Professional Documents
Culture Documents
From: L. Canepari (2005) A Handbook of Pronunciation - Lincom Europa
From: L. Canepari (2005) A Handbook of Pronunciation - Lincom Europa
Contents
p. √¤ Foreword
A Handbook of pronunciation
English, Italian, French, German,
Spanish, Portuguese, Russian, Arabic,
Hindi, Chinese, Japanese, Esperanto
1 1. Prelude
1 A synthesis of çNaturalÇ Phonetics “ Tonetics
16 Transcriptions
21 ˛e contents of the HPr (“ of the HPh)
23 Observations on phonetic terminology
28 Guide to the figures
34 Guide to di‡erent types of transcriptions
36 Transcribing by hand
37 Generic symbols
39 ˛e o‚cial IPA chart
41 canIPA “ correspondent oƒIPA symbols
46 2. English – American “ British (neutral “ other accents)
123 3. Italian
152 4. French (neutral “ other accents)
186 5. German (neutral “ other accents)
231 6. Spanish – Iberian “ American
257 7. Portuguese – Brazilian “ Lusitanian
282 8. Russian
314 9. Arabic
330 10. Hindi
345 11. Chinese – Mandarin
370 12. Japanese
392 13. Esperanto
1. Prelude
1.1.1. ©th the aid of some indispensable diagrams (mostly taken from the
HPh, but with simplifications and other modifications), we will provide the key
to adequately use the rich potentialities of Phonetics, with regard to the 12 lan-
guages dealt with in this book: English, Italian, French, German, Spanish, Portu-
guese, Russian, Arabic, Hindi, Chinese, Japanese, and Esperanto.
Above all we take into consideration voicing, vowels, consonants, and intonation
(including tones). For in-depth analyses, the reader is referred to the companion
volume HPh (A Handbook of Phonetics).
1.1.2. û 1.1 shows the glottis (which is the space between the vocal folds) and
the positions adopted during the phonation types we are interested in. In addition
to normal breathing, we have a complete stoppage (¤ the stop (ö), which techni-
cally can be neither voiced nor voiceless, but has more a‚nities with the latter
type, and will schematically be represented with Ï ), besides voicelessness ((f), Á ) and
voicing ((v), Ë ).
Furthermore, we also find lenis (or lenited] voicelessness and voicing (respective-
ly (É, v), Ù, È ), and mixed phonation (or half-voicing, (Ñ), with three schematic
û 1.1. Phonation types and positions of the glottis.
2 + +
™. lenis voicelessness Ÿ. lenis voicing
(É, s, h, ), A) Ù +
(v, z, H, ") È
3
ƒ. mixed phonation
(Ñ, Ω, ≈, Ø) {1 Î (pbX ), 2 Í
(bpX b), 3 Û (bpX )}: 1 (|'Êa, ap'Êa)÷
∆. creaky voice 2 (&aÊa'Êa)÷ 3 (aÊ'pa, 'aÊ|) ¤. falsetto
(Ca) Ú (Úma) ∏
2 a handbook of pronunciation
icons, which we will shortly see), where half of the interested phone is voiced,
while the other half is voiceless.
Generally, the çchoiceÇ –between the three of them– depends on context: after
a pause (or silence) or a voiceless C, the first half is still voiceless; whereas the sec-
ond half, which is in contact with voiced phones, is voiced: (Î ), as in German: Bett
/'bEt/ ('ÊEt). Instead, before a pause or a voiceless C, the two halves are exchanged:
(Û ), as in English: Bed /'bEd/ ('b™;fl). On the other hand, within words or phrases/
sentences, between voiced phones, the central part is the voiceless one, whereas the
two margins (each one for a fourth of the total duration) are voiced: (Í ), as in Ne-
apolitan-Italian pronunciation: dato /'dato/ ('dA;√&∂ø). We have added several par-
ticulars in û 1.1, where there are two peculiar phonation types, too: creak (Ú ) and
falsetto: (∏ ), which are useful for Chinese and Hindi respectively.
In § 4.1.7-12 of the HPh, we have explained how to verify and check if çvoiceÇ is
û 1.2. Di‡erent phonation types exemplified by some languages: American “ British English
(with mediatic British variants); Italian (with two regional variants: Naples “ Rome); French;
(Lusitanian) Portuguese; German; (Mandarin) Chinese; Hindi; Japanese.
1.1.3. As far as vowels are concerned, let us recall that from a phonetic point of
view it is more convenient to use the term vocoids, while reserving the more tradi-
tional terms for phonemes and graphemes, in addition to more general expositions,
as at the beginning of the HPh.
û 1.3-5 will help to çreconstructÇ the typical modalities for the production and
identification of vocoids, or vowel phones, which have three fundamental compo-
nents that – concisely– are: the fronting and raising of the dorsum (or central part
of the tongue), with di‡erent degrees of jaw opening, and di‡erent lip positions,
since lip rounding doubles the number of possible vocoids.
û 1.3. Orograms with the extreme points for vocoid articulations.
i π å u
û 1.4. Vocogram with the extreme points for vocoid articulations (and corresponding labio-
grams).
i u
π å
d .
n nd
de nd
ba al ro rou
un u
ro l ro
ce cen d
e
l
ra al
nt nd
nt tra
ck tra
ck tra
ck u
=
nt tr
fro rou
ce -cen
i
ba -cen
±
ba l
-
r
nt
nt
nt
ck
fro
fro
fro
Furthermore, û 1.6 will complete our general view, by providing all possible
vocoids, in orograms which contain a miniature vocogram, in order to help to see
4 a handbook of pronunciation
û 1.6. Vocoid orograms. unrounded
i Û … M {{*}}
I ¤ ¢ û {{ï}}
e Ù È X {{Ú}}
™ É ‘ x {{¢}}
E Ä å √ ∏
π Å a A å
rounded
Y y % ¯ u
è Y T ¨ U
{{∏}} ° + P o
{{π}} # ê Ö ø
{{œ}} § @ ∏ O
{{ì}} π ∏ ù Ø
1. prelude 5
i a u
i Å u
i Û … M {{*}} {¨ ˚ W} Y {% ¯} y {% ¯} u
I ¤ ¢ û {{ï}} {Ù : w} è {T ¨} Y {T ¨} U
e Ù È X {{Ú}} {È , „} {{∏}} {+ P} ° {+ P} o
™ É ‘ x {{¢}} {Í Ï ∑} {{π}} {ê Ö} # {ê Ö} ø
E Äå√∏ {É † ,} {{œ}} {@ ∏} § {@ ∏} O
π ÅaAå {Ω Ä a} {{ì}} {∏ ù} π {∏ ù} Ø
1.1.4. Also for consonants we follow a more scientific terminology, which de-
fines the consonantal phones as contoids, while reserving the more traditional
terms for phonemes and graphemes, in addition to more general expositions.
˛e three fundamental components for the production and identification of
contoids are: manners and places of articulation, and phonation types. ˛e latter
are extremely useful for contoids, since –depending on whether voicing is present
or not– they double their number, for distinctive –or phonemic– purposes, as of-
6 a handbook of pronunciation
ten happens with (t, d÷ c, G÷ f, v), which then form diphonic pairs, which are dis-
tinguished only by the di‡erent type of phonation.
˛ere are seven fundamental manners of articulation: nasal, stop, stop-strictive,
constrictive, approximant, trill, lateral. Traditionally (in spite of an actual and ob-
jective di‚culty, instead of the clearer articulatory terms that we use), some terms
of auditory origin, such as ça‡ricateÇ for stop-strictive or çfricativeÇ for constrictive,
are still widespread.
Depending on languages, some internal subdivisions are possible, as that be-
tween grooved (constrictive and stopstrictive) contoids, which are marked in com-
parison with the simple –slit– ones, from which they can be distinguished by the
addition of a groove along the longitudinal part of the tongue. For instance, (s, z÷
S, Z÷ q, Q÷ c, G) are grooved articulations.
Another important subdivision applies to trills, such as (r, K), which consist in
two rapid tappings (respectively of the tip of the tongue or of the uvula), and taps,
such as (R, ç), with just one tapping. Also flaps are possible, such as ([, ®), which
produce a rapid contact in passing to a fronter position, as can be seen from the
illustration (û 1.14.3).
û 1.9-15 present, according to the seven manners (with further internal subdivi-
sions), the contoid articulations found in dealing with the 12 languages described
in é 2-13, including the variants indicated. ˛is is why we call them çHPr con-
û 1.9.1. Nasal contoids.
˝ /m
)/m /[ /M
˝
/M /“ /"
/‹
˝
/M /˙ /n £/n
/N /¸ /º /ª
“
˝
/n /n /N /n
/˝ ˝
/~ /√ /N
˝
/∞ /” /˙ /Ñ
/\ /û /m ˝
/,
1. prelude 7
toidsÇ, in order to recall both the completeness of the HPr (in itself), and that these
consonantal articulations are not the only existing ones.
Rather, since there are çonlyÇ 200 in comparison with the 462 given in the HPh,
they are less than half (again, of course, considering them as mere çarticulationsÇ,
since potentially each one may produce a diphonic pair, with two distinct ele-
ments because of the addition of voicing. As a matter of fact, they almost double
the actual number of phones; there are 283 in the HPr, but 772 in the HPh).
/M /∫ /π /«
˝
p/b (/{ Q/Á p/b
]/7 ˝
t/d t/d 4/7
˝
T/D B/∫ ˛/Ã T/D
t/d ˝
+/_ ©/á £/8
˝
´/Ò k/g $/¢
˝
û 1.10. Stop contoids. ö
˝
›/
˝ ˝ ˝
p/ ∫/ ‡/ƒ ./…
˝ ˝
k/› %/ ∞/ w/
û 1.11.1.å.
˝
∑/ k/ (∫at) stop-strictive contoids.
8 a handbook of pronunciation
˝ ˝ ˝
q/Q q/Q ç/Ç fi/"
˝ ˝
C/‚ c/G &/1 C/G
ú/∂ º/Ÿ ˜/
˝
å/6 f/v 5/ç f/v
w/W ˝
†/∑ Q/z
/D
/ï /ø ˝
Â/J …/
˝
x/Ÿ /) ˝
X/º
¢/Ú
˝
s/z ˝
s/z †/Ã ß/fi
˝
ß/Ω ë/ò ˝
S/Z
∑/
»/« /, /y /m
˝ ˝
F/B /ñ /V /¶
/◊ Ï/ƒ ≈/¸ /⁄
˝ ˝ ˝
â/j /¥ /µ /°
∆/V ˝
W/w ˜/˜ x/‰
+
˝ ˝ ˝
/H h H ∆
/b /d /ã /%
/F /Æ /= /j
+ + + +
/S /Ã /¸ /Í
+ +
˝ + + ˝
/> /˛ /< /<
˝ “
/r /5 /5 {/D
˝ “ ˝
/R /R /R /e
+
Ó/[ r/m {V}/e /®
+ + + ˝
+
/l /R /Ï a/l
˝
+ + + +
/ı /] /∞ /l
+ + + +
˝
/$ /¯ /L Ñ/¬
+ + + +
˝ ˝
/L /ı /ù /)
In û 1.9-15, some orograms are marked with an eight-point asterisk; they indi-
cate the few articulations which actually correspond to the canonical or o‚cial
ones. Near those marked with ç“Ç another slightly di‡erent articulation appears,
which however is represented with the same symbol. In these synopses, very pre-
cise symbols are given, in order to satisfactorily couple articulations and symbols,
1. prelude 11
although for some of them more common symbols can then be used (as can be
seen in § 10.2-8 of the HPh]˘
In addition, û 1.16-7, with their labiograms, dorsograms and palatograms, are
the necessary help to distinguish –and observe well– all the nuances, which con-
tribute to di‡erentiate the contoids given there.
± +
±
+
=
≠
±
±
s˙z˙q˙Q t˙d l r
n
¿˙B˙Â˙© N˙Â
R L k˙g˙˙˙x
ë S « ¿/B
s, z÷ ß, Ω÷ S, Z †, ∑÷ Â, J÷ x, Ÿ ¸, j, w l, {L} ô !, ¡
†, ∑ s, z S, Z ‡, ƒ q, Q c, G
T, D l ô !, ¡
12 a handbook of pronunciation
û 1.18. Table of the contoids dealt with in the HPr. (@ indicates an çintermediateÇ articulation manner not present here, Ô é 10 of the HPh.]
@
`
&
F >
Ô
Æ
æ
ö >)
‰
Â
˘
ó
ƒ
ã
Ñ M
‹
_
ß
´
`
@
@
@ b { b
F
p ( p
bilabial
m[ mM‹M
b
bilabial rounded
palatalized bilabial
uvularized bilabial
labiodentalized bilabial
> fv 5 ç
]
ƒ
labiodental
V¶
7≥
√
labiodental rounded
S postalveolariz. labiodental r.
palatalized labiodental
f w†
v W ∑
uvularized labiodental
" ˙ n
◊ > ƒ
pro-dental
>‡ƒ
> fl∂
> sz sz
> ∂d
>d 7
q q
dental
Ï
t 4
QQ
d
Ã
l R
denti-alveolar
velarized dental
[5] [5] r
[R] [R] R
uvularized dental
t T
† ß
[Ï] l ≥
d D≥
D z≥
à fi≥
> £n N ¸ º ª n
. ≥
ç
alveolar
Ó ≥
r
a
≈≥
∫
¸
ô
¬
[
m≥
alveo–bilabial
labiodentalized alveolar
alveolar semi-rounded
B
∫
semi-paltaliz. alveolar semi-r.
V ≥
velarized alveolar
R
e
÷
5
ù §
ı ∞ l $
]
[n] N
uvularized alveolar
> ∂
> "
> Ω
ú
fi
postalveolar
ß
π
Í > ˛
®
Ã
e
postalveolar rounded
»
velarized postalveolar
∑ ë S « X x ‹ . / S À ¿
¯ L
> ö «
>ò Z »5 ç ¶ ÷ \ q = B
[T] [t]
Cc & C cv
[D] [d]
‚G 1Gg∆
n
postalveo-palatal protruded
postalveo-palatal over-rounded
postalveo-palatal (raised tip)
postalveo-palatal protruded
postalveo-palatal over-rounded
postalveo-prevelar
postalveo-prevelar protruded
postalveo-velar protruded
>Ѭ
> [ç
> {D
⁄ ≥
prepalatal
+
Á©
~
_
bilabialized prepalatal
labiodentalized prepalatal
√N∞ ”
> ºŸ
> ›
> âj 3 ¥
>J
>© £ ´
k% ∞
palatal
 …
n
L
,
ã % F
postpalatal
8
postpalatal rounded
prevelar
¢
Ò
Ú
provelar
µ ° V w≥
=
provelar rounded
>Ÿ )
velar
w
x
∆W
k
˙ Ñ \m ,
g
ı
ym
velar rounded
≥
j
labiodentalized velar
velar–bilabial
velar–alveolar
uvulo-postalveolariz. velar r.
<
> ˜˜ x‰ H hH ∆
∑k
º ≥
'
›
3≥
uvular
X
K≥
˜
r
pharyngealized uvular
pharyngeal
h·
laryngeal
ö
laryngeal rounded
Æ
æ
@
`
ö
&
F
‰
Â
˘
ó
Ô
ã
Ñ
_
‹
`
ß
´
1. prelude 13
Obviously, the total possible places of articulation are far more numerous than
those needed for our 12 languages. For each of them, in the respective chapter, a
table is given, which contains all the necessary contoids (and definitions) for the
neutral pronunciation of every single language.
However, the table in û 1.18 provides these contoids (except 7 complex coartic-
ulations, which appear among the orogram synopses, though), for a useful overall
comparison.
û 1.19 “ û 1.20 show the mechanisms of three particular types of explosion:
lateral, nasal, and inaudible, respectively for a stop followed by a homorganic lat-
eral or nasal contoid, or else by another di‡erent contoid, with an intermediate
phase, made up of an'articulation with two simultaneous occlusions.
û 1.19.
Lateral (å) “ nasal (å) tl, tÍ, t® (∫) tn, tó
(∫) explosions. ((Tæl, TæÍ, Tæı)) ((Tæn, Tæó))
û 1.20.
Inaudible explosion.
pt ((pæT)) kt ((kæT))
1.1.5. Intonation is the most evasive aspect of languages and (perhaps because
of this) it is generally very badly and superficially dealt with, that is when it is not
completely neglected (which would, sometimes, be preferable – rather than caus-
ing damage, or simply confusing readers).
In this synthesis, û 1.21-3 will help us to present intonation, without useless
–absurd and harmful– complications. In fact, without scaring (but also without
ba·ing or disorienting anyone), û 1.21 very intuitively shows the normal pitch
movements, in neutral British pronunciation, of the four intonemes of English
and the two most frequent preintonemes (of its four).
An intonation group includes a given number of syllables (formed by phones/pho-
nemes – so they are phono-syllables), which are combined in order to constitute
some rhythm groups, made up of the words occurring in a particular utterance.
1
See you on û 1.21.
Satur
day.
Iconic
2 tono-graphic
[Wi ll I]
se e you on d a y?
Sat ur examples.
3
[∑y won't they]
see you on Sat
urday?
4
don't] see you total di
[If I on Sat d ay… [it'll be a sa s
ur te r.]
5 don't] see you worry a
[If I on Sa t urd ay… [don't bout
it.]
14 a handbook of pronunciation
˛e most normal and frequent intonation group is composed of two parts: a pre-
intoneme and an intoneme. Obviously, the preintoneme precedes the intoneme,
which concerns the last strong stress of the intonation group. ˛ere are three marked
intonemes: conclusive /./, interrogative /?/, and suspensive /÷/. A fourth unmarked
–continuative /,/– intoneme completes the inventory. ˛e first two, as shown by the
first two examples in û 1.21, are respectively used to state or to ask something. ˛is
kind of question is defined a total question, because the answer –which regards the
whole question, in its totality– must be Yes or No (or Perhaps, I don't know, Â]˘
1.1.6. Our third example is a partial question, because it inquires about why (by
taking the rest for granted, or known) and its answer cannot be simply Yes or No.
˛us, there is only one part of the question, which is the one marked by the inter-
rogative word (such as: when, why, where, what, who, how, Â]˘
As can be seen from the movements in the third example, the intoneme that has
to be used, in partial questions, is not the interrogative one at all (as, however,
schools lead us to believe, with their grammars, and even the recordings of lan-
guage-teaching courses!). On the contrary, the appropriate intoneme is a conclu-
sive one, whereas its preintoneme is actually interrogative indeed.
As a matter of fact, a statement like ∑en you come back again (as an answer to
a question like ∑en will you tell us about it?) is intonationally di‡erent, from the
very start, from ∑en'll you come back again? In fact, even before hearing /®/ (Í) 'll,
which is the only syntactical di‡erence, we can perceive that /'wEn/ (5w™n:) is
already di‡erent in the two examples, because in the question it has a slightly ris-
ing pitch movement (while in the statement it is level), as can be seen from û 1.23.
˛e interrogative preintoneme begins with /¿ / (¿ ), whereas the normal one has
no particular symbol. (Here the space after the isolated symbol helps to identify
the preintoneme; but in actual examples it does not appear at all.)
˛e statement can di‡er also because it may have a slightly weaker stress and
di‡erent pitch: (&w™n) up to (w™n), and to /wÈn/ (wûn) as well (with a clear seg-
mental change). Here are the transcriptions of both sentences (including assimila-
tory coarticulation for /n/ (~) + heterosyllabic /j/, while /-n® j-/ = (-nÍ j-), as a com-
promise): ∑en you come back again (5w™~: j¯&khåm'bπk û'g™n:3 3), (&w™~5j¯;u, w™~-
5j¯;u, wû~5j¯;u), and ∑en'll you come back again? (¿5w™nÍ j¯&khåm'bπk û'g™n:3 3).
(As some readers might have observed, in our inter(dia)phonemic transcription
of English, we use both /l/ and /ı/, although they are not in opposition, (l, ı);
whereas, an intraphonemic kind of transcription would give only /l/, with (l, ı).
Instead, we also have /®/ (®, Í), the latter occurring before /é/ or /j/. Our transcrip-
tion is diaphonemic too, since it also shows di‡erences between American and
British English, as for instance in no /'nOU/ ('n‘;¨)a ('nø;¨)b.)
˛e last two examples illustrate the suspensive intoneme, /÷/, which is used to draw
attention to what one is going to say (or not to say) in a kind of suspense, and the
continuative one, /,/, which instead does not produce this e‡ect, as it simply divides
the utterance (just in order to continue). ˛is is done either to avoid strings which
are too long, or to subdivide them into parts which present semantic cohesion be-
tween the elements of each group, in comparison with those of another group.
1. prelude 15
1.1.7. We can now carefully see (or see again) the tonograms of the intonemes
and preintonemes given in û 1.22-3. ˛is is also useful to run trials in order to
check how much we approach –or not– what is shown there, depending on our
own spontaneous pronunciation, which may be more or less regional. It is likely
that the major di‡erences exist for the suspensive intoneme, which is the most var-
ied and çimaginativeÇ one.
In é 2-13 we always also give the imperative, /¡ /, and emphatic, /˚ /, preinto-
nemes. ˛ese are respectively used, above all, to give orders, to curse… or to ex-
claim, declare…
Lastly, let us observe that a continuative intoneme can substitute a conclusive
one, when it is needed to attenuate the impact of the latter, in order to be less
abrupt, or nicer, as also happens in partial questions addressed to strangers, such
as ∑at's the time? or ∑at's your name?, or How much does it cost?
For many other things, including paraphonics (which concerns states of mind,
attitudes…), it is necessary to refer the readers to the HPh or MaPI. We simply add
û 1.24, that shows pitch modifications in (low or mid) parentheses and in quota-
tions, that occur in the text of †e North Wind and the Sun, which we have used
for the transcriptions given at the end of each chapter (before possible appendix-
es, as for English and French).
û 1.24. Tonograms of parentheses and quotations.
1.1.8. û 1.25 helps us understand the relationship tone languages have with in-
tonation. As a matter of fact, the di‡erent tones (which are the realizations of par-
ticular tonemes, such as those of Chinese, for instance) obviously have their own
peculiar characteristics, as can be seen from û 1.26.
˛erefore, intonation is added to the existing tones, by modifying them in pre-
intonemes, according to what is shown in û 1.25. In the tonograms there, the grey
parts indicate how the preintonemes slightly deform lexical pitches (by delimiting
the available extents). In comparison with an emphatic preintoneme, /˚ /, the nor-
mal one, / /, is more compressed towards the middle part; an interrogative, /¿ /, is
raised, whereas an imperative one, /¡ /, is falling.
16 a handbook of pronunciation
Obviously, also in intonemes there are some (even greater, Ô û 13.9 of the HPh)
modifications, which are indicated by the corresponding tonograms, for Chinese
and Japanese (é 11-2; or in the phonosyntheses of the HPh, for the various tone
languages given there).
Naturally, what has been presented here about intonation holds for the lan-
guages treated, with all due di‡erences –at a tonetic level– which are indicated by
possible observations and respective tonograms. ˛ese sections might seem to be
too short, but in actual fact, they provide all that is needed, whenever one ade-
quately knows the tonetic method, which is an integral –and necessary– part of the
phonetic method (within natural phonetics).
˛e tonemic symbols, /¿ ¡ ˚ . ? ÷ ,/, can be applied to all languages, since intona-
tion functions are shared, although their actual tonetic realizations are extremely
di‡erent, as can be seen in the individual tonograms given.
/ / ( ) ((” )) /¡ / (¡ ) ((»’ ))
1 /5/ (5) >1≥ 2 /•/ ('1) >Q≥ 3 /¶/ (ç2) >5≥ 4 /6/ (63) >Z≥
Transcriptions
1.2.2. However, it is immediately clear that –dealing with descriptions and teach-
ing– indications such as ç/aj, aw/Ç are rarely faithful and highly misleading, for such
di‡erent realities as English (aÙ, aÖ) (with centralized and non-high second ele-
ments), German (ae, ao) (with peripheral and non-high second elements), and
Spanish or Italian (ai, au) (with peripheral and decidedly high second elements).
Besides, transcribing German (OY) as ç/oj/Ç would mean completely ignoring
(or concealing) the fact that –in neutral pronunciation– also the second element
is rounded (and neither fully high nor front). For English it would be desirable
–once and for all– not to continue to conceal the evident fact that not only are (™I,
‘¨/ø¨) real diphthongs, but that (Ii, ¯u/Uu) are diphthongs too.
Instead, interphonemic transcriptions seek to take into proper consideration
the characteristics of each language, even though within a less rich and less precise
symbol inventory, which is however capable of using the similarities and di‡er-
ences among the various languages in a better way.
˛erefore, we have: English /aE, aO/, German /ae, ao/, Spanish or Italian /ai, au/.
A careful analysis of the vocograms in é 3 “ 5 will surely explain the notational
di‡erence between English and German. To complete the series given above, we
have English /Ii, EI, OE, OU, Uu/, and German /OY/.
1.2.3. ˛e most e‡ective phonetic transcriptions are the taxophonic ones. ˛ey
resort to every useful symbol (among the available ones for the di‡erent phones:
vocoids and contoids) and all the most precise prosodic elements, in order to man-
age to show the necessary nuances. Only in this way is it possible to make real and
valid comparisons between di‡erent pronunciations, either regarding di‡erent lan-
guages or di‡erent areas for the same language. Otherwise, everything becomes
approximate and definitely less useful, since people may think they are working
well, while generally they are simply deceiving themselves.
Even excessive simplifications, to help learners, are not the best solution in or-
der to teach–learn pronunciation really well. For instance, when English phone-
ticians continue to use (È) even in more accurate transcriptions (perhaps with dia-
critics), for all occurrences of /È/ (and perhaps ç/ÈU/Ç too), they miss the opportuni-
ty to show reality, as when to go˚ the man˚ further˚ are rendered as ç(tÈ'gÈU, ∑È'mπn,
'f‘:∑È) (or ('fÈ:∑È))Ç in British pronunciation, instead of actual (Tû'g‘;¨, ∑È'mπ;n,
'f‘;∑å) (as happens to the quite often reported cases of ç(i:, u:)Ç for (Ii, ¯u/Uu)).
1.2.4. ˛erefore, for those who use several languages, but also for those who uti-
lize one foreign language only, the most recommendable phonemic transcriptions
are the interphonemic ones, because they use symbols in a less arbitrary way. In
fact, they do not flatten reality, by only showing what is functional, but compare
di‡erent languages better, by considering similarities and di‡erences as well.
In Japanese, it is certainly better to use /M/ rather than ç/u/Ç, even if there is no
possibility of confusing them, since this language has no /u/. However, the use of
/M/ highlights the di‡erences from other languages which do have /u/.
˛e kind of diaphonemic transcription is important, as well. It generally rests
on an interphonemic basis, although an intraphonemic basis is possible, too (but
18 a handbook of pronunciation
these languages use the dental type instead). A phone has the function to maintain
consistency between the elements of a given pronunciation: it is substance.
1.2.7. On the other hand, sounds have their value exclusively in transmission,
their function being to enable human communication through sound waves.
˛erefore, a sound is a single emission, in practice unrepeatable even by the same
person. It can oscillate quite a bit, often producing quite di‡erent realizations: it
is matter.
As can be seen in § 2.4 of the HPh, one way to allude to the fact that actual
sounds are always a bit di‡erent would be to represent them with di‡erent fonts:
o, o, o, o…
In conclusion, many di‡erent –yet similar– sounds constitute a single phone.
˛en, in the context of a particular language, several phones, not wholly identical
(but with the alternation governed by fixed and systematic rules, which can and
should be discovered and then explained simply and completely), constitute a pho-
neme. ˛e phones referred to by a given phoneme are called taxophones (or combi-
natory phones, or çallophonesÇ {a more ambiguous and less advisable term, since
it implies modifications not necessarily due to combination, but simply any sort
of di‡erence, for any reason, whether general or random}).
Sometimes, in the course of listening to recorded materials, certain sounds still
need to be disregarded in the process of constructing the phonetic inventory of
that language. ˛is is because it is possible for single speakers to occasionally pro-
duce sounds presenting abnormal deviations, whereas it is essential to consider
mainly what is more typical and frequent.
1.2.8. ˛erefore, actual sounds are practically infinite, and phonetics and pho-
nology would be decidedly complex if it were not possible to rely on the systemat-
icity of phones. In fact, the set of phones is the result of a past classification and
structuring of sounds (abstracting from the unrepeatability of the same sound).
˛is set allows us to reach recognizable types, which can in turn be represented
through precise phonetic symbols, regardless of any particular language.
˛us, what is similar in di‡erent languages and dialects is realized by phones
and their symbols, which make it possible to compare di‡erent languages (an es-
sential point in order to be able to learn and describe those languages).
˛erefore, the phonemic representation of single languages necessarily uses a se-
lection of (phonemic) symbols, with distinctive functions and purposes, even if
–of course– some symbols may be the same but with rather di‡erent phonetic val-
ues.
˛us plain phonemic symbols (generally chosen among the most common, as
the o‚cial IPA ones, ¤ International Phonetic Alphabet, or oƒIPA), do not repre-
sent pronunciation exactly, but rather the relationships between the phonemes of
a given language.
˛ey are useful for the specific purpose of keeping the current writing system
distinct from the phonic level. ˛erefore, phonemic transcriptions make it possi-
ble to avoid interference deriving from not knowing orthographic çrulesÇ, or from
20 a handbook of pronunciation
1.2.9. In the books which do not completely ignore it, intonation is usually
treated after vowels, consonants, and stress (and other prosodic characteristics,
such as length) – this is due to the greater di‚culties involved in describing intona-
tion. However, intonation should not be ignored, or relegated to the end in teach-
ing˚ since it is inseparable –in actual language– from the other elements.
˛erefore, phonetic transcriptions, in the strict sense of the word, simplify real-
ity a bit by indicating it only partially; in fact, it is as if they indicated –in any
case– a continuative intoneme, as in Italian ('tan:to2), ('tEm:po2), (a˙'ko:Ra2). (˛e
dot at medium height indicates the presence of intonation, even though of an un-
marked type. ˛e symbol (:) denotes length.)
However, in tonetic and phonotonetic transcriptions, the notation (') represents
not only stress but also a mid-range pitch (both in the context of tones and in that
of intonation) – it is thus in contrast with other signs such as (5), (ç), (6), (¶), Â.
˛e examples given –also in the chapters on the 12 languages– are of a phonet-
ic type, with possible durations, but without intonation. For instance, to a French
transcription like (pÅ'ºi) /pa'Ki/ Paris we will have to add an adequate pitch, as if
it were –say– in a conclusive intoneme. ˛us, we ought to have at least (pÅçºi)
(since another slight di‡erence exists, Ô § 4.3.5).
However, the low pitch on (çºi) depends on its occurring in the stressed sylla-
ble of a conclusive intoneme; thus, in actual fact, we ought to transcribe it as (pÅ-
çºi3 3), that gives us intonational information, which has to be added when neces-
sary (as for a conclusive intoneme), provided we already fully know the intonation
characteristics of the language.
prasegmental– level), as can be seen from the examples associated with û 1.26:
(5ma) /5ma/ ma çmotherÇ, ('1ma) /•ma/ má çhempÇ, (ç2maa) /¶ma/ m∑ çhorseÇ, (63ma)
/6ma/ mà çscold, curseǢ (In é 11, we will see that Mandarin Chinese also has a çze-
roÇ toneme and very important taxotones, too.)
