Prosody

Prosodic (or suprasegmental) properties of speech involve things that are “above” individual sounds. The distinction between the prosodic and the semgental domain is not abolutely clear. In this chapter, we will discuss three aspects of the prosody of speech: stress, accent, and intonation.

Stresses, accents, and the tonic

The terms accent and stress are often used interchangeably. This is because the two notions are related. Both refer to the prominence of a sound, usually a vowel. This prominence is manifested in the vowel being pronounced louder, with higher pitch, or longer than the others around. We are going to distinguish accent and stress. Let’s begin by briefly introducing the three notions to be discussed here.

stress In every word, some vowels are stressed, others are not stressed. This is a lexically given property of words, which means it is not fully predictable (at least, the rules for locating stressed vowels are burdened with many exceptions). Some stress patterns are more common than others. Stress does not change: if a vowel is stessed, it remains stressed, if it is not stressed, it will not become stressed.Stress seems to move in word pairs like atom átəm~atomic əthómik. The members of such word pairs are related in their meaning, but not in their phonological shape. Note that only the m is stable here, both vowels and the t of atom are different in atomic. Let us call this property of stress Stress Stability. (Some function words, like is, can, for, them, seem to behave exceptionally in this respect, we will return to them below.)

accent The regular alternation of accented and unaccented vowels provides the rhythm of speech. Accent typically falls on a vowel that is stressed lexically, but unlike stress, accent is mobile. The same word may have an accent on different vowels depending on its environment: pontóon, but póntoon brídge. Some words may lose their accent altogether (or, alternatively, never acquire an accent).

tonic The pitch pattern of a given utterance, that is, the tune or melody of the utterance, is called intonation. It is characterized by the pitch change beginning with the most prominently stressed vowel of the utterance, which is its last accent, called the tonic. In this way the tonic, accent, and stress are related: the most characteristic feature of the tune of the utterance is located on the last accent, which is also the most prominently stressed vowel.

Vowel quality

Stress in English is related to vowel quality: a reduced vowel is unstressed, a nonreduced vowel (also called full vowel) is stressed. The problem is that the reduced vowels of English may also function as nonreduced vowels: they are a proper subset of all the vowels. So any vowel may occur stressed, and, as we have seen, i ij ə əw u uw may occur unstressed, too. That is, these six vowels are ambiguous with respect to stress.There are some hints: for example, word-final ə is always unstressed: stressed short vowels do not occur at the end of any word. In many cases, stress is marked in transcriptions. Compare, for example, Kennedy kénədij and Kentucky kentə́kij, which contain the same vowels, yet in the first one the first, in the second one both of the first two vowels are stressed.Some other examples: agree əgríj vs ugly ə́glij, today tədéj vs Monday mə́ndej. We do not mark stress on the first vowel of Kentucky for two reasons. One is that the vowel e can only occur stressed anyway. The other is that, as we are going to see below, it is not stress that is marked here, but accent. In many types of transcription, different symbols are used for the stressed and the unstressed version of a vowel. For example, the symbol ʌ is very often used for stressed ə and John Wells’s Longman Pronunciation Dictionary uses i for unstressed ij. Thus Kentucky is transcribed as kenˈtʌki (with the IPA convention of placing the stress mark not on the stressed vowel, but before the stressed syllable), while Kennedy as ˈkenədi. This would make stress marking redundant, at least in these two words:or in agree əgrɪj vs ugly ʌgli. a vowel is stressed unless it is reduced, like ə or i. For such a system to work, a pair of symbols must be found for each reduced vowel: one to go in unstressed and another in stressed syllables. But transcription systems that fully implement this idea are rare — and contain a lot of vowel symbols. To illustrate such a system, here is what Dwight Bolinger does for stressed and unstressed i: inhibit ɨnhibɨt (our inhíbit), for stressed and unstressed ə: conundrum kənʌndrəm (kənə́ndrəm), and for stressed and unstressed u: volume voljɵm (vóljum) vs woman wʊmən (wúmən).

Stress patterns

We will not use distinct symbols for stressed and unstressed vowels. Many vowels that are stressed are also accented, which is marked by an acúte accent mark. The vowels e a o and also the diphthongs beginning with these vowels, ej aj aw oj, are always stressed, as are all the long monophthongs (the R vowels), iː eː əː aː uː oː. Again, the vowels that are ambiguous when not accented are i ij ə əw u uw.

