Chapter I I
Connected speech
^Connected speech 27!
■for the first part, but for the rest of the time it was pretty kuxed. t£d ourselves though. We had the car, so we were able to do some ?iNG, when it was too wet to go on the beach.
I I.I Accent
Connected speech, i.e. an utterance consisting of more than one ^i>n\ features of accentuation that are in many ways comparable with [|.(.,e jE* in polysyllabic words. Some parts of the connected utterance will \ -ugj, stand out from their environment, in the same way that certain   Iljhlcs i| polysyllabic word are more prominent than others. But accentuatio.i in ■ •rnrtftai speech differs from that in polysyllabic words because accenting i j , >hu«h3S* speech is determined largely by meaning in context. Nevertheleso .v;,m utof&rrt? are predisposed by their function in the language to be accented. TKm.    >t J? fc» words are typically main verbs, adverbs, conns, adjectives and U'uiopsirjt-y*' *F pronouns. Other categories of words, such as auxiliary verbs, (.■■njuin.l wip Kaf prepositions, pronouns, relative pronouns and articles (grammatical worte function words) are more likely to be unaccented, although they, too rM\ Ik «' exceptionally accented if the meaning requires it. •1 k\Jk
The meaning of any utterance is largely conditioned by the situation sitfifejiflto. context in which it occurs. So freedom of accentual patterning and ^ -eciallj SSJBb! the placement of the primary accent is always curtailed by the consn -in, un,—<ed tSimSti by the context. In the case of an opening remark, or when a new topic .> 11 rit. j|l^Hr duced into a conversation, there is very limited scope for variation of ii imiiil& wffi-wjg produced by accentuation. Rather more accentual freedom is possible in resource1, *?f3J|f thus, in response to the statement She came lost week, an incredulous i ■ ■> u.'i might have the pattern last weekl (i.e. 'Wasn't it the week before liu.t' l or Last week? (i.e. 'Don't you mean last month?'); or, in response to Wha> • •• Ifjj weather like?, the reply might be It rained every day (emphasising the cciw iu^| nature of the rain) or It rained every day! (where the fact of raining is emp ■-"■■:d). On the other hand much less accentual variation is possible in the MWttjjg dialogue (potential accents marked by capitals):
Did you have a good houdat? -811 Yes, very good Was the weather all right?
Wotds are usually unaccented (as noted above) and when unaccented Ify contain reduced vowels, e.g. can in /ksnju: 'kAm/.1 But most function final position keep a full vowel even when unaccented, e.g. can as in S/ Monosyllabic lexical words usually retain their full vowel value even ented positions, e.g. in We left the case in the hall, left, case and hall will ij^ir fall vowel wherever primary and secondary accents are placed.
»than one word in an utterance may receive a primary accent. A slow jar'eful sty'e °^ sPea'tmS °^en exfumts a proliferation of primary accents; .   caSUal or rapid delivery is likely to show fewer primary accents. In m-rided dialogue in a normal conversational style, the number of syllables '. ri, |UC£d vowels (or syllabic consonants) tends to exceed the number of those 'ju prominent by an accent or by the presence of a full vowel.
11.2 Prominence, accent and rhythm
■fe jiosf descriptions of English pronunciation over the last seventy years she :vion of 'stress-timing' is invoked to explain English rhythm;2 by such a ■■-.ry -stressed' syllables (including primary and secondary accents and other jv.l.ibles made prominent by 'stress'3 alone) govern the rhythm of English urci'ances, an equal amount of time being said to be taken between each a.p stressed syllables and between the last stressed syllable and the end of the unerancc, e.g.
They
However, all attempts to show such timing instrument-ally have been unsuccessful,4 it J such groups are often clearly far from isochronous (i.e. equal in duration). I "i example, in the above example the group containing the two-syllabled word '-.■iijr will be shorter than the three-syllabled groups couldn't have and chosen !■' llhough probably not in the ratio 2:3), while the group time for their (also lining three syllables) will be longer than all three if a full vowel is used it die word their.
The occurrence of full vowels5 generally predicts the rhythm of English rather re usefully than any notion of stress (besides variation of the type exemplified ihtive, there is often difficulty in deciding whether a syllable is 'stressed' or nci. when no pitch accent is present. Some might judge the full vowel on their 'ib >\ e as showing stress). For rhythmical purposes the reduced vowels are Is/,
couldn't have	chosen a	better	time for their	holiday
O   * •	• • •	9 •	#   • •	9 * *
272   Words and connected speech
Connected speech 273
v_.----« .^.i^nL ui ,U) auu i<ji \iw is regarded as a vaviar*
when they occur without a pitch accent; all other vowels are counte vowels. The one simple rule of English rhythm is the borrowing ruix*' a syllable with a reduced vowel 'borrows time' from any immediately p e syllable containing a full vowel. By the predictions of the Borrowing j;'^ vowelled syllables each take approximately an equal amount of time (a.th jSfc in practice this will be somewhat affected by the innate length of the v<-,. ^TSB the consonants in the syllable). Each syllable containing a reduced vov -i N j shorter and by the Borrowing Rule a full-vowelled syllable is itsoj f shorten ■Sr^ immediately followed by a syllable with a reduced vowel. Hspiftl
The operation of rhythm based on full-vowel timing and the Borrow p-r ba is illustrated in the following examples where a full stop marks rhe bour^fll between the syllables which are numbered and glossed as F (= Full \ovsc' R (= Reduced Vowel). Syllables with full vowels are long; those with     'W
vowels are short.
Sparrows aren't common /spa.rauz. cimt. knm.sn/ 1    2     3     4 5 F    F    F     F R
Syllables 1-3 each take approximately an equal amount of time. SyllabL 4 shortened because the following Syllable 5 'borrows' time from it, i.e. S-.liamsf? 4-5 together take approximately the same time as each of Syllables 1-3. C. 'npa^yN this with an analysis based on stress timing: there would be pitch accents, ow" /spa-/ and on fkvm-l which would count as rhythmical stresses; the two .cier-J stress groups reckoned to be equal in time would thus be /spa.raoz.anit andl /kom.an/ which seems exceedingly counter-intuitive. Alternatively stress "imimi' might consider lavoXl to be stressed, in which case the three equal groups; iu<d!d' be reckoned to be /spa.rauz/, /a:nt/ and /knm.an/; this seems to capture nr-it g! somewhat better but still leaves the two-syllabled /spa.rauz/ soundinp piiiui| longer than the other two groups. Here are some further examples:
Sparrows are plentiful /spa.rauz.a: plen.ti.ful/ 1   2   3    4   5 6 F F   F    F RR
In this example Syllables 1-3 are equal but Syllable 4 is shortened rx'i.auM%^| Syllable 5 borrows time from it, i.e. Syllables 4 and 5 are together approx nuccWl equal to each of Syllable 1-3. But only one syllable can borrow tii"^ --r-'S Syllable 6 just adds an additional short amount. This is unlike the timing pi «.i Ik by stress timing because additional reduced syllable(s) after the first oi c -i\£'_ no effect on syllables before it. ■
The dark blue pattern is the best i uf. Htrk bin: patn iz 3a best/ rf    2   3   4 5 6 7 8 s'     f   F  F R R R F
m ye j has a reduced vowel and is hence short. Syllables 2 and 3 have full "ga|s and are long. Syllable 4 has a full vowel but is shortened from long * ; use Syllable 5 with a reduced vowel borrows time from it. Syllables 6 and JKegch take the short time of syllables with reduced vowels. Finally Syllable 8 fell vowel and is long.7
»1,3 Weak forms
\i\ oal words (both monosyllables and polysyllables) generally retain their full .._n#els in connected speech and hence have a level of prominence above that of syl-i^tles w ith reduced vowels, even when no pitch prominence is associated with them. But many function words (pronouns, prepositions, auxiliary verbs, conjunctions, HftrtK1**) have different patterns according to whether they are unaccented (as is Hfa.jul) or accented (in special situations or when said in isolation). Compared "^h'i the accented (strong) forms, the unaccented weak forms of these words ^/^reductions of the length of sounds, centralisation of vowels towards /a,i,u/ { r the elision of vowels and consonants. The following list of examples presents ; the most common of these words, first in their unaccented (normal) weak form «■ lid secondly in their less usual8 accented strong form. More common weak forms f-fare aiven first (C = any consonant, V = any vowel):
:.-a ■ ■
am
an ■: and
any
(e.g. Got any money?
■are as at be
because
been
■.but
can (aux.) could •■do (aux.)
Unaccented Accented
hi /ei/
/am, m, m/ /am/
/an, n, n/ /an/
/an, an, n, n, and, nd/ /and/
C + /nil /eni/ Iqvt ni 'niAni/)
/a/ la:/
/az, z/ /az/
/at/ /at/
/bi/ /bi:/
/bikaz/, /bakaz/, /kaz/ /bjjenz/
/bin/ Mm/
C + /bat/, V + Ml + V /bAt/
C + /kan/, V + /kn/ /kan/
C + /kad/, V + /kd/ /kod/
/da/ + C, /dw, du/ + V /dui/
(e.g. do that Ida "oat/, do it /dwrt/)
274  Words and connected speech
does (aux.) Is, z, daz/
(e.g. What's lie do? /wots i: 'du:/, When's he arrive? Avenz i: a'raiv/) /of /fa/ from /fram, ftm, fm/
(e.g. The man from the .. . /5a man fin 5a/ had (aux.) /had, ad, d/
(e.g. m>/k» fe'<i done, what John had done) has (aux.) /haz, az, z, si
(e.g. What's he got, When's he got time?) have (aux.) /hav, av, v/
(e.g. What've yon done, I've done it) he /hi, i/
«tr /lis, 3:, a/
herself /ha'self, 3i'self, a'self/
him Am/ himself /im'self, iz'self/
(fe rf/rf /f himself /hi did it iz'self/) his kzl is Is, z!
(e.g. it's not /its 'not, he's coming /hi:z 'kAmirj/)
/fd/
just me must not of
(e.g. one of my
shall she
should Sir
some (adj.)9 than
that (cotvj. and rel. pron.) the
them , ____,
(e.g. Tell them to do it /tel am ta 'du: it/)
themselves /6am'selvz/
there (indef. adv.)'2 /3a/
(e.g. There were lots of them /5a wa 'tots a dam/
to (and into, onto, unto)      /ta/ + C
/tu, tw/ + V
us /as, s/
war Avaz/
/ctjas, djast/ /mi/
/mas, mast/ /nt n/ /av, v, a/ . /'wau a mai/)
/Ji/
C + /Jad/, V /sa/ /sam, srri/ /öan, ön/ /oat/
/öi/ + V,"/öa/ + C /6am, dm, am, m
m/
/friz/ Itzl
/djAst/
/mi:/
/rrust/
/nut/
AW
/JM/
/fbd/
/S3!/ /SAIll/
/5an/ /Öat/ /6i;/ /6em/
/Öem'selvz /&£:/ (rare)
/as/
AvDz/
year
yowwe// yourselves
Connected speech 275
/wi/ /wa/ /hu, u/13 /a!, ], 1/
C + /wad, ad/, V + /d/
/ju/
/ja/
/ja'self, ja'seivz/
/wi:/
/w3:/
/hu:/
Avil/
Avud/
/ju:/
/ja:, jua/
/jai'self, jafselvz/
particularly common uses of reduced forms involve auxiliary verb plus not. . ]V;[iary verb plus not are shown, for example, in the combinations he, she, it yii t and we, you, they + aren't (note also the question form aren't 1? /' a:nt ml); ■jrilarly wasn't, weren't, can't /ka:nt/, couldn't /kadnt/, doesn't, don't /daunt/ ■vii note don't know /dao'nau, da'nau/, hasn't, haven't, shan't /faint/, shouldn't v. ^ttrt/, 1*0» 't Avaunt/, wouldn't /'wadnt/. Additionally a final ltd may be lost before ■ >rd beginning with a vowel (sometimes with assimilation), e.g. /ka:n 'li:v/, &rtt Jai>/, Avaurj 'gao/. Reduced forms in pronoun plus auxiliary combinations are shown in I'm, he's, !, we're Avia/, you're /jo:/, they're /8e:/ and all subject pronouns plus will, y,ti„ld,have, had, e.g. /wi:l/, /ju:d/, /5eiv/, plus sequences like^'oa would have i lu'flavA Note also the question fonns: doyou /d3ii:/ or /d^a/, don't you /'daunlju/, s, . i'ou '"didju/, didn tyou f didnlju/, 'wouldyou /'wudju/, 'wouldn tyou /'wudntju/. Note also the mild imperative let us /lets/.
The only weak forms which can end sentences are those of pronouns. Thus f i\iliary verbs (and those main verb forms identical to auxiliaries) such as am, be. can, could, do, does, had, has, have, is, must, shall, was, were, will, t'uid retain a strong form when they occur finally even though they are unaccented, e.g. Who's got it? 'I have I'm hav/; he's not sure, but I am /hi.'znnt \o? bat 'ai am/.
Some prepositions, e.g. to, front, at, for, of, apart from having a strong form when receiving a primary accent, also keep a strong form when final and u.h~ccnted. e.g. Where have they gone to? (/tu:/, also /tu/, but not /ta/); Where's »;-• tome from? (/from/ rather than /fram/); What are you laughing for, at? (/fa:, a l. What were you thinking of? (AW). This applies, too, when prepositions <i'n- auxiliary verbs occur finally in a rhythmic group including at a 'deletion site' where the following item is understood, e.g. He looked at /at/ and solved .''/:' problem; or people who can afford to /tail (= 'do so'), buy luxuries, cf. People who can afford to Its/ buy luxuries, do so.
iume function words, not normally possessing an alternative weak form for ■ii scented occurrences, may show such reductions in rapid or casual speech, ?„- ^(/a/) don't know; I go by (lbs/) bus; Doyou know my (/ma/) brother?; for love nor (Ins/) money; two or (hi) three; ever so (/sa/) many; Scotland or (/ar/) England. These weak forms are often common only in a limited number of r i-ases, e.g. What are you doing? Avct a ja 'durrj/ or even Ivrvfs 'duirj/; and, in
276
Words and connected speech
Connected speech 277
the case of or, particularly occur in linking two numbers as in the exarnpl 'F In the case of the disyllables any, many, a qualitative prominence mav be \ \ ~^s§£g on the first syllable, i.e. /eni, meni/, but fully reduced, unaccented, for,,1" 8:,<^S be heard following a vowel in rapid speech, e.g. Have any more come \ '^^*=^ 'mo: kAm/; How many do you want? /hau mni d3u 'wrmt/. Other niono,),u>H^ function words normally retain their strong vowels in unaccented position' on, when, then, one, between, but again, although rather less commonly, ^.'Jjli vowel forms may be heard in rapid speech, especially when the word is „.|.J'^v^'| to a strongly accented syllable, e.g. What on (/an/ or /n/) earth!; When i ^"IIS*3 all's said and done; Then (/6an/) after a time; One (/wan/) always hopes; ' ... "^sM (/twin/) you and me. ' • I
The more rapid the speech the greater the tendency to reduction and (.jri,-Fi ■ I sation of unaccented words.14 Even monosyllabic lexical words may be rode,- "if*" in rapid speech, if they occur in a relatively unaccented positions ad i Lin a primary accent and especially if they contain a short vowel, e.g. t\i n ut ,t come back /"hi; 1 kam bak/ and lei in Don't get lost /daunt gat 'tast/. In n,j . 7""* may themselves be further reduced to hi, e.g. You sit over here /ju sat ixn.-       ' J He put it there /hi pat it '6s:/. The more prominent short vowels   ,-, .j^l only occasionally liable to reduction, e.g. /a/ in They all sat down on > \ i 1 /6ei o:l sat daon an 5a 'fto:/, M in We want to go /'wii want ta qao/. I niil\ -the diphthong 1ml, with its dominant central [a] element, is readily redt.l- Hi hi under weak accent, e.g. You can't go with him /ju kornt ga 'wi5 uu . nLj .1 going to do it /hi:z gana 'du: it/; / don't know /ai da'nau/ (the last twi- irt„-c 1 spelt gonna and dunno in the representation of rapid speech). j
I 1.4 Acquisition of rhythm and weak forms by 1 native learners
Such little evidence as there is suggests that some children often start oil" using the strong forms of function words. They also tend towards a constant -M length for each syllable and do not apply the Borrowing Rule (see §11.2 above), '.'j or, in more traditional terms, they have a syllable-timed rhythm.