1.3.1. ˛e HPr is about applied phonetics, since (as has been said in § 1.1.1) it
gives accurate and fairly extended descriptions of 12 languages (¤ English, Italian,
French, German, Spanish, Portuguese, Russian, Arabic, Hindi, {Mandarin} Chi-
nese, Japanese, and Esperanto) by putting into practice what is explained in the
HPh. We thus include the crucial phonemic component –constituted by function-
al phonetics– which is realized with phonemic transcriptions given together with
phonetic transcriptions, so as to show the relationship between phonemes and
phones˘
Together with these two types of notation, the o‚cial spelling of each language
is provided (or a transliteration in the case of the four Asian languages). Spelling
is given last, so that it will interfere as little as possible with the e‚cacy of the pho-
netic method and the associated transcriptions – thus the reader will come upon
the spelling only after having fixed in mind the phonic structures. As will be seen,
pronunciation variants are also given, and these will be worth studying.
Natural phonetics is an artistic science, and the best way to extend its rich poten-
tial concretely (after the necessary fundamental premises) is through applying its
techniques and knowledge to languages which are in great demand for teach-
ing–learning. In fact, these languages can be taught to others, or they can be self-
-taught (while remembering that, in order to teach adequately, it is necessary to
have learned well first).
˛erefore, it is best to begin the course of phonetic analysis with one's own lan-
guage. ˛e language one speaks is frequently a regional variant, more or less strong-
ly marked. Consequently, it is important to understand how neutral pronuncia-
tion operates and how it di‡ers from one's own pronunciation. ˛is method helps
to build consciousness of one's own speech patterns. ˛ese patterns can then be
compared with those of the neutral pronunciation, which exists in every language
of culture, even though speakers and society rarely have a consistent and precise
idea of what it is.
1.3.2. In every society, at least when there is a written language and literary pro-
duction, a neutral pronunciation exists, which is the pronunciation used by profes-
sionals in public speaking (particularly actors, presenters, and announcers of high
quality). ˛e Danish linguist Otto Jespersen declared that the èbest¶ language is
spoken when the social and regional origin of the speaker are least obvious. Neu-
tral pronunciation brings this çmiracleÇ to pass, sometimes even to the point of
making the identification of social and regional origin completely impossible.
˛e çsupportersÇ of regional pronunciations, who encourage individuals to
22 a handbook of pronunciation
maintain their çgenuineÇ pronunciation (even when they carry out çpublicÇ profes-
sions), deceive –themselves as well– because of their lack of success in bettering
their own pronunciation (a task which would have required commitment and per-
sistence).
In other cases, they fail to understand the importance of being able to free them-
selves from a heavy burden (often extremely distasteful to the individual in ques-
tion, at least on an unconscious level), all the while without renouncing their ori-
gins or identity. In fact, regional origins can be exhibited more e‡ectively by alter-
nating between one's dialect and the national language (competently, and above
all by choice). ˛e least preferable approach is to remain prisoners of an uncontrol-
lable hybrid, which is neither the o‚cial language nor a local dialect. In Italy and
the German-speaking countries, dialects di‡er largely not only in terms of pronun-
ciation but also of grammar and vocabulary too; in fact dialect –rigorously speak-
ing– refers (or should refer) to these other di‡erences, while accent predominant-
ly involves di‡erences in pronunciation.
In the other 12 chapters of the HPr, therefore, we will apply the phonetic meth-
od, which consists of a careful and rigorous comparison of the phonemes˚ phones˚
and intonation of the languages described, listening to good audio recordings, us-
ing the two types of transcription (etic and emic), analyzing the phonetic figures,
and showing commitment. ˛e last quality is clearly indispensable: it is not enough
simply to want to know – a sustained e‡ort to learn according to an e‡ective meth-
od is also required.
tation of the process employed by the human ear, which manages to compensate
in an extremely e‡ective way for the many irregularities, whether objective or inci-
dental, which can occur. ˛e results are then standardized in tonograms, by tak-
ing a sort of average of large numbers of utterances.
1.3.4. In the field of terminology, as well, scientific rigor is a great help, while
the vagueness and lack of clarity of certain obsolete traditions with little scientific
foundation is decidedly negative. On every page of the HPr (“ HPh, as well) this
requirement is addressed continually – indeed, everything becomes simpler and
easier to understand when the technical terms are clear and intuitive (even more
so than in normal speech). For example, tonic should refer only to tone –pitch–
and not to stress.
A diphthong should contain only vocalic elements, such as in ('ai), not vocalic
and consonantal elements together, as in ('ja). Were it otherwise, it would follow
that ('la, 'ma, 'sa) would be diphthongs as well – in fact, while (a, i, u) are vocalic
elements, (j, l, m, s, t, r, h) are consonants. ˛us, contrary to the opinion broad-
ly promulgated by grammarians, Italian ('ia) –for instance– is a true diphthong as
well (Ô § 5.2-3 of the HPh).
˛e concept of a phonetic syllable also remains overly subject to the influence
of writing systems and of grammatical and metrical traditions. It is natural that
books on linguistics and dialectology should speak about the phonic side of things
as well, using transcriptions. However, a minimum of rigor would be of great ben-
efit, since otherwise there is the risk of spreading and reinforcing unmitigated er-
rors, which compromise and discourage e‡ective learning.
For example, with the third millennium already here, books are still published
which give the provisory IPA table (of 1993), naturally with the mistakes includ-
ed as well (corrected in 1996), when it would have been a simple matter to down-
load, or refer to, the (currently) definitive table shown on the o‚cial website. Ob-
viously, this would only bring us to the level of the o‚cial position, which is hard-
ly satisfactory; but at least major errors and embarrassingly naive missteps could
be avoided (within the limits of the incompleteness and ingenuousness of the o‚-
cial table)… ˛ere are also those who succeed in producing brilliant achievements
X (¤ the monogram with a sort of çumbrellaÇ on top, in the place of the
like ç/c/Ç
normal /c/, or the possible ç/tXS/Ç).
As we have said in § 8.5 of the HPh, we prefer not to avoid the perfectly gram-
matical use of su‚xal comparative and superlative for front and back, contrary to
mostly non-phonetician common usage.
1.3.5. It must be quite clear that doing phonetics means giving a symbol to a
sound. ˛is task is not so trivial as the man on the street and perhaps acoustic pho-
neticians and theoretical phonologists might imagine. In fact, giving a symbol
to a sound necessarily implies several successive and linked phases, in which hear-
24 a handbook of pronunciation
ing, mimicry, kinesthesia, comparison, adjustment, and mnemonic storage are all
activated.
≈rst of all, it is necessary to be capable of su‚ciently perceiving the sound, so
that it can be identified with a precise phone which can adequately represent it.
˛e next step is to be capable of reproducing the sound using the appropriate
phone, which is carried out especially through imitation which is immediate, ¤,
occurring promptly after hearing the sound.
˛e third step, which is indispensable, is to produce the phone using kinesthe-
sia (¤ consciousness of the necessary articulatory and phonatory movements), e-
ven when the auditory stimulus is not close at hand. However, auditory memory
can still be an important guide, whether it is particular, referring to a specific
sound in a given language, or general, involving comparison with similar phones
with the help of experience in listening to and producing the phones of many lan-
guages.
In this manner, it is possible to produce a particular phone even days, months,
or years after having heard it (and to be reasonably sure of producing the correct
phone, even in the case of languages which one has never heard). ˛e secret of
good notation is for it to be realistic, and consequently, truly useful.
As a matter of fact, the fourth point, which is fundamental and decisive, is –as
our çdefinitionÇ suggests– finding a way to symbolize the specific phone by choos-
ing the most appropriate symbol out of several hundred (not just a few dozen) ele-
ments. If, after careful consideration, none of the available symbols is capable of
decently representing a particular phone, it becomes necessary to find its position
with regard to all other known phones. In this way, it will become possible to de-
cide whether it is truly a new phone in need of a new symbol. If a new symbol is
necessary, it can then be designed by following the general criteria of necessity, dis-
tinctness, and availability (as can be seen in the HPh).
1.3.6. ˛erefore, doing phonetics means managing to truly enter the phonic
system of one or more languages, thanks to a rich array of symbols as well. ˛e
symbols of the o‚cial IPA (oƒIPA = an o‡ alphabet! – Ô § 7.4 of the HPh) are not
at all su‚cient, and they create the illusion of successfully doing phonetics, while
all that has been achieved is at most a bit of phonology. Far too often, these stud-
ies proceed without the minimal understanding of what the phonetic structure in
the case in question actually is.
Clearly, it is phonology which is a part of phonetics (Ô § 1.9 and above all §
3.1.3 of the HPh), and not the other way round as some believe. In fact, within
the rubric of phonetic analysis and description, there is a functional component
as well. ˛erefore, functional phonetics (or phonemics) is an indispensable part
of phonetics, but only a part. ˛ere is little which could be done working with
phonology alone, just as there is little which could be done with just acoustics. In-
stead, what is needed is a global vision: articulatory, auditory, functional, descrip-
tive, and contrastive (while verifying various characteristics acoustically).
On this subject, it is interesting to note that the phonotonetic data of the HPh,
and of the HPr, have been compared with a sizable collection of acoustic data from
1. prelude 25
various sources, or at times, with data from a single source furnished by di‡erent
authors. In practice, the correspondence is dramatic, not only with data from a sin-
gle source, but also with data from multiple audio recordings analyzed accurately
and subsequently normalized. ‹en we speak of normalized data, we refer to the
practice of çaveragingÇ the results of multiple speakers and many utterances in var-
ious contexts, while keeping in mind phonological considerations and excluding
inappropriate samples.
˛e çdiscoveriesÇ of sociolinguistics also require normalization; otherwise there
is a significant risk of irremediably creating confusion, even when çscientificÇ da-
ta are used. ˛is fact has been responsibly demonstrated in several recent works.
We will not cite these sources (as çexamples to followÇ), simply because this ought
to be the normal way of doing things, not the çalarmistÇ or çscoopistÇ manner of
far too many publications.
At first sight, the Brazilian system could seem exactly the same as the Italian
one, as both have /i, e, E, a, O, o, u/. Instead, the two systems are di‡erent, particu-
larly since in Brazilian Portuguese (and also in Lusitanian, but with additional
di‡erences), even nasalized taxophones are expected ((i, e/™, A, 9/Ú, u), when fol-
lowed by (ö), ¤ a nasal consonant element); without these nasalized phones the
pronunciation would not be genuine (while in Italian it would be a regional one
if with nasalized vowels).
As for French, /i/ is in opposition with fourteen other vowel phonemes (/e, E, a,
O, o, u, y, °, §÷ í, Õ, Ú, ^/ and /ù/, ¤, the traditional ç/È/Ç), including the four na-
salized vowels (/í, Õ, Ú, ^/), which are true phonemes, in French, and not mere
taxophones.
Apart from these considerations, the actual phonetic realizations are not exact-
ly the same, even though we use the same phone (i). ˛e di‡erences can be seen
by comparing the vocograms (or vowel quadrilaterals) of these languages, in é 2-
13. ˛e same is true for the other elements which çcorrespondÇ.
1.3.9. If we now consider German and English, even without going into too
much detail (the details are available, of course, in é 2 “ é 5), it is clear that we
will have to take into account phonemic vowel duration. Instead, in the Romance
languages we have considered, vowel length is –practically– only phonetic (al-
though it varies considerably from language to language; consult the specific chap-
ters, or to see the di‡erences more rapidly, consult the transcriptions at the end of
those chapters).
In German and in English, the ç/i/Ç (çshort iÇ) is considerably more open than
in the Romance languages, giving (I) in German and (¤) in English, respectively.
However, even more pertinent to the present discussion of phonic systems is the
fact that in the Germanic languages, the opposition of duration is relevant (¤,
phonemic, distinctive) as well: German /I, i:/ (Schi‡˚ schief /'SIf, 'Si:f/ ('SIf, 'Si:f)); En-
glish /I, Ii/ (bit˚ beat /'bIt, 'bIit/ ('b¤T, 'bIiT) – as we have observed repeatedly, we pre-
fer a less abstract type of notation than the predominating form, which less use-
fully continues to give ç/i:/Ç, even for English, Â).
‹at emerges quite clearly from this discussion is the sorry state of those gram-
mar texts and language courses which çdescribeÇ the German /I/, or worse still, the
English one, as ça short i, as in French vite, or in Spanish listo, or in Italian fittoÇ.
‹en one considers that many Italian regional pronunciations have a vowel in fit-
to which is not at all short, the absurdity is evident! Of course, the same is true
when one tries to teach the close and short French (or Spanish, or Italian) /i/ (i)
by citing the same old misleading example: machine. Often, silence is golden…
1.3.10. Foreign and regional accents result from pronouncing a national lan-
guage according to the phono-tonetic system of a particular and recognizable area
(— of a particular and recognizable social group), especially in Italy, and in Ger-
man, Spanish, or Portuguese speaking countries. ˛erefore, it is important to be-
gin to carefully examine one's own pronunciation, so as to be in a position to work
towards the goal desired. ˛is task involves learning to analyze one's own sounds,
1. prelude 27
classifying them into precise phones (and transcribing them with adequate sym-
bols), and then seeing how these phones are in turn part of particular phonemes.
At this point, it is indispensable to have a reliable and careful description availa-
ble for the system of the language one is seeking to learn. ˛e necessary compari-
sons can then begin – they should be carried out objectively and systematically.
Obviously, it is imperative to follow an extensive series of specifically directed exer-
cises, followed by careful checks. ˛ese exercises consist of listening to good record-
ings and recording oneself to perform çmercilessÇ examinations of how much pro-
gress has been made – without cheating, otherwise all e‡ort is wasted.
For the pronunciation of Italian, Italians can rely on the seven chapters of the
MaPI treating regional pronunciations for the initial diagnosis of how regional
their pronunciation is. To (attempt to) achieve a neutral Italian pronunciation,
Italians (and foreigners as well) can use the rest of the MaPI (including the two au-
diotapes which come with it), together with the DiPI. ˛ese two çformulaeÇ refer
to two works by the present author (Manuale di pronuncia italiana {çA Handbook
of Italian PronunciationÇ} and Dizionario di pronuncia italiana {çA Dictionary of
Italian PronunciationÇ}, in the bibliography).
marked.
However, the real strength of the phonetic method resides exactly in being a-
ware of what to do in order to attain the (hopefully neutral) pronunciation of a
certain language, or of several languages. ˛is is so because we are not in the posi-
tion of a child any longer, when complete phonotonetic (and phonotonemic, and
also paraphonic) learning was right, including certain regional peculiarities, which
could be happily renounced – but that is the çbiological methodÇ!
≈nally, awareness has this further advantage as well – it allows one to choose
what is thought to be most advisable, between various possibilities.
1.4.1. ˛e orograms of the HPr (and HPh) use symbols according to certain con-
ventions, which make the orograms easier to understand (and tell apart). It is
therefore important to become familiar with these conventions, so as to be able to
make the best use possible of the rich iconic framework they provide. It is di‚cult
to understand why some çphoneticsÇ texts contain only a small number of illustra-
tions, or even none at all. It is certainly true, however, that it is better to give no
illustrations at all rather than to provide imprecise (or erroneous) ones. It would
be better still if certain books were not produced at all…
In the vowel orograms (Ô û 1.7), it is of great importance to pay careful atten-
tion to the location of the marker indicating the center of the back of the tongue.
It is even more important to take note of the precise location within the white (or
transparent) miniature vocogram, placed at the center of the oral cavity (with re-
spect to the still more precise location seen in the normal, larger vocograms), and
the shape assumed by the entire back of the tongue. In this way, the various vow-
el orograms can be compared (or just a subset of them, such as those relevant for
a particular language).
All this e‡ort should lead to a real understanding of the vocoid articulations and
of the di‡erent movements of which they are composed, so that an active panora-
ma is produced in the mind, and not simply a passive vision of things. In phonet-
ics, merely passive and memorized knowledge has little use – except to confuse
and discourage people!
Obviously, the true analysis and description of the vocoids of a given language
occur through the use of the large vocograms (vocograms in the strict sense), giv-
en that these diagrams succeed in showing nuances very precisely (as can be seen
in é 2-13, or in the phonosyntheses of é 16-23 of the HPh, as well as in the MaPI
and the various Italian regional pronunciations given therein).
1.4.2. ˛erefore, we will now consider what can be found in vocograms, which
should be observed, analyzed, and scrutinized calmly in all of their particulars. Vo-
cograms are full of details, without which it is impossible to come close to the
çspiritÇ of a language, manifested especially through vocoids, then through pitch,
and finally through contoids. Even a single millimeter makes a notable di‡erence
1. prelude 29
1.4.3. Markers can also be filled with grey in order to indicate variants (contex-
tual ones – the fundamental taxophones, pronounced using special phones; possi-
ble ones, such as those used in regional accents – geophones; or those related to var-
ious social groupings – sociophones). In neutral British English, for instance, there
are quite a few taxophones, especially including those produced when there is a
following (ı) – for example in hut˚ hull /'h√t, 'h√ı/ ('håT, 'h√ı:), or feet˚ feel ('fIiT, 'fi;Iı)
(while feeling has ('fIil¤˙)). It is impossible to continue to ignore such variations in
vocograms and phonetic transcriptions.
A representative example of Italian geophones is given with the varied realiza-
30 a handbook of pronunciation
tions of the phoneme /a/ (a), which include: (Å, A, ù, ∏, Ä, å, √). Some of these real-
izations can constitute sociophones, since they are more commonly found in
broader (instead of less broad) regional accents (these details can be found in the
MaPI in both cases). Grey markers can also have white centers in cases where they
refer to vocoids which can occur unstressed as well.
At times, it can be necessary to improvise a di‡erence in the marker or in its
shading in order to represent important realizations which depend upon the posi-
tion in the word with respect to word boundaries, stress, syllable structure, less
common use, or simple occasional variation. In this manner, it becomes possible
to avoid the use of supplementary vocograms. ˛e purpose of these special mark-
ers will be explained clearly, whether in a text placed close to the vocogram, or in
the main treatment. ˛e most common convention is the use of dashed lines, par-
ticularly for unstressed vowels which are represented by ordinary unshaded
(çwhiteÇ) markers.
For examples of these rules in practice, the reader is invited to consult the vo-
cogram of the çinternationalÇ French accent (Ô § 4.4.1), southern French (Ô §
4.4.3), or German (with its various accents: é 5), or Brazilian Portuguese, Russian,
Arabic (Ô é 7-10).
û 1.28. Markers for variants.
unstressed /È/ (û) unstressed /u/ (%)
stressed or unstressed /I, √/ (Ù, √) stressed or unstressed /U/ (P)
stressed /E/ (e, E)
1.4.5. Diphthongs are considered wide, when their figure contains a fairly long
line, and narrow, when the line is rather short. Besides these ditimbric (two-tim-
bre) diphthongs, beginning and ending with distinct vocoids, monotimbric (one-
-timbre) diphthongs also exist, with the second element in the same box as the first
element, but at a slightly di‡erent point of that box.
˛is last group consists of quite narrow diphthongs (which therefore are almost
like long vowel phonemes), and often the line is extremely short. Because the dis-
tance is so short, a dashed line can be reduced especially in such cases to a single
short segment, or to the black dot alone for diphthongs with rounded second ele-
ments.
Completing the survey, we come to vowel gemination, or vocoid doubling. ˛ese
cases involve vocoids which are neither short nor monotimbric diphthongs. A giv-
en vocoid is repeated within the same phonetic syllable, but without any move-
ment in the vocogram at all: (aa) (Ô û 1.30).
‹en the first element of a diphthong is the same as that of a monophthong
given together in the vocogram, it is possible to show the monophthong and the
diphthong together by showing the diphthong with a dashed line, instead of an
unbroken one (which would indicate the diphthong alone).
‹en variant diphthongs occur, they are denoted by grey markers together
with unbroken lines (or, in the case of an unstressed variant, the marker will be an
unshaded figure with the edge and line both dashed).
û 1.30. Short and long monophthongs and diphthongs with the same starting point (here, all
stressed).
(short or long) monophthong (short or long) monophthong
and (wide) diphthong (a, a:, ai) and (wide) diphthong with
rounded 2nd element (a, a:, au)
(short or long) monophthong (short or long) monophthong
and (wide) diphthong (a, a:, a™) and (narrow) diphthong with
rounded 2nd element (a, a:, aø)
(short or long) monophthong and monotimbric diphthong (a, a:, aa)
For example, in British English, beers /'bIÈ≤z/ ('b¤;ÈΩ) and bear(s) /'bEÈ≤{z}/
('b™;‘{Ω}) are in the same group as beer /'bIÈ≤/ ('b¤;å), even in accents which present
(éÅ, éa) for /éÈ≤/. ˛erefore, the best decisions are normally made considering
both phonemic and phonetic transcriptions.
1.4.7. In the vowel orograms (and in the vocogram) supplied in û 1.3-4, we give
the most extreme vocogram positions, with the express purpose of showing the
limits of the region of oral space used for vocoids. Instead, the orograms in û 1.6
give positions which are more commonly found in the languages of the world;
these are slightly less peripheral.
In fact, certain trapezoids (or even çvowel trianglesÇ) show all of the symbols
perfectly aligned along the edges, çthreadedÇ along the lines (rather like pearls,
since they are all circular as well), so that they extend outside of the margins. We
find such figures decidedly odd, given that the objective reality of vocograms is
quite di‡erent.
1.4.8. ©th orograms, which are fundamental for consonants, we follow sever-
al conventions, some of which are more intuitive than others. For example, it is
su‚cient to indicate nasals by showing the velum lowered, as in the cases of û
1.9.1-2. Nasalized articulations are indicated in the same way, including vocoids
(û 1.7, on the bottom). ˛ere is also nasal explosion (û 1.19, on the right).
Stops are shown with a raised velum, and as with nasals, there is contact between
two or more articulators (û 1.10).
Constrictive orograms show the articulators close to one another (û 1.12.1-2 “
û 1.32; “ û 1.1, ©, for (·)), together with a useful convention (albeit a bit less ob-
vious and objective) consisting of a horizontal black line, immediately above the
base of the orograms. ˛is line is intended to allude to the constriction – in this
case, the noise of friction produced by the flow of air through the narrowest point
of the articulation. If the line is not continuous, but divided into three segments
(as in the case of (,)), the contoid is semi-constrictive (intermediate between a con-
strictive and an approximant).
In the case of grooved constrictives (Ô û 1.32 “ û 1.12.1.∫ {and § 9.13 of the
HPh}), there is a curved line placed on the tongue blade. ˛is curve is meant to be
a reminder of the longitudinal groove that characterizes these articulations.
˛e curve (like the horizontal line) is segmented in the case of semi-constric-
tives. ˛e same marks naturally occur in the diagrams for stopstrictives as well.
1. prelude 33
slit † grooved s
In the case of approximants, there is visibly more space between the articulators,
and the horizontal line (used with constrictives) is omitted. However, there can
be a black arrow, slightly smaller than the one used for laterals, and this arrow sig-
nifies lateral contraction –lateralization– a trait accompanying and characterizing
some of the approximants (û 1.13.3). Semi-approximants have a dotted horizon-
tal line (û 1.13.2).
Trills, taps, and flaps (û 1.14.1-3) are marked with a black dot placed upon the
articulator which is in motion (whether this is the tongue tip, the uvula, or the
lips). Moreover, a dashed outline is added to trill orograms, and two dashed out-
lines to flap orograms (Ô û 1.33). In the same figure, we can see çsequential articu-
latory compositionsÇ as well, which show two rapid tappings for (r), followed by
two çopenÇ positions, with a white background.
For the tap, (R), there is only one contact, for a rapid tapping; whereas, the flap,
([), consists of three quite di‡erent (and quick) phases: firstly, the tip is brought
behind the alveolar ridge; then, while moving forwards, it touches the ridge (and
this is the pertinent articulation); finally, the third phase corresponds to the de-
tachment at every tapping, such as the two white ones in (r).
r = + + + –
û 1.33.
R = + –
Trills, (r),
taps, (R),
and flaps, ([).
[ = + +
Laterals are identifiable with an arrow placed on the most fundamental part of
the articulation. If the arrow is black, the contoid is bilateral (û 1.15.1); if white,
the contoid is unilateral (û 1.15.2). If the contoid is instead a lateral tap, a white
dot is shown (û 1.15.3). (Much more is to be found in the HPh, on other types of
laterals, with di‡erent additions.)
û 1.34.
(Bi)lateral, (l), and + +
unilateral, (ô), articulations. l ô
1.4.9. Stopstrictive orograms have a small black part, which refers to the stop
phase of these contoids. ˛e stop phase (as can be seen in û 1.11.1-2) is homorgan-
ic to the place of articulation of the constrictive phase, which immediately follows
34 a handbook of pronunciation
and constitutes the second part of these phones. ˛e result is a unitary (though
compound) phone, since its full duration corresponds to the length of other stops
or constrictives, not to the length of two phones combined in sequence.
Stopstrictives naturally have a horizontal line at the base of the figure. Grooved
stopstrictives have a curved line as well, representing the groove (û 1.11.1.∫). Stop-
-semi-constrictives have a black horizontal line divided into three parts. (Here we
indicate only the stop-strictives which occur in the 12 languages dealt with in this
book. Other languages present further types too, which are dealt with in the HPh.]
1.4.10. Palatograms (û 1.17, two lower rows) are used to show contact with the
palate throughout the course of an articulation. Palatograms are mainly useful for
certain contoids. It would be possible to employ them to add detail to the de-
scription of vocoids, particularly those which are not back. However, vocograms
and orograms are more useful in this case for purposes of description and teach-
ing. ‹en parts of the palatogram are shaded black, this convention signifies full
contact of the articulators (in the stop phase of stopstrictive contoids). Grey shad-
ing indicates instead, naturally, constrictive contact (as in the characteristic con-
strictive phase of stopstrictives). If the palatograms of the constrictives, (†, s, S) are
compared to those of the corresponding stopstrictives, (‡, q, c), this di‡erence
becomes clear immediately. It is useful to observe carefully (in the last part of every
set of diagrams, in û 1.17) the palatograms –and dorsograms– which show the
di‡erence for the voiced alveolar articulation between stop, (D), and (bi)lateral,
(l), in addition to unilateral, (ô), and constrictive, (¡), as well (although these do
not occur in the 12 languages of the HPr).
1.4.11. Dorsograms (as in û 1.17, first row) give a new perspective, which is not
longitudinal but instead transversal. ˛ese figures are particularly helpful to illus-
trate the di‡erence between slit tongue position (the unmarked position, since it
requires fewer phonic features) and grooved tongue position or lateral contraction
(or lateralization) – these are the two marked possibilities, in comparison with the
unmarked position.
‹en arrows are used in labiograms given in profile (as in û 1.5 “ û 1.16), they
indicate the direction of the characteristic movements, as produced by particular
facial muscles. Frontal labiograms require no particular explanations (Ô û 1.8 and
those just mentioned) – the vertical gap, which is steadily larger as the jaw opens,
is fairly intuitive. By carefully observing û 1.16, let us consider attentively the labi-
al di‡erences for (ë, S, «) (keeping in mind what has been said in § 4.2.4 of the HPh,
about lip-position detection). However, we think it useful to call to mind that (S)
is protruded, not simply rounded (Ô û 1.16.1).
Another useful type of diagram is the laryngogram (as in û 1.1), which should
be analyzed attentively. ˛ese laryngograms are naturally optical (as well as being
schematic and frozen in a particular instant), representing what can be seen with
a laryngoscope, or a throat-specialist's mirror. ˛ey are not acoustic laryngograms,
which measure vocal fold vibration.
1. prelude 35
1.4.12. We will now move on to tonograms, which are divided into three juxta-
posed bands, of high, mid, and low pitch (where the levels are not absolute, but
relative to the voice of each particular speaker). Both in preintonemes and in into-
nemes (Ô § 1.1.5-8 “ û 1.21-3 {and, in the HPh˚ § 6.4.5.1-4 “ § 13.8-34}), as in
tones as well (Ô û 1.26 {and, in the HPh again˚ § 6.4.4 “ § 12.17-18}), lines (or
dashes) placed at di‡erent heights and with various slopes represent stressed pho-
no-syllables. Dots, on the other hand, represent unstressed syllables, while shorter
lines/dashes represent half-stressed syllables (with secondary stress, but with pitch
indicated by the position in the tonogram).
In phonotonetic transcriptions, secondary stress is indicated by two dots placed
close together (smaller than a single dot), variously oriented according to the to-
netic necessities. Secondary stress on a medium level pitch is denoted with (&) in
order not to create confusion with the hyphen we use to show syllable boundaries.
˛is use is consistent with marking primary stress with ('). çUnstressedÇ phono-syl-
lables (or better, weakly-stressed – ¤, weaker than half-stressed syllables) with mid
level pitch are not marked in any particular way. On the other hand, in tone lan-
guages, syllables with mid pitch and weak stress are preceded by a dot placed at
medium height, (2).
1.5.1. In the HPr, there is no doubt that a great number of symbols is used.
However, these symbols are not superfluous for those who desire to do phonetics
thoroughly (and not merely çeasilyÇ, and inevitably superficially). A small num-
ber of symbols ineluctably leads to mediocrity, whereas a large number of symbols
opens the way towards the true understanding and çsavoringÇ of pronunciation.
Even if study has been methodical, concentrated, and accompanied by careful
exercises, it will remain occasionally necessary to check the value, the nature, and
the connections of certain symbols (and concepts as well) which are less frequent.
˛e best way to do so is to look for explanations in the right place, or places. In
fact, the necessary answers, verifications, and connections, as well as new perspec-
tives, can be found in the general index (contents) and in the analytical index (in-
dex), by flipping through the chapters and the sections, and by paying attention
to the tables, the lists of symbols, and the groups of figures.
˛e major categories are, of course, vowels, consonants, intonation, other prosod-
ic traits (stress, pitch, duration/length), and paraphonics.
It would be complicated –and probably useless– to try to re-explain these mat-
ters in a general synthesis (probably too compressed and complex). ˛erefore we
will merely suggest following the directions given here, emphasizing only the
meaning of the di‡erent çparenthesesÇ used to enclose the symbols.
Slashes –/ /– always denote phonemes, on a theoretical and abstract level; in-
stead, brackets –( )– are used exclusively for phones (and taxophones) – ¤, the
practical and concrete side of things, which nonetheless naturally comprise essen-
tial generalizations and normalizations, without which it would be necessary to
36 a handbook of pronunciation
Transcribing by hand
to each other, but should instead be kept separate, as in print. Moreover, symbols
should not be simplified, changing n into u, or m into æ; or l into æ or a. Dots
should not be omitted, and so ª should not be written in place of i, nor ñ for j.
çStylisticÇ additions and modifications should be avoided. ˛us, d must remain
di‡erent from ∂ and from ´, just as ä is di‡erent from the grapheme g, and h from
Ü. ˛e same goes for cases like z, which should be kept distinct from both Z and Ë.
Moreover, (L, r, R, Ÿ) are, by the same token, quite distinct from (ô/H, ¸, [, y), Â.
Naturally, small capitals should be avoided as well, since the symbols (Å, ∫, ™, I, G,
ì, ,, Ì, K) are di‡erent from (a/A, b, e/E, i/¤, g, l, n, p, r).
In conclusion, it is necessary to set aside every normal writing habit which could
lead to confusion between the symbols. ˛e best strategy to reach this essential
goal is to begin observing every symbol with great care. A çtypographic eyeÇ should
be developed which pays attention to every detail, from the size and orientation
of a stroke ((t, T, ˛, †), (B, 6)), or of a symbol ((e, Ù, È), (K, º, ˜, ‰), (A, Ø÷ a, å), (™,
É÷ E, ‘), (X, x÷ v, √÷ ©, O), (r, <, ¸), (h, H, ¥), (f, á÷ j, 3÷ J, ,÷ ã, ª)), to the presence or ab-
sence of a sort of serif, or the type of serif present: (i, I, ¤÷ u, ¯, U÷ o, ø, Ö).