We will represent a stressed vowel by s (abbreviating not only ‘stressed’, but also ‘strong’) and an unstressed vowel by w (‘weak’). That is, a w vowel is one of i ij ə əw u uw, a s vowel may be any vowel. Many words have only one stressed vowel. With words up to four syllables, we have the following possibilities:

s

cat, flap, star, cut, hurt, stir, go, flu, key, strike, sphynx

sw

carrot, happen, ugly, ticket, follow, argue, talent, panda

ws

agree, forget, contain, July, taboo, ago, pretend, behind

sww

calendar, happening, uglier, orchestra, edible, family

wsw

agenda, horizon, vanilla, continue, Alaska, retorsion

swww

category, decorative, presidency, preferable, testimony

wsww

aquarium, basilica, photography, academy, America

There are two generalizations we can draw from the data above. One is that every word must have at least one stressed vowel, so the patterns w, ww, www, wwww do not exist, let us call this Stress Requirement. The other is that one of the first two vowels of a word must be stressed, so the patterns wws, wwsw, wwws do not exist. (Note that this latter generalization also excludes the patterns ww, www, wwww, so Stress Requirement is only needed to exclude the pattern w.)

The impossibility of any English word beginning with two unstressed vowels, *#ww, is called Early Stress Requirement. ESR states that stress must occur “early”, that is, on either the first or the second vowel of any word.

Stressless “words”

There do, however, seem to be some cases where a word occurs in a sentence without a stressed vowel, which would violate the Stress Requirement. Take the following examples:

Jane can swim well dʒéjn kən swím wél
Jack saw the cat dʒák sóː ðə kát

In these sentences the auxiliary can and the determiner the occur without stress. Such a stressless form is called a weak form. That is, the auxiliary can has a strong form, kan, and a weak form, kən. The strong form occurs when the auxiliary or the determiner is stressed. In spelling this is often marked by italics or capitals.

Jane can swim well dʒéjn kán swim wel
Jack saw the cat dʒák sóː ðíj kat

With kán the sentence means that someone said Jane could not swim well, but I assert that in fact she can. With ðíj the sentence means it’s not just any cat that Jack saw, it is the very cat that we all know about.

It is only function words (auxiliaries, pronouns, conjunctions, determiners, prepositions)The complementary category is content words: nouns, adjectives, and verbs. that may have a weak form, and not all of them. For example, the pronoun my, the auxiliary may, or the preposition on does not have a weak form. The verb and the noun can also do not have a weak form, they are always stressed: in Jack and Jill can beans or Jack puts beans in the can the word can can only be kan, it cannot be kən. Content words do not have a weak form.

The weak form of a function word is not a word: it does not occur on its own. It only occurs next to another word to which it is phonologically attached. So kən only occurs in a phrase like can swim, where it forms a phonological word together with swim. The auxiliaries is iz and will wil may join the preceding subject: Joan is ʤə́wn əz, or even ʤə́wnz, Ken will kén əl, or kén ḷ. (To show this, the spelling may be Joan’s and Ken’ll.) In this case a function word contains a reduced vowel, or not even that. The weak form of a function word is called a clitic, it cliticizes to an adjacent word, called its host. A proclicit cliticizes to the following word (the cat) and enclitic to the preceding word (Ken’ll). Note that a clitic and its host phonologically behave as a single word: Ken’ll is homophonous with kennel, Joan’s with Jones, a fence with offense, or some eyes with surmise.

Since the weak forms of function words are not words in the phonological sense, we can conclude that words do not exist without a stressed vowel. So the Stress Requirement indeed holds in English.

Multiply stressed words

It should be obvious by now that if an utterance consists of several words, it will contain several stressed vowels, at least as many as there are content words in it. This is because of two reasons we have already seen: (i) Stress Stability: lexically stressed vowels do not lose their stress, and (ii) the Stress Requirement: every word must contain a stressed vowel.

