Prosodic weight: categories and continua provides a broad account of phonological weight as it affects metrical phenomena. The book treats weight not as a heavy vs. light binary distinction which languages define in different ways, but as a complex universal scale which languages can make use of categorically and gradiently, and as a whole or in parts. The author reviews literature, provides original analyses and presents new experimental and corpus data to weave a holistic picture of this complex weight scale. Well-known factors such as vowel length and presence of a coda are confirmed to be important, but evidence is also gathered for less well-known factors, such as the contents of an onset, or the sonority of onsets, codas and nuclei. The book examines the effects of weight on categorical patterns, such as many weight-sensitive stress systems, and gradient patterns, such as the tendency for heavier things to come last syntactically in many languages (called ‘prosodic end-weight’ in the book). These gradient effects reveal parts of the weight spectrum not otherwise seen.
The arguments for the effects of onset length and sonority on weight are particularly striking: though onsets are often thought to be irrelevant to weight calculations, Ryan provides evidence from a variety of languages and phenomena that they should not be ignored. The book is a pleasant read, and packed with useful information and discussion for anyone interested in weight or prosodic phenomena, both experienced researchers and novices.
The book also provides a framework for analysing the tendency for weight to align with strong positions, whether in lexical stress, phrasal stress or poetic metre. Analyses are conducted in Optimality Theory (Prince & Smolensky Reference Prince and Smolensky1993) as well as in Maximum Entropy Harmonic Grammar (Goldwater & Johnson Reference Goldwater, Johnson, Spenader, Eriksson and Dahl2003, Hayes & Wilson Reference Hayes and Wilson2008), and familiarity is assumed with both. However, the typological points revealed by the book are in general theory-neutral.
Chapter 1 introduces the concept of phonological weight, and gives a brief overview of all the different phonological processes in which weight can play a part, including those covered in depth later in the book and those not further discussed (the connection between weight and tone licensing, for example). It then discusses briefly the phonetic basis of weight, which is only returned to at the end of the book, in Chapter 6. The phonetic or psycholinguistic basis of weight as a phenomenon is not addressed in depth, and the reader is instead referred to Gordon (Reference Gordon2006) and others.
Chapter 2 treats weight scales in word-stress systems. After providing a brief introduction to the role of weight in stress assignment generally, the chapter focuses on weight scales with three or more levels. One example is Chickasaw, a language in which V rhymes are lighter than VC rhymes, which are in turn lighter than VV and VVC rhymes. Two proposals are discussed within Optimality Theory to analyse complex weight scales such as these: Coercion, previously proposed by Morén (Reference Morén2001) and others, and Prominence, a novel proposal of this book. Prominence is ultimately argued for over Coercion. The two proposals make similar predictions for weight systems with only two levels of weight.
The Coercion analysis is so named because it relies on the possibility that coda consonants are sometimes moraic (rendering the syllable heavy) and sometimes not. In brief, V < VC < VV scales exist in languages in which bimoraic syllables attract stress and long vowels are always bimoraic, but coda consonants are only sometimes moraic. They lose their moraicity, for example, when a long vowel is present in the same rhyme, thus avoiding violations of constraints against superheavy rhymes. When both a VV rhyme and a VC rhyme are present in a word, stress is drawn to the VV rhyme over the VC rhyme. A faithfulness constraint Max-μ-V prevents long vowels from being shortened, but the corresponding constraint Max-μ-C would be low-ranked in languages without contrastive geminates, like Chickasaw. Thus, when Max-μ-V is high-ranked and Max-μ-C low-ranked, the VC rhyme can surface as monomoraic, while the VV rhyme cannot. The two main problems with this account are (i) that many languages can be shown to have obligatorily moraic codas, especially languages with geminates, and (ii) even when primary stress is attracted away from a VC rhyme onto a VV rhyme, secondary or tertiary stress may still be attracted to the VC rhyme, indicating that it is still bimoraic.
The Prominence analysis resolves these issues by replacing the single Weight- to-Stress constraint with a family of constraints relating stress to weight, which fall into a stringency hierarchy, according to cross-linguistic universals. The most used example are the constraints Stress→VV and Main→VV, which demand that any stress fall on a long vowel and that main stress fall on a long vowel respectively. When these constraints are ranked together with Weight-to-Stress, which simply demands that bimoraic syllables be stressed, complex weight hierarchies are straightforward to model.
The chapter concludes with a discussion of the gradient effects of weight in stress systems, demonstrating that weight can affect stress placement statistically in languages like Hupa, Yahi and English. While heavy syllables are not always stressed in these languages, they are stressed more often than would be expected by chance. In these systems too, a more fine-grained weight scale emerges. In Yahi, for example, a coda nasal seems to be lighter than a coda liquid, which in turn is lighter than a long vowel.
Chapter 3 discusses the relationship between weight and word-minimality restrictions. Many languages seem to require that words not be shorter than two syllables, or a single heavy syllable – perhaps aligning with the minimal foot in the language for stress (McCarthy & Prince Reference McCarthy and Prince1986). The chapter does not discuss the relationship between foot size and word size in detail, though existing typologies suggest the relationship may be tenuous (Garrett Reference Garrett1999). Instead the chapter notes that, typologically, word-minimality restrictions follow a gradient scale which has much in common with the weight scale observed for stress. Languages require words to be at least VC, at least VV, and so on.