In working towards this objective, the observations in é 8-9 of the HPh will
help the reader to look at, not just see, the symbols found throughout the text. In
this way, it will prove easier to çacceptÇ the fact that (g) always has the value pres-
ent in the word get /'gEt/, and not the value in gem /'GEm/ (while gif –or ç.gifÇ–
can be pronounced either /'gIf/ or /'GIf/), Â.
1.5.4. It might prove useful, sooner or later, to have symbols available which
do not directly represent particular segments, but rather whole phonic categories.
For this task, phonetic and phonological formulae can be used, and the resulting
symbols can be employed, for example, on the edges of vocograms or tables. We
therefore provide a list of appropriate symbols of this type.
û 1.35 gives a schematic presentation of the seven fundamental manners of ar-
ticulation, for contoids. Also given are useful groupings and subdivisions, includ-
ing the distinction between obstruents and sonants (however, the mixed manners
of articulation, typical of approximants, and even more of trills and laterals, are
not included).
˛e category of obstruent contoids includes stops (but not nasals, even though
these could technically be considered stops with added nasalization), stopstrictives,
constrictives (including constrictive trills and constrictive laterals), and approxi-
mants (only the peripheral ones). ˛e sonants comprise, on the other hand, cen-
tral or lateralized approximants, besides nasals and trills (together with taps and
flaps), and laterals (including unilaterals and tapped laterals).
In various languages, for any manner of articulation except trills (and taps and
flaps), phonetically semi-… articulations are possible (¤ less tense – with no full
contact, also for nasals, stops, stopstrictives and laterals).
38 a handbook of pronunciation
é vocoid/vowel Œ semilateral C
é reduced V (in duration; = ä) µ lateral — trill/tap C
ä shortened V ¯ intense (çsyllabicÇ) lateral — trill/tap C
– nasalized V @ voiceless lateral — trill/tap C
´ devoiced V ö nasal C
V voiced lenis V M intense (çsyllabicÇ) nasal C
‚ voiceless lenis V ° voiceless nasal C
◊ half-nasalized V ù sonant (or sonorant) C
≠ rounded V % intense (çsyllabicÇ) sonant C
Ÿ unrounded V # voiceless sonant C
5 advanced V ó trill (or trill “ tap) C
Ì retracted V § voiceless trill (or trill “ tap) C
† lowered V ˘ constrictive trill C
Ï raised V Â tap C
é normal V – or under other conventions ≈ lateralized tap C
ü creaky V (or laryngealized) ‰ flap C
0 contoid/consonant „ lateralized flap C
0 reduced C (in duration; = ¸) ∑ median approximant C
¸ shortened C ß approximant C
ô glottalized voiceless C, with simultaneous ´ semi-approximant C
(ö) ã lateralized approximant C
— intense (çsyllabicÇ) C J lateralized semiapproximant C
Ò devoiced C ¡ peripheral approximant C
 voiceless lenis C – or under other conven- ô obstruent C (F, !, ò), in diphonic pairs
tions, especially diaphonemic ‡ laryngeal approximant C
C voiced lenis C â laryngeal constrictive C
= voiceless C F stop C
Ê voiced C K semi-stop C
± rounded C ƒ (slit) constrictive C
Ü unrounded C (slit) semi-constrictive C
J palatalized C _ grooved constrictive C
ˆ velarized/uvularized C ` grooved semi-constrictive C
ó advanced C ò (generic) constrictive C
ô retracted C ∞ (generic) semi-constrictive C
ú tenser/closer C ` (slit) stop-strictive C
õ less tense/close C @ (slit) stop-semi-strictive C
ó voiced creaky/laryngealized C ¤ (slit) semi-stop-strictive C
‹ lateral C Ô grooved stop-strictive C
ü constrictive lateral C & grooved stop-semi-strictive C
û voiceless lateral C I grooved semi-stop-strictive C
æ lateral tap C ! (generic) stop-strictive C
Æ unilateral C Ì (generic) stop-semi-constrictive C
1. prelude 39
1.6. For informative purposes– we also show the o‚cial IPA chart (û 1.36). It
highlights mostly its defects and global deficiencies, however.
˛e first section gives the o‚cial pulmonic consonants, which are 58 (with a
minor addition of ten çother symbolsÇ). ˛e second small section shows five dejec-
40 a handbook of pronunciation
©ø~ßø~å~†ß (~ø~-𨬵ø~¤©)
ø$™® ߥµ∫ø¬ß ¿ Voiceless alveolo-palatal fric.
Clicks Voiced implosives Ejectives ' Voiceless labial-velar fric. B Voiced alveolo-palatal fric.
Ö Bilabial ñ Bilabial « as in: w Voiced labial-velar app. ï Alveolar lateral flap
| Dental à Dental/alveol. p« Bilabial ¥ Voiced labial-palatal app. À Simultaneous S and x
! (Post)alveolar á Palatal t« Dental/alveol. … Voiceless epiglottal fric.
Ï A‡ricates and double articulat.
õ Palatoalveolar Ÿ Velar k« Velar H Voiced epiglottal fric. can be represented by two sym-
~ Alveol. lateral ä Uvular s« Alveol. fricat. ¿ Epiglottal plosive ˜ bols joined by a tie bar if necess.
tives (or clicks), five injectives (or implosives, including the theoretical uvular
one), and four examples of ejectives.
˛en the o‚cial quadrilateral follows, which tries to explain the vowels, which
are 28, in twelve pairs (for lip-position – spread or rounded), plus four isolated ele-
ments. ˛e collocation of (È, Ù, π) is highly unhappy.
˛e paltry section on tone and stress is misleading and leads one to think it is
not limited to some examples. ˛en suprasegmental signs are shown.
Lastly, we find the section on diacritics, which are thus necessary to avoid the
risk of being utterly vague and ambiguous. In fact, since only very few segmental
symbols are available, one must use some o‚cial diacritics (even two or three for
one symbol), which can be of some use when trying to hint at some important ar-
ticulatory di‡erences.
Unfortunately, this partial solution also betrays one of the fundamental princi-
ples, which used to di‡erentiate IPA from all other phonetic alphabets, precisely
for its outright refusal to use diacritical signs, simply to indicate articulatory char-
acteristics. Instead, now several diacritics are put above or under a generic sym-
bol, or after it as a superscript tiny character.
˛e chart provides nothing at all to show intonation. Even the ToBI system
(which someone seems to consider the intonation component of o‚cial IPA] is
very far from being really useful, since it fails to separate intonation both from ran-
domness and paraphonics.
1.7. For all the articulations given in the HPr (with their unitary symbols), we
thought it useful to add the corresponding oƒIPA çtranscriptionsÇ (given within
ç Ç), in order to show their çcompositionÇ (almost as in chemical or algebraic for-
mulae).
˛is will be useful both to understand the combinations of the few basic sym-
bols with so many diacritics (even if we did not use all those which could have
been necessary for absolute precision, although oƒIPA followers usually omit them
in their approximate and vague transcriptions), and to highlight that it is unthink-
able to do çdiacritical transcriptionsÇ, as all other çphonetic alphabetsÇ do. One of
the fundamental criteria of the original IPA, in fact, was to avoid articulatory dia-
critics.
˛e cases where these two IPA systems coincide are indicated with (=); whereas
a partial correspondence is marked with (=); the di‡erences, without diacritics,
with (–).
Let us recall that it is paramount to carefully observe the orograms, to constant-
ly compare them, and to find similarities also between symbols, by starting from
the o‚cial ones, from which the others have been derived (although with useful
modifications and some necessary substitutions). As we have already said, our firm
intention is to avoid articulatory diacritics and prefer unitary symbols, instead.
˛is is also done to avoid that there may be second- or third-class symbols (and
42 a handbook of pronunciation
thus second- or third-class phones, as well), because they are accompanied by dia-
critics (which are not easy to combine accurately).
(/, =) Ωh )Ç
ç(µ ∫ap
(/, j) f )Ç
ç(w ({Ó}, [) ç(Í̆ , R̆ )Ç
({V}, e)* ç(Í̆ ˆ, R̆ ˆ)Ç
Laryngeal approximant
(/, ®) ç(#̆ )Ç
(h, H) ç(hf , Hf )Ç
(/, ·) ç(Hf „)Ç Constrictive trill
(+) ç(hî≠ J)Ç
(º, ˜) ç(K#=H , K#H )Ç
(´) ç(hî≠ ˆ)Ç
(±) ç(' î≠ )Ç Lateralized flap
(=) ç(hî≠ )Ç
(≠) ç(hî≠ „)Ç (r, m) ç(R:Î˱ &, R:ÎË&)Ç
(◊ , ©) ç((R:Î˱ &ˆ, R:ÎË&ˆ)Ç
Lateralized approximant (+ 2 semi-…)
(Bi)lateral
(/, S) ç(VX< &„)Ç
(/, Ã) ç(∑¥§ &)Ç ((/, l)) ç(lY)Ç
(/, ¸) ç(<Î &)Ç ((/, Ï)) ç(lYˆ) or (ıY)Ç
(/, Í) ç(<< &)Ç ((/, R)) ç(lY:)Ç
(/, >) ç(<< &„)Ç ({a}, l) ç(lÎ≠ , Îl )Ç
(/, ˛) ç(<< &ˆ„)Ç ({ñ}, ı) ç(l≠Έ), (lÎ ˆ) or (ıÎ), (ıÎ≠)Ç
(/, <) ç(µXΩ < &„)Ç ({Ü}, ]) ç(lÎ≠fˆ), (l§Îˆ) or (ıΧ), (ıÎ≠f)Ç
(/, <) ç(µ<X &ˆ< „)Ç (/, ∞) ç(lÎ ˆ„) or (ıÎ „)Ç
(/, ‰) ç(µXhΩ < &„)Ç ((/, l)) ç(lÎ <ˆ) or (ı:Î )Ç
(/, $) ç($)Ç (=)
Trill ((L)) ç(l:¥ J)Ç
({Ñ}, ¬) ç(l¥≠ J, ¥l J)Ç
(/, r) ç(rÎ )Ç
(/, L) ç(L)Ç (=)
(/, 5)* ç(rÎ ˆ)Ç
(/, ı) ç(ì)Ç (–)
(/, 5) ç(rÎ ˆ&)Ç
({{}, D) ç(r¥+ J, ¥r J)Ç Unilateral
({'}, K) ç(K= , K)Ç (–, =)
(/, ô) ç(l@Î )Ç
Tap (/, |) ç(l@Î Ω̂)Ç
(/, ÷) ç(l@Î ˆ)Ç
({5}, R) ç(RÎ+ , ÎR )Ç
(/, R)* ç(RÎ ˆ)Ç Tap(ped) lateral
(/, R) ç(RÎ <ˆ )Ç
(/, e) ç(#)Ç (–) (/, ¬) ç(à)Ç (–)
({[}, ç) ç(R¥+ J, ¥R J)Ç (/, ») ç(àÎ:)Ç
({3}, r) ç(KW+ , KW )Ç
Semilateral
(/, ù) ç(ıf)Ç
(/, §) ç(ıf „)Ç
2. English
2.0. In this chapter we will deal with the American and British neutral accents
(or çstandard accentsÇ). We will also make a teaching proposal for an çinternation-
alÇ accent of English that could usefully be employed in pronunciation books and
pronouncing dictionaries (and in common dictionaries, too), as well as in every-
day teaching. ˛e kind of transcription we use is dia#onemic˚ expressly devised
for this kind of description, together with its corresponding #onetic and #onoto-
netic transcriptions.
Furthermore, we will also consider the American and British çmediaticÇ accents
(from non-local †√ “ radio), which are now as frequently heard as the neutral ones,
in the news, in the movies, and in songs.
To conclude, we will describe (without going into great detail, mainly by using
our usual accurate symbols and many diagrams) some other accents (with inter-
nal variations): those of Canada, Australia, New Zealand, and England (for the
latter we will give the traditional, a‡ected, and Cockney accents). In a book in
progress –E«li´ Pronunciation*– we will deal with all the native accents of En-
glish all over the world (including many non-native accents), by working directly
on a substantial number of recordings, as well.
2.1. It will be useful to list the correspondences between our diaphonemic sym-
bols and the phonemic symbols used in recent dictionaries, and especially in the
three current English pronouncing dictionaries (Longman, Cambridge, Oxford).
˛ese do not always agree for certain aspects, but we show them (between ç Ç) af-
ter the diaphonemes, with examples.
Vowels /EI/ ç/eI/Ç dZ /'dEI/ ('D™;I)
/aE/ ç/aI, √I/Ç fly /'flaE/ ('fla;Ù)
/i/ ç/i, I/Ç lady /'lEIdi/ ('l™IDi) /OE/ ç/OI/Ç boy /'bOE/ ('bø;Ù)
/I/ ç/I/Ç bit /'bIt/ ('b¤T) /aO/ ç/aU/Ç cow /'kaO/ ('kha;Ö)
/E/ ç/e, E/Ç let /'lEt/ ('l™T) /OU/ ç/ÈU, oU/Ç go /'gOU/ ('gø;¨)a ('g‘;¨)b
/π/ ç/π, a/Ç hat /'hπt/ ('hπT) /Uu/ ç/u:, u/Ç who /'hUu/ ('hU;u)a ('h¯;u)b
/√/ ç/√, È/Ç hut /'h√t/ ('h√T)a ('håT)b
/Ø/ ç/O, A:, A/Ç hot /'hØt/ ('hAT)a ('hØT)b /¢/ ç/I, È, ¢, î/Ç wi´es /'wIS¢z/ ('w¤SÈΩ)a
/U/ ç/U/Ç book /'bUk/ ('b¨k) ('w¤S¤Ω)b
/u/ ç/u, U/Ç influenza /Influ'EnzÈ/ (&¤Mfl¯- /π;/ ç/π, A:/Ç last /'lπ;st/ ('lπsT)a ('lA;sT)b
'™nz√)a (-å)b /A;/ ç/A:, π/Ç pasta /'pA;stÈ/ ('phA;sT√)a
/È/ ç/È/Ç ano`er /È'n√∑È≤/ (È'n√∑≥)a (È'nå∑- ('phπsTå)b
å)b˚ contain /kÈn'tEIn/ (khûn'Th™;In) /Ø;/ ç/Ø, O:/Ç so« /'sØ;˙/ ('sO:˙)a ('sØ;˙)b
/A:/ ç/A:, A/Ç spa /'spA:/ ('spA:) /O;/ ç/O:, Ø/Ç false /'fO;ıs/ ('fO;ıs, 'fAıs)a
/O:/ ç/O:, O/Ç law /'lO:/ ('lO:)a ('lø:)b ('fø;ıs, 'fØıs)b
/È;/ ç/√r, !:, ‘:r, Èr, ò/Ç hurry /'hÈ;<i/
/Ii/ ç/i:, i/Ç see /'sIi/ ('sI;i) ('h≥;i)a ('hå>i)b
2. english 47
/È:</ ç/‘:r, È:r, Èr, !:/Ç furry /'fÈ:<i/ ('f≥;i)a /K:/ ç/È, O:/Ç repertory /'<EpÈ≤tK:<i/ ('<™p≥-
('f‘;>i)b &Tø<i)a ('>™pÈTÈ>i, -È˛>i)b
/È:≤/ ç/‘:, È:, ‘:r, Èr, !:/Ç fur /'fÈ:≤/ ('f≥:)a /È/ ç/U, È, T, ÿ/Ç regular /'<EgjÈlÈ≤/ ('<™g-
('f‘:)b jÈl≥)a ('>™gj¨lå, -gjÈ-)b
/A:≤/ ç/A:, A:r/Ç car /'kA:≤/ ('khA:<)a ('khA:)b
/O:≤/ ç/O:, O:r/Ç door /'dO:≤/ ('Dø:<)a ('Dø:)b Consonants
/È≤/ ç/Èr, ÈÍ, ær, …r, ò/Ç wonder /'w√ndÈ≤/ /m/ ç/m/Ç some /'s√m/ ('s√m:)a ('såm:)b
('w√nD≥)a ('wånDå)b /n/ ç/n/Ç sun /'s√n/ ('s√n:)a ('sån:)b
/iÈ≤/ ç/iÈ, IÈ, iÈr, IÈr, IÈÍ, iÈÍ, i…r, iò/Ç /˙/ ç/˙/Ç su« /'s√˙/ ('s√˙:)a ('så˙:)b
happier /'hπpiÈ≤/ ('hπpi≥)a ('hπpiå)b /õ/ ç/æm, …m, m/Ç rhy`m /'<I∑õ/ ('<¤∑-
/IÈ≤/ ç/IÈ, IÈr, Ir, IÈÍ, I…r, Iò/Ç here /'hIÈ≤/ õ)a ('>-)b
('hI;≥)a ('h¤;å)b /ó/ ç/æn, …n, n/Ç cotton /'kØtó/ ('khATó)a
/EÈ≤/ ç/eÈ, E:, er, Er, eÈÍ, e…r, Eò/Ç `ere ('khØ-)b
/'∑EÈ≤/ ('∑™;≥)a ('∑™;‘)b
/UÈ≤/ ç/UÈ, UÈr, Ur, UÈÍ, U…r, Uò/Ç moor /p/ ç/p/Ç pa$ /'pπk/ ('phπk)
/'mUÈ≤/ ('mU;≥)a ('m¨;å)b (with further /b/ ç/b/Ç ba$ /'bπk/ ('bπk)
pronunciations) /t/ ç/t/Ç ‘o /'tUu/ ('ThU;u)a ('Th¯;u)b
/uÈ≤/ ç/uÈ, UÈ, uÈr, UÈr, uò, uÈÍ, UÈÍ, /d/ ç/d/Ç do /'dUu/ ('DU;u)a ('D¯;u)b
u…r, u:ò/Ç rescuer /'<EskjuÈ≤/ ('<™skj¯≥)a /k/ ç/k/Ç came /'kEIm/ ('kh™;Im)
('<™skj¯å)b /g/ ç/g/Ç game /'gEIm/ ('g™;Im)
/IiÈ≤/ ç/i:È, i:Èr, i:ò, i:ÈÍ, i:…r, i:ò/Ç seer /c/ ç/tS/Ç >ain /'cEIn/ ('ch™;In)
/'sIiÈ≤/ ('sIi≥)a ('sIiå)b /G/ ç/dZ/Ç Jane /'GEIn/ ('G™;In)
/EIÈ≤/ ç/eIÈ, eIÈr, eIò, eIÈÍ, eI…r, eIò/Ç /f/ ç/f/Ç f[ /'fjUu/ ('fj¯;u)
plZer /'plEIÈ≤/ ('phl™I≥)a ('phl™Iå)b /v/ ç/v/Ç vi[ /'vjUu/ ('vj¯;u)
/aEÈ≤/ ç/aIÈ, √IÈ, aIÈr, aIò, aIÈÍ, aI…r, /†/ ç/†/Ç wreath /'<Ii†/ ('<Ii†)a ('>Ii†)b
aIò/Ç liar /'laEÈ≤/ ('laÙ≥)a ('laÙå)b /∑/ ç/∑/Ç wreathe /'<Ii∑/ ('<I;i„)a ('>I;i„)b
/OEÈ≤/ ç/OIÈ, OIÈr, OIò, OIÈÍ, OI…r, OIò/Ç /s/ ç/s/Ç ice /'aEs/ ('aÙs)
employer /Im'plOEÈ≤/ (¤m'phløÙ≥)a (¤m- /z/ ç/z/Ç \es /'aEz/ ('a;ÙΩ)
'phløÙå)b /S/ ç/S/Ç dilution /d¢'lUuS˘n/ (DÈ'lUuSÈn)a
/aOÈ≤/ ç/aUÈ, aUÈr, aUò, aUÈÍ, aU…r, aUò/Ç (D¤'l¯uSó)b
tower /'taOÈ≤/ ('ThaÖ≥)a ('ThaÖå)b /Z/ ç/Z/Ç delusion /d¢'lUuZ˘n/ (DÈ'lUuZÈn)a
/OUÈ≤/ ç/ÈUÈ, oUÈr, oUò, oUÈÍ, oU…r/Ç (D¤'l¯uZó)b
slower /'slOUÈ≤/ ('slø¨≥)a ('sl‘¨å)b
/</ ç/r/Ç rate /'<EIt/ ('<™IT)a ('>™IT)b
/UuÈ≤/ ç/u:È, u:Èr, u:ò, u:ÈÍ, u:…r/Ç doer
/j/ ç/j/Ç yate /'jEIt/ ('j™IT)
/'dUuÈ≤/ ('DUu≥)a ('D¯uå)b
/w/ ç/w/Ç wait /'wEIt/ ('w™IT)
/˘</ ç/Èr, ær, …r, ò/Ç wonderi« /'w√nd˘- /h/ ç/h/Ç hate /'hEIt/ ('h™IT)
<I˙/ ('w√nDÈ<¤˙, -D<¤˙)a ('wånDÈ>¤˙, /l/ ç/l/Ç late /'lEIt/ ('l™IT)
-NÃ>¤˙)b
/ù/ ç/t, ÿ/Ç ci“ /'sIùi/ ('s¤[i)a ('s¤Ti)b
/I˘</ ç/IÈr, Iær, Ir/Ç heari« /'hI˘<I˙/ ('h¤<-
/≤/ ç/–, r/Ç car /'kA:≤/ ('khA:<)a ('khA:)b
¤˙)a ('h¤È>¤˙)b
/ˆ/ ç/j, –/Ç n[ /'nˆUu/ ('nU;u)a ('nj¯;u)b
/E˘</ ç/eÈr, eær, Eær, E:r/Ç cari« /'kE˘<I˙/
/·/ ç/h, –/Ç when /'·wEn/ ('w™n:÷ 'hw™n:÷
('kh™<¤˙)a ('kh™‘>¤˙)b
'W™n:)
/U˘</ ç/UÈr, Uær, Ur/Ç curi« /'kjU˘<I˙/
('khj¨<¤˙)a ('khj¨È>¤˙)b (with f. pron.) /ı/ ç/l/Ç lull /'l√ı/ ('l,ı:)a ('l√ı:)b
/Œ/ ç/È, e, E/Ç dictionary /'dIkSÈnŒ<i/ /®/ ç/l/Ç little /'lIù®/ ('l¤[®)a ('l¤T®)b
('D¤kSó&™<i)a ('D¤kSó>i)b
48 a handbook of pronunciation
Vowels
American monophthongs
2.1.2.1. û 2.1 shows the American monophthongs (the British ones are given
in û 2.2). Let us start from the eight bla$ markers, which indicate the realizations
of the following eight vowel phonemes (in stressed or unstressed syllables): (¤) /I/,
(™) /E/, (π) /π/, (A:) /A:/, (A) /Ø/, (√) /√/, (O:) /O:/, (¨) /U/. Although in the çmediaticÇ
American accent (Ô § 2.4.2.2) /A:/ and /Ø/ are often neutralized –because they can
both be realized as (A[:])– in the neutral American accent we keep them apart, for
three good reasons. ≈rstly, they are in actual fact di‡erent, even if chiefly only in
terms of length. Furthermore, in this way we can keep a diaphonemic relation
with the British neutral accent. And finally, this will help us in highlighting the
characteristics of other accents, starting from comparable bases, although they are
actually shared only by a minority of speakers. Let us stress, in fact, that a neutral
pronunciation is always learned voluntarily.
Let us now illustrate the phonemes in û 2.1. It must be recalled that the tran-
scriptions, including phonemic ones, bear a stress mark, even for monosyllabic
words, unless they are usually unstressed in sentences, like the preposition in /In/
(¤n), compared to the adverb in /'In/ ('¤n:), or to the noun inn /'In/ ('¤n:)): ('h¤T)
/'hIt/ hit, ('j™s) /'jEs/ yes, ('mπ;n) /'mπn/ man, ('fA;∑≥) /'fA:∑È≤/ fa`er, ('lAT) /'lØt/ lot,
('<√n:) /'<√n/ run, ('lO:n) /'lO:n/ lYn, ('ph¨T) /'pUt/ put.
For the sake of descriptive precision, although generally variations do not re-
quire di‡erent symbols, it is worthwhile to notice the taxophones of /Iı, Eı, πı, √ı,
Uı/; please note the two that change their symbols as well, in comparison with (™,
√): ('bEı:, 'h,ı:) /'bEı, 'h√ı/ bell˚ hull.
2.1.2.2. ˛e gr\ markers in the vocogram also show three variants of /I, U, O:/
(I, U, ø:) (compared to normal (¤, ¨, O:)). ˛ey occur in (stressed or unstressed) syl-
lables checked by /≤/: (I≥) /IÈ≤/, ([j]U≥) /[j]UÈ≤/, (ø:<) /O:≤/. ˛ere are no di‡erences in
terms of the nuclear element (or, perhaps, çnuclealÇ element) in (™≥) /EÈ≤/, (A:<)
/A:≤/, because they coincide with the black markers for /E, A:/ (™, A:). Instead, the
2. english 49
typical American realization of /È:≤/ is through an intense contoid (≥:), rather than
a vocoid as in British pronunciation ((‘:)).
Examples: ('hI;≥) /'hIÈ≤/ here, ('phjU;≥) /'pjUÈ≤/ pure, ('Dø:<) /'dO:≤/ door, ('∑™;≥)
/'∑EÈ≤/ `ere, ('fA:<) /'fA:≤/ far, ('f≥:) /'fÈ:≤/ fur˘ A possible pronunciation with (¤≥, ¨≥)
(instead of (I≥, U≥), for /IÈ≤, UÈ≤/} does not sound strange, although it is not the
most frequently heard.
For /O:</ (that is, before vowels: /O:<é/), the realization is (O:<) (di‡erent from
/O:≤/, marked by a grey marker): ('bO;<¤˙)a (->-)b /'bO:<I˙/ bori«˘ In American En-
glish the diaphonemic transcriptions /I˘<, E˘<, U˘</ correspond to /I<, E<, U</, with
their typical articulations shown by the black markers. By the way, the di‡erence
between /I˘<, E˘<, U˘</ and /IÈ≤, EÈ≤, UÈ≤/ (taken into consideration above) should
be carefully noted. Here are a few examples: ('h¤<¤˙)a ('h¤È>¤˙)b /'hI˘<I˙/ heari«,
('b™<¤˙)a ('b™‘>¤˙)b /'bE˘<I˙/ beari«, ('D¨<¤˙)a ('Dj¨È>¤˙)b /'dˆU˘<I˙/ duri«˘
It must be also noticed that, in American pronunciation, the phonemic sequence
/jUÈ≤/ has the variant /jÈ:≤/ (more typically mediatic): ('phjU;≥÷ 'phj≥:) pure (the vari-
ant /[j]È:</ occurs for /jU˘<, ˆU˘</, too): ('khj¨<iÈs÷ 'khj≥;iÈs) curious, (¤n'D¨<Èns÷ -'D≥;-
Èns) endurance, ('D¨<¤˙÷ 'D≥;¤˙) duri«˘ ˛is is also true of other consonants with a
çpalatalÇ component: ('SU;≥÷ 'S≥:) /'SUÈ≤÷ 'SÈ:≤/ sure÷ but ('phU;≥, -ø:<) /'pUÈ≤, -O:≤/ poor.
˛e di‡erence between ('mO;<n¤˙) /'mO:≤nI˙/ morni« and ('mo≥n¤˙, 'mo;<nI˙)
mourni« (which, by now, belongs only to çtraditionalÇ American pronunciation)
is no longer neutral, nor is it modern, but simply regional. As a matter of fact, they
are now both pronounced ('mø;<n¤˙) (('mø;n¤˙)b) /'mO:≤nI˙/ (and the traditional
pronunciation of mourni« could be shown diaphonemically as ç/'mOÈ≤nI˙/Ç or
ç/'mo:≤nI˙/Ç).
2.1.2.3. ˛e four white markers show the unstressed realizations of the remain-
ing three phonemes: (i) /i/, (¯) /u/, (È, û) /È/. ˛ey occur in the following contexts
– /i/ at the end of lexemes: ('l™IDi[Ω]) /'lEIdi[z]/ lady˚ ladies, ('™ni&†¤˙) /'Eni†I˙/ aJ-
`i«, and before vowels: (<i'πkT) /<i'πkt/ react÷ /u/ chiefly before vowels: (&s¤c¯'™I-
SÈn) /sIcu'EIS˘n/ situation˚ (Th¯'πıGi) /tu'πıGi/ to Algy˘ A few words ending in -ue
[value˚ ®atue˚ virtue˚ i˛ue˚ ti˛ue] may have a reduced variant, especially in a prein-
toneme, which might usefully be represented with the diaphoneme /uò/: ('vπlj¯[u],
'sTπc¯[u]) /'vπlju, 'stπcu/ value˚ ®atue˘
≈nally, /È/ is the most frequent unstressed vowel phoneme of the English lan-
guage (in particular American, Oceanian, and South African): (fÈ'ThAg<Èf≥) /fÈ'tØg-
<ÈfÈ≤/ #otogra#er. In words like ('sø¨f√)a ('s‘¨få)b /'sOUfÈ/ sofa˚ we can see that a
50 a handbook of pronunciation
final /È/ before a pause (even a short one), is realized as if it were the (unstressed)
phoneme /√/; however, if a pause is not there, this does not occur.
2.1.3.1. Let us now consider the corresponding vocogram for the Briti´ accent
(û 2.2). Here we have nine bla$ markers (for nine either stressed or unstressed
phonemes): (¤) /I/, (™) /E/, (π) /π/, (A:) /A:/, (Ø) /Ø/, (å) /√/, (ø:) /O:/, (¨) /U/, (‘:)
/È:≤/. We find: ('h¤T) /'hIt/ hit, ('j™s) /'jEs/ yes, ('mπ;n) /'mπn/ man, ('fA;∑å) /'fA:∑È≤/
fa`er, ('fA:) /'fA:≤/ far, ('lØT) /'lØt/ lot, ('>ån:) /'<√n/ run, ('lø:n) /'lO:n/ lYn, ('Dø:)
/'dO:≤/ door, ('ph¨T) /'pUt/ put˚ and ('w‘:fl) /'wÈ:≤d/ word (for American English, of
course, this last word was in the group of /≤/, as we have seen above). If we consid-
er far`er˚ /'fA:≤∑È≤/, we can see that it is pronounced exactly as fa`er˚ ('fA;∑å) (with
no /≤/ pronounced), while in American English both r /≤/ are realized: ('fA;<∑≥). For
an American pronunciation like ('fA;∑≥), for far`er, and other words, see below:
di˛imilation (§ 2.3.3.5).
