1. Terminology

The IPA provides a group of symbols for stress, length, intonation, syllabification and tone under the general heading "suprasegmentals", reflecting a conceptual division of speech into "segmental" and "suprasegmental" parts. However, as we shall see, this division is not very clean, with phonetic correlates of stress, intonation etc. often manifest in the consonants and vowels, i.e. at the segmental level. Therefore, rather than referring to "segmentals vs. suprasegmentals", I prefer the more general term "prosody" for these phenomena (among others).

2. Stress and its phonetic correlates

In English, the term "stress" is used in reference to the degree of prominence of individual syllables of single words (lexical stress), e.g. elephant (stressed - unstressed - unstressed) vs. elephantine (unstressed? - unstressed - stressed - unstressed). But not every word is actually stressed when it occurs in a sentence: some lexical stresses are picked out for sentential stress - perhaps better called "accent". E.g. "Albert went to the zoo."

IPA symbols for primary stress [] and secondary stress [] may be placed before vowels, or equally well at the beginning of each syllable (thus also providing some information about syllable boundaries): elephant [elIfnt], elephantine [elIfantIn].

Ladefoged (p. 249), like many introductory phonetics textbooks, says "the nature of stress is fairly well understood. Stressed sounds are those on which the speaker expends more muscular energy ... so that there is an additional increase in pitch." Davenport and Hannahs (p. 23), similarly, write "Stressed syllables are produced with more muscular effort, and are louder or longer than unstressed syllables." There are some problems with this view:

a) In some languages (e.g. Welsh cf. Williams 1982; Chamorro cf. Chung 1983) and even in some dialects of English (e.g. Birmingham, Tyneside, Welsh English) the pitch is lower on stressed syllables than unstressed syllables, and stressed syllables may be shorter and quieter than unstressed syllables.

b) There are other phonetic correlates of stress too. Cf. English:
 

Stressed-Unstressed			Unstressed-Stressed
[tuwIm]	"to him"	vs.	[thIm]	"to him"
[tuw]	"to her"	vs.	[th]	"to her"

In these examples, the stressed syllables have longer, closer, more rounded vowels and initial [h], whereas unstressed syllables have shorter, more central, less rounded vowels and somtimes initial [h] is absent. The phonetics of stress, then, includes not just loudness, pitch and duration, but also vowel quality, liprounding. Vowel quality differences are especially important markers of stress in English. Unstressed vowels are often said to be "reduced" to [] or [I]. Some English vowels, e.g. [], are only found in syllables with  (primary or secondary) stress: tonic [tnIk] vs. [tnIsItI]. The aspiration of voiceless aspirated stops is stronger in stressed syllables than unstressed syllables: e.g. [p^hhet^hrl] vs. [p^ht^hhroUl]

c) The phonetic realization of stress is rather language-specific. Trubetzkoy (1939) pointed out that in Bulgarian the contrast between close /u/ and mid // found in stressed syllables is neutralized in unstressed syllables, where only /u/ - the closer phoneme - occurs. (It is actually pronounced [o] in unstressed syllables. ) In contrast, in Russian, the contrast between open /a/ and mid /o/ is neutralized in unstressed syllables, where /o/ is unrounded and lowered to /a/.

Similarly, in Chamorro, mid vowel (allophones) are found in stressed closed syllables and only close vowel (allophones) occur in unstressed syllables, e.g. [sósu] "village" vs. [i susóa] "his village"; [tsótsu] "word" vs. [i tsotsómu] "your (sg.) work"; [wéswis] "to brush", [sénsin] "flesh", [néni] "baby", etc. This is the opposite of the correlation between stress and vowel height in the English examples above, in which stressed [u] is close and unstressed [] is mid. It appears then as if stress is not a well-defined phonetic feature, but an abstract phonological feature whose phonetic realization is i) manifold and ii) language-specific.

3. Duration

Length contrasts can be transcribed by doubling of letters (for simple long vs. short duration), or using length marks: [] (in print, or, just as good, [:]), [] (or [·]). There is also a diacritic for "extra short" sounds. E.g. Luganda [^\kkúlà] `treasure' vs. [kúlà] `grow up', Italian nonno [nonn] `grandfather' vs. nono [non] `ninth', or even English [kkju] `thank you/fuck you' (depending on intonation and context of use) vs. [kju] `cue/queue'.