But often there are even more stresses in a sentence, since there are words that have more than one stressed vowel. The following list contains such words. Next to each word there is a sentence (not always very meaningful) that has the same stress pattern as the word before it.

ss

campaigned, Cam reigned

sws

contradict, Cohn’s a dick

swws

catamaran, Pat had a plan

swsw

Madagascar, Maddy asked her

swwswsw

telecommunication, Terry can fetch the children

swswsww

indivisibility, Indie Vickie’s billing me

The sentence Cam reigned contains two words and two stresses. But the single word campaigned also contains two stresses. The sentence Terry can fetch the children contains three content words, and as therefore expected, three stresses. The word telecommunication has the same stress pattern: this single word has three stressed vowels. And so on in the other examples, too.

Stresses are not equally prominent, either in a sentence, or in a single word. The last one is the strongest, the first one is weaker. If there is a third stress between the two, that one is even weaker. This holds for all of the above examples. As long as this is predictable, there is no reason to mark these differences. So we do not have to explicitly mark the last stress as the most prominent one, the first one as less prominent, and the potential middle one as even less prominent. As you can see above, all stresses are marked by an s in the examples.

Accents and the tonic

The most prominent stress in a sentence, which is called the tonic, is always the last accent, which is typically (but not always) the last stress. Take, for example, this sentence in a neutral reading (that is, a reading in which nothing is contrastive, nothing is emphasized):

św w ś w Św: Terry can fetch the CHILdren (neutral reading)

We mark accents by the acute accent mark: ś. Normally every stress is also an accent, but this is not always the case, as we will see below. The last accent, the tonic, is here marked by capitalization, Ś. (Note that this is again redundant, given that the last accent is the tonic, it is always the last ś that is capitalized. Capitalization only serves to ease your comprehension.)

There are some departures from this general rule. In these cases the accent is moved away from the last stress.

contrast and given information One reason for not having an accent on a stressed vowel is contrast, which can shift the tonic (and thereby the last accent) to an earlier (or sometimes later) vowel. Partly related to contrast is the case when given information, something that the speaker expects the listener to know, is deaccented.

compounds Another instance when the last stress is not accented is compound words, which usually have an accent on the first one of the two stems, while the second stem also carries stress as per the Stress Requirement.

clash Yet another reason for moving an accent to an earlier vowel is the closeness of another accent: accents are preferably evenly distributed. This is often referred to as stress clash (recall, stress and accent are often used as synonyms). In the terminology used here it is not stresses, but accents that clash, so we will call it accent clash.

lexical Finally, some words simply have a stress pattern such that their accent is never on their last stressed vowel, although they are not compounds and no contrast or accent clash is involved.

Let us see examples for each of the above departures from the norm.

Contrast

The tonic may be on practically any word, if that word is emphasized because it contrasts with some other possibility, which is clarified after the examples below.

1. Św w s w sw: TErry can fetch the children (not Jerry)
2. św Ś s w sw: Terry CAN fetch the children (sb said he couldn’t)
3. św w Ś w sw: Terry can FETCH the children (not stretch them)

When part of a word contrasts with the same part of another word, we can encounter a similar shift of the tonic, this time not only to an earlier vowel, as in sentence 1, but also to a later one, as in sentence 2 below.

1. She’s ACcepted (not excepted)
2. I said LeNIN (not Lennon)

Although normally Lenin lénin has stress on its first vowel and not on the second, and accept əksépt on its second and not on the first, this can be overridden when a confusion is likely to occur or has already occurred. Interestingly, in this case vowel reduction is also suppressed: while accepted and excepted may be homophonous: əkséptid, in the above case accepted is pronounced ákseptid. Similarly, both Lennon and Lenin may be lénən, but in the above example the Russian dictator is lenín and the Beatle would be lenón. As if the speaker was spelling out the words aloud.

Given information

Accent is moved away from words that count as already given information. A well-known example of this is the famous James Bond introduction. The vertical bars in line 2 separates two tone units, that is, in this case we have two tonics.

1. s Ś: James BOND.
2. Ś ‖ Ś s: BOND. JAMES Bond.

The normal pattern in introducing a name is in line 1: the tonic is on the family/​last name. But James Bond does this too differently: he only utters his family name first, and then the full name. When doing so Bond is already given information, so its accent is deleted, and as a result the tonic is moved to an earlier vowel, that of James.

