interested reader to the relevant literature on English. A good bibliography may be one of the most important aspects of a book such as this. The list of references in PoE is only four pages long; it contains many references to general OT literature and to descriptions of exotic languages which are popular in the modern literature, such as Diyari and Tohono O'odham. Yet it does not refer to Harris (1994), for instance, or to Giegerich's (1992) excellent overview of the facts of English phonology, or to Fudge's (1984) well-known book with facts of word stress. There is no reference to work done within the important Lexical Phonology paradigm (Kiparsky's work is cited only twice, viz. with a 1989 article about generative metrics and an unpublished manuscipt about catalexis; the famous article by Halle and Mohanan 1985 is not mentioned at all, nor is any other work by Mohanan, or by Harris, Jespersen, Lass, Milroy, Trager, or many other important scholars in the history of English phonology). The most frequently cited author, incidentally, is Hammond; on p. 338 the work of this author is evaluated in the following way: "Hammond (to appear) offers a nice discussion of the relevance of statistical frequency [...]". The general description of The Phonology of the World's Languages has it that "each volume [...] will provide comprehensive references to recent and more classical studies of the language." To my disappointment, PoE does not fit this description.

A last reason why I would not recommend using this book as an introduction to English phonology, is that I object against its methodology. All data in the book have been extracted from work with a number of databases and were later checked against grammaticality judgements. PoE claims that this methodology "can only be seen as a step forward in terms of empirical confidence […] since most previous discussions of English syllabification and stress have been based solely on intuitions". Yet I think this claim is questionable at best, if only because it is never made clear what the databases are that this work is based on: do they contain stress markings? Do they contain material in other dialects than "the most standard dialect of American English"? Whose are the "intuitions about grammaticality" against which the database facts were checked? As a matter of fact, it seems to me that this particular methodology obscures the object of study: what is meant by 'American English' that is the topic if this book? Is it E-language or I-language in the sense of Chomsky (1986)? In the first case, it does not seem relevant to add intuitions; in the second case (and if indeed the variant described is that of the author), it hardly makes sense to mention having worked with a corpus, in particular if this eventually does not seem to have brought to light many spectacular new facts.

3. PoE as a contribution to Optimality Theory

Next to an introduction to Optimality Theory and an introduction to the phonology of English, PoE is also intended as a monograph in which several new theoretical proposals about the structure of Optimality Theory are made. These proposals are conveniently summarised in section 9.6 (PoE: 336-338).

One proposal is 'Distributional OT', a variant of OT in which alternations are disregarded. As far as I can see, PoE does not provide any arguments for his position, except that "decisions about which forms alternate with each other or are even related morphologically to each other is far more subjective", an argument which would discard most of morphology as a serious linguistic discipline for reasons of being necessarily 'subjective'.

Another proposal is that "the mora count of English syllables was constrained to be between two and three moras". I think that this is one of the places where PoE' s disregarding the literature on West Germanic becomes fatal, for this 'new' proposal has been accepted for these closely related languages by most of the scholarly community that works on these languages (even within OT; see e.g. Fery 1996 on German; Kager 1989, a work that is cited in PoE, defends this position for Dutch as well).

Other proposals concern PoE's analysis of stress. I will not go into these, except to note that the proposals are very similar to those of Burzio (1994). This is acknowledged (PoE: 276; this work is referred to as Burzio 1995 in the bibliography), but the author dismisses the relevance of this similarity on the grounds of the following two arguments: "[H]is proposal is not made in terms of OT. In addition, the details of his proposal are not worked out, and so it is difficult to compare it with the current one in specific terms." Many things could be said about these arguments. It is not clear, for instance, why it is important that Burzio's proposals do not use the technical machinery of Optimality Theory, even though they are very close to that framework in spirit. Also, it is strange to say that the details of Burzio's proposals are not worked out. Burzio (1994) is a book-length study dealing with English stress exclusively. I can see no way in which Burzio's analyses are crucially less "worked out" than those of the author of PoE or of other scholars working in the field.

4. Conclusion

In my view, cumulativity is an important aspect of scientific progress. Once a new framework is formulated, it is unadvisable to throw away all the knowledge that has been gathered in earlier times. A book series such as The Phonology of the World's Languages plays an important role in guaranteeing this type of cumulativity: knowledge about a specific language is gathered here in an accessible format for all (future generations of) phonologists to consult.

If a new framework, such as Optimality Theory, turns up, it is interesting to see how this new theory can be applied to the well-known facts and generalisations. It can be particularly fruitful to test the way a typological theory, such as Optimality Theory, works in the analysis of a single, well-known, language. In my view, this does not mean that we necessarily need a completely new analysis of every fact in the new framework: but it also should not mean that large areas of previously studied empirical domains are suddenly left out of consideration altogether. If it has been established that something can best be analysed representationally, there is no a priori reason to change this (of course, it is always possible that the old analysis was not good enough for independent reasons.)

PoE does not conform to the standards set by other volumes in the same series. In my opinion, the reason for this is that too much literature is ignored: literature on other variants of English or on other Germanic languages besides English, and theoretical literature on English in other frameworks than Optimality Theory. Many of the observations that the author of PoE found in his database were already to be found in the literature; and the same is true for many of the ideas he develops within the framework of Optimality Theory.

This is very unfortunate, because the style of the book is clear and Hammond's ideas are often very interesting. The fact that he does not take into account the many important contributions to the literature, eventually results in The Phonology of English itself being neither a major contribution to the theory, nor very suitable as a reference book, nor as a guide for beginning students of English phonology, nor as an introduction to Optimality Theory.

References

Burzio, L. 1994. Principles of English Stress. Cambridge, UK: Cambridge University Press. Cambridge Studies in Linguistics 72.

Fery, C. 'German Foot and Word Stress in OT'. In: Patrick Bye, Ove Lorentz & Curt Rice, (eds.) Nordlyd. Papers from the 2nd Workshop on Comparative Germanic Phonology 1996. Tromsø University Working Papers on Language and Linguistics, No. 24. 1996, 63-96.

Fudge, E. 1984. English Word-Stress. London: Allen and Unwin.

Giegerich, H.J. English Phonology. An Introduction. Cambridge, UK: Cambridge University Press. Cambridge Textbooks in Linguistics.

Halle, M. and K.P. Mohanan. 1985. The Segmental Phonology of Modern English. Linguistic Inquiry 16.57-116.

Harris, J. 1994. English Sound Structure. Oxford, UK and Cambridge, MA: Blackwell.

Kager, R. 1989. A Metrical Theory of Stress and Destressing in English and Dutch. Dordrecht: Foris.

McCarthy, J. and A. Prince. 1993. Generalized Alignment. Yearbook of Morphology, 79-153.

---. 1995. Faithfulness and Reduplicative Identity. In: J. Beckman, L. Dickey, and S. Urbanczyk, eds., Papers in Optimality Theory. University of Massachusetts Occasional Papers in Linguistics 18: 249-384.