I 1.5 Rhythm and weak forms-—advice to foreign learners
Rhythmical shortening of full vowels occurring before /a,i/ should be attended to: J such shortenings can be practised in pairs like short vs shorter, lead vs h tiding, % bus vs buses, wet vs wetted, John vs John looked ill, one vs one for tea. John ■ vs John '11 go etc, Those with a syllable-timed LI like Cantonese, French I. inJ . Italian, Spanish and Bantu languages, must give particular attention to sucti * shortenings.
Learners who aim at a native English accent (British or American) must learn 5 * the weak forms of function words and regard them as the regular pronunc i -l i using the strong forms only on those limited occasions where they are us.. -
-cial emphasis or contrast and in final positions). The reduction to hi ' "ords will not automatically follow from the teaching of rhythm. Even 1'"learners often do not use as many weak forms as native speakers.
l_6 Intonation
16,17
lip
it is
,.   Jkjjstic manifestation of intonation is fundamental frequency (see §3.2.1) tiLri ii perceived by listeners as pitch. Pitch changes in English have three LA I functions: (i) they signal the division of utterances into intonational 5gg: (besides pitch change, other phonetic cues often mark such boundaries, srticular, pause, final syllable lengthening and changes in the speed with i ^accented syllables are produced)—boundaries between intonational tP-rn ■. ..enerally correspond syntactically with clause and major syntactic phrase in;.rid. :ws (see further in §11.6.1.1 below); (ii)they signal syllables with primary idary accent, both in the citation of isolated words as already mentioned 1-10.2 and in the longer utterances of speech; (iii) the shape of the 1lii."> produced by pitch changes can carry various types of meaning, primarily rl'sL-vi!1 il (i.e. establishing the links between various parts of utterances) and in uni'Ml; particularly important is the pitch pattern beginning at the primary d ending at the end of the intonational phrase—often called the nuclear bould be noted that, while the variation in intonation between languages between dialects of English) is not as great as that involved in segments, nonetheless sufficient to cause a strong foreign accent and in some cases to misunderstanding. The intonation of GB is described in the following sections. Differences between GB and GA are relatively limited; differences Svrwcen GB and that of a number of northern British cities are considerable (see ■ i Jer §11.6.3 below).
11.6.1 The forms of intonation18
i:.6.1.1 Intonational phrases
l"he boundaries between intonational phrases may be indicated by a combination of internal and external factors. Most obvious among the external ones is pause: in the following example pauses can occur at the points where boundaries are .■: . icaled by / (we omit standard punctuation marks but continue to use capitals at [he beginning of sentences):
In the past five years / the way that services are delivered to the public / from both state enterprises / and private companies / has changed almost out of recognition / If we wish to make an enquiry by telephone / we have to choose between a number of options / and then between a further series of options / and so on / Even after this series of choices / we may have to listen to canned music / for a short time / or a long time / or a very long
278   Words and connected speech
time / So we may ring off and try the internet / and look website / only to be told / that if we want more information / v ring the number we have already tried.
Often, as an alternative to pause, speakers may lengthen the final syllab » v the boundary: in the piece above, for example, years, -ic, -ses, -nics . may be lengthened (such lengthening can apply both to accented and m , - , syllables and to full and reduced vowels). A boundary can also be mark? ■ "" increase in the speed of unaccented syllables following the boundar example, the intonationaf phrase beginnings and then be- and that if we art to be pronounced very rapidly and hence such syllables are also very i! -.1 , involve reduced vowels. These cues to boundaries of intonational priu.e'j JL>' not unambiguous; pause and final syllable lengthening may also be \..a hesitations, for example when a speaker has a word-finding difficulty (1i below). The 'external' cues to boundaries are supported by internal f t ,-,r particular (i) if one of the pitch patterns associated with a nuclear ton^ 1 pleted at a certain point this in itself may indicate a boundary (see ; 11 S J i below) and (ii) a jump up in the pitch height of unaccented syllables will -'.jr.»n,J».. ally only occur at boundaries. Thus the syllables So we may above an. n .. ci'\' likely to be said at a rapid tempo but will be said at a higher level than i u. ija* of the preceding very long time. This is part of the tendency for ur-in.it oril phrases to be susceptible to a declination effect, i.e. to decline in pircli imgj -. their beginning to their end, so that what are felt to be low-pitched syllah eN a1 the beginning of an intonational phrase will in fact be higher than low-pi uast ■ syllables at the end.
11.6.1.2 Primary accents
The pronunciation of single words and of longer intonational phrases ar described in terms of an obligatory primary accent and an optional secc accent. The realisation of primary accent has already been discussed in § relation to single words. There it was stated that the final pitch accent in a is usually the most prominent (and hence is referred to as the primary t while a pitch accent on an earlier syllable is referred to as a secondary acceu same sequence of secondary accents and primary accents applies to intom phrases. The final pitch accent identifies the syllable which is called the ni and begins one of a number of pitch patterns known as nuclear tones.
/1.6.1.3 Types of nuclear tone
(1) Falling nuclear tones (',)—A falling glide may start from the highes of the speaking voice and fall to the lowest pitch, marked 1 (a high r from a mid pitch to the lowest pitch, marked , (a low fall). When are high syllables before the primary accent, a high fall will involve a si
Connected speech 279
*"  u, low fall a step-down. The falling glide is most perceptible when it \?hce on a syllable containing a long vowel or diphthong or a sonorant '>C yrrUg-1*^) Creates a hi§h-level secondary accent—see §11.6.1.4 ■^low), e.g.'
He 'couldn't be seen.   He was running.
tvhm a fall occurs on a syllable containing a short vowel followed by a ^celess consonant (especially the plosives /p,t,k/), the glide is often truncated and so rapid that it is not easily perceptible, e.g.
'What have you ~got?       'What have you ,got?
u-hen svllables, collectively called the tail, follow the nucleus, the fall may be spread over a relatively high pitch on the nuclear syllable and low pitches on uV syllables of the tail, e.g.
hoibi
Sill
■lord :cent) Trie
■
hi : or -h.-rc :d it
I'd for1 gotten    It was ^yesterday
•) Rising nuclear tones (',)—Rising glides may extend from low to mid, or from low or mid to high. When the rise ends at a high point, it is marked bv ' (a high rise); when it ends at a mid point, it is marked by , (a low rise). iQsino Elides are more easily perceptible when they occur on a syllable containing a long vowel or diphthong or a sonorant, e.g.
'Can you 'come?      He's not Jll.
When a rise occurs on short syllables, particularly where the vowel is followed by a voiceless consonant (especially the plosives /p,t,k/), it must necessarily be accomplished much more rapidly, e.g.
'Can she 'cook?
280   Words and connected speech
With a tail, the rise is achieved by means of a lower pitch on th syllable with an ascending scale on the following syllables, e <■ ,,Uc
'Is it raining?       'Are you 'comfortable?
When the tail is a long one, the ascending sequence of syllables ol .-. f-j.j. may be interrupted by a middle level plateau before a final upwa J '
O
Are you 'going to London tomorrow?
(3) Falling-rising nuclear tones (fall-rise) (")—The fall and rise may he ^W" fined within one syllable, the glide beginning at about mid level aid cidi!ft«.dlfe». at the same level (or slightly above or below); in the case of a shoil '[ fcu -^T' the dip in pitch is made extremely rapidly, e.g. . TJ-*»
It's true.
It's shut.
When unaccented syllables follow the nucleus, the fall occurs on the ■ icl^a." syllable and the rise is spread over the tail (^indicates a falling si ."■>,idd1y accent—see the section on secondary accents in §11.6.1.4 below),
Connected speech 281
'You can see.
'John didn't come.
,ffl„ a fall-rise is accompanied by an added initial rise, giving a msb-S!f'^sE variant of the tone, e.g. _
"Nothing's impossible.      'You can see.
ilin   a fall mav be reinforced by an
^^^^^ - » *«* fX*™* be given extra length), e.g.
_!_£L
He wasn't alone.
It was raining.
* ris,faU on a short sy.lable followed by a tail may be realised as a low iSrfBucto syllable followed by a fall on the tad, e.g.
It's "raining.      It's 'quite "comfortable.      It's snot a "cafr-rpiilar
When full vowels occur in tails, the fall takes place on the nuclear \\ rn'c
and the rise is initiated on the last syllable carrying a full vowel. W . -
fall and the rise are on separate words, the fall-rise is indicated i f i■ f mark followed by a rise mark (i.e. ' and ,), e.g.
He x didn't "telephone.
O
He's educated
You'd "better.        It was "yesterday.
(5) Level nvclear tones (*)—The most common level tone is a mid level, which is a very frequent tone in intonational phrases which are non-final in a - ■■■■sentence. If it occurs on a single syllable that syllable will be lengthened; *.,■■.:. if a tail follows the nucleus, then the unaccented syllables remain on the same level, e.g.
o o
9 ©
o o
■^^P^rk. On my 'way to ^Manchester 11.6.1A Secondary accents
K secondary accent occurs in the pre-nuclear ^ !. usually involves a high-level pitch prommence, marked [ J e.g.
282   Words and connected speech
'When do you want it ?    We waited 'twenty minute""
More than one secondary accent may occur in the pre-nuclear posiu,, with a series of ['], To achieve prominence each succeeding seci'-H ^^Hll? involves a slightly lower level, e.g. =        JP' a<-eeti@S
It 'all 'went ,wrong.   He 'didn't 'under'stand a~tfij7,;»
A variant of high level(s) uses one or a series of glides-down, marled \ than levels, e.g.
® A.
. • ^ v Q •
I^didn't ^want "that to happen
Glides-down of this sort are more prominent than steps.
A pitch prominence (and hence a secondary accent) may also iclunuf by a step down to low pitch(es), marked [,], following initial hig'i i n, united syllables, e.g. ~"l
-Q-......Q
He didn't know how to "do it
Sometimes the low-level pitch is replaced by a glide-up, marked f.,[. I i,s s particularly common before falling nuclear tones, e.g.
Connected speech 283
clear—Unaccented syllables occurring before a nucleus (where there '"",'lecondary accent) are normally relatively low, whether the nucleus is
There were 'two of them.      And when we a rrived..
accented syllables before a secondary accent are also usually said on a dainty low pitch, the accent having prominence in relation to them, e.g.
® «
9   • •
He's 'always ,late.
It was 'after dinner..
if pre-nuclear unaccented syllables, their weak quality remaining, are said on a relatively high pitch, the effect is more emphatic and animated than if they are low in pitch, particularly if they are followed by a low nuclear tone, e.g.
It is.
There were nine of us.
\ i.| as mentioned under §11.6.1.4 above, high unaccented syllables may sometimes be used to give pitch prominence to a low accented syllable,20 e.g.
We .don't like that.
O
He ^didn't like that at "all
/1.6.1.5 The pitch of unaccented syllables
Unaccented syllables, in addition to the fact that they are said very quic u> zri
usually undergo some reduction, do not normally have any pilch pj......,at.i'
They may occur before the first accent (primary or secondary), between iinriu, or after the last (primary) accent (the nucleus).
12) After syllables with secondary accent—Unaccented syllables usually remain on almost the same pitch as a preceding syllable with secondary accent, e.g.
'All of us 'wanted to help him.
'Put it 'on the table.
'Will you be 'able to come to,morrow?
284   Words and connected speech
Addmonal prornmence ,s often given to secondary accen, , 1 descending rather than a level series of unaccented 7vliah,    - ^ is part.cularly common before a fall-rise nuclea tone e "''^
IS
You N can't do it "that way.
(3) Port-wrfeo^-Unaccented syllables following a falling nilrI a low level, e.g. g mitleuj
He's a "bachelor.
There are aJministaS^diffl^gT
After a rising nucleus, unaccented syllables continue (or effec-. ,k " . - J similarly the nse of a falling-rising nucleus may be spread oveS t unaccented syllables, e.g. P r m*
« Q •
'Put it'on the table.        It's not impossible.
-* O • «
SHU
4p* **
Has he got a solicitor?        He Moesn't Mike "criticism. 11,6.2 The functions of intonation ,
11.6.2.1 International phrasing Tu» Most commonly intonational phrases correspond with clauses. The clv.iv? may constitute a simple sentence or it may be part of a compound or coni|'lc\ uiu ']JfiH (a forward slash / marks the boundary between two international phrase-; e c He usually comes at ten o'clock. He worked hard I and passed the exam.
Because he worked hard / he passed the exam. », m
It's nice / isn't it?
But often intonational phrases correspond with smaller syntactic constituent l'u'l the clause. The subject of a clause may receive a separate intonational phrase cif its own, e.g.
Connected speech^ MS;
U^e workers / have got a rising standard of living. \' .repetitive society / is defensible. |0f industry's profits / go in taxation.
(>ce adverbials and adverbials of time and place often receive separate JJ^tional phrases, particularly when they occur at the beginnings and ends jjpises, e.g.
,1 £to to London / regularly, g-r   i ~,c government's got to give in / apparently / to every pressure from, the City.
ny view / the argument should be / how to build a partnership / between : public and private sectors.
-.Seriously / it seems to me / that the crucial issue / is . , .
^,-ln > p rare cases is it obligatory for subjects and sentence adverbials to take a . intonational phrase; but it is an option which is often taken up, particularly
fi-tj \ .R J esubject is long or its status as a new topic is highlighted, or when the mij,ji wishes to make an adverbial prominent.