This phenomenon is explained at length by Geoff Lindsey, with a followup based on the latest JB movie, and another one on superheroes.

Compounds

Many compound words also have the tonic on an earlier stressed vowel, not on the last one. Look at the following two sentences. (Note that the Early Stress Requirement does not hold for sentences: they may begin with more than one unstressed syllables.)

1. w w ś Ś: It’s a black BIRD
2. w w Ś s: It’s a BLACKbird

A blackbirdBoth of these stress patterns are possible with a neutral reading. The first one is about a bird that is black, the second about a specific type of bird, a blackbird. The spelling is also often (though not always) distinct: the compound word has no space between the two components, the phrase does. A compound word is regularly (though not always) stressed on its first stem. Note that there is no accent after the tonic, since the tonic is the last accent. So bird in sentence 2 above is stressed, but not accented.

Contrast may reverse these stress patterns, of course. Look at the following two sentences.

1. w w Ś s: It’s a BLACK bird (not a white one)
2. w w ś Ś: It’s a blackBIRD (not a blackboard)

The pronunciations of It’s a BLACK bird and It’s a BLACKbird, on the one hand, and of It’s a black BIRD and It’s a blackBIRD, on the oher, are identical, although they contain the compound blackbird and the phrase black bird, which are normally pronounced differently. So contrast may neutralize the difference between the compound and the phrase.

Accent clash

The phrase black bird is accented on its last vowel. In the larger phrase black bird’s nest the last stressed vowel is accented, but now the accent on bird and that on nest are “too close” to each other. To avoid this accent clash, accent shifts to the first vowel, that of black. Using our notation: sśś becomes śsś. It follows from this that the pronunciation of black bird’s nest is identical to that of blackbird’s nest. This means that the difference between the phrase black bírd and the compound bláckbird is again neutralized in the larger phrase black( )bird’s nest.

The result of accent shift is again that of three stresses the last one is the strongest, the first one is weaker, and the middle one is the weakest of the three.

Here are some other cases of accent shift resulting from accent clash.

s ś ś- → ś s ś-

thirteen men, sardine sandwich, left-hand drive

s w ś ś- → ś w s ś-

kangaroo park, Waterloo Station, four years older

s w ś w ś- → ś w s w ś-

academic year, fundamental error, chicken-hearted heroes

Needless to say, since accent may fall on a stressed vowel only, accent clash is not resolved when there is no earlier stressed vowel available, like in the following examples.

wś ś-

return ticket, lagoon paradise, the Bern ticket

More multiply stressed words

We have seen that compound words often have a stressed vowel after their last accent. Here are some more examples.

śs

airport, toothpaste, limestone

śws

basketball, honeydew, thunderstorm

śsw

grasshopper, hamburger, bookkeeper

śwsw

superpower, butterfinger, watercolour

It is not only compounds,Etymologically hamburger is not a compound word, but it behaves like one. but also one-morpheme words that may have the same pattern. Here are some examples.

śs

robot, ally, decade

śws

apricot, satisfy, renegade

śsw

ancestor, Chewbacca, orgasm

śwsw

alligator, pomegranate, Schwarzenegger

These words thus have stress patterns typical of compound words: robot is śs like rowboat, apricot is śws like chimneypot, ancestor is śsw like grasshopper, alligator is śwsw like superpower, etc. This means that the accent may be missing from the last stress of a word not only because of contrast/​given information, being part of a compound, or being pushed to the front because of an accent clash, but also lexically: some words just happen to have an unaccented stress after their accent.

Accent patterns

In the following we are going to look at all the possible patterns that two- and three-syllable words may exhibit.

Two-syllable words

Two-syllable words are of four types. Because of the Stress Requirement, at least one vowel must be stressed in any word. But it is also possible that both are stressed. In the latter case an accent may fall either on the first or the second vowel.

The following list give examples for each pattern together with the approximate percentage of the given pattern among all two-syllable words. So we see that more than half of these words are stressed on their first syllable and since their second vowel is unstressed, the accent is also on the first one (św). Almost a third of them also have initial accent but their second vowel is also stressed. The second stress is not normally accented in these words (śs).Accent may of course fall on the second vowel in case of a contrast: it’s TuesDAY, not Chewbacca. A smaller set have an unstressed vowel in their first syllable and stress on the second (wś), and there is an even smaller one with stress on both syllables, of which the second is accented in isolation (sś). (This is the group which is subject to accent shift.)