Let us quickly add that even for (ø:) /O:≤/, as in ('wø:[Ω]) /'wO:≤[z]/ war(s)˚ by now,
2. english 51
f
/¢/ = /I/ (¤) /È[≤]/ (È, û*, åò|)
/O:/ (ø:), /O;/ (ø:, Ø)
/E[ı]/ (™, ™ı) /È:</ (‘:>), /È:≤/ (‘:), /È;</ (‘:>)
f
2.1.3.2. For British English, /IÈ≤, EÈ≤, UÈ≤/ are given in û 2.3: ('h¤;å) /'hIÈ≤/ here,
('phj¨;å) /'pjUÈ≤/ pure, ('∑™;‘) /'∑EÈ≤/ `ere˘ We have already seen that in British En-
glish a simple vocoid, with no contoid, occurs in cases like ('f‘:) /'fÈ:≤/ fur˘ It is to
be noticed that, in this type of pronunciation, the phonemic sequence /[j]UÈ≤/, by
this time, is almost exclusively substituted with its variant /[j]O:≤/: ('phjø:÷ 'phj¨;å)
pure, ('phø:÷ 'ph¨;å) poor, ('Sø:÷ 'S¨;å) sure˘
It is curious to note that, in the British pronunciation of the sixties, a similar
trend was very strong, but it soon stopped: ('phj‘:, 'khj‘;>iÈs, ¤n'Dj‘;>Èns, 'Dj‘;>¤˙, 'S‘:)
pure˚ curious˚ endurance˚ during˚ sure (at that time, of course, the neutral pronun-
ciation had (¤é) ç/Ié/Ç for modern (ié) /ié/: ('khj‘;>¤Ès), curious). Another curiosity
is that, in the first half of the twentieth century, the same pronunciation (j‘:) /jÈ:≤/
(which today no longer occurs, except as an old-fashioned variant) was a variant of
/IÈ≤, I˘</: ('D¤;å÷ 'Dj‘:) dear˚ ('h¤;å÷ 'hj‘:) here˚ (s¤n's¤;å÷ -'sj‘:) sincere˚ ('khl¤È>Èns÷ 'khlj‘;-
>Èns) clearance÷ however, it still remains as a secondary variant in ('j¤;å÷ 'j‘:) year˘
2.1.3.3. Also in the British accent, the four white markers show the realizations
of the remaining phonemes, always in unstressed positions: (i) /i/, (¯) /u/, (È, û)
/È/. ˛ey occur in the same contexts – /i/ at the end of a lexeme: ('l™IDi[Ω]) /'lEIdi[z]/
lady˚ ladies, ('™ni&†¤˙) /'Eni†I˙/ aJ`i«, and before a vowel: (>i'πkT) /<i'πkt/ react˘
/u/ occurs chiefly before vowels: (&s¤c¯'™ISó) /sIcu'EIS˘n/ situation˚ (Th¯'ÅıGi) /tu-
'πıGi/ to Algy˘ ˛e possibility of finding /'vπlju/ value (§ 2.1.2.3) is less common
in British English.
≈nally, /È/ is the most frequent unstressed vowel phoneme in British English,
too, also because of the vocalization of /È≤, é:≤/): (fÈ'ThØg>Èf‘Ω) /fÈ'tØg<ÈfÈ≤z/ #o-
togra#ers ((fÈ'ThAg<Èf≥Ü)a)˚ ('>¤cÈfl) /'<IcÈ≤d/ Ri>ard (('<¤c≥fl)a), (Thû'g‘;¨) /tÈ'gOU/
to go ((Thû'gø;¨)a). Of course, we also find (å|) /È, È≤/ (before pauses): ('s‘¨få)
/'sOUfÈ/ sofa {('sø¨f√)a}, ('A;fTå) /'π;ftÈ≤/ after {('π[;]fT≥)a}.
2.1.3.5. It is useful to recall here (although this is also true of the preceding cas-
es of /È≤/) that, in neutral British pronunciation (as well as Oceanian and South
African), the normal realization of /È≤/ is (È): (phÈ'fø;mÈns) /pÈ≤'fO:≤mÈns/ perform-
ance˚ (&πsÈ'Th™;In) /πsÈ≤'tEIn/ ascertain˚ (hÈ'håzbÈnfl) /hÈ≤'h√zbÈnd/ her husband˘ Of
course, this holds good unless in absolute final position before a pause, (å|), or final
in a rhythm group, or with the grammemes /zò, dò/, (‘ò, ‘Ωò, ‘flò): ('h¤È>¤˙) /'hI˘<I˙/
heari«˚ ('h¤;‘ w¤∑'j¯;u) /'hIÈ≤ wI∑'jUu/ here wi` you, ('b¤;‘Ω) /'bIÈ≤z/ beers, ('h¤;È> Èn-
'∑™;‘) /'hIÈ≤ Èn{d}'∑EÈ≤/ here and `ere˘
˛e modern neutral British pronunciation of /EÈ≤/ does not change any longer
according to context, but it is always (™‘) (with a strong tendency to a long mon-
ophthong, through (™É), up to ç(™:) /E:/Ç, as has, for a few generations, already hap-
pened to the previous ç/OÈ≤/Ç): ('b™;‘[Ω]) /'bEÈ≤[z]/ bear(s)˘ For /A:≤, È:≤, O:≤/, as we
have seen, /≤/ is lost and realized through a çzeroÇ phone, (`): ('khA:[Ω]) /'kA:≤[z]/
car(s), ('f‘:[Ω]) /'fÈ:≤[z]/ fur(s), ('Dø:[Ω]) /'dO:≤[z]/ door(s).
In absolute final position and before pauses, we find: ('b¤;å|) /'bIÈ≤|/ beer,
('khj¨;å|) (and (-jø:|)) /'kjUÈ≤|/ cure and (fÈ'ThØg>Èfå|) /fÈ'tØg<ÈfÈ≤|/ #otogra#er˘
But, if final in a rhythm group, we have: (È'b¤;‘ fÈ'mI;i) {(È'bI;≥ f≥'mI;i)a} /È'bIÈ≤ fÈ≤-
û 2.3. British /éÈ/ diphthongs.
/IÈ, IÈ≤, I˘</ (¤È>é, ¤‘z+, ¤‘D+, ¤å+|) /új/ˆûUÈ, -UÈ≤, -U˘</ (újû¨È>é, újû¨‘z+,
újû¨‘D+, újû¨å+|) (+ /újûO:≤, -</}
'mIi/ a beer for me, (&∑πk'khj¨;È> '¤zô 'g¨;fl, -jø:> '¤z-) {(&∑πk'khjU;≥ '¤zô 'g¨;fl, -j≥: '¤z-)a}
/∑πt'kjUÈ≤ 'Izót 'gUd/ `at cure isn't good, (∑ÈfÈ'ThØg>Èf‘ 'D¤D¤T) ((∑ÈfÈ'ThAg<Èf≥ 'D¤D-
¤T)a) /∑ÈfÈ'tØg<ÈfÈ≤ 'dIdIt/ `e #otogra#er did it.
Diphthongs
2.1.4.1. Let us now consider the seven phonemic diphthongs of American En-
glish. ˛ey have ten realizations, which are necessary for a good pronunciation (û
2.4). ˛e bla$ markers stand for the seven phonemes, while the three gr\ ones
show taxophones, or contextual variants. Meanwhile, we will see: (Ii) /Ii/, (™I) /EI/,
(aÙ) /aE/, (øÙ) /OE/, (aÖ) /aO/, (ø¨) /OU/, (Uu) /Uu/: ('ThI;i) /'tIi/ tea, ('D™;I) /'dEI/ dZ,
('ha;Ù) /'haE/ high, ('bø;Ù) /'bOE/ boy, ('na;Ö) /'naO/ now, ('gø;¨) /'gOU/ go, ('hU;u)
/'hUu/ who˘ Besides, we need: (iIı) /Iiı/, (j¯u) /jUu/, (Uu, ¯u) /ˆUu/, ([j]uUı) /[j]Uuı/:
('fi;Iı, 'fiI®) /'fIiı/ feel, ('j¯u†) /'jUu†/ you`, ('nU;u, 'n¯;u) /'nˆUu/ n[, ('ju;Uı, 'juU®)
/'jUuı/ yule, ('khu;Uı, 'khuU®) /'kUuı/ cool˘
˛e other diphthongs, even if followed by /ı/, do not change much their com-
ponents, apart from those with front second elements; besides (except for /OU/,
which has only /ı/, in neutral pronunciation), they freely fluctuate between /ı/ and
/®/ (with a possible realization of /®/ as /Èı/ (ûı)): ('s™;¤ı, 's™¤®) /'sEIı/ sail, ('fa;Éı,
'faÉ®) /'faEı/ file, ('bø;Éı, 'bøÉ®) /'bOEı/ boil, ('fa;Öı, 'faÖı) /'faOı/ fowl÷ but only
('sø;¨ı) /'sOUı/ soul.
û 2.4. American diphthongs.
/Iiı/ (iIı) /jUu/ (j¯u, jUu)
/Ii/ (Ii) /[ˆ]Uu/ (Uu, ¯u), /[j/ˆ]Uuı/ ([j/`]uUı)
'G‘;¨2 5w‘¨˙ 'g‘;¨3 3) /'nOU. 'GOU, 'wOUnt 'gOU./ No, Joe won't go. ˛e first element of
/OU/ is central and unrounded, (Ԭ), while in American pronunciation it is back
and rounded, (ø¨): ('nø;¨23 'Gø;¨2 'wø¨˙ 'gø;¨23)a. In the British accent, at the be-
ginning of the twentieth century, (o¨) was widespread; until the fifties it was (P¨),
always with lip rounding, while (Ȩ, ‘¨), at that time, sounded rather a‡ected.
û 2.5. British diphthongs.
/Iiı/ (iIı) /[j/ˆ]Uu/ ([j]¯u)
/Ii/ (Ii) /[j/ˆ]Uuı/ ([j]uUı)
/OE[ı]/ (øÙ, øÉı)
/EI[ı]/ (™I, ™¤ı)
/OU[ı]/ (‘¨, Ö¨ı)
/aE[ı]/ (aÙ, aÉı) /aO[ı]/ (aÖ[ı])
2.1.4.3. In the vocogram showing /IÈ≤, EÈ≤, [j]UÈ≤/ (û 2.3), there are also /aEÈ≤,
aOÈ≤/. As a matter of fact, in a typical British pronunciation (besides remaining sta-
ble, as in American English), both can frequently reduce to ç/aÈ≤/Ç (a;È) (and
(a;å|)): ('faÙ‘Ω, 'faÙå|) and ('fa;‘Ω, 'fa;å|) /'faEÈ≤[z]/ fire(s). Otherwise, /aOÈ≤/ can be-
come ç/A:È≤/Ç (A;È) (and (A;å|)), up to coincide with /A:[≤]/: ('ThaÖ‘Ω, 'ThaÖå|),
('Tha;‘Ω, 'Tha;å|) (also ('ThA;‘Ω, 'ThA;å|), besides ('ThA:[Ω])) /'taOÈ≤[z]/ tower(s).
In this kind of pronunciation (sometimes defined çsmoothingÇ), even the rarer
/EIÈ≤, OEÈ≤, OUÈ≤/ can always be lessened, respectively, to ç/E:È≤, O:È≤, È:È≤/Ç. ˛us
ç/E:È≤/Ç (up to coincide with /EÈ≤/): ('l™I‘Ω, 'l™Iå|), ('l™;‘Ω, 'l™;å|), ('l™;‘[Ω]) (and also
('l™;É[Ω], 'l™:[Ω])) /'lEIÈ≤[z]/ lZer(s); ç/O:È≤/Ç (up to coincide with the old ç/OÈ≤/Ç): (¤m-
'phløÙ‘Ω, -øÙå|), (-ø;‘[Ω], -ø;å|), /Im'plOEÈ≤[z]/ employer(s); ç/È:È≤/Ç (up to coincide
with /È:≤/): ('m‘¨‘Ω, 'm‘¨å|), ('m‘;‘Ω, 'm‘;å|), ('m‘:[Ω]) /'mOUÈ≤[z]/ mower(s) (Ô
('faÙ≥[Ü], 'ThaÖ≥[Ü], 'l™I≥[Ü], ¤m'phløÙ≥[Ü], 'mø¨≥[Ü])a).
Vowel diaphonemes
2.1.5.3. Two other diaphonemes, /π;, Ø;/, are more important in distinguishing
between the American and British modern neutral accents. ˛e first, /π;/, shows
the di‡erence between /π/a (but it often behaves like a long monophthong ç/π:/Ç)
and /A:/b (chiefly before /f, †, s/ and before /ö0/), as in: ('g<π[;]sp)a ('g>A;sp)b
/'g<π;sp/ grasp˚ ('lπ[;]sT)a ('lA;sT)b /'lπ;st/ la®˚ ('phπ[;]s)a ('phA;s)b /'pπ;s/ pa˛˚ ('hπ[;]f)a
('hA;f)b /'hπ;f/ half˚ ('π[;]fT≥)a ('A;fTå)b /'π;ftÈ≤/ after˚ ('phπ[;]†)a ('phA;†)b /'pπ;†/ pa`˘
56 a handbook of pronunciation
2.1.5.5. ˛ere are another couple of diaphonemes, /A;, O;/, which are relatively
less significant, because /A;/ is used especially in words of foreign origin, written
with an a\ ('phA;sT√)a ('phπsTå)b /'pA;stÈ/ pa®a, (&vIi™T'nA:m, -™ö'-)a (-'nπ;m)b /vIiEt-
'nA;m/ Vietnam˘
‹ile /O;/ occurs in particular in words written with ausC˚ aunC˚ alC\ (O;'sTI;≥,
A-)a (ø;'sT¤;å, Ø-)b /O;'stIÈ≤/ au®ere, ('hO;nT, 'hAnT)a ('hø;nT)b /'hO;nt/ haunt˚ ('sO;ıT,
'sAıT)a ('sø;ıT÷ 'sØıT)b /'sO;ıt/ salt˘ As the examples show, /O;/ concerns, above all,
American English. On the other hand, /A;/ may present twofold possibilities, in
both accents, according to words and to speakers.
˛e (socio)diaphoneme /¢/
2.1.6.1. ˛e last vowel diaphoneme we must consider is /¢/ = /È, I/. ˛is refers
to the alternation in the realizations of /¢/: between /È/ and /I/. Clearly, /È/ (È) pre-
vails in the American accent, while /I/ (¤) prevails in the Briti´ one, even if things
are a little more complicated. Indeed, in American English, too, there are cases of
2. english 57
/¢/ = /I/, chiefly in a more refind and more conservative way of speaking. By the
same token, in British English, there are cases of /¢/ = /È/, chiefly in a less refined
and more innovative way of speaking. So, we find a greater convergence at a more
up-to-date and modern level.
Here are a few examples: (sÈ'vI;≥)a (s¤'v¤;å)b /s¢'vIÈ≤/ severe, ('™ksÈGÈnT, '™gz-)a
(-¤G-)b /'Eks¢GÈnt, 'Egz-/ exigent, (&khAmp<È'h™~SÈn)a (&khØmp>¤'h™~Só)b /kØmp<¢-
'hEnS˘n/.
2.1.6.2. It may be a good idea to take stock of the situation about some (real or
seeming) su‚xes and prefixes, because we still find old-fashioned and outdated tran-
scriptions, especially in bilingual dictionaries. ˛e modern neutral pronunciation,
British too, by now, has /È/ (while /I/ sounds quite pompous) in: -ace ('phπlÈs)
/'pπlÈs/ palace÷ -ate ('chAklÈT)a ('chØklÈT)b /'cØklÈt/ >ocolate÷ -iI ('hπpÈli) /'hπpÈ-
li/ happiI÷ -i“ ('khwAn[[]È[i)a ('khwØnTÈTi)b /'kwØnùÈùi/ quanti“˘
‹en /-Èùi/ is preceded by /s/, it can often become /-sti/. So, a notation like /-s˘ùi/
includes both possibilities, while excluding that the two diaphonemes /˘, ù/ may
work together, because of contextual incompatibilities. ˛at means that, if /˘/ falls,
then /ù/ automatically becomes /t/, because it is preceded by /s/): nece˛i“˚ univer-
si“˚ veloci“ /nÈ'sEs˘ùi, jUun¢'vÈ:≤s˘ùi, vÈ'lØs˘ùi/.
Besides, we have: -le˛ ('hø¨plÈs)a ('h‘¨plÈs)b /'hOUplÈs/ hopele˛÷ -ne˛ ('g¨DnÈs)
/'gUdnÈs/ goodne˛˘ For -e˛˚ /Ès/ prevails, chiefly in American English, while in Brit-
ish English /Is/ is also possible (actre˛˚ waitre˛); in some cases, /Es/ too (du>e˛);
for prince˛˚ we have: ('ph<¤nsÈs, -™s)a (p>¤n's™s, 'ph>¤ns™s)b. To end with, -let ('b<™Is-
lÈT)a ('b>-)b /'b<EIslÈt/ bracelet÷ -ret ('skA;<lÈT)a ('skA;lÈT)b /'skA:≤lÈt/ scarlet˘
2.1.6.3. We have /¢/ (which means, mainly /È/ in American, but /I/ in Briti´
English] for: -ed ('w™I[Èfl)a ('w™IT¤fl)b /'wEIù¢d/ waited÷ -es ('hø;<ßÈΩ)a ('hø;s¤Ω)b /'hO:≤-
s¢z/ horses÷ -e® ('b¤gûsT)a ('b¤g¤sT)b /'bIgIst/ bigge®÷ -et ('Th¤kûT)a (-¤T)b /'tIk¢t/ ti$et˚
('vEıvÈT)a ('v™ıv¤T)b /'vEıv¢t/ velvet (but ('¤nT<ÈsT, '¤n[[]È&<™sT)a ('¤n˛>ÈsT, '¤nTÈ&>™sT)b
/'Int<Èst, 'InùÈ<Est/ intere®]÷ -ite ('ApÈzÈT, -s-)a ('ØpÈz¤T, -s-)b /'ØpÈz¢t, -s-/ opposite˘
Besides: -ice ('AfÈs, 'O;-)a ('Øf¤s)b /'Ø;f¢s/ o‚ce÷ -ine (¤g'zπmÈn)a (-¤n)b /Ig'zπm¢n/
examine÷ -i« ('v™<È&faÙ)a ('v™>¤&faÙ)b /'vE<¢faE/ veri«˘
Furthermore: be- (bû'kh√m:)a (b¤'khåm:)b /b¢'k√m/ become÷ de- (DÈ'mπ;nfl,
-π:nfl)a (D¤'mA:nfl)b /d¢'mπ;nd/ demand (but ède¶ /dIi-/: (DIi'n™Ic≥)a (-cå)b /dIi'nEI-
cÈ≤/ denature]÷ pre- (ph<È'Th™n:fl)a (ph>¤-)b /p<¢'tEnd/ pretend÷ re- (<È'ThaÙ≥)a (>¤-
'ThaÙå)b /<¢'taEÈ≤/ retire (but çre-¶ /<Ii-/: (<Ii'g™;In)a (>Ii-)b /<Ii'gEIn/ regain]˘
In words like become˚ demand˚ pretend˚ retire˚ eleven˚ we could add that /¢/ has
a possible variant /i/ (or even /Ii/). ˛is, generally, belongs to a formal American
pronunciation; while, the British one is at the opposite side. ˛erefore, it is safer
to stick to what we have just said. Of course, everyone will decide for themselves,
especially through the regular consultation of a reliable pronunciation dictionary
(but it is much better to look up words, regularly, in more than one dictionary).
To end with, most internal -e-˚ -i- (in unchecked syllables), generally, have /¢/
(namely, as a trend, /È/ in American and /I/ in British pronunciation): ('™lÈmÈnT)a
('™l¤mÈnT)b /'El¢mÈnt/ element˚ (&¤nsTÈ'ThUuSÈn)a (&¤nsT¤'Thj¯uSó)b /Inst¢'tˆUuS˘n/ in®i-
tution˘
58 a handbook of pronunciation
2.1.6.4. On the contrary, regularly we have /I/ in: -ic(s) (fÈ'n™[¤k[s])a (-T¤k[s])b /fÈ-
'nEùIk[s]/ #onetic(s)÷ -ical (sÈ'Th¤<¤k®)a (->¤-)b /sÈ'tI<Ik®/ sa“rical÷ -i« ('sTπnD¤˙)
/'stπndI˙/ ®andi«÷ -´ip ('f<™~S¤p)a ('f>-)b /'f<EndSIp/ friend´ip÷ -ive (¤n'Th™ns¤Ñ)
/In'tEnsIv/ intensive˘
˛is group includes: -age ('v¤l¤ä) /'vIlIG/ village÷ -i´ ('¤˙gl¤S) /'I˙glIS/ Engli´÷ -i®
('l¤˙gw¤sT) /'lI˙gwIst/ li«ui®÷ e- and i- (at the beginning of words, in unchecked
syllables): (¤'l™vÈn) /I'lEvÈn/ eleven˚ (¤'Thπ¬jÈn) /I'tπljÈn/ Italian÷ im- and in- (in
checked syllables): (¤m'phø;<TóT, -<öóT)a (-ø;T-÷ -öóT)b /Im'pO:≤tót/ important˚ (¤n-
'sT™;fl) /In'stEd/ in®ead˘ In some American pronunciations, all these words may
have either /I/ or /È/.
As far as /¢/ is concerned, Oceanian and South African English are more like
American English.
Usually, /È/ is unstressed, being the weak vowel par excellence. ˛ere are two
forms, however, that are very often heard even with stressed /È/ (except in formal
pronunciation): ('ch¤ıD<Èn, 'chûı-, 'ch®-, 'ch¨ı-)a (-Ã>-)b /'cIıd<Èn, 'cÈı-, 'cUı-/ >il-
dren˚ ('DÈzóT, 'DÇóT, -óö) /'d√zót, 'dÈ-/ doesn't (this is given in an example in § 2.6.4,
too). Let us consider also the possibility of çrestressingÇ for emphasis, as in: Oh, I
didn't buy it: it wasn't the ('∑I;i, '∑Èö) dre˛, it was ju® a ('ö™;I, 'öÈö) dre˛ (example
adapted from a phonetics newsgroup).
Consonants
2.2.0. At the beginning of this section, we will show the table of the consonan-
tal articulations of (American and British) neutral English: û 2.6. It is useful to
make regular reference to it, in order to thoroughly understand the English conso-
nantal system.
û 1.9-15 show the orograms of all the contoids needed to describe English (and
the other languages dealt with in HPr), including secondary, occasional, or re-
gional variants, arranged according to their manner of articulation.
prevelar round.
postalveopalatal
velarized alveol.
{postalveolariz.}
velar rounded
postalveolar
protruded
labiodental
prepalatal
laryngeal
alveolar
bilabial
palatal
dental
velar
Nasals
2.2.1.1. English has three nasal phonemes\ /m, n, ˙/. ˛e velar phoneme does
not occur at the beginning of English words, but it is normal in internal and final
positions: ('n™;Im) /'nEIm/ name, ('†¤n:) /'†In/ `in, ('†¤˙:) /'†I˙/ `i«, ('bπ;˙) /'bπ˙/
ba«, (bû'kh√m¤˙)a (b¤'khåm¤˙)b /b¢'k√mI˙/ becomi«, ('s¤˙¤˙) /'sI˙I˙/ si«i«,
('lO:˙ û'gø;¨, 'lA;˙)a ('lØ;˙ û'g‘;¨)b /'lØ;˙ È'gOU/ lo« ago, ('sT<™˙[k]†ó¤˙)a ('s˛>-)b
/'st<E˙†óI˙/ ®re«`eni«˘
In unstressed syllables, after /t, d÷ †, ∑÷ s, z/, there is a typical intense (çsyllabicÇ)
realization of /n/ {(ó) /ó/; after /†, ∑/, we find ((2))}. It is definitely worthwhile to
use it in (dia)phonemic transcriptions, as well, although clearly its origin is /Èn/:
('khATó, -öó)a ('khØTó, -öó)b /'kØtó/ cotton, ('w¨Dó) /'wUdó/ wooden, ('≥;†ó)a ('‘;†ó)b
((-†2)) /'È:≤†ó/ ear`en, ('mø:< ∑ó'∑πT)a ('mø:)b ((∑2-)) /'mO:≤ ∑Èn'∑πt/ more `an `at,
('fπ[;]só)a ('fA;só)b /'fπ;só/ fa®en, ('chø¨zó)a ('ch‘¨zó)b /'cOUzó/ >osen˘ After /0t,
0d/, it is best to use /Èn/: ('w¤nsTÈn) /'wInstÈn/ Win®on˚ ('hø¨ıDÈn)a ('hÖ¨ı-)b
/'hOUıdÈn/ Holden˘
Furthermore, we have /˘n/ (= (Èn)a (ó)b) after /S, Z/: (ph<È&n√nsi'™ISÈn)a (-å-, -Só)b
/p<Èn√nsi'EIS˘n/ pronunciation˚ (phÈ'<¤ZÈn)a (->¤Zó)b /pÈ'<IZ˘n/ Parisian˘ In a prein-
toneme, (Èn) can easily become (ó), chiefly after /S, Z÷ c, G/.
After other consonants, /Èn/ (Èn, ûn) is more usual: ('l√~cÈn)a ('lå-)b /'l√ncÈn/
lun>eon˚ (<È'l¤GÈn)a (>¤-)b /<¢'lIGÈn/ religion˚ ('<¤bÈn)a ('>¤bÈn)b /'<IbÈn/ ribbon,
('√~jÈn)a ('å~jÈn)b /'√njÈn/ onion, ('Th™Ikûn) /'tEIkÈn/ taken˘
‹en speed is higher, especially in a preintoneme, we can go as far as (có, Gó÷
põ, bõ÷ kô, gô): ('l√~cÈn, -có)a ('lå-)b˚ (<È'l¤GÈn, -Gó)a (>¤-)b˚ ('ø¨pÈn, -põ)a
('‘¨-)b /'OUpÈn/ open˚ ('<¤bÈn, -bõ)a ('>¤-)b˚ ('Th™Ikûn, -kô), ('ø;<gûn, -gô)a ('ø;g-)b
/'O:≤gÈn/ organ˘ On the other hand, when speed is lower, or for emphasis, /ó/ can
easily become (Èn): ('l¤só, -sÈn) /'lIsó/ li®en˘
Regularly, we have: (n[D]Èn) /n[d]Èn/: ('l™nÈn) /'lEnÈn/ Lennon, ('l√nDÈn)a ('lån-)b
/'l√ndÈn/ London÷ but, ('wAnTó, -nöó, -nTÈn, 'wO;n-)a ('wØnTÈn÷ -nöó)b /'wØ;nt˘n/
wanton˘
2.2.1.2. A˛imilation is very important, and it must not be neglected either in
the description of languages, or in teaching and learning. Let us notice that, of the
three English nasal phonemes, the two marked ones, /m, ˙/, resist well; while, the
unmarked one, /n/, undergoes several changes, contrary to what phonemic tran-
scriptions generally seem to indicate.
However, proceeding in an organized manner, we have: ('D<I;imfl)a ('Ã>-)b
/'d<Iimd/ dreamed, ('D<™mT)a ('Ã>-)b /'d<Emt/ dreamt, ('s√m&ThaÙmΩ)a ('såm-)b /'s√m-
taEmz/ sometimes, ('s™;Im 'kha;Ùnfl) /'sEIm 'kaEnd/ same kind, (sÈm'G™li) /sÈm'GEli/
some jelI˚ ('s™;IM 'vøÙs, -m '6øÙs) /'sEIm 'vOEs/ same voice, ('s™;IM 'fπkT, -M] 'f-, -m
'åπkT) /'sEIm 'fπkt/ same fact, ('kh√Mf≥T, -M]f-, -må-)a ('khåMfÈT, -M]f-, -må-)b
/'k√mfÈ≤t/ comfort˘ As can be seen, only with labiodentals, /m/ shows a slight let-
-up, becoming labiodental, (Mf÷ M]f), but this happens just in trivial cases from a
lexical-semantic point of view. Otherwise, /f, v/ may become bilabial (constric-
tives): (må, m6).
60 a handbook of pronunciation
Our examples also show that at present a homorganic (labiodental) stop may
very often be inserted into the sequence (mf) (¤ only with voiceless /f/); but a too
frequent use is better avoided. ˛is homorganic insertion can happen with other
sequences too: ('wø;<m†, -mp†)a ('wø;m-)b /'wO:≤m†/ warm`, ('™mTi, '™mpTi) /'Em-
ti/ emp“, ('s¤msó, 's¤mpsó) /'sImsó/ Sim(p)son.
Even /˙/ resists well: ('<O:˙fl, '<A;˙fl)a ('>Ø;˙fl)b /'<Ø;˙d/ wro«ed, ('w¤˙&bπk) /'wI˙-
bπk/ wi«ba$˘ Only for the su‚x -i« (¤˙) /I˙/ (but at a non-neutral level, not to
be followed), can we have /In, ¢n, ó/: ('s™l¤˙÷ ≠-¤n÷ ≠-Èn) /'sElI˙/ selli«. We can also
find: ('l™˙†, -˙k†) /'lE˙†/ le«`˚ ('kh¤˙sTÈn, -˙ks-, -˙z-, -Tó) /'kInst˘n/ Ki«®on˚ (π˙-
'zaÙÈ[i, π˙g-)a (-Ti)b /π˙'zaEÈùi/ anxie“, ('π˙SÈs, 'π˙k-) /'π˙SÈs/ anxious.
2.2.1.3. On the contrary, except in a very accurate way of speaking (even too
accurate!), /n/ assimilates to a following contoid: (¤m'blπk) /Im'blπk/ in bla$˚ (¤M-
'vaÙT) /In'vaEt/ invite˚ (¤M'f™kT÷ ¤M]-) /In'fEkt/ infect˚ (¤˙∑È'bAks, ¤˙˙È-)a (-Øks)b /In-
∑È'bØks/ in `e box˚ ('Th™˙†÷ -˙t†) /'tEn†/ ten`˚ ('Th™ns÷ -˙ts) /'tEns/ tense˚ ('ph¤ns≥,
-ts≥)a (-å)b /'pInsÈ≤/ pincer (for ((˙)) see below).
For some time, /nzé/ has been simplified (but /ndzé/ can always be restored, ac-
cording to current spelling): ('w¤nz≥, -ndz≥)a (-å)b /'wInzÈ≤/ Windsor˚ ('l¤nzi, -dzi)
/'lInzi/ Linds\˘ Even in /nzò/, a /d/ can be inserted: ('khl™n:Ω, -n:dΩ) /'klEnz/ cleanse˚
('khl™nz¤˙, -ndz¤˙) /'klEnzI˙/ cleansi«. But this is less and less recommendable,
specially with grammemes: (¤'Thπ¬jÈnΩ÷ -ndΩ÷ È-) /I'tπljÈnz, È-/ Italians˚ ('GA;nΩ÷
-ndΩ)a ('GØ;-)b /'GØnz/ John's˘
Furthermore, we have (~) (postalveopalatal, ((n))): ('¤~c) /'Inc/ in>˚ (¤~'chaÙn√)a
(-å)b /In'caEnÈ/ in ≥ina˚ ('™I~G®) /'EInG®/ a«el˚ (È'Th™~SÈn÷ -~cÈn÷ -~+SÈn)a (-~Só÷
-~có÷ -~+Só)b attention˘ But we find (~) (prepalatal) before /j/ (if this is hetero-
syllabic): (khûm'phπ~jÈn) /kÈm'pπnjÈn/ companion˚ ('√~jÈn)a ('å-)b /'√njÈn/ onion˘
Besides: ('h™n<i)a ('h™N>i)b /'hEn<i/ Henry˚ (¤n'Th<¤ns¤k)a (¤N'˛h>-)b /In't<InsIk/ in-
trinsic˚ ('Th™˙ 'g≥:ıΩ, 'g≥;®Ω)a ('g‘:ıΩ)b /'tEn 'gÈ:≤ız/ ten girls˚ Â. Also (¤m'w¤n[m]≥)a
(-nTå)b /In'wInùÈ≤/ in winter should be noticed.
A better transcription for (~c, ~G, ~+S) would be ((nc, nG, ntS)), with ((n)) – and
even ç((t))Ç (for a çpostalveopalatal stopÇ). It was stated above that dental ((˙)) could
be represented with (n), as well, chiefly in (n∑), because the simple fact that /n∑/
has (dental) (∑) allows us to infer that we automatically have ((˙∑)), by assimilation.
On the contrary, a special symbol would be more important in ((˙˙)), to show
that it is not (nn) (alveolar, but dental, coming from ((˙∑)) /n∑/), as, for instance,
in: (¤n'nø;¨&Tha;Ùm)a (-‘;¨-)b /In'nOUtaEm/ in no time. On the other hand, for some
speakers, a less completely assimilated realization produces exactly (nn). ˛is is
not very di‡erent auditorily, so more precise symbols could even be avoided, with
no real drawbacks.
In lexical composition, as well as for the negative prefix un-, in a slow — careful
way of speaking, people try to keep (n), while –currently– assimilation to the place
of articulation of a following consonant is quite regular.