-  ij.uerj less common, types of structure which commonly form a separate jatcnattonal phrase are parentheticals (including vocatives and appositives) and lei constructions, e.g.
Lacy / will you please stop making that noise.
•rofessor Bull / the Head of the Department / declared his support.
lata'/ this will really amaze you / actually got the highest marks.
rhis will be achieved by hard work / by brainpower / by interactive subtlety
■ ■ I and by keeping to deadlines.
\i ■i-restrictive relative clauses, which are of course semanticaUy similar to parentheticals, also regularly take a separate phrase, while restrictive relatives io not, e.g.
The old man / who was clearly very upset / denied the charge. The man who appeared in the dock / looked very ill.
Although there are tendencies, and some obligations, in the assignment of inton-itional phrases, there remains considerable flexibility. Where clauses are short, hey may be combined into one group, e.g.
'I don't think he will/
While subjects are often separate phrases, objects are generally not. Nevertheless •I fronted object or an object in a parallel structure may be so phrased, e.g.
This / you really ought to see.
I like him / but I loathe / and detest / his wife.
«4*1
286   Words and connected speech
Connected speech 287
Besides the probabilistic correlations with grammatical units then- i "-3 be a length constraint; stadies have suggested that in conversation"-'-'!'0' around half the international phrases will be 3-4 words in |en, .'■ 111 under 10 per cent of cases will they be over 8 words in length21 I Vn."'!*' from prepared texts, intonational phrases are likely to be longc ■ ., i to be at least partly governed by punctuation. '' w'i
I 1.6.2.2 Primary accents and new information
In previous sections intonational phrases were said to have one -u n r\ (= nucleus), at which point begins one of a number of nuclua t.-general terms the nucleus falls on the most prominent syllable (ain ^ prominent word) in an intonational phrase. In more particular term- *n ■nuri marks the end of the new information. Old (sometimes referre I i ■„ information is that which has either been mentioned before hi t'ru. nn** intonational phrases or which is in the listener's consciousness I '(.ji|3= „ presence in the surrounding physical environment.
Sometimes intonational phrases consist wholly of new information ». such phrases occur out-of-the-blue or in response to 'What happened ' lr where the intonational phrase is wholly new the nucleus falls on rtia relent syllable of the last lexical item22 (lexical item here means nouns: verb., jdieuMH adverbs and the word 'item' rather than 'word' has been used because co'iMjj^i lexical phrases like wind up and child abuse are involved), e.g.
Jane's had a 'baby.
Something happened on Sunday which was quite un'usual. I don't really want to af tend. He was accused of'dividend stripping.
There are some exceptions to the rule of the last lexical item. One grou- i ■! etc^p-i tions concerns intonational phrases having an intransitive verb or ieib phe&Ss„ whose subject is non-human or which loosely involves (dis)appearar
That "building's falling down. A 'doberman's on the prowl. The "dog barked, (cf. The man 'swore.) I heard a "bird sing.
Another group of exceptions concerns certain types of adverbial in fii'.il r»'--i 'u1'- | Sentence adverbials (i.e. those which modify the whole sentence) and ,-.K-rtWj| of time usually do not take the nucleus in this position, e.g.
I go to'Manchester usually. It wasn't a very nice 'day unfortunately. There's been a 'mix-up possibly. He didn't sue'eeed however.
•   in some cases to having the adverb at the end of a sentence with-''""^'■nt is 10 divide the sentence into two intonational phrases with the : *" ''\'ao a separate phrase on its own, e.g.
I. 1
|0- Manchester / .usually, f..- a very nice 'day / un fortunately.
" • ■sentence adverbials which are usually classified as conjuncts, e.g. incidentally, »   cannot take a sole nucleus in this way, but must have a separate nucleus -^ir own or are non-nuclear, i.e.
t goto Manchester incidentally, (not possible) (net'dentally I go to Manchester, (not possible) inci'dentally / I "go to Manchester, (possible) [ went to "Manchester incidentally, (possible)
-e other types of expression, which are similar to adverbials in that they are ,j nature of afterthoughts, are also common in final position with no accent; [ example, vocatives and direct speech markers, e.g.
■Don't you a'gree, Peter? Don't be a 'fool, he said.
£Vit i i4d information occurs at the end of the sentence, then this will be unac-Bfed e.g.
i V, -iy don't you invite John to the party?) rfivause 1 don't "like John.
.ttehad a long 'wait.)
Yuit.mean we had a 'very long wait.
jjr f~e last example there is obviously some element of contrast present—between rUig and very long. Sometimes the nucleus may fall on a contrasted item even f i.' l a later item in the intonational phrase is new, e.g.
John is quite a tall man / but his brother's very 'short.
h "ertaui, very limited, cases, the whole of an intonational phrase comprises old ■-.ion lation. One such case concerns echoes, i.e. where a second speaker echoes jething a first speaker has just said; and the accentuation of the second speaker ■ I follow that of the first, e.g.
(I couldn't 'do it). You couldn't 'do it?
(This time we went to Ireland). Oh you went to "Ireland / ,did you?
- u
288  Words and connected speech
i 1.6.2.3 Primary accents on function words
At the beginning of this section it was noted that the p n falls on a lexical item (norms, verbs, adjectives, adverbs i lncr some special cases when the primary accent falls on a fun> i,. t H'Jjj commonly applies to auxiliary verbs and prepositions. Sometimes word highlights a contrast of tense or negation (sometimes th presupposed):
(Why haven't you 'phoned me?) But 1 'have phoned you. (Why don't you 'love me?) But I 'do love you. i (I wish I was in "England.) But we 'are in England. s (This book ought to be re'printed.) But it's 'been reprinted i
Preposition accenting is another way of giving special emphasis '| h=n an alternative accenting which is much less emphatic; in the follcm-Ly e* compare the accents on to, or, with and of with more nninatk.<>l .iccuS
sensible, name, belongings and many:
(Why are you doing 'that?) It's the only sensible thing 'to do. ! 'Acted a lot / ,did you? / Can't say I remember your 'face / 'oi jourl [Announcement] Please make .sure /you have all your belong ujfc The mistakes you made were not "big / but there were a great ,-i.ir. ot (I a'greed with the decision.) But do you accept that there wort ^ aspects 'to the case?
In question word questions extra emphasis (showing greater interesl cr iirt is often produced by putting the nucleus on the verb be or an auxiliar. \c-
(We can't let "that happen.) What 'should be done? (That's a very expensive 'ring.) How much 'is it? (I expect the 'family to come to night.) How 'is your wife? j [Thinking you recognise someone] Who 'is that man over ther J
Connected speech 289
,ei,t.-(arun / 'too. jtarian / 'too. tapered "too / that no 'women were coming, ig^fiirm's going too. Jpfight:;ioo.
ijpyoti live in that area / ,too? fees 'too they do that, i". $i« area 'too he showed his mettle, ifcn -oo agreed to come. Hit a vegetarian / as 'well.
also-occurs in the verb phrases might/may as well where it can be yclear
jJIj rltgftt as well have done it 'anyway, y iniv as well finish 'now. itet too also functions as an intensifier as in too hard)
£ is used rather than too following a negative:
11.6.2.4 Focusing adverbs
There are a number of focusing particles (i.e. those which focus m 'ij iglR-parricular words or group of words) which govern where the nucleus i^urs, either taking the nucleus themselves or projecting it onto a following re n
(1) Nucleus on too and as well in their meaning 'in addition'
Too with the meaning 'in addition' takes the nucleus and o:> ir tikt- i separate intonational phrase (IP), particularly when final in the I:1 ' 11> '"*rl IP, too will copy the tone of the preceding IP. As well (with virtua i- rl . -i-t'i-S meaning) also occurs in final position and must take a nucleus
[ m not a vegetarian / 'either, not a vegetarian / 'either. j&tE that either may also occur as a conjunction, e.g. Either you can do 'this I
can do "that) fi'Even, not even (inclusive adverbs)
- Even focuses on all that occurs between even and a following nucleus (which may be immediate or at a distance):
Even 'John > Not. even 'John agreed. He didn't even say goodbye. ■ He even pawned his wife's 'jewels. : Not even a hundred 'pounds would make me do that. He didn't even say he was 'sorry. I'm not even 'slightly amused.
I: Not. .. at all (emphatic negative) Not... at all can be non-nuclear or nuclear but, if it is nuclear, the nucleus
: falls on all rather than not:
It's not like that at all. : He's not at all friendly.
290   Words and connected speech
Connected speech 291
(5) Only, not only, if only, only because (as restricting Only and only phrases are commonly non-nuclear an | scope of focus stretches from the only up to the nucl -■■ ' *at<*"' Only 'three.
Only 1 know how to do that, " Only ten of them turned 'up. \ He only did it for 'your sake. It's not only'John who said that.
If only we'd thought of that be'fore. a I'll a'gree / but only because it's you that's putting it fom^, i' On the other hand only can be nuclear and backward-looking. =■ , We were allowed two tickets'only, '
(6) Also (additive) * Like only, also is commonly non-nuclear and the scope c. rntUl>, up to the nucleus whereas a nucleus on also is backwar.l-mukiy focuses on the matter that precedes:
He's also written to the 'chairman. Also under scrutiny was the 'president of the company. He also passed the e'xam. He 'also passed the exam.
(7) Enough (reinforcing adverb or adjective) In initial position enough often follows a sentence adverb !"-t>o-.* cn i strangely, curiously and oddly) and takes the nucleus, whirca.-, Ji positions it is usually non-nuclear:
Strangely e nough / it doesn't 'work like that. Oddly e"nough /1 a'gree with you. He wasn't big enough to 'reach it. He be'haves nicely enough. There wasn't enough 'space.
(Note that enough may also be a pronoun, I've had "more than enough, ffi e"nough.)
I1.6.2.5 Some special accentings _
In cleft sentences the nucleus may fall on the focused item but it may also . post- or pre-focally, cf.
It was 'Tom who suggested it. It was Tom who su'ggested it. It's 'always Tom who suggests things.
Ir"     r Mousing puts the nucleus on a pronoun, e.g.
v said that they planted the bomb. *       ,ud 'it was responsible.
inouns are commonly nuclear:
s on my OTke shift them'selves. * *   the car that sells it'self.
■ 3 number of generally vague words (which may nevertheless be precise £it ,'u|gf context) which commonly take the primary accent when it might ^->r 1 to be placed on a more specific (preceding) word. Chief among these
,e.g-
faant-as a regular "thing. g;t ji.iir 'things.
* i„i<.i'tirtg also shows up in the phrases that's the "thing, a close-run "thing, iright "thing. Similar accenting commonly occurs on stuff, matter, person, tre<mdplace in phrases like where s my 'stuff, what's the 'matter, she's «ynice 'person.
jjiinal numerals are also commonly accented and take precedence over the ue words just mentioned, e.g. the 'first thing to do, the 'next time you come.
f:-4 2;6 The meanings of tones
nost all primary accents in words and longer utterances have up to now been !S*n pi fied using the high fall nuclear tone (marked '). This is the way in which -d > .-ual words are usually cited in isolation. Moreover, in all styles of English ipit\!i simple falls to pitch (whether from a high or mid starting-point) account for the majority of nuclear tones (generally estimated around 50 per cent).25 ftnplc rises and fall-rises are generally estimated to account for a further 40 per
Eat of tones. The preponderance of falls is usually slightly higher in conversation r in i.iher types of speech, e.g. in scripted reading. Since rises and fall-rises ,h. hni used as a cohesive device signalling more to follow, it is not surprising , -ii. In-, are more frequently used in reading where they will often indicate that
-- m e is not yet finished. l The meanings of nuclear tones are sometimes more discoursal in nature (e.g. fcthey indicate links or the absence of links between successive intonational j phrases), sometimes more attitudinal (e.g. they indicate the speaker's doubt or •■■certain >v about what he is saying) and sometimes more semantic (e.g. they co-■ occip 'vt1! lexical meanings which are reinforcing or limiting—this is particularly ' toe case with adverbials). In general the meanings of tones are not directly '-Era-rnnical, but grammar may indirectly be involved in two ways: (i) some
Connected speech 293
292  Words and connected speech
attitudes are inherently more associated with questions; in pari1(. |r which often has a meaning of surprise, frequently marks an e< In, ^ ' also previous section); and (ii) the attitudinal and discoursal mi niri--"* will vary somewhat according to the syntactic sentence-type (e.g lll j *C', interrogative, yes/no-interrogative) with which the intonational -n- ^H^* Because of the variation in meaning according to sentence-typ. pi-iV, r the description of the meanings of nuclear tones24 which tb'iiw^ according to the following categories: (1) major declaratives, (_. M,' tives, (3) yes/no-interrogatives, (4) wh-interrogatives, (5) h .._nl
oaä
acta
(6) imperatives, (7) exclamatives, (8) social formulas. In genet.' h,|.n tones (whether .low fall, 'high fall, or "rise-fall) are separative, iti3\\r. Jt assertive; whereas rlow rise, 'high rise and "fall-rise are continu nv „■ hd!1 and non-assertive. Level tones (most common among these bei        i j<lj belong with the rising tones in the sorts of meanings they con-..'
The examples are given as isolated utterances or precede!1 i.i i,,^ 'setting'. It should be remembered that the attitudinal meaninj ..[ ai ,Jt) must always be interpreted within a context, both of the situation and ,■ speaker's personality. It may well happen that an intonation v I ic \ iv one set of circumstances might, for instance, be offensive or par .i.ts: used by another person or in other circumstances.
»Ittel
(1) Major declaratives
Major declarative refers to those cases where the intonat on. I phiw: i relates with an independent clause, with the main clause in complo- enters with the last clause in compound sentences, and with th.ii r u ul au> ffAn these which is remaindered when a separate intonational pi i    ■, MM.ateft-' an adverbial or a subject or some other part of the clause,. < ,v&.j
He didn't go.
I took an overcoat I because it was raining. I took the car / and drove to London. The first man on the moon / was Neil Armstrong. Usually / we do it this way. We do it this way I usually.
In major declaratives falling tones are the least attitudinally maiU.: . 1 "le "ries; with the high fall expressing more liveliness and involvement th?i ill ■ I \a ill
It's a very nice 'garden, Of 'course it is. It's a very dull ,book. The parcel arrived on , Thursday.
Fall-rise is common on major declaratives with a variety ■ I particular, reservation, contradiction, contrast and warning as ii
I L Mil i""1-
flflHT
EL.. •„ 'wife / even if I don't like him. & the twenty fifth today, isn't it?) Twenty "sixth. - j-,': succeed / but "Philip did. H^iii don't do it / JohnTl be very "cross.
are common on echoes (as already mentioned at the end of the previ-
.'" n5 and on declarative questions: IP"?1"
1*. ,,j jt in blue.) You did it in 'blue? didn't go?
fj'tones are less common. The low rise with only other low syllables before ^- with no preceding pitch accent) is complaining:
IVoU mustn't S° a,way-
ttesreas- with a high pitch before it, it is encouraging or even patronising (this Bfirice is very common in speech to children):
*i¥m»'U'only over.do things, frtere's 'no point in .rushing. fi"jb sequence is frequent on imperatives (see below) with a similar sort of JSiKa.rig. Finally, the least common nuclear tone25 is the rise-fall. Its meaning Ifisii.-iiy involves an element of being very impressed, or, conversely, being very f§5?inpressed and hence indignant or even sarcastic:
IIj % the head of a big firm in "London. Olvm'deed / How "nice for you.