św (54%)

carry, clever, parrot, ceiling, value, distant, youngest, Adam, bishop, treason, pleasure

śs (30%)

robot, ally, decade, alcove, elbow, Tuesday, centaur, torment (noun)

wś (9%)

adopt, July, derive, taboo, agree, believe, canoe

sś (7%)

sardine, fourteen, maltreat, IQ, torment (verb), Chinese

Three-syllable words

Three-syllable words may have one, two, or three stresses. The last type is the least common.

One stressed vowel

We have already seen that three-syllable words containing only one stress are of two types, śww and wśw. These are also among the most common patterns. The third possibility, wwś is ruled out by the Early Stress Requirement

śww (28%)

animal, bachelor, possible

wśw (18%)

abandon, component, develop

wwś

—

Two stressed vowels

Words with two (or three) stresses exhaust all possibilities. As we can see in the frequencies below, the patterns śsw and sśw, which contain adjacent stresses are more common than swś, which does not. This shows that stresses may “clash”, it is only accents that must not. This also shows that stress is stable: we cannot get rid of a stress clash, since stressed vowels remain stressed and unstressed ones do not become stressed.

Accent shift is most common in words with the pattern swś (Lemonade Joe is śws ś). Curiously, sśw words rarely exhibit accent shift.

śws (20%)

apricot, catalogue, compensate

swś (6%)

avatar, lemonade, recommend

śsw (9%)

ancestor, Manchester, phantasm

sśw (11%)

Manhattan, September, uncertain

wśs (2%)

amortize, Beyoncé, molybdate

wsś (.07%)

divorcee, escapee, returnee

Three stressed vowels

This pattern is not very common, about 6% of all three-syllable words contain stress on all their vowels. Here are some examples.

śss (3%)

adumbrate, demarcate, zoophile

sśs (1%)

dehydrate, misconduct, tripartite

ssś (2%)

addressee, chimpanzee, Vietnam

Intonation

Vowels are not only distinguished by their relative prominence (loudness or length), but also by their pitch (the differences in the frequency of the sound wave, the height of the voice). This is called tone.

In so-called tone languages, tone can distinguish lexical items from each other. The sound string ma in Mandarin may carry four different tones, distinguishing four different words (source of audio).

1. mā ‘mother’ (level tone)
2. má ‘hemp’ (rising tone)
3. mǎ ‘horse’ (falling-rising tone)
4. mà ‘scold’ (falling tone)

The tone marks are iconic: they represent the pitch levels with their shape. So the symbol for rising tone moves upwards, the symbol for falling tone moves downwards, the symbol for falling-rising tone first moves down, then up, and the symbol for level tone stays on the same level.

English is not a tone language, here tones do not distinguish lexical words, but modalities, attitudes, completeness, noncompleteness. Such a language is called an intonation language.

Parts of the tone unit

Tone, that is, pitch change starts on the tonic, the last accent of a tone unit (often, though not necessarily a sentence). Thus the tonic is the most important, and only obligatory part of a tone unit. If there are syllables after the tonic, they constitute the tail. If there is an accent before the tonic, the part from this accent up to the tonic is the head. Anything before the head is the prehead. The tonic contains only one syllable, the other parts of a tone unit may contain more than one syllable. The head and the tonic contain (and begin with) an accent, the prehead and the tail do not contain any accented vowel.