As an actual compromise, here we will show that it is possible to maintain an
apical contact while adding a secondary coarticulation (with no full contact) – bi-
labial, (º); labiodental, (¸); velar, (n): (&√mbÈ'lIivÈb®, &√º-)a (&åmb¤-, &åº-)b /√nb¢-
2. english 61
Stops
2.2.2.1. ˛ere are three diphonic pairs of ®ops (¤ pairs of both a voiceless and
a voiced articulation): /p, b÷ t, d÷ k, g/. Of course, /t, d/ are alveolar (T, D): (ThÈ'D™;I)
/tÈ'dEI/ todZ, ('D™sTÈ&TUuT)a ('D™sT¤&Tj¯uT)b /'dEst¢tˆUut/ de®itute˘ However, before /†,
∑÷ s, z/, /t, d/ become dental (t, d): ('™It†) /'EIt†/ eigh`, ('w¤∂†) /'wId†/ wid`,
('hπts) /'hπts/ ha_, ('h™;dΩ) /'hEdz/ heads˘
On the contrary, before /</ (>)b, in British English, /t, d/ become postalveolar,
always due to assimilation: ('Th<a;Ù)a ('˛h>a;Ù)b /'t<aE/ try, ('kh√nT<i)a ('khåN˛>i)b
/'k√nt<i/ country, ('D<¤˙k)a ('Ã>¤˙k)b /'d<I˙k/ drink˘
For /t<, d</, however, several pronunciations are possible, mostly with /t, d/ real-
ized as stop-strictives (or ça‡ricatesÇ): in American English (ç[h]<, Ç<÷ C[h]<, ‚<÷
c[h]<, G<)a; in British English (ú[h]>, ∂>÷ fi[h]>, ">÷ C[h]>, ‚>÷ c[h]>, G>)b.
Furthermore, a (homorganic) constrictive realization is possible for /</ (which is
then, respectively, alveolar or postalveolar rounded, (ï, ø)): (T[h]ï, Dï)a (˛[h]ø, Ãø)b.
So: ('Th<a;Ù, 'çh<-, 'Ch<-, 'ch<-, 'Thï-)a ('˛h>a;Ù, 'úh>-, 'fih>-, 'Ch>-, 'ch>-, '˛hø-)b /'t<aE/
try, ('kh√nT<i, -ç<i, -C<i, -c<i, -Tïi)a ('khåN˛>i, -ú>i, -fi>i, -C>i, -c>i, -˛øi)b /'k√nt<i/
country, ('D<¤˙k, 'Ç<-, '‚<-, 'G<-, 'Dï-)a ('Ã>¤˙k, '∂>-, '">-, '‚>-, 'G>-, 'Ãø-)b /'d<I˙k/
drink˘
All these pronunciations are possible as neutral ones too, although opinions re-
garding their correctness may be influenced by spelling. On the other hand, from
a structural point of view, /c<, G</ could represent a fitting parallelism with /S</,
as in ('S<¤˙k)a ('S>-)b /'S<I˙k/ ´rink˘
2.2.2.2. Some other transformations of /t, d/ are much more significant; indeed,
although to foreigners (T, D) might seem more çmarkedÇ, actually, in the natives'
phonological system, /t, d/ are an unmarked diphonic pair of stops (as happens to
/n/ in comparison with /m, ˙/). From an articulatory point of view, /t, d/ are liable
to assimilation; not to çcomplicateÇ things, but rather to make them easier.
So, /t, d/, before /p, b, m, w/, generally, become (p, b): (&∑πp'mπ;n) /∑πt'mπn/
`at man˚ (&∑πp'bø;Ù) /∑πt'bOE/ `at boy˚ ('nAp 'w√n:)a ('nØp 'wån:)b /'nØt 'w√n/ not
one; likewise, before /k, g/, they become (k, g): ('wAk kû~jÈ'DU;u, 'w√-)a ('wØk kû~-
j¨'D¯;u)b /'wØt kÈnjÈ'dUu/ what can you do?˚ (&∑πk'g≥:ı, -≥;®)a (-‘:ı)b /∑πt'gÈ:≤ı/ `at
girl˘
62 a handbook of pronunciation
Even the rare sequences /pf, bv/ present some kind of assimilation (in one direc-
tion or in the other): ('kh√på¨ı, -]f¨ı)a ('khå-)b /'k√pfUı/ cupful, ('Ab6iÈs, 'A7v-)a
('Ø-)b /'ØbviÈs/ obvious˘
Prevelar articulations, which are automatic by coarticulation, need not be ex-
pressly written down: ('†¤˙k¤˙, 'g™T) {(('†¤”´¤˙, 'Ò™T))} /'†I˙kI˙, 'gEt/ `inki«˚ get˘
2.2.2.3. One fundamental thing, already seen in previous examples, which must
not be neglected in learning and teaching, is that, in stressed syllables, initial /p,
t, k/ are çaspiratedÇ (unless they are preceded by /s/ in the same syllable and in a
same lexeme), also after silence (even in an unstressed syllable) – (0h) /0/: (ThÈ-
'D™;I) /tÈ'dEI/ todZ, ('phIik) /'pIik/ peak (but: ('spIik) /'spIik/ speak]˚ ('Th™Ik) /'tEIk/
take (but we have: ('sT™Ik) /'stEIk/ ®ake]˚ ('kh™;≥)a ('kh™;‘)b /'kEÈ≤/ care (but: ('sk™;≥)a
('sk™;‘)b /'skEÈ≤/ scare]. Nevertheless, one should notice: (m¤s'ph<¤n[[]Èfl)a (m¤s-
'ph>¤nT¤fl)b /mIs'p<Inù¢d/ misprinted (with di‡erent phono-syllables and di‡erent
morphemes).
2.2.3.2. Let us now turn to the contexts where /t/ (T) becomes /ù/ ([, m), in nor-
mal (not slow, nor particularly careful) speaking.
Between a stressed (or unstressed) vowel and another vowel, or (≥, ®): ('b™[i)a
('b™Ti)b /'bEùi/ Bet“, ('vIi[ø¨)a ('vIiT‘¨)b /'vIiùOU/ veto, (&v¤zÈ'b¤lÈ[i)a (-ÈTi)b /vIzÈ'bIlÈ-
2. english 63
ùi/ visibili“, (&π[È'm¤sT¤k)a (&πTÈ-)b /πùÈ'mIstIk/ atomi®ic, ('l™m≥)a ('l™Tå)b /'lEùÈ≤/ let-
ter, ('l¤[®)a ('l¤T®)b /'lIù®/ little˘
Between /n, ≤, ı/ and a vowel, or (≥, ®) (remembering that, as our examples show,
([, m) may often be dropped after /n/, (n[[], n[m])): ('bπn[[]Èm)a ('bπnTÈm)b /'bπnù-
Èm/ bantam, ('w¤n[m]≥)a ('w¤nTå)b /'wInùÈ≤/ winter˚ ('mπn[[]®)a ('mπnT®)b /'mπnù®/
mantel, ('†≥;[i)a ('†‘;Ti)b /'†È:≤ùi/ `ir“, ('phA;<[i)a ('phA;Ti)b /'pA:≤ùi/ par“, ('SEım≥, -T≥)a
('S™ıTå)b /'SEıùÈ≤/ ´elter˘
Even before a stressed vowel (provided it is heterosyllabic): (phÈ'Th™I[ø¨, phÈ[-
'™I[ø¨)a (phÈ'Th™IT‘¨)b /pÈ'tEIùOU÷ pÈù'EIùOU/ potato, (È['O:ı)a (È'Thø:ı)b /Èù'O:ı, È'tO:ı/ at
all (it should be noted that there is a di‡erence, between the two accents, for at all]˘
˛e sequences /nù, ≤ù, ıù/ may even have çfusedÇ realizations, (å, m, e): ('ph™Iå¤˙)a
/'pEInùI˙/ painti«, ('phA;mi)a /'pA:≤ùi/ par“, ('fO;ei)a /'fO:ıùi/ faul“.
However, in American English, when speed is reduced, or when more attention
is paid to the way of speaking, /ù/ ([) becomes /t/ (T): ('l™m≥, 'l™T≥) /'lEùÈ≤/ letter˚
('w¤n[m]≥, -nT≥) /'wInùÈ≤/ winter˘ ˛e same goes, even in a normal manner of speak-
ing, for /ıù/: ('SEım≥, -ıT≥) /'SEıùÈ≤/ ´elter, ('O;ım≥, -ıT≥, 'Aı-) /'O;ıùÈ≤/ alter˘ ˛is happens
even to less common words, such as: ('vIi[ø¨, -Tø¨) /'vIiùOU/ veto, ('phl™I[ø¨, -Tø¨)
/'plEIùOU/ Plato˚ (d¤'fIi[&¤zõ, -&T¤zõ) /dI'fIitIzõ/ defeatism.
Also in Australia, New Zealand, South Africa (and in towns in Wales and Ire-
land) /ù/ is ([, m); while, Scotland is well-known for /ù/ = (ö), even between vowels
and before (®) /®/ (and even before its typical pronunciation of (ÈR) /È≤/). ˛e same
change, /ù/ = (ö), occurs even in broad accents in England, in particular, in Lon-
don, Birmingham, Â.
Unexplosion
2.2.4.1. Notably, in English, stops (both voiced and voiceless) are unreleased,
chiefly after /é[m, n, ˙, ≤, ı]/, before pauses or consonants. ˛is means that their
third phase (¤ their o‡-glide) is inaudible, incomplete. It is useful to put the dia-
critic (æ) after a proper symbol, to show unreleased contoids, especially at first and,
of course, when it is the subject in question, as here.
˛erefore, (voiced or voiceless) stops are unreleased after (stressed or unstressed)
vowels, even followed by homorganic N (/mp, mb÷ nt, nd÷ ˙k, ˙g/), or by /≤, ı/.
˛is holds good except in very slow or careful pronunciation.
Here are some examples, although it is to be remembered that, when single
words are said in isolation, before a pause, it is more usual to produce (and let
hear) the o‡-glide, indicated by ($): ('bA;Êæ)a ('bØ;Êæ)b /'bØb/ Bob˚ ('kh√pæ)a ('khåpæ)b
/'k√p/ cup˚ ('khπmpæ) /'kπmp/ camp÷ ('hπTæ) /'hπt/ hat˚ ('hπ;nflæ) /'hπnd/ hand˚
('<™;flæ)a ('>™;flæ)b /'<Ed/ red˚ ('Sø;<Tæ)a ('Sø;Tæ)b /'SO:≤t/ ´ort˚ ('fO;ıTæ, 'fAıTæ)a ('fø;ıTæ, 'fØıTæ)b
/'fO;ıt/ fault˚ ('blπkæ) /'blπk/ bla$˚ ('bπ˙kæ) /'bπ˙k/ bank˚ ('D¤;âæ) /'dIg/ dig. ‹en
said in isolation, as plain examples, before a pause, they are released\ ('bA;Ê$)a
('bØ;Ê$)b˚ ('hπ;nfl$)˚ ('Sø;<T$)a ('Sø;T$)b˚ ('bπ˙k$), Â.
Before a consonant, they are unreleased: ('πkæT) /'πkt/ act˚ ('πkæT≥)a (-å)b /'πktÈ≤/
actor˚ ('<√;bæfl)a ('>å;bæfl)b /'<√bd/ rubbed, ('πTæk¤nsó, 'πkæk-) /'πtkInsó/ Atkinson,
64 a handbook of pronunciation
('b™;gæfl) /'bEgd/ begged˚ ('sTApæT)a ('sTØpæT)b /'stØpt/ ®opped, ('hATæ&dAâ, -O;â, -öæ-)a ('hØTæ-
DØâ, -öæ-)b /'hØtdØ;g/ hotdog˚ ('sTApæ 'DUu¤˙)a ('sTØpæ 'D¯u¤˙)b /'stØp 'dUuI˙/ ®op doi«˘
More examples: ('Gπkæ 'phi;Iı, -iI®) /'Gπk 'pIiı/ Ja$ Peel, ('bA;bæ 'gø;¨Ω)a ('bØ;bæ
'g‘;¨Ω)b /'bØb 'gOUz/ Bob goes˚ ('skATælÈnfl)a ('skØTæ-)b /'skØtlÈnd/ Scotland˚ (¤pæ'baÙts÷
¤Tæ'b-) /It'baEts/ it bites˚ (¤Tæ'Th™Iks, ¤öæ'Th-) /It'tEIks/ it takes˚ (¤kæ'khπ<iΩ÷ ¤Tæ'kh-)a (->-)b
/It'kπ<iz/ it carries˘
Intermediate articulations are possible between the two extreme ones, above all
when people pay special attention to their speech (although this must not lead us
to think that these pronunciations are necessarily çbetterÇ). In fact, for /t, d/ (+ /p,
b÷ k, g/, in addition to (pp, pb÷ Êp, bb÷ kk, kg÷ âk, gg)), the alveolar contact can
be maintained, by adding either a bilabial, (B, ∫), or a velar, (Ú, ¢), coarticulation.
A bilabial or velar articulation is also possible, to which an alveolar coarticulation
can be added: respectively (Q, Á) or ($, ¢).
2.2.4.2. ˛e so-called çnasalÇ and çlateralÇ explosions are included in this group.
It is essential that transitions from /t, d/ to /n, ó÷ l, ®/ are direct, with no o‡-glide
similar to çaspirationÇ, and even with no insertion of vocoids.
˛erefore, we have: ('ph¤Tæni) /'pItni/ Pitn\, ('w¨Dæó) /'wUdó/ wooden, (khûm-
'phlIiTæli, -öæli) /kÈm'plIitli/ completeI, ('l¤[®)a ('l¤Tæ®)b /'lIù®/ little, ('sw¤nDæ®) /'swInd®/
~indle.
2.2.5.1. It is a good thing to include the laryngeal (or glottal) stop (ö) in the
symbols inventory of the English phonological system, even if, strictly speaking,
there are no (classical) minimal pairs, in order to declare its phonological status.
˛e fact is that it is important, too, to have (ö) from a descriptive and teaching
point of view.
In the (American and British) neutral pronunciation (ö) is used, when there is
some emphasis, before vowels, especially stressed ones: (¤ts'öO;f®)a (-'öø;-)b /Its'[ö]O:f®/
it's Yful! In British pronunciation, (ö) may be used even to avoid the insertion of
a non-etymological /</ (çintrusiveÇ, at the end of § 2.2.9.4): ('lø: öÈn'ø;Då) /'lO: Èn[d]-
'O:≤dÈ≤/ lY and order˚ instead of the frequent ('lø:> Èn'ø;Då) (('lO: Èn'ø;<D≥)a).
-ö-, -T-)a (-ø:ı)b /'fUtbO:ı/ football, ('khπk&g√T, -ö-, -T-)a (-å-)b /'kπtg√t/ catgut˚ ('aÖt '∑™;≥,
-ö)a (-™;‘)b /'aOt '∑EÈ≤/ out `ere, ('n√T&SEı, -ö-)a ('nåT&S™ı, -ö-)b /'n√tSEı/ nut´ell˘
Let us also notice: (aÙ'khπ[;]nT {'DUu¤T}, -nö, -n {-È-, -ö})a (-A;n- {-¯u¤T, -ö})b /aE-
'kπ;nt {'dUuIt}/ I can't [do it]˚ (wi'wø¨nT {'DUu¤T}, -nö, -n {-È-, -ö})a (-‘¨n- {-¯u¤T, -ö})b
/wi'wOUnt {'dUuIt}/ we won't [do it]˘ In absolute final position, (ö) is not used, ex-
cept for /It, Èt/, and this only in informal speech: (wi'wAnT, -O;nT)a (-ØnT)b /wi-
'wØ;nt/ we want˚ (wi'wAn[[]¤T, -ÈT, -ö)a (-ØnT¤T, -¤ö)b /wi'wØ;nùIt/ we want it˚ ('Th¤kÈT,
-ö)a (-¤T, -¤ö)b /'tIk¢t/ ti$et.
However, the change /t/ = (ö) is less frequent before /h/: ('™IT 'hπts÷ -ö) /'EIt 'hπts/
eight ha_˘ It sometimes occurs in /étó/: ('khATó, -öó)a ('khØTó÷ -öó)b /'kØtó/ cotton˘
In /ntó/, it is more frequent (but it is only possible with /≤, ı/): ('sk<πnTó, -öó,
-[[]Èn)a ('sk>πnTÈn÷ -öó)b /'sk<πnt˘n/ Scranton˚ ('khl¤nTó, -öó, -TÈn)a ('khl¤nTÈn÷ -öó)b
/'klInt˘n/ Clinton˚ ('EıTó, -öó, -[Èn)a ('™ıTó, -TÈn÷ -öó)b /'klInt˘n/ Elton˘
/t/ = (ö) is no neutral pronunciation, before /é, È≤, ®/: ('b™[i)a ('b™Ti)b /'bEùi/ Bet“,
('b™m≥)a ('b™Tå)b /'bEùÈ≤/ better, ('l¤[®)a ('l¤T®)b /'lIù®/ little ((≠'b™öi), (≠'b™ö≥)a, (≠'b™öå)b,
(≠'l¤ö®)). Nevertheless, it can occur in /tò'é/: (&nAT'IivÈn, &nA['-, &nAö'-)a ('nØ-)b /nØt'Ii-
vÈn/ not even˘ It is accepted even for it˚ before a pause: ('Th™Ik¤T, -¤ö) /'tEIkIt/ take it˘
‹ereas, it is possible, for /p/, (only) before /p, b/, and for /k/, (only) before /k,
g/ (otherwise, it is not neutral): ('sø¨p 'phaÖD≥, -ö)a ('s‘¨-, -Då)b /'sOUp 'paOdÈ≤/ soap
powder, ('b¨k&kh™Is, -ö&kh-) /'bUkkEIs/ bookcase, ('bπk 'gA;<Dó, -ö)a ('gA;Dó)b /'bπk
'gA:≤dó/ ba$ garden˘
British glottalization
Lenitions
È≤/ ladder (Ô ('lπm≥) /'lπùÈ≤/ latter˚ which may have a slight di‡erence in duration,
(ò) – shorter than a semi-chrone, (;)), ('<aÙD≥÷ '<aòÙm≥) /'<aEdÈ≤/ rider (Ô ('<aÙm≥) /'<aEùÈ≤/
writer]˚ ('b™Di÷ 'b™ò[i) /'bEdi/ beddy (Ô ('b™[i) /'bEùi/ Bet“]˘ However, such a pronun-
ciation may not be considered neutral, though very widespread, but only çmediat-
icÇ.
Another (and neutral) way to keep a di‡erence, partially recovers the voiceless-
ness of /ù/: ('lπb≥÷ '<aÙb≥÷ 'b™Ai), or –better still– totally: ('lπr≥÷ '<aÙr≥÷ 'b™Ói) (as men-
tioned and illustrated above), with no lengthening of the vocoid before /d/ ([, m).
In informal Briti´ pronunciation, chiefly in monosyllables of low semantic val-
ue, /tòé/ may be realized as ([): (Thû&g™TÈ'ph>™zóT, -&g™[È-) /tÈgEùÈ'p<Ezót/ to get a
present˚ (&gØTÈ'mπc, &gØ[È-) /gØùÈ'mπc/ got a mab˚ (&khwaÙTÈ&b¤TÈÑ'˛h>πf¤k,
&khwaÙ[È&b¤[È-) /kwaEùÈbIùÈv't<πfIk/ quite a bit of tra‚c˚ (¤Tû'kh‘;flTÈmi, ¤[-) /ItÈ-
'kÈ:≤tÈmi/ it occurred to me˚ (&sø;T*'GØ;Ê, &sø;[*-) /sO:≤ùÈv'GØb/ sort of job˚ ('wØT È-
'ph¤Ti, 'wØ[ È-) /wØùÈ'pIùi/ what a pi“˚ ('nØT ‘¨nli'∑¤s, 'nØ[ ‘¨n-) /nØùOUnli'∑Is/ not
onI `is˚ (bÈT'ø;ıs‘¨ '∑πT, bÈ['-) /bÈù'O:ısOU '∑πt/ but also `at˚ (bÈTaÙ'D¯;u, bÈ[aÙ-)
/bÈùaE'dUu/ but I do.
2.2.7.1. English has just one (diphonic) pair of stopstrictives, (c, G) /c, G/. For
segments, or phones, articulatory terms are preferred over auditory ones (and, of
course, acoustic ones), because they are much more adequate and clear, generally
self-explanatory.
For this reason, we are happy to avoid ça‡ricateÇ, in favor of a more descriptive
and tangible (even checkable) term, such as pre®opped con®rictive, which we will
presently reduce to ®op-®rictive, after explaining that they are unitary phones, or
çsoundsÇ, in that they have a total duration comparable to that of any other sin-
gle phone, like (p, t, T, k) or (f, s, S, x), not like the sum of two of them (as in (ts,
kx)).
In addition, they must be homorganic (¤ produced at the same place of articul-
ation). So, the first half of a stop-strictive consonantal phone is a short stop, while
its second half is a short çfricativeÇ one (or, better, a constrictive one). ˛e place
of articulation is determined by the second component, to which the first one is
just a mere closure, correctly at the same place (even if no actual stop phone exists
at that place, in any real language).
˛e simpler and more convenient way to symbolize stop-strictive phones is by
means of two çmonographedÇ symbols. Of course, the second one is the more spe-
cific, so the first can be a looser one, because its only function is to show a closure,
which may be generically labial, pre-lingual or post-lingual. For this reason, the
stop phases of the various possible stop-strictive phones, are su‚ciently shown by
using simply (p, b÷ t, d÷ k, g).
As we said, the only (diphonic) pair of stop-strictives of the English language is
(c, G) /c, G/. In stressed syllables (or after pauses, even in the rare cases of un-
stressed syllable, as in ≥aucerian), /c/ is çaspiratedÇ, as /p, t, k/ are (although most
68 a handbook of pronunciation
native phoneticians do not say that, in the least): ('ch¤mni) /'cImni/ >imn\.
Usually, /c, G/ have (a slight) labial protrusion, and, most often, they are artic-
ulated with the tongue tip in a high position (but we need not really use special
symbols, such as ((c, g))).
Besides, /G/ (as any other voiced phoneme in diphonic pairs) is partially de-
voiced before a pause or before a voiceless consonant: ('G√;ä)a ('Gå;ä)b /'G√G/
judge.
‹ile English stops are very often inaudibly released, English /c, G/ always
show an audible plosion, even when they occur before themselves (notice that we
prefer to mark this plosion only here, by means of (0$)): ('wAc$ 'kh™≥fÍi÷ 'wO;c$)a
('wØc$ 'kh™‘fÍi)b /'wØ;c 'kEÈ≤fÈli/ wab carefulI˚ ('w¤c$ 'chI;iΩ÷ 'W¤c$÷ 'hw-) /'·wIc
'cIiz/ whi> >eese˚ (È'lA:<ä$ 'ch≥;c)a (È'lA:ä$ 'ch‘;c)b /È'lA:≤G 'cÈ:≤c/ a large >ur>.
˛e only possible reduction may be in changing the first stop-strictive with the
corresponding constrictive (çfricativeÇ) one, (S, Z): ('w¤S 'chI;iΩ÷ 'W¤S÷ 'hw-), (È'lA:<Ë
'ch≥;c)a (È'lA:Ë 'ch‘;c)b˘ ˛is can also occur with /nG, nc/ before /c, G/ (seen that
/nG, nc/ have a less frequent variant /nZ, nS/): ('O;<È~G 'GUus, -~Z, 'A<-)a ('Ø>¤~G
'G¯us, -~Z)b /'Ø;<¢nG 'GUus/ ora«e juice˘
For British English, we must add that /c/, as well as /p, t, k/, can show the two
kind of glottalization we saw (§ 2.2.6.1-2), with the same frequence and degree of
advisability. But, for /c/, it occurs even before vowels: ('f™c, 'f™¸, 'f™öc) /'fEc/ feb,
('b™~c, -~¸, -~öc) /'bEnc/ ben>, ('s‘;cT, -¸T, -öcT) /'sÈ:≤ct/ sear>ed, ('s‘;cmi, -¸mi,
-öcmi) /'sÈ:≤cmi/ sear> me, ('f™c¤T, -¸¤T, -öc¤T) /'fEcIt/ feb it, ('Th¤icå, -¸å, -öcå)
/'tIicÈ≤/ tea>er.
≈nally, chiefly in British English, /c/ may become (öS), before a pause or a C\
('f™öS, 'b™~öS, 's‘;öST, 's‘;öSmi) (examples we have already seen) and ('khπc÷ 'khπöS)
/'kπc/ cab˚ (¿5w¤c 'b¨k3 3÷ ¿5w¤öS 'b¨k3 3) /'w·Ic 'bUk/ whi> book?
2.2.8.1. Also for this manner of articulation (as for the stop-strictive one, rather
than ça‡ricateÇ), we prefer to use an articulatory term, because of its greater clari-
ty.
˛ere are four (diphonic) pairs, /f, v÷ †, ∑÷ s, z÷ S, Z/ (f, v÷ †, ∑÷ s, z÷ S, Z). For /f,
v/ readers are referred to what has been said about /pf, bv/ (§ 2.2.2.2). We now add
some examples that show the frequent reduction or dropping of /v/: (aÙ'khπ[;]m
bÈ'lI;iv&∑πT, -iV&∑-, -i&∑-)a (-A;m b¤-)b /aE'kπ;nù b¢'lIiv∑πt/ I can't believe `at˚ ('g¤vmi
'fa;ÙÑ, 'g¤Vmi, 'g¤mi) /'gIvmi 'faEv/ give me five!˚ (aÙÑ'Sø;<TóD¤T, aÙV-, aÙ-)a (-ø;T-)b
/aEv'SO:≤tód¢t/ I've ´ortened it (in spite of an information loss in comparison with
I ´ortened it), (∑™≥'la;Ùvz ≥'Th™<Èb®, -ÙVz, -Ùz, ∑≥-)a (∑™È-, È'Th™>¤-)b /∑EÈ≤'laEvz È≤'tE<-
¢b®/ `eir lives are terrible (in spite of the ambiguity with `eir lies are terrible),
('fø:≥2 'fa;ÙÑ2 's¤ks2 's™vè2 '™IT2÷ 'fa;ÙV2 's¤k2 's™bõ2÷ 'fa;Ù2 's¤k2 's™õ2)a ('fø:2)b /'fO:≤ 'faEv 'sIks
'sEvó 'EIt/ …four, five, six, seven, eight… And, let us note: ('G™å 'ph¤T, 'fa;Ù6 'b¤ts)
/'GEf 'pIt, 'faEv 'bIts/ Je‡ Pit˚ five bits.
/†, ∑/ are slit dental (whereas in American English a çprodentalÇ or çinterdentalÇ
2. english 69
articulation is possible, perhaps more often indeed, which may be transcribed with
((w, W)); however, since their auditory impression is not very di‡erent, the o‚cial sym-
bols, (†, ∑), may be su‚cient. It is important for foreigners to acquire this slit ar-
ticulation well: ('†¤˙k) /'†I˙k/ `ink, ('∑¤s) /'∑Is/ `is, ('w¤;„) /'wI∑/ wi`. ‹en a sen-
tence begins with a more or less pleonastic I `ink, generally, a semi-constrictive /†/
is used: (√&fl¤˙k)a (å-)b; the same can occur for the /ò∑/, (∂), of grammemes: `e˚ `is…
Besides, in quick informal speech, /ò∑/ in forms such as `e˚ `at˚ `is˚ bo` is
normally assimilated: ('w√ts zÈ'Tha;Ùm, 'wAts)a ('wØts)b /'wØts ∑È'taEm/ what's `e
time?, ('w√ts 'zπT, 'wAts)a ('wØts)b /'wØts '∑πt/ what's `at?, ('bø¨s 'sa;ÙdΩ)a ('b‘¨s)b
/'bOU† 'saEdz/ bo` sides˚ (hiz∑û'w™Im≥, hiz∂û-, hizzû-)a (-Tå)b /hiz∑È'w™ItÈ≤/ he's `e
waiter˚ (¤˙∑È'mø;<n¤˙, ¤˙∂È-, ¤˙˙È-)a (-ø;n-)b /In∑È'mO:≥nI˙/ in `e morni«˚ ('O:Ï ∑È-
'Tha;Ùm, 'O:Ï∂È-, 'O:ÏlÈ-)a ('ø:Ï)b /'O:ı ∑È'taEm/ all `e time˘
2.2.8.2. /s, z/ are grooved dental constrictives, usually pronounced with the tip
of the tongue raised, ¤ çdenti-alveolarÇ, so that they could be transcribed with (s,
z), especially for comparative purposes, in order to emphasize the di‡erence be-
tween (s, z), pronounced with the tip of the tongue lowered. On the other hand,
native speakers themselves may indi‡erently have one articulation or another, even
vacillating, so the plain symbols can safely be used: ('s™;Im) /'sEIm/ same, ('<ø¨zÈΩ)a
('>‘¨z¤Ω)b /'<OUz¢z/ roses, ('s¤z≥Ü)a ('s¤z‘Ω)b /'sIzÈ≤z/ sci˛ors˘
For word-initial sm-˚ sn-˚ sl- (as well as for the non-autochthonous sr-), /s/ (s) is
normal (contrary to some other languages): ('smO:ı)a ('smø:ı)b /'smO:ı/ small˚
('sn™Ik) /'snEIk/ snake, ('slIip) /'slIip/ sleep ((&s<Ii'lA;˙k√, s<¤-, &S<Ii-)a (-π˙kå)b /s<Ii'lA;˙-
kÈ, s<I-, S<Ii-/ Sri Lanka).
For dis- followed by a voiced stop, there are several possibilities, both phonetic
and phonemic: (&D¤sbÈ'lI;iÑ, -sÊ-) /dIsbÈ'lIiv/ disbelieve˚ (D¤s'D™;In, -s'fl-, -Ω'D-, -z'D-, -'sT-)
/dIs'dEIn, -z'd-, -s't-/ disdain˚ (D¤s'g√sT, -s'â-, -Ω'g-, -z'g-, -'sk-)a (-åsT)b /dIs'g√st, -z'g-, -'sk-/
disgu®˚ (D¤s'ga;ÙΩ, -s'â-, -Ω'g-, -z'g-, -'sk-) /dIs'gaEz, -z'g-, -'sk-/ disguise˘
Let us now observe (but only here) that /s, z/ preceded by one or more conso-
nants, are usually articulated as (dental/denti-alveolar) approximants, (∂, d), rather
than as constrictives: (('SAp∂))a (('SØp∂))b /'SØps/ ´ops˚ (('†π˙k∂)) /'†π˙ks/ `anks˚
(('bEı:D))a (('b™ı:D))b /'bEız/ bells˚ (('hπn:dD)) /'hπndz/ hands. But it is su‚cient to tran-
scribe: ('SAps)a ('SØps)b, ('†π˙ks, 'bEı:Ω, 'hπ;ndΩ)a ('b™ı:Ω)b.
In American English, the sequences /≤s, ≤z/ are realized as (<ß, <fi) (with alveolar
constrictives): ('f≥;ßT) /'fÈ:≤st/ fir®˚ ('ph≥;ß) /'pÈ:≤s/ purse˚ (h≥'ßk≥;T) /hÈ≤'skÈ:≤t/ her skirt˚
(f≥'ßmO:ı 'phIip®) /fÈ≤'smO:ı 'pIip®/ for small people˚ ('Dø:<Ü) /'dO:≤z/ doors˚ ('f≥:Ü) /'fÈ:≤z/
furs˘ In British English, /s<, z</ generally become (ß>, fi>): ('nj¯ufi&>iIı) /'nˆUuz<Iiı/
n[sreel.