£j| R -^-ft'l is often used for gossip: -   Have you heard? / Jill's "pregnant.
.'I M-nor declaratives
Under minor declaratives are included all those parts of declarative sentences ,., which were excluded under (1) above. Most of these occur in sentence
t.......non-final positions, e.g. subjects, adverbials, the first clause of compound
sentences and often the subordinate clause of complex sentences. Th* tones used on these intonational phrases are usually from the rising group: All-rise, low rise and mid level. Fall-rise again carries its common meaning of'contrast'. The difference between the other two tones in non-final position has to.do with style: low rise is the most oratorical and is also typical of read-IBS aloud, whereas the mid level carries no other meaning other than that of ncn-fmaiity, which is perhaps why it alone of these three tones occurs only in son-iirmi position:
294  Words and connected speech
Will
What I'd "like / is a drink of tea. The "best person to do it / wotdd be Bill Bailey. The "crucial issue / is that. . . We took the ,car / and drove to Birmingham, On my way to 'work / it started to rain. Unfortunately / it doesn't work like that, (cf. Unfortunately / it doesn't work like that.)
Most adverbials which have a separate intonational phrase win ■ but there are a number of adverbials of a particularly assertive 1 rt commonly take a falling tone (e.g. literally, certainly, honest, course, besides): I
Be'sides / he's had time to think about it. By the 'way / what do you think of the new chap?
As indicated by the last example, some adverbials can occur be) i as well as declaratives. Adverbials also frequently occur foi ,iv, ir? thl clause; in these cases the rise which occurs is almost always      lLSc rbjjtjl falling type again takes a fall): : -,
I went to Canada / last .year. ~^f% It didn't work / unfortunately.
He turned bright red / 'literally. 1^ J
In the case of final subordinate clauses two sequences of tones i re pcsifo«| If the previous main clause has a fall, then the subordinate claiiM. will tsfegt a low rise. Alternatively the main clause may take a fall-rise and the >jburunjj| clause the fall, cf. JB
I began to feel 'ill / because I hadn't had enough to ,eat. "i 1 began to feel "ill / because I hadn't had enough to 'eat. I
1
(3) Yes/no-interrogatives I In GB the more usual and more polite way of asking yes/no qui'.tiiii * is" with the low rise (although a high rise is more frequent in Genera' \TeriOMlS if a potentially accented syllable is available before the nucleus 'her lbiS£ will take a high pitch: .j
(It's going to rain I'm afraid.) Do you 'really .think so? * (I'm really enjoying myself.) Is 'this your .first visit to Londot "i (The large size costs a pound.) Is 'that the ,new price? -
A falling tone (high fall or low fall) on a yes/no-interrogative marks 1 j1- tr^Xj pressing:
Connected speech 295
1 .   ,.011 remember where I left my new shoes?) Are they in the 'wardrobe? ■S.*wplained it all to me.) But do you under'stand it? Jp01.((jnd my pen anywhere.) Are you sure you brought it 'with you?
~-A      is often used to mark a yes/no-interrogative as an exclamation:
E1§f-W' .   .]. didn't even leave a message.) Now isn't that peculiar! r^' ~r .roing to Spain tomorrow.) Aren't you "lucky! i%)ye rt fused to help me.) Would you believe it!
wfr.interrogatives
Whe usual tone on wh-interrogatives is falling (low fall or high fall):
i ■ (S ie wants you to send an apology.) What's it got to do with 'her? fr... (Y;iu mustn't tell her.) Why 'not?
,sni. didn't get the job.) How do you know? f"^
jiiu.TJtive tone on such interrogatives is the low rise (like yes/no inter-&Kil"vS, it is more likely to be a high rise in General American). The use of kcr%' is more tentative:
£   (We're olT on Thursday.) What time do you ,start?
(I'm afraid it didn't work.) Why did you do it .that way?
fcaVtf -vnterragatives can be used with high rise to ask for repetition:
(He's completely irresponsible.) 'What did you say? (Her name was Pettigrew before she was married.) 'What did yon say she .....was called?
Tag-interrogatives
Tag-interrogatives consist of a sequence of an auxiliary verb and a pronoun : appended to a preceding declarative. They are most commonly negative if . a. preceding statement is positive, and vice versa (called 'reversed polarity' tugs). Such tags have two common alternatives; a falling tone (high fall or low fall) or a rising tone (usually low rise). Both types of tone expect agreement, the fall inviting or demanding it, the rise leaving open the possibility of disagreement:
("'s a long way from the shops.) It's right on the outskirts / 'isn't it? (Ihad a lovely time,) Yes / The day did go well / 'didn't it? (Lend me your copy of Shakespeare.) You will look after it / 'won't you? (Where did 1 put my golf clubs?) You left them in the garage / didn't you? (He asked me to drive him there.) But you won't be able to go / .will you? (Who was that woman he was with?) It was his sister / .wasn't it?
296   Words and connected speech
Connected speech 297
Another type of tag has constant polarity. This type only ha,'. |, tones are impossible). The meaning conveyed is in the natun. i'-" "*! echo of a statement from the preceding speaker: ' fi
(1 think he's going to emigrate.) So he won't marry her / WOn' /j§ (Rachel's gone away with John.) She's still seeing him / is sbe^-^
A variation on this type involves a pronoun and auxiliary verb omiii-d I previous main clause (it is heard on TV or radio as a detective's in& technique), e.g. I
Watch a lot of television / ,do you?
Come on his bike / did he? -i Didn't give it any thought / .didn't you? i
(6) Imperatives ^ Abrupt imperatives have a falling tone. Polite imperatives, whkh are ail
„,,„„„.,1-:—•<- - -i - !■.•■• "Jl
suggesting that the listener has a right to refuse, are said wiih (most frequently low rise and sometimes fall-rise):
(I've decided to lend him my car.) Don't be such a silly 'fool;" (What should I do now?) Go and wash the car. (You shouldn't have spent all that money.) Don't be ,angry about it (I'm afraid I've had enough of you.) Give me another .chance; -jsteJ (r have a very delicate job to do here.) Be "careful. . I
The use of a rising tone rather than a falling tone softens the impcjuve. times the rising tone is combined with a tag: -fc^i
(Can I have some more wine?) Help your'self / .won't you?
(Her nerves are terrible.) See if you can 'help / .will you?  : ' jt'
(I'm doing my best.) ^Well / hurry'up / .can't you? hJF
(7) Exclamatrves ^ jc
Exclamatives (i.e. those sentences having the syntactic form of ar i.uiaeiJlSWJf" i.e. an initial question word and no verb) take a falling to.ie unci ifS®?* rise-fall):
What a beautiful May! How 'stupid he is! What a very silly I tc ffa\ What a palaver! « -fe^
Similarly individual words, particularly nouns and adjectives, can he ' ^ca*£j£^ clamatory force by the use of a falling tone, e.g. \
'Nonsense! You 'idiot! 'Marvellous!
J*,'-Hicult'to g>ve ru'es f°r tne intonation of social formulas because it is *    ix|icre native speakers of English often have idiosyncratic habits. It 4& a'Ll| ^ ., generally true that falling tones generally show sincerity, while U ones are usedm situations where a formulaic pleasantry is appropriate. * = h-'.kyou is appropriately said with a rise on being given a ticket, while '■ \iH i*1 <n°re likely if a genuine favour has been done, and a low fall fxhe fli8"er 'n nan<^ 's coring. Oood morning with a high fall is sincere-i ii: . ffld sometimes inappropriately so!), with a low fall is brusque Mil' •• low rise is polite (and possibly overly so!). This greeting (and '*\ ~. ,;| crO can also be said with the 'stylised' tone which involves a step "!^'ai   t'n 'eve' *° 01   'eve' '3egilui'r)g on me accented syllable, thus;
Good "morning.
f       ^ pill.., a special one often used on vocatives and on jocular sentences,26 e.j
"Johnny  You've split your "pants!
ni! .jr t'lences
F" In the preceding subsections it has become apparent that some lonal sequences %s. lf,, common in successive intonational phrases, particularly those within iiu- N.j.r.-ice. The most common sequence involves a tone from the rising group (low rise, fall-rise or mid level) followed by a fall (high fall or low "v: I":.!' i, .'iii. ihc syntactic sequences are sentence adverbial or noun-clause subject "   l'l.irwcd by a clause remainder and an initial subordinate or co-ordinate clause:
I ii ■ urtunalely / it didn't 'work.
Ihc 'usual excuse / is that there's not enough 'time.
Bi-Ljuse he gave up too "early / he lost two thousand 'pounds.
Ml vaked everything on .winning / and ended up with 'nothing.
A Sicon.! sequence involves a falling tone (high fall or low fall) followed by a sslil i-ne (almost always low rise, occasionally fall-rise): the syntactic sequences main clause plus adverbial or main clause plus 'open' tag:
1 deny the whole thing / usually. : That's the best way to 'do it / .isn't it?
298  Words and connected speech
Connected speech 299
A third common sequence is a falling tone (low fall high falh ing tone (low fall, high fall); this is common with sentenc'    '" a,-'*«e reinforcing type, with 'closed' tags, and with co-ordinate clC '^"■"'''^ independent of one another: '   - -1 lieh an
I go to 'London / 'regularly. It was a beautiful 'day / 'wasn't it? (What should I do?) Take up 'singing / write a 'book / do an / buy a 'bicycle / 'anything!
1 "naß
I Z.6.2.7 The use of secondary accents
Secondary accents are produced by pitch prominences which m i j- b-'ftnil nucleus, i.e. they are pre-nuclear accents. As the name implies h?-, c less to meaning than primary accents. The first secondary accent ii .-ii'ipt, phrase often serves to mark the beginning of the new information, . ";. i the 'way to the 'station. Where there is a series of secondary accent, ■ piomirl being achieved by 'stepping' the pitch'—see §11.6.1.4 above), ttu _: -lv. like that on way in the example above, serve only to divide the n;v, infun into chunks. This sequence of secondary accents which steps <1kv n n common type of pre-nuclear accenting in GB and can occur btfuic all iiugB&i tones. The other most frequent type of sequence involves one or a sjne-. <i( <mSr (see §11.6.1.4 above); this gives more emphasis to the words takr? the gecS and is particularly common before the fall-rise nuclear tone, e.g. h •■vasn: like 'that. *
■mos
11.6.3 Regional variation in intonation
Within the UK the most marked variation compared with GB concert!1-1 it- m«£g§, extensive use of rising tones in many northern cities (sometimes uTensi^f*?! as Urban North British); it is reported for Belfast, Birmingham, (ilxsgMnH Liverpool, Manchester and Newcastle.27 Rises in these cities arc moie l.cqup» on declaratives than in GB and indeed in some of them may typici.ll.. be fe^t most frequent tone on declaratives. The rise may be a continuous on; cr njjt j consist of a rise followed by a plateau (with an optional downdrifl ai \ht arSJ).-Belfast28 (and Derry)2' has both types and has the most regular uw i'f of all the cities, falling tones being much less frequent than rising ontga speakers of all educational classes (where in the other cities the typi.u! iisteni tones may be lost in more educated speakers). In the south of Englaii". ttiMMr, tion does not vary much from GB, but one type of tag in.o.i-ilio'i llMlt'Y particularly common in popular London, concerns the use of a falling tag pure!} jl as a response, as in '"^™
i.on
t.s a[i this 'mess?) B: I've been doing the 'cleaning / 'haven't I "h did you get up so early?) B: Because the 'postman came banging Ehe door / 'didn't he
pointing to a bottle] What's all this?) B: I use it for cleaning the silver •don't 1?
American has less difference from GB than that in northern British " entioncl above; one notable difference is the increased use of high rise psed tolow rise, particularly for yes/no-interrogatives. Australia and New [ are most notable for the intonational change whereby a high rise is ligly t'sed on declaratives;""1 it seems to be used as a check (hence 'check-jEpj by the speaker that the listener is paying attention and understanding, Kfafy when the speaker is presenting information. This usage is now com-j heard in England:31 often high rises are used when falls might be expected, ^iiet 'Soma in town yesterday I and she didn t look very well / and she %:cshehad 'shingles.
i.4 Pitch range
fc-osc^edixtg sections §11.6.1.3 and §11.6.2.3 falls were divided into high falls and jk falU and rises were divided into low rises and high rises. This sort of variation life height of tones concerns accent range. However, there is another sort of ration in height which refers to the pitch of the whole intonational phrase, pied as the interval between the lowest and highest pitches. Speakers may pase the width of the normal pitch range of their intonational phrases by raising Mitch of the highest pitches. Such variation in the width of the pitch range 'intonational phrases is often referred to as ksy. The most common use of key
JiiCems the delimitation of paratones (the spoken equivalent of written paragraphs), -. ti paralone being marked by a wide key in the first IP and the end of a Jaratvi© by a narrow key in the last IP32 (and often followed by an extended pause). P . • ics often correspond with topics (defined in the widest sense); the most %&m"ous use of such differences of key, paratones and topics, is in newsreading. '* TVere is yet another sort of variation in pitch range which involves raising or jdjr*sring both low and high pitches, so that the pitch overall is lower or higher 1 iu'' ot wider. A low register is used for parentheticals (marked here by doubling jidic boundary marker), e.g.
I ran into Jane last week // by the way / did you know she has three children? // and she said....
H i register is in general associated with greater emotional tension but never-ess has to some extent become conventionalised. For example, the adoption a Tittle girl voice' may be used to signal helplessness.
Connected speech
301
300   Words and connected speech
11.6.5 Intonation and punctuation
Punctuational usage is generally prescribed in manuals of punctuation i those put out by publishing houses) according to grammar rather "faj • Indeed in some cases a punctuation mark has no correlate in inttiti- '* apostrophe marking possession or elision, and the use of spaces bet v. However, in many cases punctuation is as related to intonation as- t> Punctuation marks serve both to delimit, e.g. to mark the end < f or to specify, e.g. that a sentence is a question. In both areas there .ti-= -la. relations between punctuation and intonation. The use of any of t.iu $M comma, semi-colon, colon, full stop, question mark and exclamation i,aTj, a correlates with a boundary between international phrases. Howi-it: ;L; syntactic positions where such an international break is very comr- ,.n oia the use of a punctuation mark (especially the comma) is genera.'!; cro»l This applies particularly to the position between subject aud verb \. reJL in §11.6.2.1 above, this is a very common position for an intona'i.injl fw the subject is either particularly long or contrastive. Hence, althouo.i = in this position is generally proscribed by manuals of punctual im. the#i nevertheless be the tendency to insert a comma here because < I" tfo intuition about intonational-punctuational correlations, e.g.