Look at the following tone units. As before, we mark accents by the acute accent mark (two on digraphs) and the parts of the tone units are marked by colours. We assume the neutral reading of each tone unit, that is, the tonic is not moved due to contrast or given information.

tonic only

Gó.
Mé.
Tén.

prehead+tonic

Shall Í?
Did you gó?
The Políce.

tonic+tail

Húrry.
Públish it.
Kíll him.
Twénty.

prehead+tonic+tail

She’s slééping.
She míssed him.
The Béátles.
Tomórrow.

head+tonic

Jáne slééps.
Ámy slééps.
Ámy has slépt.
Ámy is her párents’ ónly chíld.
Thírtéén.
Síxty síx.
Twó thousand fíve hundred séventy níne.
Jámes Bónd.
Thómas Móóre.
Hénry the Éíghth.
Wáshington DĆ.

head+tonic+tail

Jáne’s slééping.
Jáne was slééping.
Jáne míssed him.
Síx twénty.
Twénty twénty.
Twó thousand fíve hundred séventy.
Bób Dýlan.
Dávid Bówie.
Quéén Elízabeth.
Évery Chrístian líon-hearted mán will shów you.

prehead+head+tonic

The chíldren sléép.
Chicágo díed.
Some partícipants have made sérious mistákes.
Toníght’s the níght.

prehead+head+tonic+tail

The chíldren are slééping.
The chíldren are slééping in the gárden.
Conécticut Wíldcats.
Patrícia took gárdening as a hóbby.

Types of tone

The most common types of tone are the following.

Falling tone/long fall

The falling tone (or long fall) is typical of completed statements (1), wh-questions (2), confirmation-seeking tag questions (3), exclamations (4). We use tone marks like for Mandarin above, the falling tone is represented by a downslope (technically a grave accent mark): `. Accents are here marked by underlining (because the acute accent mark will be used for rising tone). We use colours as above.

1. You always make such a `mess, John.
2. What can you see on the `blackboard?
3. That was a nasty sur ̌ prise ‖ `wasn’t it?
4. What a di`saster!

What looks odd in sentence 1 is the lack of an accent on the content word John. This is because this name here is a vocative (referring to the person addressed), and vocatives lack accent.

Sentence 3 contains two tone units, the second one is a confirmation-seeking tag question: the speaker expects the listener to agree with them.

High rising tone/long rise

The high rising tone (or long rise) is typical of polar (that is, yes-no) questions (1–3), echo questions (4–5), and information-seeking tag questions (6). It is marked by an unslope (technically an acute accent mark): ́ .

1. Is it you who made such a ́ mess, John?
2. Are you ́ ready?
3. You’re ́ ready?
4. A: Brmqxck has arrived to Brunswick. B:  ́ Who has arrived to Brunswick?
5. A: John has arrived to Brmqxck. B:  ́ Where has John arrived?
6. This is `Schubert, ‖ ́ isn’t it?

Sentences 2 and 3 show that a yes-no question is possible without subject–auxiliary inversion, the tone itself identifies an utterance as a question.

The second sentences in 4 and 5 are echo questions: B has not heard clearly who it was that has arrived and where John has arrived, so they want clarification. In this case the tonic is on the wh-word, and every accent after it is deleted.

The second tone unit in sentence 6 is an information-seeking tag question. The speaker is uncertain of the truth of what they have just said, and asks the listener if it was correct or not.

Low rising tone/short rise

The low rising tone (or short rise) is typical of uncertainty (1–2) and of parentheticals (3–4). This is marked by a lowered upslope: _/.

1. Maybe the baby’s a_/wake?
2. It looks very much like the _/other one…
3. To`day, ‖ says <sub>/</sub>Mary, ‖ we’ll `leave.
4. `Richard, ‖ often called the <sub>/</sub>Lionheart, ‖ returned from Je`rusalem.

Sentences 1 and 2 illustrate how the spelling struggles to indicate the uncertainty signalled by the low rising tone.

Falling-rising tone/fall-rise

The falling-rising tone (or fall-rise) is typical of incompleteness or non-finality (1–3), lists (4–5), personal opinions (6–7).

1. When we ̌ get there, ‖ we’ll have a good `meal.
2. I’d `buy you it, ‖ if I could a ̌ fford it.
3. I opened the ̌ door ‖ and `went in.
4. ̌ One, ‖ ̌ two, ‖ `three.
5. ̌ One, ‖ ̌ two, ‖ ̌ three…
6. John could ̌ do it for us.
7. He’s not ̌ that bad.

The falling tone at the end of the list in sentence 4 indicates that the list is complete, the speaker has stopped counting. In sentence 5, on the other hand, the fall-rise indicates that the list is not complete, further numbers may be expected, even if eventually they are never added.

show me the questions
let’s go back to the contents page