The bulk of the chapter discusses the interaction of word-minimality with resyllabification across word boundaries. In languages where resyllabification crosses word boundaries, prosodic words still strive to maintain their minimum size on the surface. If a coda consonant, necessary for satisfying word-minimality, is resyllabified onto a following word, how does the language repair this? Three languages are discussed: Latin, in which subminimal words are allowed to exist on the surface after resyllabification, though they are avoided in poetry, Tamil, in which words rendered subminimal by resyllabification are repaired via gemination of the resyllabified consonant or vowel lengthening, and Māhārāṣṭrī Prakrit, in which resyllabification is mostly blocked when it would create a subminimal word.
Chapter 4 discusses the role of weight in poetic metre, first giving a brief overview of categorical weight restrictions across languages (most languages treat V as light and VC and VV as heavy). Next, the chapter discusses variable weight, the ability of certain syllables to scan as either heavy or light, depending on phonological processes that may apply in a given context. For example, a cluster may sometimes syllabify as an onset (e.g. Latin su.pre.mum), and sometimes as a coda + onset (sup.re.mum).
Next, gradient effects of the same weight scale established throughout the book are discussed. It is noted that when mismatches occur in a metre, for example a heavy in a light position, not all mismatches occur equally often. Instead, heavies in light position tend to be on the lighter side of heavy, and lights in heavy position tend to be on the heavier side of light. Based on this observation, we see the emergence of a gradient scale similar to the one observed in Chapter 2: coda nasals are heavier than coda obstruents, long vowels are heavier than short vowels with a coda, etc. More detail emerges here too regarding onsets: they exhibit the opposite pattern from codas with respect to sonority. While a more sonorous coda makes a syllable heavier, a less sonorous onset makes it heavier: an onsetless syllable will scan as lighter than an onset nasal, which is lighter than an onset obstruent.
Finally, the chapter foreshadows Chapter 5 in discussing the preference for heavier syllables to occur closer to the end of a line of poetry. Here we see the effects of nucleus sonority: the more sonorous the nucleus, the more likely it is to show up line-finally.
Chapter 5 discusses the phenomenon of prosodic end-weight. It focuses mainly on binomial expressions (e.g. bread and butter) and compounds, finding that heavier forms tend to come later in the expression. Here, a variety of distinct elements contributing to the weight of a word are independently elucidated, including vowel length and vowel, coda and onset sonority, as well as others which were not relevant in Chapters 2–4, namely location of the stressed vowel, and number of syllables. Corpus evidence, statistical modelling and in some cases novel experimental data are used to argue for the effects of this gradient weight scale, independently of other factors known to influence ordering in binomials and compounds, such as animacy and proximity to the speaker. Though some effects, like syllable count, have no bearing on the weight scale used in stress systems, word-minimality and poetic metre, most aspects of the weight scale seen in this chapter align perfectly with those seen in previous chapters. It becomes inescapable that there is a single general weight scale referred to by various elements of the grammar, even poetic forms, rather than separate weight scales used for different purposes.
The chapter makes the argument that prosodic end-weight is the result of a grammatical preference to align weight with phrasal stress, not unlike the preference to align weight with word stress in Chapter 2, or with strong metrical positions in Chapter 4. A new family of constraints is proposed, building from the Stress→VV and Main→VV constraints in Chapter 2. These constraints demand that particular phonological elements be found in positions with phrasal stress, such as VV→φ s, which demands that a long vowel be in a phrasally strong position, and V→φ s, which demands that any vowel, short or long, be in a phrasally strong position. These constraints appear in a stringency hierarchy.
This proposal that prosodic end-weight is the result of aligning phonological weight with phrasal stress makes the prediction that languages with left-aligned phrasal stress should exhibit beginning-weight. Some arguments are presented that this is in fact the case, including new corpus evidence from Turkish, demonstrating beginning-weight effects of syllable count.
Chapter 6 briefly treats issues not discussed in detail in the previous chapters, such as the nature of gradience and the phonetics of weight, including whether weight is actually aligned to syllables, moras or something else. (The author argues for ‘something else’, namely the ‘p-centre interval’, but does not go into enough detail to be convincing. Rather, the reader is referred to Ryan Reference Ryan2014.)
If one criticism of the book might be raised, it would be that there is an overreliance on examples and argumentation based on Latin. Latin is treated throughout as a living or recent language in which things like vowel duration, resyllabification and grammaticality judgements can be easily known, rather than a defunct language whose native speakers were dead long before there was a field resembling modern phonology. Sceptical readers would have benefited from a more detailed account of how we can know, at this historical distance, which vowels were long, which syntactic elements were clitics, which consonants were resyllabified, etc.
Ryan's style is overall quite clear, though readers may find themselves turning to a dictionary more often than usual for translations of Latin phrases, or definitions of extremely low-frequency words. The overall quality of typesetting in the book is very high. Readers will find fonts and overall layout, as well as examples, graphs and tables, crisp and enjoyable to read. A list of abbreviations is provided at the front, and a constraint list provided in an appendix. Impressively, no typos at all came to my attention, despite the large number of complex OT tableaux.
This book is not a comprehensive treatment of phonological weight in all its domains. It does not cover the relationship between weight and tone, compensatory alternations such as lengthening that preserve the weight of a syllable, weight-based restrictions on syllable structure or weight requirements for reduplication or allomorph selection. Instead, the book is a valuable contribution to the study of weight as it affects prosodic phenomena. It provides extensive evidence for a variety of universal or semi-universal factors affecting weight calculations, and a framework for formal analysis in Optimality Theory of weight-to-strength mappings, whether in lexical stress, phrasal stress or poetic metre. This book will provide a firm foundation to anyone researching any of the particular phenomena treated in Chapters 2–5, or phonological weight in general.