The best way to do this, would be to ask him first.
Specification concerns the marking of a sentence-type as a q<ies:iw. i» exclamation (the use of the full stop marks the sentence as a nor.-i'acsticjy non-exclamation, but may be used for different sentence-types, c .■ qxatj command and request). Such specifiers have poor correlations with i ulea. The full stop usually correlates with a fall but by no means always l que mark correlates regularly with a high rise in the case of 'declarative qudsj (§11.6.2.6(1)), frequently with a low rise in yes/no-questions andapproxirf half the time in tag-questions and frequently with a falling tone in .vh-qnest Only in the case of the exclamation mark is there a regular corn I iiioi. w least an overall category of tone—it always correlates with a fallin= loi.e (j| high fall or rise-fall).
11.6.6 Acquisition of intonation by native learners3' ".
Many babies are excellent mimics of intonation and may produce Hngiisli-'oundifJ intonation patterns on nonsense syllables (often called 'jargon i iioualicfl i£ the late stages of their pre-linguistic babbling.34 At this same time t ie\ ta| also have a distinction of meaning in their use of a fall versus a n.-. 'ir-i syllables—typically the fall is naming while the rise is requestin" I veil the children who do not have this distinction during the babbling peri id mil !-,en* ally acquire it during the period of one-word utterances (which i p — Ls
"■'"'nine months during a child's second year). At the two-word stage e capable of varying nucleus placement, although whether this is •  ailing new m^      information or whether it is more rigidly tied J§|ift types of sentence is not clear. (It is, for example often the case •K^gives have the accent on the possessor while locatives have it on the
Icf 'Daddy car 'Daddy's car' vs Daddy 'car 'Daddy's in the car'). Mlalown about the later acquisition of intonation, although it is likely that H>c of the fall-rise tone are learnt early. But a full mastery of the more fiances of international meaning may not be acquired until the age of ten
* we1""";
j f intonation—advice to foreign learners
jfwe<gh learner should pay particular attention to:
mL ^-ttieving a better style in reading aloud by appropriately dividing his speech intonational phrases. Such division may be done in English in ways >      similar to his native language (especially in the case of most other 'ifTiJOpean languages) but nevertheless the learner should note the frequency ** n irt- which sentence adverbials and the subjects of sentences are given their
I. o*n intonational phrases. 4 Ki'ing the nucleus on the focal point in the sentence. Some languages (like ' Frereh, Italian and Spanish) more regularly have the nucleus on the last word in us intonational phrase. This may sometimes mean accenting old information - occurring at the end of a phrase, which is incorrect in English, e.g.
Would you like to come to London with me tomorrow? No/1 don't like LONdon.
j&t) I.sing appropriate nuclear tones. Learners should note that the fall-rise especially on a single word) is rare in most languages but very frequent in ■English for a range of attitudinal meanings on declaratives and for subjects with their own intonational phrase. Fall-rise is also frequent on sentence «dverbials in initial position, although low rise is the usual tone in final position (but those exceptional falling adverbs, like definitely, which take low fail or high fall in any position must also be noted).
.-4:i j\ cruse of simple falling tones (especially high falls), together with an overuse '.of tildes-down in pre-nuclear positions, will produce an excessively aggressive ■sftw ...while conversely an overuse of simple rising tones (including fall-rises «5X glides-up in pre-nuclear position, which are uncommon in GB) will sound '-exi.o"5ively tentative. The overuse of falls is typical of north Germans while an wenwe of rises is typical of Scandinavians.
Connected speech 303
.-BS,
1 1.7 Hesitations
Pause was stated in §11.6.1.1 to be one of a number of phoneti are used to mark boundaries between intonational phrases. Bir , i occur in other positions: (a) before words of high lexical conte.i >c"",t a low probability of occurrence in a particular context (and whei.' (.„ ~'1 often searching for the right word); and (b) after the first word in , r ^ phrase, where it appears to be a pause for planning, i.e. the spe.ilr. u',0 that he has something to say but has not yet planned it in detai' '" |i, ml'JJ (b) may be also used more deliberately as ways of attempting i.i r-jB ruption. Both differ from pauses at intonational phrase boundaiij'    r. 7^ being filled rather than unfilled (i.e. silent) pauses. In GB filled |        " 1,3 ally filled with [s] or [m] or a combination of the two, e.g.
I don't agree with that /1 [si] think it would be better if. .. You see / the myth is / and I'm [s:] I can see from the applaud Well [ami] I don't think I will.
This type of filled pause used in GB is not necessarily the typ„- o- Hire used in other dialects of English or in other languages. Scotti-1- r,u,i g" [ei], French uses [0:] and Russian uses [n:]. A quite dratnafii Uian^e c4j"*33| produced in the degree to which a person's speech sounds native-1.|..-> c\ wWffBJFQ
the correct type of filled hesitations.
I 1.8 Voice quality37 :<.'MJ The area of voice quality has very little scientific work associr.red v .ih it$M£4 ticularly on a cross-linguistic basis and this section is perforce .n.1 \cr* bysfanmf^ atic. Some reference has already been made to the topic under ;* X . . Oie.JffipfcM term 'voice quality' refers to positions of the vocal organs wlin.1 cl 'immBj speakers' voices on a long-term basis. Long-term tendencies ii 1 ■M'lu.iir.g tongue and the soft palate are referred to as articulatory setting 111.1 -i- nldlVfP^. to positions of the vocal cords are called phonation types. In soitu U ii.- lanauige^^ a phonation type may co-occur with a lexical tone (e.g. two torn-. ,1 .t) rwdkM^ tinguished in some varieties of Vietnamese, one simple high n >e uiu 1' i-- biijlk,^ rise with accompanying creak). But more generally a particunr mhii quiHjB:~j may be characteristic of an individual, of a particular language 01 i''jk»l. ™ff[lk be used within a language to convey a particular attitude or crrniiiMi
The articulatory setting of a language or dialect may differ trim C-B So joined languages like Spanish may have a tendency to hold the tongue uiv'rj litfwwfe^j in the mouth, while others like Russian may have a tendency 1 1 ln'ld n ,urt^3kijj back in the mouth. Nasalisation may be characteristic of 1111» -pukers ot£_ ' American English, while denasal voice (which may lead to a I-'i-ltvA 3 resonance in nasal consonants) is frequently said to occur in 1 ■, 11x11 las^r i and lax are labels which apply to the muscular tension in the whH'" i»' in*-" xot^Ji
..generally said to be lax (making it sound 'mellow'), while French n ' '*, , dre said to be tense (making them sound 'metallic' or 'strident'). '    !.i commonly described phonation types are creaky voice, breathy voice, voice (sometimes called harsh voice and involving the false vocal Pu . jj^tue the vocal cords), whispery voice, falsetto (dividing the vocal 10 two halves and hence raising the pitch by an octave), and raised-j j0wered-larynx voices. Breathy voice is said to be used by many af Danish and Dutch, creaky voice is used by many speakers of GB ■ rticularly by speakers of CGB, while ventricular voice is a characteristic speakers of Scottish English and speakers of Cockney. Within GB (and * ^ y wider in English) some phonation types are associated with certain styles m>£ ■ 1|, ;Ijd emotions: breathy voice is often called 'bedroom voice', whispery JjjfJ?8^ ,,,metimes called 'stage whisper' or 'library voice', ventricular voice is
ilcll with anger and lowered-larynx is called sepulchral or 'vicar s voice .
fc-'fh ijVi ,] ..t the sign ' here and throughout the book shows the place of the primary o.k.k. il Vs the symbol" implies, in the citation forms of words this is usually a falling at, in this chapter a falling accent becomes one of a number of different tones -'•      11 ■ the intonational system. See §11.6.1 and note 16.
IJ& I'lk- l"t5V Abercrombie (1967), Halliday (1967) and numerous TESOL textbooks. \ 1. I, ->\. • made prominent by stress alone refers to a subset of full-vowelled syllables > be prominent rhythmically. . 4 , „ npson (1980), Roach (1982) and Dauer (1983).
i I j I Vowel rhythm has affinities with Metrical Phonology (Liberman & Prince, 1977; "' 1 la- Js. 1995) which has a level of strong and weak syllables and extrametrieality. ~$ [-i--vowel!ed rhythm and the Borrowing Rule were first put forward in Bolrnger 1) and applied to TESOL in Faber (1986). Instrumental confirmation is given in -Jpsoh (1980).
""" ^ mevarieties of English, notably Caribbean English as an LI and Indian English as a U2. are marked by a much lesser use of reduced syllables and hence the rhythm
■searer to that traditionally labelled 'syllable-timed'. *8 if the words listed with weak forms above occur in the two hundred most frequent ids-in the spoken part of the British National Corpus (Leech et at., 2001). ., 9 Some does not occur in a weak form when used as a pronoun, e.g. /sAm mart 'sei/, .''aid 'laik SAm/.
■I That as a demonstrative adjective or pronoun lakes a full vowel, e.g. that man /5al man/, that's the one /5ats 63 'wAn/.
There may be new tendency to use /5s/ before vowels among younger speakers. This was first reported for American speech in Todaka (1992). Windsor Lewis (2013: §3.7.1.3) discusses its possibility of occurrence but only with an obligatory [?] before the vowel.
As a demonstrative adverb, there will have a full vowel, e.g. there's the book /3e:z 33 hoi.-.
'" ^A weak form with Ih/ would normally be used when unaccented but following a pause. "      Shockey (2003) for many instances of hyperreductions in casual or rapid speech. Windsor Lewis (2013: §4.7) for priorities for foreign learners in the learning of faction words.
304  Words and connected speech
16 Much recent work on the form of English intonation has been done i * tradition represented by Pierrehumbert (1980/87) and PterretiumberT "-j^ "r* (1990), which decomposes pitch contours into sequences of high and "' also detaches phrase tones and terminal tones at the end of intonat '"' However, the treatment of intonational meaning in this newer framework i- - .™ at a general level and the nuclear tone approach is retained here beta ",rcir" easier to treat in this setting the local meanings produced by associa I tones and syntactic types, 1!
17 Refer to the companion website for a reading of all the examples in tf ,1
18 When combined with text, tones are marked using what are usually, c" 1 I stress marks', though in this book they are more accurately called 1 marks'. All marks indicate a following accented syllable. Individual ma-iconic) indicate pitch patterns starting at the accent. For a more contr. ■ ■ *' of the intonation of GB. see O'Connor & Arnold (1973), Cruttenden (!<■,'"„ Jva (2006).
19 The type of transcription used in the diagrams is called 'interlinear tone ■ " I indicates a syllable. A large filled dot indicates a pitch accent, either pnmar >.irr«I| A filled ring sometimes has a tail indicating pitch movement on the act       „ j| itself. A large unfilled ring indicates a syllable made prominent by havinj vowel. Small rings indicate unaccented syllables.
20 These high unaccented syllables are sometimes marked with a high-leS si have not used this mark because we prefer to preserve the general rati :fj marks in the text always apply to accented syllables.
21 Quirk et al. (1964; 683), Crystal (1969: 256) and Altenberg (1987: 25i 3
22 Altenberg (1987: 174) found the nucleus occurring on the last lexical item t ~* ;i3 of cases.
23 Quirk et al. (1964: 681), Crystal (1969: 225) and Altenberg (1987: 36 ! the somewhat higher figure reported in §12.5(1).
24 For further information on the meanings of tones, see Halbday (1967) > 11 or!iw$ Arnold (1973), Cruttenden (1997) and Wells (2006).
25 For the frequency of nuclear tones, see Quirk et al. (1964: 681), Crysu 11959' and Altenberg (1987: 37).
26 For details of this tone in English, see Ladd (1978b).
27 Cruttenden (1995, 2001, 2007).
28 Lowry (2002),
29 McElholm (1986),
30 Guy et al. (1986), Britain (1992). J
31 Bradford (1997). Shobbrook & House (2003), '-,
32 See, in particular, Brazil (1975. 1978, 1985),
33 Crystal (1986).
34 Peters (1977).
35 Cruttenden (1974, 1985).
36 Cruttenden (1997: 30-32).
37 See Laver (1980) and Henton & B la don (1988),
38 The terms are also often used to distinguish between the short and lo " ' i <;■= «i English (see §8.4.1(9) above and Chomsky & Halle (1968)). ..'t-J
12
tfds in connected speech
». | citation forms and connected speech
■ f . ds as separable linguistic units are recognised in the sophisticated written ' ff-in of Enalish by the use of spaces between words. Although in the continuous f rrcjn of speech there are no pauses between words corresponding to such written
inai.-s words nevertheless show their independence by their ability to stand 'thnc, e-g- as repHes to questions and when being referred to or cited. Differences i exist between the pronunciation of words in their cited, isolate forms and "tfc;n pronunciation in connected speech, when they are subject to influences from
■ rf.ier; surrounding sounds and from larger accentual and rhythmic patterns. The jjjgerences may concern the word as a whole, e.g. weak forms in an unaccented
Ljitiation; or they may concern a word's accentual pattern, e.g. loss or movement 'of an accent due to its position in a larger accentual pattern; or they may involve jjjke sounds used at word boundaries as in assimilations, elisions and liaisons. 'This variation between isolate forms and context-influenced forms often depends ion the style of speech. The style generally described in this book is slow but cdioal. i.e: it is not rapid and it is not careful. A rapid style will produce many rp exchanges from citation forms than are described in this chapter (although 'they are sometimes mentioned as such) while a careful style will produce fewer changes than are described.
12.2 Neutralisation of weak forms
'Ae jave. seen already (§11.3) that a number of function words may have different pronunciations when they are accented (or said in isolation) and when, more : "iicalfy, they are unaccented. Such is the reduction in the unaccented forms that words which are distinct when said in isolation may be neutralised (see §5.3.4 above) when unaccented. Such neutralisation generally causes no problem to listeners because of the high rate of redundancy (see §1.3.1) of meaningful cues; only rarely does the context allow a variety of interpretations of an unaccented r.The examples of neutralisation which follow might occur in casual speech -id are almost certain to occur in speech which is both casual and rapid.
306   Words and connected speech
hi = unaccented are, a (and, in rapid speech, her, or, of) The 'plays are "poor He 'plays a 'poor man She 'wants a 'dog She 'wants her 'dog 'One or 'two of them are 'coming 'Two 'books are 'mine 'Two 'books of 'mine
/9V/ =
unaccented have (aux.), o/1
'Some 0/ rone (piece). , . 'Some have rwon ... The 'boys of'Etaa 'fish The 'boys have 'eaten "fish
far/ - unaccented are, or
'Ten or 'under 'Ten are "under
16s! = unaccented the (and, in rapid speech, there) The 'seams are "crooked There 'seems a 'chance
/s/ = unaccented is, has, does
'What's ('s = does or is) he Tike? 'What's ('s = has) he 'lost?
tzl = unaccented is, has, does
'Where's ('s = has) he 'put it?