More often, /S Z/ are produced with the tip of the tongue raised, ¤ as çapico-
-postalveo-palatal roundedÇ contoids, so that more suitable symbols, contrastive-
ly, could be ((x, ç)). But, as several natives pronounce them with the tip of the
tongue lowered, (S, Z) will be su‚cient. ˛e most important thing to keep in mind
is that they have a certain degree of lip protrusion (indeed, they must be labeled
as protruded]: ('S¤p) /'SIp/ ´ip, ('phl™Z≥)a (-å)b /'plEZÈ≤/ pleasure˘
For /s, z/, too, assimilation is rather important. As a matter of fact, /s, z/ = /S, Z/
70 a handbook of pronunciation
(S, Z) before /S÷ c, G÷ j/: (∑¤S'SAp)a (-Øp)b /∑Is'SØp/ `is ´op, (∑¤S'GA:<)a (-A:)b /∑Is-
'GA:≤/ `is jar, (∑¤S'jI;≥)a (-¤;å)b /∑Is'jIÈ≤/ `is year, (∑IiË'SU;uΩ)a (-¯;uΩ)b /∑Iiz'SUuz/ `ese
´oes, ('hπËSi, 'hπSSi) /'hπzSi/ has ´e?÷ even /sc/ = (Sc) is possible: ('khw™scÈn, -Sc-)
/'kwEsc˘n/ que®ion˘
Generally, with you˚ your˚ there is a complete fusion between elements: (aÙ'm¤S√,
-¯)a (aÙ'm¤S¯÷ -å)b /aE'mIsju/ I mi˛ you, (hi'nIiD Z≥'hEıp, -G ≥-÷ -U≥-)a (ZÈ'h™ıp, -G È-÷
-ø;-)b /hi'nIidz jK:≤'hEıp/ he needs your help, (ÈZÈ's™;I, ¶È-÷ &π-, -¯-) /⁄zju'sEI/ as you sZ.
Let us also consider: (È'Z™T÷ π-) /⁄z'jEt/ as yet. (Making use of diaphonemes like /⁄,
K:/ may help in reducing the space of a phonemic transcription, chiefly in a dic-
tionary.)
Approximants
2.2.9.2. It is proved that the American /</ is a prevelar approximant, with a very
slight –and (almost) uninfluential– raising of the tip of the tongue towards the
postalveolar region. Instead, the British sound is decidedly postalveolar, (>), in the
specific meaning of an area after the alveolar one, approached by the tip of the
tongue (not by the lamina, as in the IPA o‚cial point of view). It is actually an
apico-postalveolar articulation.
It will be very important to observe the orograms of these two approximants
very carefully (û 1.13.3). Both of them are laterally contracted, just as real lateral
phones, but there is no contact with the roof of the mouth (as, instead, with real
laterals). ˛e absence of such a lateral contraction would simply deprive these ar-
2. english 71
ticulations of their typical timbre, which is so similar (in these two appoximant
phones), even though their actual articulations are relatively very di‡erent.
In addition, both (<) and (>) show a certain amount of lip rounding (more evi-
dent in stressed syllables), which –changing both towards a duller timbre– con-
tributes in making them less di‡erent auditorily, while remaining articulatorily ra-
ther di‡erent.
2.2.9.3. Once the exact articulations are clear, it is easy to understand why, for
/t<, d</, the British pronunciation regularly undergoes assimilation, giving (˛>, Ã>).
On the other hand, the fact that the auditory impression is so similar for these two
types of phones, may explain why, even in the American pronunciation, (˛>, Ã>)
can be used, besides the more usual ones, (T<, D<).
Certainly, it is very strange that the majority of phoneticians (even native ones)
keep on using the symbol (>) to hint at the American type, which is far from be-
ing postaveolar. By the way, the term po®alveolar corresponds to the o‚cial one
çretroflexÇ, which picturesquely tries hard to pass itself o‡ as a real point of articul-
ation, while, in fact, it is at most just a very peculiar articulatory modification.
But, as is well known, good kinesthetic, auditory (and even acoustic) skills are not
the same for all people…
Up to now, we have seen several examples of /</ (<)a (>)b, and several others will
follow. Let us remember only that our diaphonemic transcription rigorously dis-
tinguishes between /</, which is always pronounced in the two accents, and /≤/,
which is pronounced, as such, only in American English. As a matter of fact, in
British English, /≤/ corresponds to çzeroÇ, as r is pronounced only before vowels:
('<I;≥)a ('>¤;å)b /'<IÈ≤/ rear˚ ('<™;≥)a ('>™;‘)b /'<EÈ≤/ rare˚ ('<ø:<)a ('>ø:)b /'<O:≤/ roar.
In American pronunciation, /È≤/, preceded by vowels or consonants, is realized
as (≥). It is the same even for /˘</ (and, by and large, for /È</); /È:≤, È:<, È;</ are real-
ized as (≥:): ('m≥;D≥)a ('m‘;Då)b /'mÈ:≤dÈ≤/ murder˚ ('m≥;DÈ<≥, -D≥≥)a ('m‘;DÈ>å, -Ã>å)b
/'mÈ:≤dÈ<È≤, -d˘<È≤/ murderer˘ Also (È<, È:<)a, for (≥, ≥:)a are acceptable, even if less
frequent.
In a broad New Zealand accent, /</ is (˛), instead of (>), usual also in neutral
New Zealand English. As in American English (except in typical Southern, East-
ern, and Black accents) and Canadian English, also Irish English (in the whole is-
land) has /≤/ = /</. ˛e same goes both for an area in the South Island in New Zea-
land and for the West Country in the southwest of England (as well as for some
more limited areas in the North of England). A typical Scottish accent, usually, has
/<, ≤/ = (R).
As a speech defect, /</ is realized as a labiodental (V). ˛is is so widespread, espe-
cially in Great Britain, that someone considers it to be normal (all the more so be-
cause it is frequent in the mediatic British accent).
˛e other approximants
Laterals
di‡erence between (®, Í), and the actual fact that, before (a heterosyllabic) /j/, not
only does (ı) not occur (as, instead, it does before any other consonants)˚ but it is
prepalatal, (¬), ('m¤¬-jÈn). In other contexts, it is alveolar, (l), as in: ('lUusaÙT)a ('l¯u-,
'lj¯u-)b /'lUusaEt, 'lˆ-/ leucite˘ Even between a stressed short vowel and another vow-
el, which makes one syllable with /l/, in the neutral (American or British) pronun-
ciation, (l) (not (ı)) is used: ('fAl-i)a ('fØl-i)b /'fØli/ folI˚ ('fAl-ø¨)a ('fØl-‘¨)b /'fØlOU/
follow˚ ('Th™l-¤m) /'tElIm/ tell him˚ ('fIil¤T) /'fIilIt/ feel it, ('fO:l 'aÖT)a ('fø:l)b /'fO:l 'aOt/
fall out˚ ('b¤l '™vÈnΩ) /'bIl 'EvÈnz/ Bill Evans.
2.2.10.2. ˛eir distribution is in any case rather simple, once it has been clear-
ly explained. As we have said, in diaphonemic transcriptions, it is convenient to
use the two primary symbols, /l, ı/; although (to be precise) it would be useful to
use at least four symbols, (l, ¬, ], ı) (besides (¯) in British {and British-like} pronun-
ciation, before /<, t<, d</), in addition to three intense ones, (Í, –, ®). Furthermore,
at least in specific descriptions, the taxophones with dental contact, ((l, Ï)), should
be indicated, and also those with lip-rounding after V or C which have a labial
component): ((t, ¥, 3, ∞, ®÷ ¬, ÿ)). Especially the coarticulation of labialization is au-
tomatic; therefore, it is necessary to mark it only where it is being explained, in or-
der to draw attention to the phenomenon.
Before V˚ we regularly have /l/ (l): ('lI;iÑ) /'lIiv/ leave, ('lπ[;]sT)a ('lA;sT)b /'lπ;st/ la®,
('l¨k) /'lUk/ look, ('lø;Ùn) /'lOEn/ loin. To be rigorous, lip-rounding also occurs be-
fore rounded V, by coarticulation: (('t¨k, 'tø;Ùn)); however, a special symbol –like
((t))– is not needed, since it is absolutely inevitable to prepare the lips for the round-
ed vocoids that follow, within the syllable.
In fact, an articulation of /l/ without lip-rounding, ((l)), would somehow be per-
ceived as something çstrangeÇ, exactly as for /k, g/ followed either by front V or
by /j/, or else by rounded V or by /w/: the articulations ((´, Ò)) and ((k, g)), respec-
tively, are natural and automatic: (('´h¤T)) ('kh¤T) /'kIt/ kit, (('Ò™T)) ('g™T) /'gEt/ get,
(('´âj¯;uÊ)) ('khj¯;uÊ) /'kjUub/ cube, (('kWø:<))a ('khø:<) (-ø:)b /'kO:≤/ core, (('gUus))a
(('g¯us))b ('g-) /'gUus/ goose, (('kWaÙT)) ('kwaÙT) /'kwaEt/ quite.
˛erefore, in particular for /®ò/, an çobjectiveÇ pronunciation, obtained by juxta-
posing /p/ and /®/, for instance, would produce an e‡ect that may perplex native
speakers. Strictly speaking, in fact, (p®) would have something less in comparison
with the genuine ((p¬)), as in (('phIip¬)) ('phIip®) /'pIip®/ people.
2.2.10.4. We will now report, though not recommend, the frequent insertion
of a homorganic stop before /†, s, S/ (not /f/) preceded by /ı/: ('f¤ı†i÷ -ıt†i) ((-Ït†i))
/'fIı†i/ fil`y˚ ('fO;ıs÷ -ıts, 'fAı-)a ('fø;ıs÷ -ıts, 'fØı-)b ((-ÿts)) /'fO;ıs/ false˚ ('wEıS÷ -ıTS, -ıc)a
('w™ıS÷ -ıTS, -ıc)b /'wEıS/ Wel´˘ ˛is occurs with /z/, as well, but more rarely: ('bEı:Ω÷
-ı:dΩ)a ('b™ı:Ω÷ -ı:dΩ)b /'bEız/ bells˚ ('fi;IıΩ÷ -ıdΩ, 'fiI®-) /'fIiız/ feels. On the other hand,
the reverse simplification may be heard, too, as in: ('fi;IıdΩ÷ -ıΩ, 'fiI®-) ((-ÏΩ)) /'fIiıdz/
fields˚ ('bEıts, -ıs)a ('b™ıts, -ıs)b ((-Ïs)) /'bEıts/ bel_. Especially when there are possi-
ble ambiguities (and grammemes >-s˚ -'s≥), spelling is better respected.
In non-neutral American pronunciation, /l/ can be realized as (]) (sometimes
even (ı)) before V or /j/. ˛is can also occur in New Zealand, in northern Wales,
but most of all in Scotland. On the other hand, in Ireland, in South Africa, and
in southern Wales, usually, (l) may be heard, even before C or pauses.
2.2.11.1. ≈nally, there are some more or less isolated words, which are pro-
nounced di‡erently in the two neutral accents. Others often show both pronunci-
ations. In American pronunciation, word-initial syllables are commonly full, ¤
76 a handbook of pronunciation
not pronounced with /È/: /vEI'kEIS˘n/a /vÈ-/b vacation˚ /nOU'vEmbÈ≤/a /nÈ-/b novem-
ber˘ Even the article a often has a full timbre (which could sound too formal in
British pronunciation): /∑È≤zÈ'gUd '<Iizó, -zEI-/a /-zÈ-/b `ere's a good reason˘
Furthermore, bisyllabic verbs in -ate˚ more frequently in American English, bear
initial stress: /'dIktEIt, -'tEIt/a /-'tEIt/b dictate, /'vaEb<EIt/a /-'b<EIt/b vibrate. ˛e su‚x
-ile is /-˘ı÷ -aEı/a /-aEı/b: /'hØstAŒı/ ('hAsT®, -aÉı)a ('hØsTaÉı)b ho®ile (and mobile as
well), /'fÈ:≤ùAŒı/ ('f≥;[®÷ -aÉı)a ('f‘;TaÉı)b fertile˚ /'mIsAŒı/ ('m¤s®)a ('m¤saÉı)b mi˛ile
(and reptile˚ ®erile]˘ But we have: /'dØs®÷ -aEı/a /'dOUsaEı, 'dØsaEı/b docile and /'GUu-
v˘n®, -ÈnaEı/a /-ÈnaEı, -˘n®/b juvenile. We also find: /'p<Ø-/a /'p<OU-/b pro-: /'p<ØsEsI˙,
-È-/a /'p<OU-, 'p<Ø-/b proce˛i«, /'p<Øg<Es, -Ès/a /'p<OU-, 'p<Ø-/b progre˛ (but: /'p<OU-
g<πm/ programa, -mmeb).
To end with, let us see: /sIvÈl¢'zEIS˘n÷ -laE-/a /-laE-, -l¢-/b civilization, /'kπp¢lŒ<i/a
/kÈ'pIl˘<i/b capillary, /'kO:<ÈlŒ<i, 'kØ-/a /kÈ'<Øl˘<i/b corollary, /'lπb˘<ÈtK:<i/a /lÈ'bØ<È-
t˘<i/b laboratory, /'nOUb√di, -bØ-, -bÈ-/a ('nø¨&b√Di, -&bADi, -bÈDi)a /-Ø-, -È-/b ('n‘¨&bØ-
Di, -bÈDi)b nobody (and aJ-˚ every-˚ some-]˘ Let us also notice this (seemingly
çcuriousÇ) word: ç/'aE˘≤n/Ça ('aÙ≥n, 'a;Ù<n)a ç/'aEÈ≤n/Çb ('aÙÈn)b iron˘
2.2.11.2. Here is a collection of over 100 interesting cases, which we present ex-
ceptionally in alphabetic order, giving current spelling first. However, we must be
aware that, not infrequently, speakers of one of the two accents (a and b) use pro-
nunciations shown for the other accent. For single words, this is quite obvious:
˛ere is an additional short list of words that in American English may often
have also /O:/ (O:)a, besides the more usual /Ø/ (A)a (Ø)b: >ocolate˚ doll˚ dolI˚ dol-
#in˚ god˚ golf˚ gone˚ mo$˚ on˚ resolve˚ revolver˚ rolf˚ solvate˚ solve˚ ®omp˚ ~amp˘
Structures
2.3.0. In this section, we will deal with various macro-segmental aspects, includ-
ing intonation.
2.3.1.1. As can be seen from many of the previous transcriptions, English un-
stressed syllables almost always have /È, ¢, I/: (mÈ'nA[ÈnÈs)a (-'nØTÈ-)b /mÈ'nØùÈnÈs/
monotonous, (ph<È'v™n[[]Èfl)a (ph>¤'v™nT¤fl)b /p<¢'v™nù¢d/ prevented, ('™DÈm≥Ü)a (-¤-
T‘Ω)b /'Ed¢ùÈ≤z/ editors, ('m¤nÈsT≥)a (-¤sTå)b /'mIn¢stÈ≤/ mini®er˘ On the contrary, in
foreign accents of English, unstressed syllables too often show full vowels.
Indeed, many vowels (and consonants) may disappear entirely in comparison
with spelling, as in: ('kh√Mf≥[Èb®, -fT≥-, -f≥-÷ -mT≥-)a (-åMfÈTÈ-, -fTÈ-, -fÈ-÷ -mTÈ-)b /'k√n-
fÈ≤ùÈb®, -mf-/ comfortable, ('mA;<g<ÈT, -≥ÈT)a ('mA;g>ÈT, -ˇÈT)b /'mA:≤g˘<Èt/ Margaret,
(&j¯unÈ'v≥;ßÈ[i, -ßTi)a (-'v‘;sÈTi, -sTi)b /jUun¢'vÈ:≤s˘ùi/ universi“˘
However, not all unstressed syllables have vowel reduction or fall: ('khAm™nT)a
('khØ-)b /'kØmEnt/ comment, ('πsfO;ıT, -A;ıT)a /'πsfO:ıt, -A:ıt/ (-ÅıT, -ø;ıT)b /'πsfπıt,
-O:ıt/ as#alt, ('khUupAn÷ 'khj-)a ('kh¯upØn)b /'kUupØn/ coupon˘
Only regular consultation of a pronunciation dictionary (or, better, dictionar-
ies) can give the exact structure of words and sentences, in English as in any oth-
er language.
In English sentences, respect of the reduction, or weakening, of many gram-
memes (or functional words) is vital: articles, prepositions, conjunctions, auxil-
iary and modal verbs, some pronouns and some other forms. ˛ere are not many
reduced forms (using a clearer term than the traditional one çweak formsÇ) – about
eighty – but they are the most frequent ones. ˛ey are listed below (with exam-
ples), in alphabetical order, for easy consultation.
2.3.1.2. ˛e examples given illustrate various elements simultaneously. It is
worthwhile to observe them very carefully and to consider all the variants given
(but only in phonetic transcription, for the sake of space, in a very economical
way, while full transcriptions would have been more monotonous, and would not
be able to show the same things with precision):
[`\'d]˚ (¤[Èbb¤n'D√n:, -Db-, -™n-÷ -Èn-÷ ¤ThÈ-)a (¤TÈ-, -¤n-, -Iin-, -ån:)b it had been
done [it'd]˚ (∑È'mπ;n ûg'gO:n, ÈD-, -A;n)a (-Ø;n)b `e man had gone˚ (Si'hπfl TÈ-
'sT™;I, -πT) /Si'hπd tÈ'stEI/ ´e had to ®Z (+ (-π[ È-)a (-πT È-)b /-πù È-/);
has\ (hÈzi'gO:n, hπ-, -A;n)a (-Ø;n)b has he gone?˚ (Siz'D√n:, SiÈz-÷ SihÈz-)a (-ån:)b ´e
has done [´e's]˚ (¤ts'b¤n:, ¤[Èz-÷ ¤thÈz-)a (¤ts-, -I;in, -¤n:, ¤TÈz-)b it has been [it's]˚
('Gø:<G ÈΩ'kh√m:, Ès-÷ -ä hÈ-)a ('Gø:G, -åm:)b George has come˚ (hi'hπΩ TÈ-
'sT™;I, -πs) he has to ®Z÷
have\ (hÈvjÈ'sIin¤T, hπ-, -j¯-) have you seen it?˚ (∑™Iv'gO:n, -A;n÷ ∑™IÈ-÷ -hÈ-)a
(-Ø;n)b `\ have gone [`\'ve]˚ (wiÑ'kh√m:, -f'k-÷ wiÈ-÷ wihÈ-)a (-åm:)b we have
come [we've]˚ (wi'hπÑ TÈ'sT™;I, -πf) /wi'hπv tÈ'stEI/ we have to ®Z˚ (jÈ'S¨DÈÑ,
j¯-, -πÑ, -&hπÑ) you ´ould have˚ (jÈ'S¨DÈ 'D√n:, -*, -Èv, j¯-)a (-ån:)b you ´ould
have done [´ould 've] + (-'kh¨-, -'w¨-, -'m√sT-a/-å-b, -'m™I-, -'maÙ[-a/-T-b) could˚
would˚ mu®˚ mZ˚ might÷
he\ (hi'w™nT) he went˚ ('w™ni 'sO;¤T, w™-, wû-)a ('sø;¤T)b when he sY it˚ ('hπzi) has
he?÷
her\ (h≥'a;ÙΩ)a (hÈ>'-)b her \es˚ ('Th™l≥)a (-å)b tell her˚ ('g¤v≥ ≥'hπT)a (-È>‘-, -È hÈ-)b
give her her hat˚ (Th¯≥'m√∑≥, ThÈh≥-, Thw-)a (ThÈhÈ'må∑å, Th¯‘-, Th¯È-, Thw-)b
to her mo`er÷
him\ (aÙ'sO;¤m÷ -Èm)a (-ø;¤m)b I sY him˚ ('l™[¤m '¤n:÷ -Èm÷ -õ)a ('l™T-)b let him in÷
his\ (h¤Ω'ph™n:) his pen˚ (hi'Th¨k ¤z'b¨k) he took his book˚ (h¤Z'j¯u†) his you`÷
I\ (aÙ'sI;i, √Ù-)a (aÙ-, åÙ-)b I see˚ (aÉı'Th™Ik, ,ı-÷ -®-÷ -ûı-÷ -w®-)a (aÉı-, √ı-÷ -®-÷ -ûı-÷
-w®-)b I will take [I'll]˚ ('ha;Ö kûDaÙ'sT™;I, -√-÷ -È-)a (-aÙ-, -å-÷ -È-)b how could I
®Z?÷
if\ (¤faÙ'm™;I, Èf-, ≥-) if I mZ˚ (¤fjÈ's™;I&sø¨, Èf-, ≥-, -¯-)a (-‘¨)b if you sZ so˚ (¤f'nAö
f≥'j¯;u, Èf-, ≥-, -T)a (-nØö fÈ'-)b if not for you÷
in\ (¤n'l√nDÈn)a (-å-)b in London˚ (¤m'phπ<Ès)a (->¤s)b in Paris˚ (¤˙'khπnÈD√)a
(-å)b in Canada˚ (hi'sπ[ ¤n∑È'ch™;≥, ¤˙˙È-, ¤nnÈ÷ -T ó∑È-÷ -T ónÈ-)a (-πT, -™;‘)b he
sat in `e >air˚ ('b<™Ik¤[ ¤n'ThU;u÷ -¤T ó-)a ('b>-, -¤T, -¯;u)b break it in ‘o˚ (aÙ-
m¤nÈ'h≥;i, √-÷ -mnÈ-÷ aÙÈm-)a (-å>i, å-)b I am in a hurry [I'm]÷
is\ (¤z¤ö'j¯;u, -T'j-÷ Ǥ-) is it you?˚ (¤ts'mI;i÷ ¤[¤z-)a (¤ts-÷ ¤T¤z-)b it is me [it's]˚ ('kh¤mΩ
'hI;≥÷ -m ¤-)a ('h¤;å)b Kim is here [Kim's]˚ ('∑¤s ¤zÈ'fa;Ùn 'D™;I, '∑¤s ÇÈ'-÷ '∑¤s sÈ'-) `is
is a fine dZ, ('<ø;¨z ¤zÈ'naÙs 'w¨mÈn, '<ø;¨z ÇÈ'-, '<ø;¨z zÈ'-)a ('>‘;¨z)b Rose is a
nice woman, (∑¤s'D¤S ¤z'waÙT, Ç'-, s'-, ∑Ès-) `is di´ is white÷
it\ (¤z¤ö'Th<U;u, Èz-, Ç, -T'T-)a (-'˛h>¯;u, -˛'˛-)b is it true?˚ (¤ts'O:ı '<aÙT, ts'-, s'-÷ ¤[¤z-)a
(-ø:ı '>-÷ ¤T¤z-)b it is all right [it's]˚ ('j™s ¤['¤;Ω÷ j™'sT¤;Ω)a (¤T'¤;Ω÷ j™'sT¤;Ω)b yes it is˚
('Th™Ik¤T, -ûT)a (-¤T)b take it˚ (¤[®biû'g¨;∂ '†¤˙:, T®-÷ ¤pw®-÷ ¤Tw®-)a (¤T®-)b it will
be a good `i« [it'll]˚ (¤[Èbbi'naÙs, Twû-, -Db-÷ ¤pw-÷ ¤Tw-)a (¤TÈ-, …)b it would
be nice [it'd]÷
i_\ (¤ts'Th™;¤ı) i_ tail˚ (¤T'hπ;D ¤ts'ThIi† 'b<ø¨kûn, Èts-)a ('b>‘¨-)b it had i_ tee`
broken÷
ju®\ (aÙvG√s[T]'sIin¤m, -GÈs-) (-Gå-, -GÈ-)b I have ju® seen him [I've]˚ (Siz'G√s-
gO:n û'w™;I, -Ès-, -A;n)a (-å-, -È-, -Ø;n)b ´e has ju® gone YZ [´e's]˚ (G√s'laÙk
¤Ω's√n:, GÈs-, -T'l-)a (Gås-, -ån:)b ju® like his son÷
maJ\ ('m™ni 'bø;ÙΩ) maJ boys˚ ('ha;Ö m™ni'mø:<, mÈni-, mni-)a (-ø:)b how
maJ more?÷
2. english 81
mZ\ (jÈm™I'π[;]sk≥, j¯-÷ -mi-)a (-A;skå)b you mZ ask her˚ (wim™I'gø;¨÷ -mû-)a
(-‘;¨)b we mZ go˚ (Sim™I'sT™;I÷ -mÈ-) ´e mZ ®Z÷
me\ (hi'khO;ıbmi, -Dmi)a (-ø;ı-)b he called me˚ ('Sø¨mi ∑û'w™;I)a ('S‘¨mi)b ´ow
me `e wZ÷
mu®\ (aÙmÈs'gø;¨, -m#-, √Ù-)a (åÙ-, -‘;¨)b I mu® go˚ (¤pmÈsbi'D√n:, ¤ö-)a (-ån:)b it
mu® be done˚ (SimÈs'ph™;I) ´e mu® pZ÷
my\ (maÙ'h™;fl, m√Ù-, m√-)a (måÙ-, må-)b my head˚ ('nAT TÈmaÙ'nAl¤ä, -√Ù-, -√-,
-È-÷ -¤-÷ -i-, -ö)a ('nØ-, -åÙ-, -å-)b not to my knowledge÷
no\ (nø¨'mø:< D¯'a;Ù÷ nÈ-)a (n‘¨'mø:÷ nÈ-)b no more do I˚ (nø¨'√∑≥ '<Iizó÷ n¯-)a
(n‘¨'å∑È '>-÷ n¯-)b no o`er reason˚ (¤tsnø¨'g¨;fl÷ -nÈ-)a (-‘¨-÷ -nÈ-)b it's no good÷
nor\ (&nIi∑≥'f¤S nø;<'fl™S, n≥-÷ &naÙ-)a (&naÙ∑È-, nø;-, nÈ-÷ &nIi-)b nei`er fi´ nor fle´˚
(&nIi∑≥'hI;i n≥'a;Ù, nO;<-÷ &naÙ-)a (&naÙ∑È-, nÈ>-, nø;>-÷ &nIi-)b nei`er he nor I÷
not, n't\ (¤['¤zóT)a (¤T'¤-)b it isn't˚ (aÙ'wø¨nT, √-)a (å-, -‘¨nT)b I won't˚ (hi'D√zóT,
-'DÈ-)a (-å-, -'DÈ-)b he doesn't˚ (wi'Dø¨m 'phl™;I)a (-‘¨-)b we don't plZ˚ (¤['¤zô
'g¨;fl)a (¤T'-)b it isn't good˚ ('¤zói, -óTi) isn't he?˚ ('¤zó¤ö, -ó¤T) isn't it?˚ (hi'j¯usó-
&T¯) he used not to÷
of\ (∑È'f¤f† Èv'm™;I, *'m-) `e fif` of MZ˚ (û'kh√p ÈÑ'ThI;i, -È'T-, -≥'T-)a (-åp)b a cup
of tea˚ ('f≥;sT Èv'O:ı, *'-)a ('f‘;-, -ø:ı)b fir® of all÷
on (the most reduced form only occurs when no ambiguity with in is possible):
(&√pO;n∑È'bAks, -An-÷ -Èn-, -nnÈ-)a (&åpØ-, -Øks)b up on `e box˚ (&¤tsO;mmaÙ-
'sa;Ùfl, -Am-, -m√Ù-÷ -mÈ-÷ ¤[¤z-)a (-Øm-, -måÙ-÷ ¤T¤z-)b it is on my side [it's]˚ ('w√[-
Èn '≥;†, 'wA-, -O;n, -An)a ('wØTÈn '‘;†, -Øn)b what on ear`!÷
once\ (w√ns'mø:<, wûn-)a (wåns'mø:, wûn-)b once more (= again] – (&w√ns-
'mø:<)a (&wåns'mø:)b once more (= one more time]÷
one(s) (the form without /w/ may be judged as dialectal or regional): (È'b¤g&w√n,
-wûn÷ -ûn)a (-wå-)b a big one˚ (&∑πtsû'g¨;b&w√n, -wûn, -D-÷ -DÈn÷ -Dó÷ &∑π[¤-
zÈ-)a (-wå-÷ &∑πT¤zÈ-)b `at is a good one [`at's]˚ (w√n'O;ıwûΩ 'hø¨ps, wû-, -¤Ω,
-™IΩ, -O;w-)a (wån'ø;ıw¤Ω 'h‘¨ps, -ûΩ, ™IΩ-, -ø;w-)b one alwZs hopes˚ (&∑ø¨z≥-
'naÙs&w√nΩ, -wûnΩ÷ -sÈnΩ)a (&∑‘¨zÈ-, -wå-÷ -sÈnΩ)b `ose are nice ones÷
or\ (ThÈ'D™;I ≥TÈ'mO;<ø¨, ø;<T-÷ -mA<-)a (ÈT-, ø;T-, -Ø>‘¨)b todZ or tomorrow˚ ('ThU;u
≥'†<I;i 'pha;ÖndΩ÷ ø;<-)a ('Th¯;u ø;-, È'†-, ->-)b ‘o or `ree pounds˚ (≥'Eıs)a (È>-
'™ıs)b or else÷
our\ (A;<'sku;Uı)a (A;-)b our (ool˚ (&∑Iiz≥A;<'ph™n:Ω)a (-zÈ>A;-)b `ese are our pens÷
per\ ('f¤fTi p≥'s™nT)a (pÈ-)b fif“ per cent˚ ('fa;ÙÑ p≥'s™mp p≥'πnÈm, -mö, -nT)a (pÈ's-,
pÈ>'π-)b five per cent per annum÷
´all (in American English it is a stylistic choice): (&SπlaÙ'Th™Ik¤T, -√Ù-)a (SÈlaÙ-
'Th™Ik¤T, SÍaÙ-, -åÙ-, -å-)b ´all I take it?˚ (aÙS®'DU;u, aÉı-, ,ı-)a (-¯;u, √ı-)b I ´all
do˚ (S®wi'gø;¨, Sûwi-, Swi-)a (-‘;¨)b ´all we go?÷
´e\ (Si'w™nT) ´e went˚ (Si'hπΩT¯, -sT¯) ´e has to˚ ('hπzóSi, -óTSi) hasn't ´e?÷
´ould\ (SÈDi'kh√m:)a (-åm:)b ´ould he come˚ (jÈSÈD'DU;u¤T, j¯-)a (-'D¯u¤T)b you
´ould do it˚ (aÙSûg'gø;¨&naÖ, -Sg-, -ÈD'g-, √-)a (-‘;¨-, å-)b I ´ould go now˚ (aÙ-
SÈ∂'†¤˙k&sø¨, -St'-, √-)a (å-, -‘¨)b I ´ould `ink so˚ (∑È[iSûâ'kh√m 'aÖT, -Èfl'-, -iS-
'kh-)a (-Ti-, -å-)b `at he ´ould come out÷
sir\ ('j™ss≥, -s-)a (-å)b yes, sir˚ ('nø¨s≥)a ('n‘¨så)b no, sir˚ (s≥'GA;n)a (sÈ'GØ;n)b Sir
John˚ (s≥'πıf<Èfl)a (sÈ>'Åıf>¤fl)b Sir Alfred˚ (s≥'chA:<ıΩ, -A;<®Ω)a (sÈ'chA:ıΩ, sÈ-,
82 a handbook of pronunciation
-åÙm, -åm, -Èm)b `e fir® time I went `ere˚ ('†<I;i&ThaÙmΩ 'fø:< ≥'ThwEı:Ñ, -T√Ù-,
-T√-, -TÈ-)a ('†>-, 'fø:> È-, -åÙ-, -å-, -È-, -™ı:Ñ)b `ree times four are ‘elve÷
to\ (ThÈ'l√nDÈn)a (-å-)b to London˚ (ThÈ'sku;Uı) to (ool˚ (Th¯'¤˙glÈnfl, 'Thw¤-) to
E«land˚ (ThÈ'j¯;u, Th¤'j-, Thi'j-) to you˚ (Th¯'π;n) to Ann˚ ('kh√mT¯¤T, -Tw¤T)a
('khåm-)b come to it˚ (jÈ'hπÑT¯, -fT¯, j¯-) you have to˚ (Thû'g¤;Ñ) to give˚ (Thû-
'w¤n:, Th¯-) to win˚ (Th¯'IiT÷ ThÈ-)a (÷ ThÈ'ö-)b to eat˚ (Th¯'O;f≥, -'A-÷ ThÈ-)a (÷ ThÈ-
'öØfå)b to o‡er – for to˚ before consonants˚ (È, û) can be very short; and, be-
fore voiceless consonants, they are often devoiced: ('Th™n TÈ'fa;ÙÑ) ten to five;
up\ ('m™Ik√p j≥'ma;Ùnfl, jU≥-÷ -ûp-)a (-åpjÈ-, -jø;-÷ -ûp-)b make up your mind˚ (∑≥z-
'w√n √p'∑™;≥÷ Èp-÷ ∑≥¤z-)a (∑Èz'wån åp'∑™;‘÷ Èp-÷ ∑È>¤z-)b `ere is one up `ere
[`ere's]÷
upon\ ('w√n È&pO;nÈ'n√∑≥, È&pA-, È&p√-, ÈpÈ-)a ('wån È&pØnÈ'nå∑å, ÈpÈ-)b one upon
ano`er˚ ('la;Ùn ÈpO;n'la;Ùn, -pA-, -p√-, -pÈ-)a (-pØ-, -pÈ-)b line upon line÷
us\ ('Th™lÈs) tell us˚ (l™ts'gø;¨, l™ös-, l™s-)a (-‘;¨)b let's go! – but: ('l™[Ès 'gø;¨)a ('l™T-
Ès 'g‘;¨)b let us go÷
was\ (aÙwûz'<O:˙, -A;˙)a (-'>Ø;˙)b I was wro«˚ (hiwÇÈ'f<™n:fl)a (->-)b he was a
friend÷
we\ (wi'm™;I) we mZ˚ ('A;<nTwi, -mpwi, -möwi)a ('A;n-, 'A;m-)b aren't we?÷
were\ (∑™Iw≥'¤ı:)a (-wû>-)b `\ were ill˚ (w≥jÈ'∑™;≥, -j¯-)a (wû-, -™;‘)b were you
`ere?÷
what\ ('sI;i w√TjÈv'D√n:, wA-, wû-, -TS-, -c-, -¯v-÷ hw-÷ W-÷ -j¯È-÷ j¯hÈ-)a (wØ-, wå-,
…)b see what you have done! [you've]˚ (hi'nU;u w√[i'wAn[[]Èfl, wA-, wû-, -'wO;n-÷
hw-÷ W-)a (-'nj¯;u wØTi'wØnT¤fl, wå-, …)b he kn[ what he wanted˚ (&w√m≥jÈ-
's™I¤˙, -j¯, &wA-, wû-÷ hw-÷ W-÷ -™;¤˙)a (&wØTÈ-, …)b what are you sZi«?
