1: Introduction
Prosody is basically all the parts of speech which don't fit neatly into the concept of a segment, things like rhythm and pitch.The 'prosodic hierarchy': prosodic phrases are made up of prosodic words1 are made up of
feet are made up of syllables.
The constituents which make up each of these levels can vary in length, segment quality and/or structural complexity. Together, these form 'quantity'. Various phonological processes may be sensitive to quantity.
2: The syllable and quantity-sensitivity
The syllable is usually where quantity-sensitivity happens, though not in all languages. Phenomena which are quantity-sensitive at the syllable level include stress, tone and poetic meter.2.1 An example: Latin stress
If the penultimate syllable is heavy, it is stressed. If not, the antepenult is stressed instead.Heavy syllables are CVV or CVC. Light syllables are CV (consonant-vowel).
This distinction can be observed in Latin verse e.g. Horace's odes, which has the same pattern of heavy-light syllables on each line, even though heavy could mean long vowels, diphthongs or closed syllables.
2.2 A typology of weight patterns
Whilst the Latin division is found in many languages - including Classical Arabic, Finish, Gothic, and Hausa - there are also languages which count only CVV syllables as heavy, and CVC as light with CV, such as Tiberian Hebrew.No language is found with heavy CVC but light CVV syllables.
There are also languages in which certain types of consonants pattern with vowels, such that CVC1 is heavy, whilst others do not, such that CVC2 is light.
For example, Kwakw'alaq treats syllables ending in sonorants (approximants like /w/ and nasals like /m/) as heavy, and syllables ending in obstruents (plosives like /p/ or fricatives like /s/) as light.
The consonant types are split according to the sonority hierarchy: the split is not a specific place, but any type more sonorous than the split will pattern with vowels to form heavy syllables, and any less sonorous type patterns with an empty coda to form light syllables.
Moving all CVC into either heavy (or into light) is therefore just the extreme end of the spectrum.
Onsets don't count towards syllable weight (except in few languages, where they do).
2.3 Examples of weight patterns
In languages which do have quantity-sensitive stress, stress is attracted to heavy syllables.In a survey of ~400 languages (Gordon 2006), ~45% were quantity sensitive, of which just over 30% counted CVC as heavy; just under 30% counted CVC as light; and 3% split partway through the sonority hierarchy. (The rest had different weight patterns, which this chapter doesn't look at.)
Tone can also be sensitive to weight. In such systems, light syllables can only have 1 tone, whereas heavy ones can bear multiple tones (called contour tones).
Looking at the same survey again, for quantity-sensitive tone, there are lots of languages which count CVC as light, or only some CVC as heavy, but very few which count all CVC as heavy, because it is phonetically very difficult to pronounce multiple tones on a single short vowel.
3: Representation of syllable quantity
There are 2 main representational approaches. In the first, the configuration is different for light and heavy syllables (the way in which the consonants and vowels combine to make higher level structure has different branching.)In the second, syllables are made up of constituents called 'morae' which are weight-bearing. Syllables with 1 more are light, syllables with more than 1 are heavy. A mora is made up of different combinations of vowels and consonants, depending on the language.
How exactly the mapping from segments to morae works depends on the theoretical approach.
4: Vowel length and quantity-sensitivity
Some languages do not have quantity-sensitivity at the syllable level, like Bulgarian and Modern Greek. These languages also do not have vowel length (i.e. have no distinction between long vowels and short vowels), which raises questions about the relationship between the two.
5: Binary or multivalued weight distinctions?
So far, we've only looked at light vs heavy, a binary distinction.
What about CVVC or CVCC syllables?
In Latin, CVVC and CVCC count as heavy, the same as CVV and CVC.
But in Hindi, there are 3 weight levels, and CVVC and CVCC are super-heavy, in constrast to heavy CVV and CVC syllables.
Some languages - like Turkish - disallow CVVC or CVCC syllables altogether.
Languages which treat CVC as light also treat CVCC as light, and CVVC as heavy, the same as CVV - C is simply not weight-bearing.