'Where's ('s = is) he 'going?
'Where's ('s = does) it 'go? (rapid speech)
fax/ = unaccented as, has
'How 'much has he 'done? As 'much as he 'can
/an/ = unaccented and, an
'On and 'off
'On an 'off-chance
■Hj
Words in connected speech. 307
"■'faj = unaccented and, not
linjdn'the ,do it? /'didn i ,du; it/ fie 'did and he 'didn't /hi 'didn i'didnt/
"a/ = unaccented had, would yd ('d = had, would) 'put it there
'jjj Variation in the accentual patterns of words
[#Wl-.-n   word (simple or compound) pattern consists in isolation of a primary Wi.-L-.-t preceded by a secondary accent, the primary accent may be lost completely connected speech, another primary accent follows closely in the next
■lur'teen, but 'thirteen "pounds \VcSP01inster, but 'Westminster 'Abbey t,:llJ'grown, but a 'full-grown 'man at-er'noon, but 'afternoon 'tea
ie secondan. accent in the word rather than the primary may be lost when lil.u'word with a secondary accent immediately precedes, e.g.
'eight thirteen; 'near Wesťmřnster; 'not full 'grown; 'Friday afternoon
Sjch examples, and the others in this section, confirm the tendency in English . tp avoid adjacent accented syllables.
it is in order to avoid accents on adjacent syllables that 'accent shift' occurs phrases such as 'Chinese "restaurant (but Chi'nese), 'outside "world (but ťside). Where the accents are separated by unaccented syllables, the accent .shift is variable (though pronunciations with the shift are probably more common l.:dii those without), e.g. diplo'matic, diplo'matic "incident or 'diplomatic 'incident; uama'rme, aquama'rine ti'ara, 'aquamarine ti'ara.
This tendency to the alternation of accented and unaccented syllables is so ongthat the accent may be shifted in the case of certain words whose citation to contains only one, later, accent but where a full vowel occurs in the preced-
l!J syllable, e.g. or'nate but 'ornate 'carvings; u'nique but 'unique "features;
.'i:d di'rect but 'direct 'access. The alternation tendency extends into longer ' erances and may be seen in examples such as i'dea but The 'idea 'pleases %'recom 'mend but / can 'recommend "several; and in phrasal verbs such 'come 'out, 'get "in, e.g. The 'pictures 'didn't come "out, but They 'came t "well and 'What 'time will 'you get 'in? but 'What 'time will you 'get in •m 'work?
308   Words and connected speech
Words in connected speech 309
12.4 Phonetic variations within words and at boundar)Jj|
Our phonological units, the phonemes, represent abstractions from actual nl 'I reality. If the phoneme Itl is given a convenient, generalised label—a alveolar plosive—it is nevertheless true that the actual phonetic realisati-1 .J- . !§] consonant depends on the nature of the context, e.g. Itl is aspirated when befrlal a vowel (except after hi) as in [then]; it is dental, rather than alveolar, whmiL adjacent to /6/ as in [wrt8]. Besides these variations within words, such \r.    . mtm also occurs at word boundaries (and at morpheme boundaries in compound a !*fi* complex words) where tendencies towards co-articulation or assimilai ion haviJtfil to be noted. T
Assimilations at boundaries, like those within words, may be men-i- , f allophonic kind; or they may be of such an extent that a change of phoneme involved, when comparing the pronunciation of a word in isolation v. i.h igfeffl pronunciation in a particular context Influence at word and morpheme bmindanes^ii functions predominantly in a regressive or anticipatory direction, i.e. features jf >sS one sound are anticipated in the articulation of the preceding sound; less frequently "'"i it is progressive or perseverative, i.e. one sound influences the followirc. smi< or it is coalescent, i.e. a fusion of forms takes place. ■ 1
12.4.) Allophonic variations
Since the actual realisation of any phoneme is at least slightly different in e\ 1 context, it is necessary to give examples only of those variants which exhin i "■■ striking changes. The same types of allophonic variation, involving a change .|i place of articulation, voicing, lip position, or position of the soft palate, maj found within the word and also at word boundaries:
(1) Place of articulation |j
(a) within word: <k Itl—dental in eighth (influence of [9]) J| Ik!—advanced (pre-velar) in key (influence of [i;]) i In/—dental in tenth (influence of [0]) T /ml—labiodental in nymph, infant (influence of [f ]) ■ I si—retracted in result (influence of [lj) 'i
(b) at word boundaries: Jk /t/—dental in not that (influence of [5]) 7i Idl—dental in hide them (influence of [5]) fc /ml—labiodental in ten forks, come for me (influence of ff ])
(2) Voice—devoicing of continuants following a voiceless consonant, j|
(a) within word: JL /l,r,wj/—devoiced following voiceless consonants, e.g cry, plight, q\ queue
/m,n,n/—slightly devoiced following voiceless consonants, e.g. smoke,
snow, mutton, open /'aopm/, bacon /'beikrj/ . at word boundaries (only in close-knit sequences): /l,r,w,j/—devoiced following voiceless consonants, e.g. at last jytjatst], atrest [s'trest], atonce [s'twAns], see to /f ['si;tv/it], thank you ['Oarjkju;].
\ote also the devoicing of word-final voiced plosive or fricative consonants before silence, and of fricatives when followed by a voiceless consonant; and of uOrd'iriitial voiced fricative or plosive consonants when preceded by silence, 0 in What can you give? ([y]); Can you breathe? ([§]); It's his ([z]); near the . bridge (rdj]); They've ([y]) come; with ([§]) some; He's ([?]) seen it; George ■ ([$]) ca,r' (M) v?ry good ([d]); ([§]) there; ([?]) Zinc does ([?]). ,i) Lip position—under the influence of adjacent vowels or semi-vowels.
~ (a) within word:
lip-spread lip-rounded2 pea, heap        pool, hoop, upward tea, beat two, boot, twice, outward
keep, speak      cool, spook, quite, backward mean, seem      moon, horn, somewhat knee, seen        noon, onward leave, feel        bloom, fool, always read rude, route
feel, leaf        fool, roof seat, geese       soon, goose, sweep sheet, leash      shoot, douche, dishwasher he, heat who, whom, hoot
(b) at word boundaries, e.g. /t,k,n,n,l,s/ are somewhat labialised in such cases as that one, thick one, thin one, wrong one, shall we, this way; a rounded vowel (as opposed to semi-vowel) in an adjacent word does not seem to exert the same labialising influence, e.g. IwJ does not labialise Isl markedly in Who said that? nor does IxJ in this ought to. (4) Nasal resonance4—resulting particularly from regressive but also from progressive lowering of the soft palate in the vicinity of a nasal consonant.
(a) within word: nasalisation of vowel preceding /m/ in ham and /til in and, of vowel between nasal consonants in man, men, innermost, and of short vowels on each side of the nasal consonant in any, sunny, summer, singer; also IV in such situations as in helmet, wrongly; and possible slight nasalisation of vowel following /m,n/, as in meal, now.
(b) at word boundaries: vowels may sometimes be nasalised somewhat by the boundary nasal consonant of an adjacent word, especially when an adjacent nasal consonant also occurs in the word containing the vowel,
Ipl
Itl
Ikl
Im I
In/
IV
hP
1ft
Is/
IS' /hi
310   Words and connected speech
e.g. the first /a/ in bring another, or hi in come in, but son.c-> with no adjacent nasal consonant in the word containii, ■ |., (usually unaccented), e.g. /a/ in come along, wait for mt night. Approximants may also be nasalised by a nasal ii word, e.g. Ill in tell me.
J2.4.2 Phonemic variations
Different phoneme selection within the same word may occur (either oetwS two speakers or between different styles of speech in the same speaker) depejjjj ing on the degree of assimilatory pressure felt by the speaker, e.jj !..>£,; 'LJ^ be /lerjG/, AerjkO/, or fiend/, encounter may have /in/ or /irj/ in the first syllafe disgrace may have final /s/ or Izl in the first syllable, absolutely may hi final Pal or Ipl in the first syllable and issue may have medial /sj/ o'- a coal form /J7. Historically a phonemic change within a word can sen due to assimilation to surrounding sounds, e.g. by labialisation /w i. :' or /wo:/ (swan, water) or by coalescence /tr,er,ur/ —» /ar/ —► In; influence of the post-vocalic hi (first, earth, curse) and /sj,zj/ —» /j    ; mati^a* vision).
Many phonemic changes occur in connected speech at word bcnrr. it-, (j J changes as compared with the phonemic pattern of words' citation '■uu o ^ucli phonemic variation is found in changes within the pairs of vou_\l >.n. phonemes and especially in changes involving modification of die *Lce articulation. *
12.4.3 Voiced/voiceless variations
Word-final voiced fricatives followed by a word-initial voiceless consonan: xa: with some speakers be realised as the corresponding voiceless fricative. 11 k two words form part of a close-knit group, Thus the final 16/ of with nu> cic replaced by IW in with thanks; the final Izl of was by /s/ in He was sent;unA the final Ivl of of we've, by If I in of course, We've found it. Such a chance M a| voiceless fricative is an extension of the allophonic devoicing of such con; c:unki mentioned in §12.4.1(2). The phonemic change in such examples will be con-pletrj in that a preceding long vowel or diphthong will be realised in the reduce " s appropriate to a syllable closed by a voiceless consonant (See §8.4.1 Vuei) (4)-(8), 9.2.1(5), 9.4(4)). ;
The weak form of is or has is Is! or Izl according to the final conson nil «d the preceding word, cf. the cat's paw, the cat's gone /kats/ vs the dog's p*-dog's gone /dogz/.
It is unusual in GB for word-final /b,d,g/ to be influenced in the same ■ following voiceless consonants, though voiceless forms may be he id in 1'Lh-contexts in the speech of some parts of northern England, e.g. the Id! ol ." «** time and the Igl of big case may be realised as /t,k/.
Words in connected speech 311
... be noted that word- or morpheme-final voiceless consonants in English " "'^'assimilate to their voiced counterparts: such pronunciations of nice boy, ^zjress, half-done, they both do, wishbone, birthday, as /'naiz 'bai, 'blag "dres. jarv'dAn. 8ei 'bau5 'du:, wr^baun, 'b3:5dei/ do not occur in GB.
2 4 4 Nasality and labialisation
Tftorierriic assimilations involving nasality (i.e. anticipation or continuation of ■.. lowered soft palate position) would be likely to show Pol (or Ivl) ~* /ml, IAI tzl or75/) -+ In/, Igl —> /rj/, such changes being based on roughly homorganic ,runi articulations; nasalisation of other sounds, e.g. Ill or vowels, is never v loDemic, there being no nasalised counterparts with approximately homorganic 'luiith articulation. Such phonemic nasalisation as does occur concerns mainly !'.,L jiveolars, especially adjacent to the negative not often written n't. A preceding voiced consonant, most commonly a plosive, becomes a nasal (and at the same ■i -ii the final M may be elided). These changes are characteristic only of rapid - KJfih, e.g.
IAI —* Ai/—He wouldn t do it /hi 'wunn(t) 'du: it/, good news /'gun "nju:z/ -'■■/al'—<- A}/ —> /rj/—He couldn't go /hi 'korjrj(k) "gao/ j IAI —► Pol —* An/—Good morning /gum 'momrn/ Ivl —> Iml—You can have mine /ju krj ham 'mam/ Izl —» In!—He doesn't know /hi 'dAnn(t) 'nau/ 73/ —» In!—He wasn't there /hi 'wonn(t) 'mil
Pe nasalised assimilated form may itself be elided, giving /'gu nju:z, gu 'momm, ■i 'wrm 'ne:l etc. Note also / don't know /ai da'nau/, sometimes written / dimno. The extension of labialisation produces no changes of a phonemic kind, since -position is not a distinctive feature opposing any two phonemes in GB. hi -.■■1 lad come nearest to having an opposition of lip action, but the lip-rounding <hl is very slight and open and, in any case, there is some difference of tongue position and a considerable difference of length. Where Av/ precedes a vowel of s'Iail type (and, therefore, might be expected to exert a rounding influence), Ji'her labialisation has become established at an earlier stage of the development ut the language (e.g. in was, what, war, water) or two pronunciations are today r-'imitted. e.g. qualm /kwaim/ or /kwa:m/, quaff/kwa:f/ or /kwrjf/. Labialisation 11 /<x\l involving a phonemic change to A>/ or h:l does not extend beyond word iM'indaries, e.g. in two arms or The car won i go. But some confusion may :ur between a strongly centralised form of/ao/ and hi! in a labial context, cf. ?y weren't wanted and They won 1 want it. Such confusion may also occur en speakers used a labialised form of Irl (see §9.7.2) in They weren't right vs ■ -ley won t write.
3 I 2   Words and connected speech
Words in connected speech 313
12.4.S Variations of place
The most common phonemic changes at word boundaries conrj.pi place of articulation, particularly involving de-alveolarisation. Th0i; are normal in casual speech, speakers are usually unaware that t ii made. The phenomenon is essentially the same as that resulting i j assimilation of place. Electropalatographic research5 shows: \ asshnilations of place are rarely complete, e.g. in an assimLdt | an apparent change from alveolar to labial, as in bad boy —> L,u| residual articulation on the teeth ridge may accompany the labuj (See §9.2.6(2), 9.6.2(2).)
1 «-'>aa
lllllll
fllP
i pho r i:'Vd ■ k-y:,
^iculJ
(1) Regressive for anticipatory) assimilation: instability affinal Word-final /t,d,n,s,z/ readily assimilate to the place of the W initial consonant while retaining the original voicing, /t.d n ■' ■ . bilabials before bilabial consonants and by velars before veV!!'''*'* /s,z/ are replaced by palato-alveolars before consonants conrn i,n feature:6
ioiiait •it
Itl —> /p/ before /p,b,m/, e.g. that pen, that boy, that mar, i.rt, p-n «
'boi, 3ap 'man/ 4&SSi>-—* Ikl before /k,g/, e.g. that cup, that girl /dak 'kAp, uua, uj:I> iTSE Id/ —> Pol before /p,b,m/, e.g. good pen, good boy, good ,<, ■• ls'K
gob 'boi, gub 'man/ <&■!&>'' —> Iql before /k,g/, e.g. good concert, good girl /g<x, kr»-.ssi. '93!l/
/n/ —* /m/ before /p,b,m/, e.g. ten players, ten boys, ten m : itn PieKSfifc tern 'boiz, tern 'men/ 1--3S=!