[what're]˚ (&w√TDjÈ 's™;I, -TGÈ, -TSÈ, -cÈ, -GÈ, -¯, &wA-, wû-÷ hw-÷ W-)a (&wØ-, …)b
what do you sZ? [d'you]˚ (&w√TDjÈ'DU;u, -TGÈ, -TSÈ, -cÈ, -GÈ, &w√DÈjÈ-, &w√DjÈ,
-¯, &wA-, wû-÷ hw-÷ W-)a (&wØ-, …)b what do you do? [d'you]÷
when (not interr.): (Èmw™naÙ'sO;¤T, õwû-, -nÈ'-)a (-ø;-)b and when I sY it…˚
(&sø¨w™~jû'g™ö '∑™;≥, -wû-, -T-)a (&s‘¨-, -™;‘)b so when you get `ere…÷
where (not interr.): (∑È'phl™Is w™≥iwûΩ'fa;Önfl, w≥-)a (w™‘>i-, wû>i-)b `e place
where he was found˚ (û'kh√nT<i w™≥'phIip® 's¤˙:, w≥-)a (-åN˛>i w™‘-, wû-)b a
country where people si«÷
who\ (∑È'mπ;n ¯'D¤D¤T, h¯-) `e man who did it÷
will\ ('∑π[® 'DU;u÷ '∑πpw®÷ -Tw®)a (-T® 'D¯;u, …)b `at will do [`at'll]˚ (jûı'sI;i,
j¨ı-, juı-÷ j¯ûı÷ j¯w®) you will see [you'll]˚ (w¤l¤p'w≥;k, -ö'w-, -T'w-÷ wûl-)a
(-‘;k)b will it work?˚ (¤[®bi'mI;i÷ ¤pw®-÷ ¤Tw®-)a (¤T®-)b it will be me [it'll]˚
('GA;m wûıbi'hI;≥, -n ®bi-)a ('GØ;-, -¤;å)b John will be here [ John'll]˚ (∑È'ch≥;c
[w]ûıbi'f¨ı:, w®-)a (-‘;c)b `e >ur> will be full÷
would\ (wûD¤pbi'g¨;fl, -öb-, -Tb-) would it be good?˚ (∑™ID'DU;u÷ ∑™IÈD-÷ -wûD-)a
(-¯;u)b `\ would do [`\'d]˚ (hiâ'kh√m:, -fl'k-÷ hiûâ÷ hiÈfl-÷ hiwû-)a (-åm:)b
he would come [he'd]˚ (¤[Èbbi'naÙs, -Db-÷ ¤pwû-÷ ¤Twû-)a (¤TÈ-)b it would be nice
[it'd], ('GI;im wûd's™;I&sø¨, -n Èd-)a (-‘¨)b Jean would sZ so÷
you\ (¤fjÈ'DU;u, -j¯-)a (-¯;u)b if you do˚ (aÉı'Th™¬j√, -j¯, ,ı-÷ aÉûı-÷ aÙwûı-)a …(-j¯,
√ı-)b I will tell you [I'll]˚ ('†π˙kj¯, -j√)a (-j¯)b `ank you˚ ('A;<~c¯, -√)a ('A;~-
c¯)b aren't you?˚ ('D¤DócÈ 'sI;i, -c¯, -óTS-) didn't you see?˚ ('D¤Dócû 'khw¤T, -¯,
84 a handbook of pronunciation
2.3.1.4. Here we will make some examples of compounds with reduced second
elements, especially in British English: ('sT<O:&b™<i)a ('s˛>ø;bˇi, -b>i)b ®rYberry
/-bŒ<i/ (in particular with monosyllabic roots, Ô § 2.3.5), ('kh√b≥fl)a ('khåbÈfl)b cup-
board, ('wEıkûm)a ('w™ı-)b welcome, ('s√nD™I, -Di)a ('sån-)b sunday (often /-dEI/ in
an intoneme, but /-di/ in a preintoneme, ç/-dŒE/Ç);
('phIim≥&b≥ø¨)a (-TÈbˇå, -b>å)b Peterborough /-b˘<KÈ/, ('™DÈm&b≥√)a (-¤mb>å, -bˇå)b
Edinburgh /-b˘<È/, ('nU;u&b™<i)a ('nj¯ubˇi, -b>i)b N[bury /-bŒ<i/ (in particular
with monosyllabic roots, Ô § 2.3.5), ('l™sT≥)a ('l™sTå)b Leicester, ('w¨sT≥)a ('w¨sTå)b
Worcester, ('nø;<fûk)a ('nø;fûk)b Norfolk, ('Aksf≥fl)a ('ØksfÈfl)b Oxford;
('khπsÈm) Casham, ('D≥;Èm)a ('Då>Èm)b Durham /È;</ and ('kh√n¤˙&hπm)a ('khån-
¤˙ûm)b Cunni«ham /-·⁄m/, ('¤˙glÈnD) E«land, ('phø¨smÈn)a ('ph‘¨s-)b po®-
man, ('G™n[[]®mÈn)a (-nT®-)b gentlemen, ('s™<È&mø¨ni)a ('s™>ÈmÈni)b ceremony /-mKÈ-
ni/, ('phl¤mȆ) PImou$;
('sO;s&phπn)a ('sø;spÈn)b saucepan /-p⁄n/, ('nAns™ns, -sÈns)a ('nØnsÈns)b nonsense
/-sŒns/, ('jø;<kSI≥, -S≥)a ('jø;kSå, -S¤å)b Yorkshire /-SEÈ≤/, ('hπnsÈm) handsome, ('h™nD≥-
sÈn)a (-DÈ-)b Henderson, ('fø¨ksTÈn, -&sTø¨n)a ('f‘¨ksTÈn)b Folkestone /-stKÈn/.
Furthermore: ('laÙ&b<™<i, -b<È<i, -b≥i, -b<i)a ('laÙb>È>i, -b>ˇi, bˇi, -b>i)b library˚
('kh√sTÈ&m™<i)a ('khåsTÈmÈ>i, -Èmˇi, -Èm>i)b cu®omary˚ (DÈ'<™kTÈ<i, -T≥i, -T<i, DaÙ-)a
(D¤'>™kTÈ>i, -˛ˇi, -˛>i, DaÙ-)b directory˚ ('Dø;<mÈ&Tø<i)a ('Dø;mÈ˛>i, -TÈ>i)b dormitory.
Taxophonics
2.3.2.1. From the examples given thus far, the use of #onetic duration for the
various English phones will be su‚ciently clear. However, we will summarize its
2. english 85
characteristics. In stressed syllables, the long vowels (/A:, O:, È:/, and the possible
long ones from the diaphonemes /A;, π;, Ø;, O;, È;/) as well as the diphthongs (/Ii,
EI, aE, OE, aO, OU, Uu/) undergo a little shortening –half-´orteni«, indeed: from
(é:, é;é) to (é;, éé)– when they are followed, within the same word or rhythm
group, by at least one of the following three elements: (1) a voicele˛ consonant, or
(2) an un®re˛ed vowel (and that changes them into diphthongs or triphthongs,
(éé, ééé)), or (3) a whole un®re˛ed Ällable. ˛e second elements of compounds
have secondary-stressed syllables; so they have no influence on length.
˛erefore, we have: ('phl™;I) /'plEI/ plZ, ('phl™;IΩ) /'plEIz/ plZs, ('phl™;Ifl) /'plEId/
plZed, ('phl™;I&bπk) /'plEIbπk/ plZba$, but ('phl™I¤˙) (or, possibly, ('phl™;¤˙))
/'plEII˙/ plZi«, ('phl™I≥)a (-å)b /'plEIÈ≤/ plZer, ('phl™IÈb®) /'plEIÈb®/ plZable, and
('phl™IT) /'plEIt/ plate, ('phl™Its) /'plEIts/ plates, ('phl™I[Èfl)a (-T¤fl)b /'plEIù¢d/ plated,
('phl™I[¤˙)a (-T¤˙)b /'plEIùI˙/ plati«, ('ph™InT) /'pEInt/ paint˘
Equally: ('khA:<)a (-A:)b /'kA:≤/ car, ('khA:<Ü)a (-A:Ω)b /'kA:≤z/ cars, ('khA:<fl)a (-A:fl)b
/'kA:≤d/ card, ('khA:<D&bø;<fl, -[b]&b-)a ('khA:D&bø;fl, -b&b-)b /'kA:≤dbO:≤d/ cardboard, but:
('khA;<T)a (-A;T)b /'kA:≤t/ cart, ('khA;<bÈn)a (-A;b-)b /'kA:≤bÈn/ carbon, ('khA;<D¤˙)a
(-A;D-)b /'kA:≤dI˙/ cardi«, ('khπ[;]nT)a (-A;nT)b /'kπ;nt/ can't˘
Besides, also unstressed or half-stressed syllables shorten, as seen in cardboard (a
true compound, as to cupboard /'k√bÈ≤d/, that is crystallized, by now, so that a
more suitable spelling for the latter could certainly be ècubbard¶]\ (phA;<'Th¤sÈ-
pÈnT)a (also (ph≥-)a) (-A;'Th¤s¤-)b /pA:≤'tIs¢pÈnt/ participant˚ (phA;<'Th¤sÈ&p™IT)a (also
(ph≥-)a) (-A;'Th¤s¤-)b /pA:≤'tIs¢pEIt/ participate˚ ('m™;I&D™I) /'mEIdEI/ MZdZ˘ Besides:
('w™Isp™Ip≥&bπ[;]skûT)a (-È&bA;sk¤T)b /'wEIstpEIpÈ≤-bπ;sk¢t/ wa®epaper basket˘
˛ese degrees of length hold good both in intonemes and preintonemes.
2.3.2.2. For the English stressed vowels (/I, E, π, √, Ø, U/), there is another inter-
esting fact about phonetic length. In syllables checked by final voiced consonants,
occurring in diphonic pairs (¤ /b, d, g, G÷ v, ∑, z, Z/), short stressed vowels under-
go a little lengthening – half-le«`eni«, indeed: from (é) to (é;)– ('l¤;fl) /'lId/ lid
(but ('l¤T) /'lIt/ lit], ('mπ;fl) /'mπd/ mad (but ('mπT) /'mπt/ mat]˚ ('b√;Ω)a ('bå;Ω)b
/'b√z/ buzz (but ('b√s)a ('bås)b /'b√s/ bus]˘ ˛ese degrees of length hold good even
in both intonemes and preintonemes.
On the other hand, when final stressed syllables are checked by an isolated
voiced consonant (¤ not forming a diphonic pair – that is /m, n, ˙÷ ı/), instead of
the vocoid, the contoid is a little lengthened (but only in intonemes, before paus-
es): ('Th™n:) /'tEn/ ten, ('j√˙:)a ('jå˙:)b /'j√˙/ you«, ('b¨ı:) /'bUı/ bull (but ('b¨ı&hø;<n)a
(-ø;n)b /'bUıhO:≤n/ bullhorn˚ (∑È'b¨ı '<π;n)a ('>-)b /∑È'bUı '<πn/ `e bull ran).
In both accents, though, there is an exception to the exception, insofar as /π, Ø/
are half-lengthened (in a preintoneme, too), instead of a following contoid:
('GA;n)a ('GØ;n)b /'GØn/ John, ('bπ;˙) /'bπ˙/ ba«, ('hπ;nfl) /'hπnd/ hand˘
In the sequences /IÈ≤, EÈ≤, UÈ≤/, the first element is half-lengthened, both in into-
nemes and preintonemes; this occurs before vowels as well, if final in rhythm groups:
('hI;≥)a ('h¤;å)b /'hIÈ≤/ here, ('∑™;≥)a ('∑™;‘)b /'∑EÈ≤/ `ere, ('phjU;≥, 'phj≥:)a ('phj¨;å,
'phjø:)b /'pjUÈ≤/ pure˚ ('hI;≥ Èn'∑™;≥)a ('h¤;È> Èn'∑™;‘)b /'hI˘< Èn[d]'∑EÈ≤/ here and `ere˘
We should notice that, in American English, /IÈ≤, EÈ≤, UÈ≤/ followed by vowels,
86 a handbook of pronunciation
within words or rhythm groups, become ç/I<, E<, U</Ç: ('sp¤<¤T) both for /'spI˘<It/
('sp¤È>¤T)b spear it and for /'spI<It/ ('sp¤>¤T)b spirit˚ ('h¤<¤˙)a ('h¤È>¤˙)b /'hI˘<I˙/ hear-
i«˘
2.3.2.3. Even as far as (partial) devoici« is concerned, the examples thus far
will have already been a clear general survey. A short summary is, however, useful,
because there are also some particular remarks to be made, only here, even if we
need not mark them all in our transcriptions.
˛e devoicing of voiced diphonic consonants (/b, d, g÷ G÷ v, ∑, z, Z/), before
pauses or before voiceless consonants (çpo®devoicingÇ), is very important: ('bA;Ê)a
('bØ;Ê)b /'bØb/ Bob˚ ('Dπ;fl) /'dπd/ dad˚ ('G√;ä)a ('Gå;ä)b /'G√G/ judge˚ ('v≥:Ñ)a
('v‘:Ñ)b /'vÈ:≤v/ verve÷ ('bA;Ê 's™;Ω)a ('bØ;-)b /'bØb 'sEz/ Bob sZs˚ ('Dπ;fl 'Th¨k¤T) /'dπd
'tUkIt/ dad took it.
Of less importance is their devoicing after pauses or after voiceless consonants
(çpredevoicingÇ), which is slighter, too. It is true that for some speakers it is as
strong as postdevoicing, but it is usually less evident, and we need not mark it in
our transcriptions (although they could be shown by means of a dot under a sym-
bol: (b, a) (or above: (g, Z)): ('bA;Ê)a ('bØ;Ê)b /'bØb/ Bob, Â.
A dot could be used even after çaspiratedÇ /p, t, k, c/, but we will do that only
here, because (h) is su‚cient: (('ph&™;I)) /'plEI/ plZ˚ (('khwaÙT)) /'kwaEt/ quite˘ It is
the same also for the other voiceless consonants (although their devoicing is only
slight, and therefore usually it need not be marked): (('fˆ¯;u)) /'fjUu/ f[˚ (('†≤™T))a
(('†>-))b /'†<Et/ `reat˚ (('s2ø:<))a (('s2ø:))b /'snO:≤/ snore˘
Everyday-speech simplifications
And there are combinations, too: ('l¤T<Èli, -[È<È-)a ('l¤˛>Èli, -TÈ>È-)b /'lIù˘<Èli/ lite-
ralI˚ (ph≥'Th¤kjÈl≥li, -kj≥li, -kli)a (phÈ-, -kjÈlÈ-÷ -kj¨lÈ-)b /pÈ≤'tIkjÈlÈ≤li/ particularI˚
('ph<Abli, -bb-, -bÈb-, &ph<Ali)a ('ph>Ø-)b /'p<ØbÈbli/ probabI˘
2.3.3.3. Some other cases: (wûı, w®, wû {,ı'Th™¬jû&w√T, -ö})a ({√ı-, -Ø-})b /wEı,
wÈı, w®/ well, [I'll tell you what]˚ (≥&ø¨nli√'kh¨;fl, -ni-)a (-‘¨-, -å'-)b /IfOUnliaE'kUd/ if
onI I could˚ ('skj¯uzmi, ¤k'-) /Ik'skjUuzmi/ excuse me!˚ (baÙ'ba;Ù, b√-, bÈ-, &baÙ'ba;Ù)a
(bå-)b /baE'baE/ bye-bye˚ (g¨b'ba;Ù, gûb-, g¨'-, gû'-÷ -D'b-, 'ba;Ù) /gUd'baE/ goodbye˚
(g¨b'mø;<n¤˙, gûb-, g¨'-, gû'-÷ -D'm-, 'mø;<n¤˙)a (-ø;n-, 'mø;n-)b /gUd'mO:≤nI˙/ good
morni«˘
Besides: ('j™s, 'j™hs, 'j™h, 'j™Ès, 'j™È, 'j™ö, —j™, —jÈs), and ('j™√, 'jπ√, 'ja:, 'jA:)a (-å)b,
('j™û, 'j™pæ, 'j√pæ, 'j√öæ, —j√)a (-åpæ, -åöæ, -å)b /'jEs÷ 'jEÈ÷ 'jEp÷ 'j√p/ yes!˚ ('nø;¨, 'nø¨ö,
'nø¨p)a ('n‘-)b /'nOU/ no!˚ (&O;ı'<aÙT, O;'-, -ö)a (&ø;ı-, ø;-)b /O:ı'<aEt/ all right˚ (khûm'A;n,
-'O:n)a (-'Ø;n)b /k√m'Ø;n/ come on!˚ (khûm'¤n:) /k√m'In/ come in!˚ (khûm'I;≥, -'H-)a
(-¤;å)b /k√m'hIÈ≤/ come here!
More examples: (s¤ö'Da;Ön, s¤-, s¤T-) /sIt'daOn/ sit down˚ (√Dó'ø;¨, aÙDø¨'nø;¨)a
(åD'ó‘;¨, aÙD‘¨'n‘;¨)b /aEdOUn[t]'nOU/ I don't know˚ (√D¤ó'ø;¨, aÙ&D¤Dó'ø;¨)a (åD¤ó-
'‘;¨, aÙ&D¤Dó'‘;¨)b /aEdIdó[t]'nOU/ I didn't know˚ (√'sp™k&sø¨, aÙ¤k-)a (å'sp™k&s‘¨)b
/aEIk'spEktsOU/ I expect so˚ (aÙ'spø¨Ω, √sÈ-)a (-‘¨Ω, åsÈ-)b /aEsÈ'pOUz/ I suppose˚ (√'†¤˙k
j≥'<aÙT, -˙ö)a (å-, jÈ'>-)b /aE'†I˙k jÈ≤'<aEt/ I `ink you're right˘
≈nally: (aÙ˙gûnÈ'DUu¤T, å˙û-÷ -&gø¨nÈ-, -√-, -O;-, -A-, -ÈT)a (å-, -¯u¤T÷ -&g‘¨nÈ-, -å-,
-ø;-, -Ø-)b /aEmgÈnÈ'dUuIt÷ -gOUnÈ-÷ -'gOUI˙tÈ-/ I'm goi« to do it [gonna do it]˚ (hizgû-
n¯'πD¤T÷ -&gø¨n¯-, -√-, -O;-, -A-, -ÈT)a (-&g‘¨n¯-, -å-, -ø;-, -Ø-)b /hizgÈnu'πdIt÷ -OU-÷ -'gOUI˙-
tÈ-/ he's goi« to add it˚ (wi&wAnÈ's™;I, -wO;-, -nT-, -nö-)a (-wØ-)b /wiwØ;nùÈ'sEI/ we want
to sZ [wanna]˚ (wi&wAn¯'IiT, -wO;-, -nT-)a (-ØnT-)b /wiwØ;nùu'Iit/ we want to eat˚ ('s√m-
†¤˙ 'D¤f<ÈnT, 's√mpõ)a ('såm-, ->ÈnT, 'såmpõ)b /'s√m†I˙ 'dIf<Ènt/ some`i« di‡erent˚
88 a handbook of pronunciation
(f¨ı'f¤ı:, fûı'-, f¨'-, fÈ'-) /fUı'fIı/ fulfill˚ (¤M'vaÙ<ÈnmÈnT, -ÈmmÈ-, -ÈmÈ-, -aÙ<mÈ-, ™-,
È-)a (->Ènm-, ->ÈmmÈ-, ->ÈmÈ-, -aÙÈmÈ-, -aÈmÈ-, ™-, È-)b /In'vaE<ÈnmÈnt, E-, È-/ en-
vironment˚ ('g√v≥nmÈnT, -v≥m-, -vÈm-, -võ-, -vm-÷ -bõ-÷ -mm-)a ('gåvÈnm-, -vÈmm-,
-vÈm-, -võ-, -vm-÷ -bõ-÷ -mm-)b /'g√vÈ≤nmÈnt/ government.
American English has, too: ('khπnDÈ&D™IT, 'khπnÈ-, -DÈT) /'kπnd¢dEIt, -d¢t/ candi-
date˚ (πnT'A;<kT¤k, πn['-, πn'-, -'A;<T¤k) /πnù'A:≤ktIk/ antarctic˘
American dissimilation of r
2.3.3.5. To simplify the articulation of words and rhythm groups with two /≤/'s,
American pronunciation can have variants with /`/ for the first /≤/, although not
very frequently, even in stressed syllables: ('fA;<∑≥, 'fA;∑≥) /'fA:≤∑È≤/ far`er˚ ('f≥;∑≥,
'fÈ;∑≥) /'fÈ:≤∑È≤/ fur`er˚ ('ø;<D≥, 'ø;D≥) /'O:≤dÈ≤/ order˚ ('m≥;D≥, 'mÈ;D≥) /'mÈ;≤dÈ≤/ mur-
der˚ ('khø;<n≥, 'khø;n≥) /'kO:≤nÈ≤/ corner˚ (†≥'mAmÈm≥, †È'-) /†È≤'mØm¢ùÈ≤/ `ermom-
eter˚ ('fø;<w≥D, 'fø;w≥D) /'fO:≤wÈ≤d/ forward˚ ('fø;<&w≥;D, 'fø;&w≥;D) /'fO:≤wÈ:≤d/ for[ord˚
(s≥'ph<a;ÙΩ, sÈ'-) /sÈ≤'p<aEz/ surprise˚ ('g√v≥n≥, -vÈn≥, -vn≥) /'g√vÈ≤nÈ≤/ governor˚ (ph≥-
'Th¤kjÈl≥, phÈ'-) /pÈ≤'tIkjÈlÈ≤/ particular˘
Also: ('khπn[m]≥&b™<i, -È&b-) {('khπnTÈbˇi, -b>i, -&b™>i)b} /'kπnùÈ≤bŒ<i/ Canterbury˚
('wO;m≥&b™<i, -[È-, 'wA-) {('wø;TÈbˇi, -b>i, -&b™>i)b} /'wO;ùÈ≤bŒ<i/ Waterbury˚ ('Am≥&b≥;n, 'A[-
È-) /'ØùÈ≤bÈ:≤n/ Otterburn˚ ('b≥;n≥fl, 'bÈ;-, b≥'nA:<fl, bÈ-) /'bÈ:≤nÈ≤d, bÈ≤'nA:≤d/ Ber-
nard˚ ('khπm≥&p¤l≥, -[È-) /'kπùÈ≤pIlÈ≤/ caterpillar˚ ('EıD≥'b™<i, -DÈ&-) {('™ıDÈ&b™>i, -Èbˇi, -È-
b>i)b} /'EıdÈ≤bŒ<i/ elderberry˚ ('<™z≥&vwA;<, -zÈ&-÷ -&vwø;<) /'<EzÈ≤vwA:≤/ reservoir˚ ('nø;<-
†<Èp, 'nø;†-) /'nO:≤†<Èp/ Nor`rup˚ ('s√∑≥n≥, 's√∑Èn≥) /'s√∑È≤nÈ≤/ Sou`erner˚ ('m≥;DÈ-
<≥, -D≥≥, 'mÈ;D-) /'mÈ:≤dÈ<È≤/ murderer˚ ('nø;<∑≥n≥, 'nø;<∑Èn≥, 'nø;∑Èn≥) /'nO:≤∑È≤nÈ≤/
Nor`erner˘
Less systematically, dissimilation is possible even in rhythm groups: (h≥'hA;<T,
hÈ'-) /hÈ≤'hA:≤t/ her heart˚ (j≥'m≥;ßi, jÈ'-) /jÈ≤'mÈ:≤si/ your mercy˚ (A;<'phø;<c, A;'-) /A:≤-
'pO:≤c/ our por>˚ (∑™≥'f≥:mΩ, ∑™È'-, ∑≥'-, ∑È'-) /∑EÈ≤'fÈ:≤mz/ `eir firms˚ (∑™≥'D≥;[i, ∑™È'-,
∑≥'-, ∑È'-) /∑EÈ≤'dÈ:≤ùi/ `\'re dirty˚ (f≥'w≥:fl, fÈ'-) /fÈ≤'wÈ:≤d/ for word˚ (ø;<'hI;≥, ø;'-, ≥'-,
È'-, -'H-) /O:≤'hIÈ≤, È≤'-/ or here˘
In addition to /≤/, the following examples will show dissimilation for /</ (which
is used in British pronunciation too, due to a kind of simplification, even by anal-
ogy): ('laÙ&b<™<i, -b<È<i, -b≥i, -b<i, -&b™-, -bÈ-)a ('laÙb>È>i, -b>ˇi, bˇi, -b>i, -bÈ>i)b /'laE-
b<Œ<i/ library˚ ('f™b<¯&™<i, -b<È<i, -bj¯-, -bjÈ-)a ('f™b>¯È>i, ->¨>i, -b>ˇi, bˇi, -b>i, -bj¨>i,
-bjÈ-)b /'fEb<uŒ<i, -bju-/ February˚ ('s™k<È&T™<i, -kû-)a ('s™k>È˛>i, -È&T™>i, -kû-)b /'sEk<È-
tŒ<i/ secretary.
In addition: (sTÈ'nAg<Èf≥÷ -gû-)a (-Øg>È-÷ -gû-)b /stÈ'nØg<ÈfÈ≤/ ®enogra#er˚ (fÈ-
'ThAg<Èf≥÷ -gû-)a (-Øg>Èfå÷ -gû-)b /fÈ'tØg<ÈfÈ≤/ #otogra#er˚ (ph<È'f™s≥÷ phÈ-÷ ph≥-)a
(ph>È'f™så÷ phÈ-)b /p<È'fEsÈ≤/ profe˛or˚ (ph<È&n√nsi'™ISÈn÷ phÈ-÷ ph≥-)a (ph>È&nånsi'™I-
Só÷ phÈ-)b /p<Èn√nsi'EIS˘n/ pronunciation˚ (ph<È'phπ<È&Tø<i÷ phÈ-÷ 'ph<™p≥È-, 'ph<™-
p<È-)a (ph>¤'phπ>È˛>i, ph>È-÷ phÈ-)b /p<¢'pπ<ÈtK<i÷ 'p<Ep˘<ÈtK:<i/ preparatory˘
All in all, we can see that dissimilation mostly occurs with: ('ø;<) /'O:≤/, usually,
in stressed syllables (where, even if /≤/ is not pronounced, words do not become
ambiguous, because the vowel timbre alone is distinctive; see the examples above),
and with: (≥) /È≤/ in unstressed syllables: surprise˚ particular˚ caterpillar˚ governor˚
`ermometer…
Morphonological remark
Simply, we have:
/d/ (D) after voiced phonemes (¤ vowels, diphthongs, and voiced consonants,
except the very /d/): /'plEId/ plZed, /'hÈ;<id/ hurried, /'bA:≤d/ barred, /'<√bd/ rubbed,
/'G√Gd/ judged, /'plπnd/ planned÷
/t/ (T) after voiceless consonants (except the very /t/): /'stØpt/ ®opped, /'wØ;St/
wa´ed, /'swIct/ ~ibed, /'lπ;ft/ laughed÷
/¢d/ (ÈD)a (¤D)b after /t, d/, in order to be able to pronounce them: /'wEIù¢d/ wait-
ed, /'nIid¢d/ needed, /'stA:≤ù¢d/ ®arted˘
≈nally, we have:
/z/ (z) after voiced phonemes (¤ vowels, diphthongs, and voiced consonants,
except the grooved ones, /z, Z, G/): /'gOUz/ goes, /'flaEz/ flies, /'dEIz/ dZs, /'lEIdiz/
ladies, /'<√bz/ rubs, /'wEIvz/ waves, /'plπnz/ plans, /'GØnz/ John's, /'<IcÈ≤dz/ Ri>-
ard's÷
/s/ (s) after voiceless consonants (except the grooved ones, /s, S, c/): /'tØps/ tops,
/'<aEts/ writes, /'b<E†s/ brea`s, /'GEfs/ Je‡'s, /maE'pE˘<Ènts/ my paren_'÷
/¢z/ (Èz)a (¤z)b after /s, z÷ S, Z÷ c, G/, in order to be able to pronounce them: /'kIs-
¢z/ ki˛es, /'<OUz¢z/ roses, /'dIS¢z/ di´es, /'swIc¢z/ ~ibes, /'πks¢z/ axes, /'cA:≤ız¢z/
≥arles's˘
Stress
2.3.4.1. We know that (the position of) stress may be distinctive, in English:
('¤mpø;<T)a (-ø;T)b import (noun, adj.), (¤m'phø;<T)a (-ø;T)b import (verb); ('ph<™z-
óT)a (->-)b present (noun, adj.), (p<¤'z™nT)a (p>-)b present (verb).