Avoid confusing phonetic duration (measured e.g. in milliseconds) with phonological length (e.g. long vs. short vowels and consonants). Many durational differences in English are purely allophonic. Note that the duration of the short vowel /a/ can easily exceed that of the long vowel /i/ because more open vowels have a longer duration in English than closer vowels.

4. Pitch, tone and intonation

`Pitch' refers to the perception of relative frequency (e.g. perceptually high-pitched or low-pitched). Tone refers to significant (i.e. meaningful, constrastive, phonemic) constrasts between words signalled by pitch differences. Tone may be lexical, as in Mandarin Chinese:
 

Tone number	Description		IPA transcription example	Meaning
1	high level		[má]	`mother'
2	high rising		[m¯á]	`hemp'
3	low (falling+)rising		[màá]	`horse'
4	high fall		[mâ]	`scold'
"no tone/neutral tone"	(depends on preceding syllable)		[ma]	(question marker)

Or grammatical tone, as in many African languages, e.g. Edo:
 

Tense	Monosyllabic verbs	Disyllabic verbs
Timeless	[ì mà] `I show'	[ì hrùlè] `I run'
Continuous	[í mà] `I am showing'	[í hrùlé] `I am running'
Past	[ì má] `I showed'	[ì hrúlè] `I ran'

However, as with stress, there may also be non-pitch aspects of tone. Lexical tones are often related to durational, phonatory and vowel quality distinctions as well as frequency distinctions. For example, Mandarin Chinese tone 3 (low rise) is long with creaky voice, Hunanese tone 2 has breathy or chesty voice. Tibetan tone 1 words have voiceless initial consonants whereas tone 2 words have voiced beginnings. Long vowels in tone 4 or 5 open syllables in Thai are checked by a final glottal stop.

Intonation refers to the rise and fall of voice pitch over entire phrases and sentences, even in non-tone languages, such as English:

5. Prosody reconsidered

The preceding paragraphs have shown that many more features than duration and pitch are (or can be) prosodic. What, then, do we mean by `prosody'?

a) Features (or groups of features) not located at a single place in the sequence of consonants and vowels (e.g. stress, tone).

b) For example, (groups of) features associated with a whole syllable, word or phrase.

c) For example, features of the boundaries of syllables and words (e.g. assimilation, linking, absence vs. presence of initial [h] in the `to him/her' examples above). `Grenzsignale'.

6. What else can be a prosody?

Virtually anything! Some examples:

a) Place of articulation (cf. place assimilation at word-junctures).

b) Manner of articulation (cf. initial consonant mutation e.g. in Welsh), including ...

c) Degree of stricture (e.g. spirantization of final stops as a boundary feature)

d) Voicing (cf. voicing assimilation at word-junctures and in initial mutation)

e) Retroflexion (e.g. Sanskrit)

f) Frontness and backness (e.g. umlaut, vowel harmony)

g) Openness and closeness (e.g. vowel harmony)

h) Centrality and peripherality (cf. English stress above).

i) Aspiration, whisper (cf. stress above, Sanskrit). In English [h] may only occur once in a word. Apparent exceptions like jojoba are loan-words.

j) Nasality e.g. in Terena, Sundanese, and in the Urhobo examples in previous handouts.

k) Glottality (e.g. [] in vowel-initial words in English or German, Danish stød).

l) Lip-rounding (vowel harmony; cf. also discussion of stress above).
 

References

Chung, S. (1983) Transderivational relationships in Chamorro phonology. Language 59 (1). 35-66.

Ladefoged, P. (1993) A Course in Phonetics. (Third Edition) Harcourt Brace Jovanovich.

Trubetzkoy, N. S. (1939) Grundzüge der Phonologie. Travaux du cercle linguistique de Prague 7.

Williams, B. (1982) The Problem of Stress in Welsh. Cambridge Papers in Phonetics and Experimental Linguistics Vol. 1. Department of Linguistics, University of Cambridge.