In/ —»/rj/ before /k,g/, e.g. teM cwps, fen giWs /ten 'kAp-. 'i n
(As a result of word-final assimilations, /rj/ may be preceded b> \n -than /i,e,a,D,A/. Thus /rj/ can occur after long vowels as a result o~ rvir.l'atffiCjP. e.g. I've been /bi'.rj/ gardening, She 'II soon /su:n/ come, his own    ■< .'i!1". afr-Wftj Assimilations to alveolars and between labials and velars ni^i- -n-netiiasssM"^ be heard in rapid speech, e.g. same night /sem 'nart/, king ChurU' ki i jEtja&aM same kind /seirj 'kamd/, blackmail /'blapmeil/. ,  T5 ,
/s/ —»/JV7 before /J.^djj/, e.g. rtis s&op, cross channel, thL .w... ".'i >*■*•■-< /5if 'Jbp, kruf "Jatsl. 81/ 'd^Adj, 5if 'jia/. _, ^h^j
/z/ —► /y before /ftf.ctjj/, e.g. thoseyoung men /5ao3 'JAn 'nu.i .    .' !*jfjjjJ /'1(1:3 Jbp/, ffowe churches /Sotrj 'fy.ftzl, has she? /Tia3 M -' .u.rjl.yj
Other assimilations involving fricatives may occur in rapid speech: assimilate to /s,z/, e.g. / loathe singing /ai looz 'srrjm/, What's th : zs 'tarm/, fla? ffe />os; come? /haz z3 .psus kAm/.
■1
may wits
e a high frequency of occurrence in word-final position, especially J"S-   „1 'md so their assimilation leads to many neutralisations in con-
quickly), /raip 'p£:z/ (right or ripe 'hop manjug/ (hot or top manure),
f ^pjech, e.g. /ran 'kwikli/ (ran or
,,fs), /laik 'kriim/ (/ite or light cream), /' ;t. (Pari'! Show ovparish show), /wrjf Jo: 'weit/ (fF/Wj or JFafcA jw -i .with a neutralisation to a labiodental articulation, /'grerp vain/ (grecrt J.'-',, .(»«), [rAtt) 13 P: 'niAni] (care or rum for your money).
Iveoiar consonants /t,d,n/ are adjacent in clusters or sequences sus-3 to assimilation, all (or none) of them will undergo the assimilation, e.g. jjomp.' be late, He won't /waurjk/ come, I didn't /digrjk/ go, He found 1 both, a kind /kaing/ gift, red and black /reb m "blak/. Elision may also these clusters (see §12.4.6 below).
's 9
^pa-
Coalescence o//t,d,s,z/ with l]l
■f/he-process which has led historically to earlier /t,d,s,z/ + /j/ giving /jf.chjj^/ nedially in a word (nature, grandeur, mission, vision—§9.3.1) may operate n casual speech at word boundaries, e.g.
.....Itl + Ijl—what you want /wrrtfu 'wont/
ldl+ /j/—Would you? /'wuchju/ Is! + 1)1—in case you need it /rrj keifu 'ni:d it/ Izl + 1)1—Has your letter come? /ha3o: 'lets kAm/, as yet /a'3et/
Iiiftjj coalescence is more complete in the case of /t,d/ + /j/ (especially in question . ■ didn't you?, could you?); in the case of /s,z/ + 1)1, the coalescence i^ati      may be marked by extra length of friction, e.g. Don't miss your train
4.'asji r miff31 'hem/.
i.sfiil speech, some GB speakers may use somewhat artificial, uncoalesced, Jj&ins within words, e.g. nature, question, unfortunate, soldier /'nertjs, 'kvvestjan, ffit'ftiitjonat,: 'sauldja/. Such speakers will also avoid coalescences at word bound-tees: yet other careful speakers, who use the normal coalesced forms within £*c-dv. 1 tay consciously avoid them at word boundaries. (See also §12.5 below.)
H3) Progressive (or perseverative)
W Progressive assimilation is relatively uncommon. It may occur when a ; pl-i^tve is followed by a syllabic nasal and the nasal undergoes assimilation jf- to the same place of articulation as the preceding plosive, e.g. lul —► /ml f rfKi /p,b/, happen, urban /'hapm, '3:bm/; and In/ —* /n/ after /k,g/ in second f    chance, organ as /sekrj 'Ifatus, "o:gn/.
j12.4.6 Elision
hum word-interna! elisions (see §10.8) and those associated with weak |5brms, other elision of sounds occurs in rapid speech, especially at or in the
• vicinity of word boundaries.
314   Words and connected speech
Words in connected speech   3IS
(1) Vowels
(a) Allophonic variation—When one syllable ends with a cl
(i.e. one whose second element is closer than its first in Glf '" dirt and the next syllable begins with a vowel, the second " : diphthong may be elided. Word-internal examples of the'""™1"11 in §8.11 (e.g. hyaena /hai'iins/ smoothed to [ha'ima]) "iret'Wc
4m
neutralisation, thus layer Heist with smoothing is the sai-'= mower /'maoa/ with smoothing is the same as rmrrh 'i? ' smoothing occurs across word hmmdaripe » «      ■ .:■ In: ^ifi
(b)
smoothing occurs across word boundaries, e.g. go awe ■ may as well /ai mr.z 'wel/, / enjoy it /ai in' djo: rt/, try ag or [tra: 'gen].
Phonemic elision—Initial hi is often elided particularly.' h?i ia-OLXif by a continuant and preceded by a word-final consona i ,„ rr n»-1i^B3g^i for the loss oils/ frequently being made by the syllubici-, oi 'MtSfW tinuant), e.g. not alone [not 'laon], get another [get n mm| r:v juSSw [t\a 1 "on], /re was annoyed [hi waz 'noid]. When fin. I i j^ciij'-i3«ll
following linking Itl (see §12.4.7) and word-initial vow I - qgy elided, e.g. a/rer a while /atfrre 'wail/, a matter of/at, nz ? '. 'fakt/, father and son /faiSran 'sAn/, over and crůove /gi:\ ,ir 3-ju3~3Ti When any is unaccented /el may be elided following a;i j,i U5 aJFar sonant, e.g. don't put any. . . /'putni/ or /'potni/. Wotd-i.TtuHv v>H|f"$ syllables containing Is/ may be elided, e.g. before I go. fy.r nr ^aajtafe, between you and me /twiin ju am 'mí/.
(2) Consonants—In addition to the loss of Ihl in pronominal weak trnri aifcsifej other consonantal elisions typical of weak forms (see §11.3. il t aKeolkr ^ plosives are apt to be elided. Such elision appears to take place .:nn 'sadtw"^* when Itl or lál is the middle one of three consonants. Any ciirs.Tio.nl mtylfe-appear in third position, though elision of the alveolar plosive .s .jljti-.Ajt rare before Ihl and /j/. Thus elision is common in the sequenc. w K.;iefS- X"! continuant + HI, or voiced continuant + lál (e.g. /-st, -ft, -Jl, n.l - J,      y , -fld, -vd/) followed by a word with an initial consonant,8 c ,g. >?l >.' Aw. *astá <yi back, last chance, first light, west region, just one; left turn, sc. . >•! > >'^§,j^|j> wheel, drift by, soft roes; mashed potatoes,finished nowjinishe. • '...'t pui&kt a them; bend back, tinned meat, lend-lease, found five, send roun '. V, •d,vc& TF, hold tight, old man, cold lunch, bold face, world religion; refusta •• i'-, s;.."^^* past, caused losses, raised gently; loathed beer; moved back, loved ňoWinpQĚm saved runs, served sherry. Similarly, word-final clusters of voi iL"-' :,irslvř«.Jjs, or affricate + Itl or voiced plosive or affricate + lál (e.g. /-pt -ki. -It -IA -gd, -djd/) may lose the final alveolar stop when the followi'i; '\'">rd "HJj* an initial consonant, e.g. kept quiet, helped me, stopped apt .'      • ••it!*& ^ well; liked jam, thanked me, looked like, looked fine, picked ••   ' icfrtí ^ Paris, fetched me, reached Rome, parched throat; robbed " ■/'■ ■ 'Abed » gently, grabbed them; lagged behind, dragged down, begged o    - hit%**
■urged them, arranged roses, judged fairly. (In the sequence /-skt/,.^ :i " r)ier than Itl is often elided, e.g. risked prison, asked them?) I he Hnal ". .. s /-nt, -It/, which are the only alveolar sequences which involve a ; c'u*\j. 0f voicing, are less prone to elision, the /XI often remaining as [?]: ' f"    nf down. Elision of a plosive medial in three or more is to be expected, e "    because of the normal lack of release of a stop in such a. situation, cue to its presence is likely to be the total duration of closure. iM be seen that in many cases, e.g. in / walked back, They seemed glad, . i's    0f word-final It/ or Id! eliminates the phonetic cue to past tense, . nsation for which is made by the general context.
(.IMP-"
f final It/or lál is rarer before initial Ihl, e.g. the alveolar stops are more »■ x -rh retained in kept hold, worked hard, East Ham, reached home, gift horse, ,, jme, grabbed hold, round here, bald head, jugged hare, changed horses, S^jii. i hJudi, moved house. Final /t,d/ followed by a word beginning with /j/ jj;■ u, ^iiy kept in a coalesced form, i.e. as /if/ and /dj/, e.g. helped you, liked wfciv r-i-' left you, grabbed you, first use, lost youth. S fl'- r °^ tne ntg&tive /-nt/ is often elided (see also §12.4.4), particularly j ,-illables, before a following consonant, e.g. You mustn't lose it /jo mAsn ti, Doesn't she know? /ckzn Ji 'nau/, and sometimes before a vowel, ;fj „ ]dn't he come? /wodn i 'kAm/, You mitstn't over-eat /ju mAsn aovsr' i;t/. ■tie— lmon is the omission of the stops in the negative /-nt/ component of inrtii'i'illables. e.g. He won't do it /hi waon 'du: it/.
(luMf rs oi word-final Itl and word-initial Itl or lál are sometimes simplified : in r..pl speech, e.g. I've got to go /aiv qots 'gao/, What do you want? /wods ju i *t i' rr /wndsu 'wont/, and less commonly lál before It/ or lál, e.g. We could
•r w ko 'trai/, They should do it /5ei    'du: it/. .   71  elision of one of a boundary cluster of only two consonants sometimes -c.n> n rapid speech, e.g. He went away /hi wen o'wei/, / want to come ._. i1 cm kAm/ (< /ai 'wonta kAm/, which frequently occurs), Give me a cake mi 9 -keik/, Let me come in /lemi kAm 'ml, Get me some paper /gemi sm reips/;. as well as the very reduced forms of I'm going to /'aim gsna, 'airjana, ir'&irjna/.-The Nt in of can be elided in rapid speech before a consonant, e.g. a /■it wofcake Is pits s 'keik/, (see also note 41 to §9.4.3(2)). Clusters in adverbs t-r.d with -ly are also liable to reduction in casual speech, e.g. stupidly »- "stjuipili/, openly /'aupani/, certainly /'s3;tni/.
12.4.7 Liaison
: ■ Linking Itl—As has been mentioned in §9.7.2(2)(a), GB introduces word-final post-vocalic Itl as a linking form when the following word begins with a vowel (and in a some cases in morpheme-final position before a suffix as: in bore /bo;/ boring /boirm/). The vowel endings to which an Itl link ■may be added are /£;,a;,o:/ and those single or complex vowels which may
Words in connected speech \ 3(7:
have a final [a] (/a,n,3:,ia,ua/), e.g. in far off four acei, „,, out, fur inside, near it, secure everything. Prescriplivjstl — the use of linking M to those cases where there is an <r>    -i ^'OB nevertheless many examples of linking /r/ occur where tfieiL i,* s!>=' the spelling, such /r/s being labelled as 'intrusive'. Such /r/s ar-'r particularly in the case of [o] endings, e.g. Russia and China,- \ f-,, , ^jL drama and music /draimar am 'mjulzik/, idea of /ai'drar       ">' jj Pakistan /indiar an pa:ki'sta:n/, area- o/"agreement /lc:nar -v . «i and rather less frequently after final /a:,o:/ e.g. law and order | -,;T ~ j*10* awe-inspiring /"oir inspaianrj/, raw rra'orc /rair 'Anjan/. Spe lin ness remains an inhibiting factor in the use of linking hi, general tendency among GB speakers is to use Irl links, even- , —among those who object most strongly.9 The comparative r; ■ 11_-. ,j contexts for 'intrusive' /r/s following /ai,a:/ tends to make 'Vs,!^ ft" aware of the 'correct' forms; thus I saw it /ai "sa:r it/, drcn.i'r rl-j'jmjBfe are generally disapproved of, though those who avoid such rr i:uiciMfc59vJ have to make a conscious effort to do so. The focusing .M > Iter.- or^^ 'intrusive' /r/s as an undesirable speech habit has led to tlv u<e fa speakers of a pause or glottal stop in such cases of vowel I iniu>. Ahh'aiSF5 result that, in avoiding 'intrusive' /r/s, they have also abandoned 1 .-u-/r/s in favour of a glottal stop or a glide between the abutting \.>v ei=] s.Lja£-j secure it [sifkjoa ?rt], War and Peace [wo: ?and 'pits]. As mi-'lvr m i\iceli^5-in those regions where post-vocalic III is pronounced and ;.o.'i ;'irn*%fflfe-1' identified as separate word forms in isolation, the tenderi..* v 11 irudys«i; intrusive /r/s is less marked than in GB or in. GB-influenced r.pe-, 11 speecesf
■ Aft 1
The same process is in operation whether the Irl link insert.-1 1, I ,»tonealfci-,' justified (linking) or not (intrusive). The examples below derm ii^-rate tl if 3t$|pj environment is phonetically comparable whether the Irl link is in .eiled oefrajkl
a suffix or before a separate word and whether it is linking or it 1 ii-n?
idea of it ai'drar av rt/
raw egg ra:r eg
the spa at Bath 5s spair at baiOv
iJI
stir	stirring	stir it in
/St31	'st3:nrj	'st3ir it 'ml
dear	dearer	my dear Anna
/dia	'drars	mat diar 'ana
roar	roaring	roar angrily
Ira:	'roirin	ra:r arjgrali
star	starry	a star in the sky
/sta:	'stairi	a stair in 5s 'skat
'strtajjjj n hvisish
phe insertion of M is obligatory before a suffix beginning with a vowel, vfiefe the Irl is historical, e.g. boring.
■he insertion of hi is optional, though generally present, before an hrnediately following word beginning with a vowel, where the hi is tstorical, e.g. pour it, over and over again /stiver and auvar a'gen/. Jier [a] an inserted Irl, even though not historical, is generally used efore a following word beginning with a vowel, e.g. vanilla essence /"Millar "esans/, vodka and tonic /vndkar an 'tomk/. .fter /a:/ and hi/ an inserted Irl, when not historical, is often avoided efore a following vowel, e.g. nougat and chocolate /nu:ga:r an 'foklit/, raw in the wind/stroir in 8a 'wind/.
I he insertion of hi before a suffix, where the hi is not historical, is uften strongly stigmatised, e.g. strawy /'strari/, gnawing /noinrj/.
