Conclusion and further issues

This chapter concludes, summarizing key findings concerning prosodic weight and raising issues for further research. It compares three approaches to weight-sensitivity in phonology, namely, categorical mapping constraints (such as WEIGHT-TO-STRESS), phonetic discriminant analysis (which identifies optimal criteria), and direct phonetics–phonology interface approaches (such as $t$-to-Stress, where $t$ is a numerical weight percept). Advantages and pathologies of each approach are discussed, pointing towards a possible eventual synthesis. The chapter also includes sections treating the opacity of weight criteria, the domain of the weight percept (which is argued to include parts of onsets by way of p-centers), and the varying degrees of categoricity vs. gradience found in different types of phenomena, where the incidence of gradience tends to correlate with domain size.

Prosodic end-weight and the stress–weight interface

Prosodic end-weight refers to the specifically phonological aspect of end-weight, as emerges when one controls for other factors influencing word order, such as frequency, semantics, and syntactic complexity. Eight principles of prosodic end-weight are established, all aligning with the typology of weight more generally, suggesting that prosodic end-weight reflects bona fide phonological weight as opposed to raw complexity or duration. Several possible explanations for prosodic end-weight are considered, including final lengthening, complexity deferral, phonotactic or rhythmic optimization, and phrasal or nuclear stress. Phrasal stress is argued to be the core explanation for prosodic end-weight. Thus, weight-stress mapping operates both within words and in phrasal prosody. Weight-mapping constraints from earlier in the book are extended to phrasal contexts. This analysis predicts, evidently correctly, that some languages, such as Turkish, should exhibit prosodic beginning-weight rather than end-weight.

Prosodic minimality in isolation and in context

Prosodic minimality refers to the minimum size requirements that languages impose on prosodic words. To date, nearly all research on prosodic minimality considers the prosodic word in isolation. This chapter summarizes this literature but focuses rather on the phonological analysis of minima in the context of larger prosodic constituents, a domain that reveals new issues. In particular, resyllabification across words can threaten minima (as when CVC words resyllabify), to which languages can respond either by suppressing resyllabification if it threatens minimality, by allowing resyllabification but repairing the word through lengthening, or by letting the resulting degenerate word stand as such. Case studies of Prakrit, Tamil, and Latin illustrate these three possibilities, respectively. Tamil is of further interest because only a subset of its coda consonants contribute to minimality. Evidence converges from across systems that its two rhotics fail to bear weight, despite being highly sonorous coda consonants.

Introduction

A variety of phonological phenomena are sensitive to weight, including stress, prosodic minimality, meter, end-weight, tone licensing, compensatory lengthening, syllable structure, reduplication, and allomorphy. After a preamble situating prosodic weight historically with examples from Sanskrit, this introductory chapter surveys weight-sensitive systems and the universals of weight that hold across them. For example, length and complexity contribute to weight in every part of the syllable, including the onset, which is traditionally regarded as being irrelevant for weight. Sonority also affects weight across phenomena, with greater sonority contributing to weight in the syllable rime but detracting from it in the onset. Weight further applies to prosodic constituents above the syllable. Finally, the scope and aims of the book are summarized, including an outline of the subsequent chapters on metrical weight systems as well as an introduction to the constraint family employed to relate weight to prominence across systems.

Quantitative meter

Quantitative meters regulate the distribution of syllable weight or moras in verse constituents. They generally involve a binary criterion for weight, but often add to it sensitivity to gradient, intracategorial weight. A distinction is drawn between variable weight, which can involve optional processes (such as variable cluster syllabification or vowel shortening in hiatus) and gradient weight, in which phonological structure is fixed but the meter evinces sensitivity to a weight continuum. For example, superheavy syllables are sometimes avoided in cadences, as in Sanskrit. Second, different position types sometimes exhibit different tolerances for heavier or lighter heavy syllables, permitting the diagnosis of an intraheavy continuum, as illustrated for Finnish, Greek, and Tamil. Finally, line-final position can favor heavier heavies or lighter lights. This chapter also considers prospects for Interval Theory, by which the weight domain spans the left edges of successive vowels.

Weight scales for stress

Stress placement in words is often affected by syllable weight, stress being attracted to heavy syllables. Weight in such cases is usually binary (heavy vs. light) but often instantiates more complex scales. This chapter focuses especially on the phonological analysis of ternary and higher order scales, featuring case studies of several languages set in Optimality Theory. It argues that such scales must be analyzed in terms of vowel prominence rather than moraic coercion or coda prominence. The relation of geminates to stress also features prominently, as it is maintained that geminates can be analyzed as uniformly moraic for stress. Finally, several cases of gradient weight for stress are surveyed, including English. In these systems, stress placement responds statistically to weight, which manifests a fine-grained continuum rather than a simple categorical opposition and often includes onset and sonority effects.