No language has been found which both splits consontants between forming heavy codas and light codas, and has super-heavy syllables.
The moraic representation can cope with super-heavy syllables, providing not all segments must belong to a mora. E.g. in Latin, the extra segments are assumed to be extra-moraic, outside of a mora, and hence don't affect the weight.
6: Inconsistencies
Different processes imply different weights
E.g. in Kiowa3, vowels are shortened in all closed syllables; but contour tones can appear on some CVC syllables (where the coda is a sonorant).
E.g. in Lhasa Tibetan, only CVV syllables are heavy, but both CVV and some CVC syllables can bear contour tones.
Different contexts imply different weights
E.g. in Arabic, word-final consonants do not contribute to weight
E.g. in Chugach4, CVC syllables are heavy word-initially, but light elsewhere
This can either be treated as syllables having variable weight, depending on the context; or as syllables in certain positions being 'extra-metrical' i.e. not being counted.
This can either be treated as syllables having variable weight, depending on the context; or as syllables in certain positions being 'extra-metrical' i.e. not being counted.
7: The foot and quantity-sensitivity
Syllables are grouped into feet. Feet tend to be binary. In quantity-sensitive systems, how syllables are grouped depends on their weight.
A 'trochaic' foot consists of either 1 heavy syllable or 2 light syllables. In moraic theory, these are both 2 mora. 'Trochaic' means that weight falls on the first syllable of the foot.
An 'iambic' foot consists of either 1 heavy syllable, 2 light syllables, or 1 light followed by 1 heavy syllable.
'Iambic' means that weight falls on the second syllable of the foot.
Trochaic feet are optimally 2 light syllables, whereas iambic feet are optimally 1 light then 1 heavy. So trochaic feet are even, whereas iambic feet are uneven (Hayes 1995). For more on this, see my post On the perception of rhythmic grouping.
8: Scalar quantity systems
"Binary quantity systems are based primarily on grouping syllables into feet", whereas scalar quantity systems are based on a notion of prominence. Syllable weight is the main dimension of prominence, but there is also evidence for tone or vowel height as prominence dimensions.
9: Higher levels of prosodic hierarchy
Prosodic words
The minimum size of a prosodic word is usually a single foot. Some languages allow prosodic words to be minimally a single syllable.
"There are no known cases of a prosodic word minimally branching into two feet", which we'd expect given examples of minimal foot size being two morae.
So prosodic words are apparently not quantity sensitive.
Prosodic phrases
Evidence from Italian, French and English shows binary branching in prosodic phrases (i.e. you don't just combine 3 or more prosodic words together, but combine two to make a phrase, then combine this further).
Certain phrase types in Serbo-Croatian must contain at least 2 prosodic words.
So prosodic phrases are sensitive to quantity: there are minima and maxima for the constituent parts.
10: Markedness
Heavy syllables are marked, whereas light feet are marked; it is not heaviness or lightness itself which is marked.
Heavy phrases are unmarked, though constituent size at this level is mostly determined by morphosyntax (i.e. what meanings you want to combine!)
The chapter gives examples of constraints describing this in Optimality Theory, but offers no physical or cognitive explanation of these instances of markedness.
Footnotes
1 Prosodic words are different from just 'words' in that they don't always align with spaces. Some items like 'a' and 'the' tend to attach themselves to other words (called 'cliticising'). The word and its clitic then behave like a single unit when it comes to prosody.2 Kwakw'ala is a Wakashan language, spoken around Vancouver Island, Canada, by the Kwakwaka'wakw.
3 Kiowa is a Tanoan language, spoken in Oklahoma.
4 Chugach is a dialect of Pacific Yupik, spoken in Alaska.
References
Hayes, Bruce (1995) Metrical stress theory: Principles and case studies. University of Chicago PressZec, Draga (2011) Quantity-sensitivity. Blackwell Chapter 57, pp. 1335-1361
Gordon, Matthew (2006) Syllable weight: Phonetics, phonology, typology. London: Routledge