There appears to be some graduation in the likelihood of occurrence if rsertrf Irl, as follows:
|L ,.._,[kally (as well as historically) the resulting hi closes the syllable rather
1«. I;.- ig initial in the next, e.g. the hi of more ice /mo:r 'ais/ is shorter than ~- .it nore rice /ma: 'rais/, the latter also being associated with accent onset v,nble pitch change (see further in §12.4.8 below).
-
C i 1 ii-ing [J,w]—In vocalic junctures where the first word ends in /ill, hi, lit, ^ ei. /ai/, or hi/, a slight linking [j] may be heard between the two vowels, , my arms [mat "'aimz], may ask [mei 'ja:sk], he ought [hi 'jo:t], annoy
Mr. fonoi 'jai6a], beauty and [bjulti ''and]. But this is not sufficient w he equated with phonemic /j/; indeed there are minimal pairs which > ■ -trate the difference between linking [j] and phonemic /j/, my ears 'haz] vs my years [mai 'jiaz], and / earn [ai 'j3:n] vs I yearn [ai ~J3in], '■■in iarly a linking [w] may be heard between a final At:/, Isal and /ao/ : '1 a following vowel, e.g. window open [windao ""aupan], now and then "nu :*oud "Sen], you aren't [ju: 'wa;nt]; and minimal pairs illustrating li 1'. ng [w] and phonemic IsnI can be found, e.g. two-eyed [tu: '"aid] vs . ■■ wide [tu: 'waid]. Alternative pronunciations, more frequent in rapid ch, in the case of the sequences of diphthong plus following vowel, ■1 vulve the absorption of the second element of the diphthong, i.e. of pr-   the fi] in the case of /ei,ar,ai/ and of the [u] in the case of /au,au/, .uvdng renderings like annoy Arthur /ana: 'a:5a/, my ears /ma: 'raz/, *-    window open /'winds aupan/ or / 'wmdar sopan/ (see further under §8.11 (8) above).
\ h \ei another possibility, the linking ['] or [™] may be replaced by a glottal * stop. This is most common before a vowel beginning an accented syllable, e.g. zvsr/ angry [veri '?arjgri] (see further §9.2.8), However, a glottal stop in such
"'■•es is not so often used as in some other languages, e.g. German, and is usually
'"'Siated in English with some degree of emphasis.
318  Words and connected speech
Words in connected speech 319
■c"arji l;,lli8tVfi
(3) Other boundaries—It is unusual for a word-final consonant -(. i
as initial in a word beginning with an accented vowel, the ide-tSi?^ words being retained (see §12.4.8). Thus, run off, give in, !e<^ """''vi /rA'nnf, gfvm, le'sofh/ (shown because the nuclear tone, us ,-1 citation, does not begin on the consonant); and get up, look ■■■ are not usually [ge 'thAp, lu 'k'aut, sh> 'p'o-igjoirj] (the plos strong aspiration characteristic of an accented syllable-initial no,,,. 4. or two phrases in common use do, however, show such fran-icr -home, not at all are often pronounced [a 'thaum, nut a T\->--" i| ™ considered as constituting, in effect, composite word form, ' *'
12.4.S Juncture
As we have seen in the previous sections words may be considerable ml at boundaries by factors like assimilation and elision. Nevertheless sdn'e p|u features may be retained which mark word or morpheme boundaries (geil referred to as juncture). Thus, the phonemic sequence /piistoiks/ may mean 1 talks or pea stalks according to the different word boundaries (i.e. lor.s H or /pi: + stalks/). In this case, if the boundary occurs between /s/, [,d ;t'^« words peace and talks are established by the reduced /ill (in a syllable cloi<3| a voiceless consonant) and by the aspiration of III; on the other hand, if tin- biureS occurs between /hi and Is/, this may be signalled by the relatively full icngjltiffli (in an open word-final syllable) and by the unaspirated allophone of/l < foilovij /s/ in the same syllable) as well as a stronger /s/ word-initially tl .111 i-.,ird-fi<s The following examples illustrate various ways in which phonetic cueu mark word boundaries:
(a) I scream ice cream
(b) why choose white shoes
(c) a name an aim
/ai skrkm/ /ais kri:m/
/wai tfu:z/ /wart Ju:z/
/a nenn/ /an enn/
1'i.ig ls.il, strong /s/, little d^\ui£jig eff iduced /ai/, weak /s/, de^ o '
long /ai/, short [/] as element i-r tp reduced /ai/, long /J7 *
relatively long /n/ word-initially relatively short In! word-finally 1 possibility of glottal stop before;:.
The glottal stop before a vowel beginning an accented syllable in     "uit exam?; is optional and generally not used unless emphasis is required (see §'>.- 8;. Similarly, a sequence of words may be distinguished from a single »uf(t
ues are only potentially distinctive and may not be present in cons'*'1"1 Jocularly rapid) speech or may have only slight phonetic value. In any ■ctcd {0 word identification are merely additional to the large number
by Ibc context
5»
j j Stylistic variation10
(a) nitrate night-rate
(b) illegal ill eagle
/nartrert/ /nait reit/
Mi:gl/ III i:gl/
devoiced hi little devoicing of Irl
. clear [1] before vowel : dark [I] in word-final position possibility of glottal stop before ,
K» ^ features of connected speech discussed in this chapter are common in !",u i) speech of native speakers of English and the lack of such features »'he-abnormal. But GB is not a monolithic accent and displays considerable *J '    even within the speech of one speaker, particularly in the use of the detailed here. Many factors influence this variation and a major factor 1, ,jf discourse, e.g. whether a speaker is being careful or casual, slow or «r j Moreover the average rate of delivery" differs from speaker to speaker ''a „ ..-,s of discourse style.
'donation
1:1 all styles of speech, simple falls in pitch (whether from a high or a mid starting-point) account for the majority of nuclear tones, between 60 per cent tad 70 per cent in most conversations.12 The falling-rising nuclear tone tccounts on average for roughly 20 per cent. Thus it may be seen that speech ixhibiting a large number of rises or rise-falls is conspicuous in this respect, 'usual speech has longer intonational phrases and contains fewer accented jyllables than careful speech. Careful speech often shows a concentration >f fall-rises or simple rises, e.g. If you pull them off' I and put them in a yjass of rwater / they grow little 'roots I and then / you plant them in soil I mdthey ,grow I and then you've got a 'nother spider plant. 1 Weak forms
Hie use of strong and weak forms does not appear to be a matter of style accept insofar as the more frequent occurrence of strong forms in more weft)I speech results from additional accents. The use of strong and weak Mis is entirely regular in both careful and casual styles of speech: weak orms occur unless the grammatical word is accented. Since IPs are shorter n careful speech, there will be more accents and hence more strong forms. .mking Ixl
Is with weak forms, linking Ixl is frequent in all styles of speech, though an Ixl link is not necessarily used on every occasion where such an insertion would be possible. (See §12.4.7.) Its occurrence is of no stylistic significance. (The avoidance of so-called intrusive Ixl results from a deliberate carefulness shown by some speakers.) ) Assimilation
Assimilations occur in all styles of speech. But unassimilated forms generally occur more often than assimilated forms, which tend to increase in frequency in the more casual style of speech. But rate of utterance on its
"MO
320  Words and connected speech
own does not govern the use of assimilation. One speaker 'jAtirj/ for fust shutting when speaking carefully, neverthel't h for horse show when speaking rapidly. Speakers use palat,,.,,! ''. lations (of the kind /'speif JXtl/ for space shuttle) and bilah , |' ^ (of the kind /6ap 'p3isn/ for that person) less commonly tli„ii -uT^' assimilations (of the kind /Jb:k 'kAt/ for stor/ ci/f). Such lf|Jr ' is also more common than coalescent assimilations (such i as in /nauti'dsntsmsn/ for noted yachtsman or Izl +       ,  ^ ^ for because you). But coalescence is frequent in common the auxiliary plus pronoun of phrases like didyou, can't yoi< d.iW^ and may occur even in careful speaking, e.g. Would you     u ,UA /'wudju'laik s'kAp 3v ,ti:/. "
(5) Elision i Elisions do show some correlation with rate of delivery, h all MJ become more frequent as the rate of utterance increases,  m WJ, careful speech they are almost entirely regular (e.g. alveolar plusr/ed be elided interconsonantally, hi in pre-nuclear unaccented v,iLbir» in unaccented non-initial grammatical words—see §12.4.6 i>i (..1-ualj they are less rule-bound and may contain unpredictable elisi 11« ■ uch a| of /I/ and IN in Well, that's all right /we 'ats o: ,ratt/.
(6) Co-occurrence of phonemic features of connected speech i The occurrence of hi links, elisions and assimilationsupturn^ sense that when the appropriate phonetic environments occit, \1u,j p; may or may not operate. If such processes do operate, ilu. w.ll the regular patterns described in §§12.4.5-12.4.7. Uiteraiur efur sii both assimilation and elision in conjunction together. In wmd-liMi pos| after the elision of a final Itl or Id/ the remaining fricamc oi OanQk be assimilated to the initial consonant of the following i'.chI. eg. nf shop /kbuzd 'Jnp —> klauz 'Jr>p —> klaU3 'jbp/, hand ma,' hand met nan 'meid —» ham 'meid/ and just shutting /d$Ast 'JAtirj -d5Aj' JXtirj/.
(7) Plosive release An important type of (non-phonemic) variation concer i: plosives, particularly the voiceless series. As noted in §9. usually has an inaudible release when followed by anothr -.n But in careful speech, there is a marked increase in the m mt released plosives, e.g. / looked quizzical [ai Iukth 'kwizikal | Vt their final stops more than men.13
12.6 Frequency of occurrence of monosyllabic and
polysyllabic words if
The following percentages of occurrence of words with different -lunibs syllables were found in one corpus of conversations:14 1 syllal-.e -K- I"-1' *cr%
ilie i
Words in connected speech 321
_ 15 per cent; 3 syllables—2.7 per cent; 4 syllables—0.3 per cent, vllabtes—0.03 per cent. When the 1,000 most common words used W" 14 it was calculated that some 15 per cent admit of the kind of i:" Variability mentioned in §10.9 and §11.3. Half of such words permit-^i'It flic variation were monosyllables whose phonemic structure depended "^foe degree °f accellt placed upon them, i.e. most words with phonemic were function words.
m-- ■■
jjj""Ady''e to 'ore'8n learners
fc^gti learners need not attempt to reproduce in their speech all the special forms of words mentioned in the foregoing sections. But those aiming speaker competence should observe the rules concerning weak forms, Rife! cultivate the correct variations of word accentual patterns and should Biffs proper use of liaison forms, avoiding in particular an excess of pre-vocalic gtifstops. m addition, they should be aware of the English assimilatory jjjgBcies governing words in context, so as to avoid un-English assimilations {like that /ai laig ,6at/ (incorrect voicing) or / was there /ai ws5 'Set/ fekect dental modification of the place of articulation). In listening to native Htakeis they should be aware of the types of assimilation and elision which nas. teen described above; otherwise they will find it difficult to understand HRLj] of ordinary colloquial English. This knowledge is particularly important BLa>.- a second language is often learned on a basis of isolate word forms ifeeas in conversation these will be frequently modified. f*he foreign learner is recommended to aim at a relatively careful pronunciation Jcngksh in.his own speech but to be aware of the features which characterise
■is f-sual pronunciation, particularly by native speakers. The following dialogue sjjiss some of the differences which may be found between a more careful fed i more casual pronunciation:
What
do   you  think  we   should  do    this evening
'wot      du: jui 'Ornk  wi: Jbd
(2) 'wwfeu 'enjk  wi Jbd
B.       How   many of us   will  there be?
(1) 'hau   meni 'dv as   wil   oe; bn
(2) 'hau   mni sv as   1 3»
'du: Sis i'.vnuj 'du:   8as iivnirj
'bi:
There are  the  two   of us,    and probably the two
(1) 5er o     &8           3v as/   and   pretabh 5, tu;
(2) 591- a     &>   'tu:   av as!  m     'probbli 3« tu.
:      girl5 from next     door. That'll  be  four  of us/ already
(1)  gv.tz from 'nekst  'do:/ Sail      bi  ~fo:t ov os o:l red.
(2  gailz &m   neks    'do:/ 6*1      bi   "fb.r sv as/ o: redl
322  Words and connected speech
I
(1) ai
(2) a(i)
I've
(1) aiv
(2) a(i)v
I
(1) at
(2) a(i)
I
(1) ai
(2) a(i)
think '9irjk 6irjk
only aonli acini
they' : 5ei ; 5e:r
tallied 'to At 'ta:k
wonder 'w/uidsr 'wAndr
if if
if
to
t3
ta
we wi: wi
l nice i   'nais jatj i   'naij jAg
them once, Sam ,WAns/ dm wAns/
young couple, .kApl/ ,kApl/
but bat bat
can kan kn
try
'trai "tea:
and
an
m
should [bd
Jg
book some
'btsk sam
'bok sm
go 'gau 'gau
to ta ta
seats
,si:ts/
silts/
Notes
1 Native speakers often make written mistakes of the sort 'I could of j'i i „■ this neutralisation,
2 This will apply only for those speakers who have appreciable rounding o and semi-vowels. See §8.9. LI for the development of unrounding of/u:/
3 For some speakers Irl lias inherent labialisation and will not be lip-spread a lip-spread vowei.
4 See Corm (1990), See nasopharynx opening videos 6.5-10, 8.11. 9.25. 10.|] 0
companion website.
5 See Nolan & Kerswill (1990). They also found girls less likely to assii < I i t.i»' and /n/ more likely to assimilate than Id/.
6 See also §12.5 for stylistic variation in the frequency of assimilatior
7 Byrd (1992b) found around 78 per cent of sequences of /s,z/ plus     -„\-.ic?.l \i palato-alveolar articulation only (in the timit database of American I'ngli' n effect from syntax, sex, or dialect,
8 Deterding (2005) in a study of newsreaders on the BBC World Service foil common in both suffix and stem-final /t,d/ and most common before ihiu i. ikwij% fricatives and nasals.
9 Noted as long ago as Sweet (1890: ix): 'Thus J know as a fact thai mo speakers of Southern English insert an hi in idea(r) of, India(r) Office. deny it'. A century earlier Sheridan deplored intrusive Itl: 'Anothei ■ u. the letter r to all proper names ending in a unaccented, as Belinda   I' Belinda, Dorinda' (1762: 46).
10 The information on speech in this section is based on Ramsaran (197X) who useddai drawn from twenty hours of recorded conversation involving six GB speakers.
11 The slowest rate of utterance recorded in conversation in Ramsaran (1978) was 1 sylls/min (3.1 sylls/sec, 7.6 segs/sec) and the fastest was 324 sylls/min (5 4 syl 13.4 segs/sec). Byrd (1992a) found men speaking 6.2 per cent faster than womer
12 These figures, taken from Ramsaran (1978) are slightly higher than trios.-§11.6.2.6.
13 Byrd (1992c).
14 Berry (1953).
15 Gimson (1969).
-1 ii.
addii iiJar, lof '
Kef"