English sentences usually keep the stresses of their words well, even in lexical
monosyllables, while grammatical monosyllables lack any stress (as, in general, do
polysyllabic unstressed syllables): ('sπmz 'bO;t '†<I;i 'nU;u 'smO:ı 'blπk 'khπts)a ('bø;t
'†>I;i 'nj¯;u 'smø:ı)b Sam has bought `ree n[ small bla$ ca_÷ but we have: (ó∑≥-
wÇÈ'lA:<ä 'kh<a;ÖD *'phIip®)a (-ûwÇÈ'lA:ä)b and `ere was a large crowd of people,
(bÈ[&¤ts*∑û'g<™I[ÈsT ¤m'phø;<Tós, -öós)a (bÈT&-, -'g>™IT-, -ø;-)b but it's of `e greate® im-
portance˘
However, in long words (and in sentences as well), many syllables with full vow-
els generally receive secondary stresses (especially when they occur near unstressed
and reduced syllables): (&ø¨v≥'™sTÈ&m™IT)a (&‘¨vÈ>'™sT¤-)b overe®imate, (&v¤zÈ'b¤lÈ[i)a
(-Ti)b visibili“, (&ph≥;pÈn&D¤kjÈ'lπ<È[i)a (&ph‘;-, ->ÈTi)b perpendiculari“, ('<™kûg&naÙΩ)a
('>™-)b recognize, ('m™ks¤&kø¨)a (-‘¨)b Mexico.
In compounds, the more frequent structure is ('à&à) (more rarely (&à'à): (&f¤f'ThI;in)
fifteen). Sometimes, even ('à'à) occurs, as in çcollocationsÇ (or occasional –or free–
compounds, which are, then, modifiable): ('blπk&b≥;fl)a (-‘;fl)b bla$bird (but ('blπk
'b≥:fl)a ('b‘:fl)b bla$ bird), ('b¨ı&DAâ, -O;â)a (-Øâ)b bulldog˘
Of course, there are also many instances like: ('¤˙gl¤S&ThIic≥)a (-å)b E«li´ tea>er
ça teacher of EnglishÇ and ('¤˙gl¤S 'ThIic≥)a (-å)b E«li´ tea>er ça teacher who is
EnglishÇ.
2. english 91
2.3.4.2. Let us now consider compounds such as fir® cla˛ (noun and adverb)
and fir®-cla˛ (adjective), and the collocation fir® cla˛˚ in a sentence like `at was
`e fir® cla˛ to be considered˘ From a phonetic point of view, they are alike: ('f≥;ß[T]
'khlπ[;]s)a ('f‘;s[T] 'khlA;s)b; however, from a phonemic point of view, and for teach-
ing and lexicographical purposes as well, it could be very useful to distinguish
them as: /'fÈ:≤st'klπ;s/ (compounds: ç/'à'à/Ç) and /'fÈ:≤st 'klπ;s/ (collocation: ç/'à 'à/Ç).
Besides, patterns are flexibly structured. As a matter of fact, we have: (&f¤f'ThI;in)
fifteen and ('ph™;Iä f¤f'ThI;in) page fifteen˚ but ('f¤f&ThIim 'ph™;IGÈΩ)a (-¤Ω)b fifteen
pages÷ ('b<π;n[D] 'nU;u)a ('nj¯;u)b brandn[˚ but (È'b<π;n[D]&nUu kûm'phj¯um≥)a
(-&nj¯u, -Tå)b a brandn[ computer˘
Moreover: ('s™kûn[D] 'hπ;nfl) secondhand˚ but ('s™kûn&hπnfl 'khlø;¨[∑]Ω)a
(-‘;¨[∑]Ω)b secondhand clo`es÷ and (∑[™]≥'O:ı &s™kûn'hπ;nfl)a (∑™‘>'ø:ı)b `\'re all sec-
ondhand÷ also (&πfT≥'nU;un)a (&A;fTÈ'n¯;un)b afternoon and (&g¨DπfT≥'nU;un, g¨D&-,
gûD&-)a (&g¨DA;fTÈ'n¯;un, g¨D&-, gûD&-)b good afternoon˚ but ('πfT≥&nU;un 'ThI;i)a ('A;f-
TÈ&n¯;un 'ThI;i)b afternoon tea˘
A few cases can vary according to speech rate, but also whether they occur in
intonemes or preintonemes, as well as according to personal choices. Here, we will
make use of di‡erent degrees of intermediate stress, too, which (without an em-
phatic one, (")) are, in descending order: ('), (“), (&), (÷), ( ). It is worthwhile observ-
ing nuances carefully: (&D™mÈn'sT<™ISÈn '™ks≥&saÙzÈΩ, 'D™mÈn&s-, -'sT<™ISÈn“™ks≥&saÙzÈΩ)a
(-'sT>™ISó '™ksÈ&saÙz¤Ω)…b demon®ration exercises˚ ('™lÈ&v™Im≥ 'Ap≥&™Im≥, '™lÈ&v™Im≥“Ap≥&™I-
m≥, '™lÈv™Im≥&Ap≥÷™Im≥)a (-TÈ> 'ØpÈ&>™ITå)…b elevator operator˚ ('laÙT&haÖs 'khIip≥, 'laÙT-
&haÖs“khIip≥, 'laÙThaÖs&khIip≥)a (-å)…b ligh`ouse keeper÷ let us notice: ('laÙT 'haÖs-
&khIip≥)a (-å)b light housekeeper˘
teÇ for /vIzÈ'bIlÈùi/ (&v¤zÈ'b¤lÈ[i)a (-ÈTi)b visibility˘ But some American dictionaries
are misleading, because they mark secondary stress for most unstressed syllables
bearing full vowels.
Regrettably, mainly dictionaries published in the ¨º (even pronouncing diction-
aries) do not use secondary stress wisely enough. As a matter of fact, a collocation
like ('s™nT<® 'hIi[¤˙)a (-˛>-, -T¤˙)b /'sEnt<® 'hIiùI˙/ central heati« is, usually, represent-
ed as */&sentrÈl 'hi:tI˙/, exactly like (&s™nT<È'l¤sT¤k)a (-˛>-)b /sEnt<È'lIstIk/ centrali®ic
(their */&sentrÈ'lIstIk/).
However, the more they mark the better, even when things are predictable, pro-
vided they do so in an exact and accurate way. Indeed, teaching transcriptions, es-
pecially for beginners, should show several characteristics, with no absurd and
groundless fear that they may confuse. In reality, too simple a transcription is less
useful and, sometimes, misleading.
Intonation
2.3.6. ˛ere are several and quite varied que®ion tags, or tag que®ions˚ in En-
glish, while, other languages generally have fixed formulas. In the English lan-
guage, they are morphologically determined (by modifying auxiliary and modal
verbs and changing their positive/negative polarity). ˛ey have two di‡erent func-
tions: confirmations of somebody's suppositions (by means of conclusive into-
nemes), or actual que®ions, to really ask something, for lack of any certainty.
Let us see a few examples: It's cold todZ, isn't it? or You're American, aren't you?
or †\ aren't well, are `\? For confirmation, we will have: /Its'kOUıd tÈ'dEI. ¿'Isó{ù}-
¢t./, /jÈ<È'mE<¢kÈn. ¿'A:≤ntjUu./, /∑EI'A:≤nt 'wEı. ¿'A:≤∑EI./; while, for information: /Its-
'kOUıd tÈ'dEI. ¿'Isó{t}¢t?/, /jÈ<È'mE<¢kÈn. ¿'A:≤ntjUu?/, /∑EI'A:≤nt 'wEı. ¿'A:≤∑EI?/.
Other accents
2.4.0. ˛is section is likely to be very important for descriptive and communi-
cative purposes, as it deals with pronunciations going beyond the neutral accents.
94 a handbook of pronunciation
f
f
('håT)i ('h√T)a ('håT)b /'h√t/ hut, ('hØT)i ('hAT)a ('hØT)b /'hØt/ hot, ('sø:)i ('sO:)a ('sø:)b
/'sO:/ sY, (&s¤cu'™¤SÈn)i (&s¤c¯'™ISÈn)a (&s¤c¯'™ISó)b /sIcu'EIS˘n/ situation˘
For /È:≤/, we have (È:‰): ('wÈ:‰fl)i ('w≥:fl)a ('w‘:fl)b /'wÈ:≤d/ word÷ for /È:<, È;</ (È:<):
('hÈ;<i)i ('h≥;i)a ('hå>i)b /'hÈ;<i/ hurry÷ for /È≤/, in unstressed syllables, we find (ȉ):
('b<å∑ȉ[Ω])i ('b<å∑≥[Ω])i (-≥[Ü])a (-å, -‘Ω)b /'b<√∑È≤[z]/ bro`er[s]˘
For /IÈ≤, EÈ≤, UÈ≤/ we have (é;ȉ): ('h¤;ȉ)i ('hI;≥)a ('h¤;å)b /'hIÈ≤/ hear, ('∑™;ȉ)i ('∑™;≥)a
('∑™;‘)b /'∑EÈ≤/ `ere, ('ph¨;ȉ)i ('phU;≥)a ('ph¨;å)b /'pUÈ≤/ poor÷ and for /I˘<, E˘<, U˘</ we
have (¤<, ™<, ¨<): ('h¤<¤˙)i ('h¤<¤˙)a ('h¤È>¤˙)b /'hI˘<I˙/ heari«, ('m™<i)i ('m™<i)a ('m™‘-
>i)b /'mE˘<i/ Mary, ('Dj¨<¤˙)i ('D¨<¤˙, 'D≥;-)a ('Dj¨È>¤˙, 'Djø;>-)b /'dˆU˘<I˙/ duri«˘
2.4.1.4. As to diphthongs, it is su‚cient to notice: /Ii, Uu, OU/ (ii, uu, ø¨) (no-
tice that (ii, uu) are diphthongs, although very narrow): ('Thi;i)i ('ThI;i)a:b /'tIi/ tea,
('fju;u)i ('fj¯;u)a:b /'fjUu/ f[, ('Thu;u)i ('ThU;u)a ('Th¯;u)b /'tUu/ ‘o, ('nø;¨)i ('nø;¨)a
('n‘;¨)b /'nOU/ no. Besides, for /aE, OE/, a second element like (™í) is enough (Ô û
2.10, to avoid introducing/learning another vocoid; so much so that a pronunci-
ation with (™í) is much better than çforeignÇ *(ai, Oi), and it is near the native
one): ('ga;™)i ('ga;Ù)a:b /'gaE/ guy, ('bø;™)i ('bø;Ù)a:b /'bOE/ boy˘ Little has to be said
about /EI, aO/: ('D™;¤)i ('D™;I)a:b /'dEI/ dZ, ('na;ø)i (na;Ö)a:b /'naO/ now˘
Neither the (û) taxophone of /È/ near velars, nor other taxophones of /I, E, π,
U÷ √, Ii, Uu÷ EI, aE, OE/ + /ı, ®/ are necessary, but their use makes one's pronunciation
more çauthenticÇ, although this çinternationalÇ model is already definitely more
realistic than those o‡ered in so many textbooks, even by English-speaking phone-
ticians (where, among other thing, ç/i:, u:/Ç are still indicated). ˛us, taxophones
like (û÷ ¤#, ™#, π#, ¨#÷ √@, iI, uU÷ aÉ, øÉ) (but (™I, ™¤ù)) can really change one's pro-
nunciation from simple çinternationalÇ (for foreigners) into çnative internationalÇ
pronunciation. Instead, the use of (i:, u:) clearly indicates either a non çnative-likeÇ
pronunciation, or one which is regional or socially inappropriate.
96 a handbook of pronunciation
For /ı, ®/, it is better to use (ù, æ) (semilateral, û 1.15.1), which, articulatorily,
are decidedly simpler than (ı, ®) (in case, even velar can be used –velar semilater-
al (), £)– which, auditorily, are decidedly better than (l, Í)).
As for consonants, su‚ce to say that (ö) is acceptable for /t/ (§ 2.2.2.4) and that
([) may be good for /ù/, mainly after vowels, while, (T) is more recommendable, af-
ter /n, ≤, ı/: ('b™Ti, -[i)i ('b™[i)a ('b™Ti)b /'bEùi/ Bet“, ('Thw™nTi)i ('Thw™n[[]i)a ('Thw™n-
Ti)b /'twEnùi/ ‘en“˘
˛us, for r˚ we have: /</ (<), /≤/ (‰). Besides, /·w/ simply corresponds to /w/:
('w™n:)i /'·wEn/ when, and /ˆ/ to /j/: ('Thju;un)i /'tˆUun/ tune, ('Dju;un)i /'dˆUun/ dune,
('nju;u)i /'nˆUu/ n[˘
˛e intonation of çinternationalÇ English has a restrained and more general
movement, as can be seen from û 2.11 (although the first stressed syllable in a pre-
intoneme could be half-high, (Ì), instead of just raised mid).
2.4.1.5. Both the ç®πÇ and the American models have some problems of social
acceptability. ç®πÇ, though still very widely used, both in the ∫∫© news and in
some kinds of British sitcom and movies, has always had a strong connotation of
artificial a‡ectation, which makes it disagreeable to many native speakers. ®π is
generally associated with a çhighÇ social position (™ members of the aristocracy,
of the higher clergy or military ranks, Tory MP's, prestigious university professors,
Â), and a certain age group (over 50 years of age). Clothing, too, should be su‚-
ciently formal, to be suitable for the ®π accent. If these conditions are lacking –¤
for common natives– the British neutral accent could prove to be definitely inap-
propriate.
It is to be said that, paradoxically, even an impersonation of an ®π speaker (even
only partially successful, especially if belonging to certain particular varieties) may
give rise to negative feelings from British listeners belonging to the middle or
working class. Indeed, these people might find a few snobbish phonetic traits, easi-
ly recognizable as marked ones, even if mixed with foreign traits, especially if the
speakers are young(er).
Against these (empirically checked) di‚culties, it might be better to choose the
American neutral accent. ˛is is certainly recommendable to learners in North
America, but not in the British Isles, where most people would consider it inap-
propriate.
We reckon that this dilemma may be faced, in a practical and diplomatic way,
aiming at acquiring an çinternationalÇ accent, which –although it might seem to
be çnobody's accentÇ– would have none of the possible negative connotations of
the two mediatic accents (we will be dealing with below), or even of the two na-
tional neutral ones (which some people seem to consider çnobody's accentÇ, as
well, since –in both nations– only about 3% of native speakers actually use them).
£Mediatic∞ accents
2.4.2.0. In addition to the two neutral accents, the American and the British
ones, and to the international accent, we believe it is important to show the two
kinds of accent actually used by American and British native speakers. Naturally,
çmediaticÇ pronunciations are often used by North-American people in the mid-
western states, and by English people in the southeastern counties, respectively, as
well as by most spoken-word mass media (¤ radio and television) that do not use
the neutral accents nor more local ones.
˛erefore, û 2.12-18 (which speak for themselves) must be carefully analyzed,
comparing them with the neutral ones, in order to capture the di‡erences, which
are sometimes not slight! In ordinary people's opinion, mainly if their own pro-
nunciation is directly concerned, these mediatic accents are thought to be less pe-
culiar than the neutral ones.
On the other hand, as everyone knows, neutral pronunciation –in percentage
terms– is used the least by native speakers; but, it is the one generally aimed at by
advanced foreign learners, except for more or less frequent interferences, especial-
ly from their mother tongue, and individual peculiarities.
For this reason, we do not hesitate to show these actual realities, although we
will not get to the point of recommending an active usage by foreigners. On the
contrary, a passive usage is certainly welcome, to really understand native speak-
ers, when they talk… ças they canÇ. Actually, this happens every day, all over the
world, because school and society usually ignore (good) pronunciation.
However, as these mediatic pronunciations are really very widespread, and of-
ten considered as çalmostÇ neutral (or, at least, less a‡ected and less artificial than
the neutral accent), many people would be willing to declare them to be neutral.
˛ey would do so, on the one hand, in opposition to pronunciations which are
more recognizable as local ones, and, on the other hand, to an çunsubstantialÇ neu-
tral kind of pronunciation… It is no rare fact that some çmediaticÇ speakers fluc-
tuate towards the neutral type (or away from it), for some words.
Here we wish to briefly draw attention to some details, with reference only to the
respective (American or British) variant, for a direct (çinternalÇ) comparison. In the
light of what has been seen so far, it will not be di‚cult to find the corresponding
neutral forms of the other accent. On the contrary, it will be a very useful exercise.
98 a handbook of pronunciation
guished by labialization (which is slight since the two vocoids are low). ˛us many
speakers can unify them (but, productively — perceptively, things are rather com-
plex and oscillating) by also obtaining ('sA;<i, 's˙:˙)m;a (('sA<i)a /'sØ<i/ sorry˚ ('sO:˙)a
/'sØ;˙/ so«), all the more so because, for /O:/, some çintermediateÇ articulations be-
tween /O:/ and /A:/ are decidedly common: (ùA), as in ('sŒ:˙, 'sŒ;˙˙)m;a /'sØ;˙/ so«
and ('sù:, 'sù;A)m;a /'sO:/ sY˘
Besides, we have /È/ (x)m;a (ç/√/Ç), not only when final before a pause, but even
after a pause, even if near velar(ized) C, where –in the neutral accent– we find
(û)a: (x'ph]√;Ù) /È'plaE/ appI˚ (x's√;Ùfl) /È'saEd/ aside˚ (x'ThímpT) /È'tEmpt/ attempt˚
(xÊ'Th™;In) /Èb'tEIn/ obtain˚ (x']Ô;sn) /È'lOUn/ alone˚ (x'gÖ;¨) /È'gOU/ ago˚ (x'w™;I) /È-
'wEI/ YZ˚ (x'<π;Ônfl) /È'<aOnd/ around˚ (x'kh>:) /È'kÈ:≤/ occur˘
/π[ı]/ (π, πı), /π;/ (πÄ) /A:, Ø/ (A:), /Ø;/ (ù:, A:),
/Ø;</ (A:<, ù:<, ø:<)
‡
2.4.2.3. û 2.13 gives the diphthongs. Let us observe the di‡erences for /Ii, aE,
aO, OU, [j]Uu, Uuı/ (¤i, √Ù, πÖ, Ö¨, [j]%¯, ¯¨ı)m;a (Ii, aÙ, aÖ, ø¨, Uu, j¯u, Uuı)a:
('b¤;i)m;a ('bI;i)a /'bIi/ bee˚ ('g√;Ù)m;a ('ga;Ù)a /'gaE/ guy˚ ('Dπ;Ôn)m;a ('Da;Ön)a /'daOn/
down˚ ('nÔ;s)m;a ('nø;¨)a /'nOU/ no˚ ('Th%;¯)m;a ('ThU;u)a /'tUu/ ‘o˚ ('nß;*)m;a ('nU;u)a
/'nˆUu/ n[˚ ('fj¯;¨ı)m;a ('fj¯;uı)a /'fjUuı/ fuel˘
further variants are fairly frequent, which produce: /I˙/ (i˙), /A:≤=, A:≤’0é/ (A<):
('khi˙:, 'si˙i˙)m;a for ('kh¤˙:, 's¤˙¤˙)a /'kI˙, 'sI˙I˙/ ki«˚ si«i« and ('phA<k, 'khA<m>,
'DA<]i˙)m;a for ('phA;<k, 'khA;<m≥, 'DA;<l¤˙)a /'pA:≤k, 'kA:≤ùÈ≤/ park˚ carter, darli«.
2.4.2.5. As far as C are concerned, keeping in mind that these observations are
general (and not absolutely çobligatoryÇ for all speakers — words) and that our
transcriptions are çnormalizedÇ, let us say that for /l[j]éı/ we have (][j]éı): (']¢ıT)m;a
('l¤ıT)a /'lIıt/ lilt˚ (']Öı:)m;a ('l,ı:)a /'l√ı/ lull˚ ('mK]j'n)m;a ('m¤¬jÈn)a /'mIljÈn/ million˚
('n˙v–ûsT)m;a ('nAvÍÈsT)a /'nØvÍ¢st/ noveli®˘
Besides, /</ has a uvularized velar rounded articulation, which is darker (or
çharderÇ): ('<¤;>)m;a ('<I;≥)a /'<IÈ≤/ rear˚ ('<ø:<)m;a ('<ø:<)a /'<O:</ roar˚ ('w’n≥)m;a ('w¤n≥)a
/'wInÈ≤/ winner˘ Please note that generally (<) exerts on /È/ the same retracting and
raising e‡ect of /k, g, ˙÷ ı, w/: (Thû'<™;In)m;a (ThÈ'<™;In)a /tÈ'<EIn/ to rain.
As we have seen, a N nasalizes the following V˘ It often also nasalizes the preced-
ing V (as we indicate). Even intense C in contact are nasalized. ‹en we find syl-
lables with /{é}é/ + /mp˘, nt˘, ˙k˘/, we very frequently have ({–}–) + (Mp, Möp)
(∫T, ∫öT, ∫ö) («k, «ök) (which are too often described simply as ç(–0)Ç à la françai-
se): ('kh™’Möp)m;a ('khπmp)a /'kπmp/ camp˚ ('sT≈Möp)m;a ('sT√mp)a /'st√mp/ ®ump˚
('kh™’∫ö, -∫öT)m;a ('khπ[;]nT)a /'kπ;nt/ can't˚ ('DÔs∫ö, -∫öT)m;a ('Dø¨nT)a /'dOUnt/
don't˚ ('†I«ök)m;a ('†¤˙k)a /'†I˙k/ `ink˚ ('D<™’«ök)m;a ('D<π˙k)a /'d<π˙k/ drank˘
/ù/ behaves as in neutral American pronunciation, but often this use spreads to
/d/ as well, so that (']πm>)m;a can represent either ('lπm≥)a /'lπùÈ≤/ latter or ('lπD≥)a
/'lπdÈ≤/ ladder÷ (&≈n≥'ßT™;’nfl, &≈nm>-)m;a (&√nD≥'ßTπ;nfl)a /√ndÈ≤'stπnd/ under®and˘
˛e same goes for /ˆ/ = (`), as seen in n[÷ and it is also possible to have /·w/:
˛erefore, foreigners must be familiar with the mediatic British accent too, but
with no real need to actually use it. However, often, the native speakers' concep-
tion of this accent is subjective and contrasting. In fact, if on the one hand they
may consider it to be neutral, or almost neutral, with respect to çcommonÇ peo-
102 a handbook of pronunciation
ple, when çpublicÇ people are concerned –as politicians are– the same pronuncia-
tion can be defined as çCockneyÇ, that is very dialectal (and hardly appropriate).
But we do know that linguistic opinions are very colorful and personal.
2.4.3.2. As can be seen from û 2.16, /π, Ø, O:/ are higher, (E, O, oo). ˛e last one
is doubled (since it moves upwards slightly), or diphthongized (downwards) when
lexeme final, or followed by the /zò, dò/ (oø[fl/Ω]ò) grammemes. Also /E/ is higher,
though remaining within its own box, (™°) (to be better distinguished from /π/ (E)):
('[h]Eöç)m;b ('hπT)b /'hπt/ hat˚ ('<Oök)m;b ('>Øk)b /'<Øk/ ro$˚ ('lo;on)m;b ('lø:n)b /'lO:n/
lYn˚ ('lo;ø)m;b ('lø:)b /'lO:/ lY˚ ('mo;ø)m;b ('mø:)b /'mO:≤/ more˚ ('lo;øΩ)m;b ('lø:Ω)b
/'lO:z/ lYs˚ ('pho;øfl)m;b ('phø:fl)b /'pO:≤d/ pored˘ In the vocogram, the diphthongs due
to the vocalization of /ı/ are marked in grey: /Iı, Eı, πı, √ı, Uı/. Note the di‡erent
symbols: ('bÙ;P[§])m;b ('b¤ı:)b /'bIı/ bill˚ ('bÉ;P[§])m;b ('b™ı:)b /'bEı/ bell˚ ('Ä;P[§])m;b ('Å;ı)b
/'πı/ Al˚ ('kh√P[§]öç)m;b ('kh√ıT)b /'k√ıt/ cult, ('bU;P[§])m;b ('b¨ı:)b /'bUı/ bull.
˛ose who systematically realize /π/ as (Å) are influenced by a sort of strategy
in reaction to the mediatic (and Cockney proper) closer articulation, which is real-
ized as (E), but it does not belong to neutral pronunciation. In addition, it can
cause confusion with /√/, not neutralization (as some think), since these realiza-
tions are uttered by di‡erent speakers, not the same, although in the same places.
Besides, /u/ is fronter, (%): (&s¤öc%'ÄISÈn)m;b (&s¤c¯'™ISó)b /sIcu'EIS˘n/ situation;
/√, È[≤]|/ are lower (and fronter), (aï): ('baöça)m;b ('båTå)b /'b√ùÈ≤/ butter˚ ('b¤;a)m;b
('b¤;å)b /'bIÈ≤/ beer; whereas, /È:≤, A:{≤}/ are diphthongized, (È‘, Aå): ('fÈ;‘[Ω])m;b
('f‘:[Ω])b /'fÈ:≤[z]/ fur[s]˚ ('khA;å[Ω])m;b ('khA:[Ω])b /'kA:≤[z]/ car[s]˘
2.4.3.3. û 2.17 gives us the positions of /IÈ≤, I˘<÷ EÈ≤, E˘</. We have just seen
/IÈ≤|/ (beer]÷ instead, for /I˘</ (and /I˘<, IÈ≤/ not before pauses) more often we have
(¤:): ('[h]¤;<¤n, -¤È-)m;b ('h¤È>¤˙)b /'hI˘<I˙/ heari«˚ ('[h]¤:< Èn'∑™:, -¤È-, -™;‘)m;b ('h¤;È> Èn-
'∑™;‘)b /'hI˘< Èn[d]'∑EÈ≤/ here and `ere˚ ('[h]¤: 'khamz ∑È'san:, '[h]¤;‘, &[h]¤;'-, &[h]¤È'-)m;b
('h¤;‘ 'khåmz ∑È'sån:, &h¤È'-)b /'hIÈ≤ 'k√mz ∑È's√n/ here comes `e sun. It is to be noted
that even /E˘<, EÈ≤/ are more often long monophthongs than narrow diphthongs:
('m™;<ÙI, 'm™‘-)m;b ('m™‘>i)b /'mE˘<i/ Mary (beside `ere, just seen); /UÈ≤/ practically
becomes /O:≤/: ('pho;ø)m;b ('phø:, 'ph¨;å)b /'pUÈ≤/ poor˘
Generally, in this accent, for /ééÈ/ sequences we have (ééÈ, ééa|): ('sÙIa)m;b
('sIiå)b /'sIiÈ≤/ seer (and also (AÙ'DÙIa)m;b (aÙ'D¤;å)b /aE'dIÈ/ idea), ('DT%a)m;b ('D¯uå)b
/aEÈ≤/ (AÙÈ, AÙaò|), /aOÈ≤/ (πPÈ, πPaò|), /EIÈ≤/ (ÄIÈ, ÄIaò|), /OUÈ≤/ (x¨È, x¨aò|), /OEÈ≤/ (oÙÈ, oÙaò|),
/IiÈ≤/ (ÙIÈ, ÙIaò|, ÙI<é), /UuÈ≤/ (T%È, T%aò|, T%<é), /A:≤/ (Aå) + ç/A:È≤/Ç (AåÈ, Aåaò|), /éé˘<é/ (éé[È]<é)
2. english 103
2.4.3.5. Even for çmediaticÇ British English, neutralizations are frequent be-
fore /ı/ (although less frequent than in Cockney, the typical and popular –and less
educated– dialect and accent of the East End of London). In fact, Ô û 2.19, we of-
ten find ('f¤;P[§])m;b both for ('f¤ı:)b /'fIı/ fill and ('fi;Iı)b /'fIiı/ feel÷ in addition,
('[h]a;P[§])m;b both for ('hÅ;ı)b /'hπı/ Hal and ('h™;¤ı)b /'hEIı/ hail˚ and both for
('ha;Öı)b /'haOı/ howl and ('h√ı:)b /'h√ı/ Hull, as well. Besides, we find ('DO;U[§])m;b
both for ('DØ;ı)b /'dØı/ doll and ('DÖ;¨ı)b /'dOUı/ dole÷ and ('f¨;U[§])m;b both for
('f¨ı:)b /'fUı/ full and ('fu;Uı)b /'fUuı/ fool (the last one is possible for ('fo;U[§])m;b
('fø:ı)b /'fO:ı/ fall too]˘ (We will deal with (§) shortly.)
2.4.3.6. For the consonants, the strong preglottalization of /p, k, c/ must not
be forgotten as it practically occurs in all the cases indicated in § 2.2.6.1-2 “ §
2.2.7.1; in addition, it is important to mention the massive substitution of /t/ with
(ö), in all the cases indicated in § 2.2.5.2; whereas we can say that, generally, (ç)
replaces (T) /t, ù/ of neutral pronunciation. ˛erefore: ('D<Oöp)m;b ('Ã>Øp)b /'d<Øp/
drop˚ ('D<Oöpç)m;b ('Ã>ØpT)b /'d<Øpt/ dropped˚ ('phEök)m;b ('phπk)b /'pπk/ pa$˚
('phEökç)m;b ('phπkT)b /'pπkt/ pa$ed˚ ('<ÙIöc, -öS)m;b ('>Iic, -öS)b /'<Iic/ rea>˚
('<ÙIöcç, -öSç)m;b ('>IicT, -öST)b /'<Iict/ rea>ed˚ ('<AÙöça)m;b ('>aÙTå)b /'<aEùÈ≤/ writer÷
('f¨ö&boU[§])m;b ('f¨p&bø;ı, -ö-, -T-)b /'fUtbO:ı/ football˚ ('g™ö 'Dπ;Pn)m;b ('g™T 'Da;Ön, -ö)b
/'gEt 'daOn/ get down˚ ('lÄIölÙI)m;b ('l™ITli, -öli)b /'lEItli/ lateI˚ ('çhAåöç)m;b ('ThA;T)b
/'tA:≤t/ tart˚ ('çh™sça, -ßça)m;b ('Th™sTå)b /'tEstÈ≤/ tester˘
˛e preglottalization of /p, t, k/ (and of /c/, as in neutral British English) also
occurs between V˚ both within words and in sentences: ('ph™öpa)m;b ('ph™på)b
/'pEpÈ≤/ pepper˚ ('b™öça)m;b ('b™Tå)b /'bEùÈ≤/ better˚ ('bÄIöka)m;b ('b™Ikå)b /'bEIkÈ≤/
baker˚ ('l™öç¤ç, -¤ö)m;b ('l™T¤T, -¤ö)b /'lEùIt/ let it˘ In the /st, sc/ sequences, there is the
possible variant /s/ (ß) (by further assimilating to the following contoid, which is
typical of broader accents): ('sçOöp, 'ßç-)m;b ('sTØp)b /'stØp/ ®op˚ ('sç<Ä;I~ä, 'ßç<-,
'߲>-)m;b ('s˛>™;I~ä)b /'st<EInG/ ®ra«e˚ ('khw™scÈn, -ßc-, -ëcÈn, -ScÈn)m;b ('khw™scó,
-Scó)b /'kwEsc˘n, -Sc-/ que®ion˘
For /ù/, ([) can occur, too: ('ph<¤öçÙI, 'ph<¤[ÙI÷ 'çhw™nöçÙI, -™n[ÙI)m;b ('ph>¤Ti,
'Thw™nTi)b /'p<Iùi, 'twEnùi/ pret“˚ ‘en“˘ Even (ùöó) (in addition to (ùöçÈn, -ó)):
('khl¤nöó, '™P[§]öó, 'bÈ‘öó)m;b ('khl¤nTó, '™ıTó, 'b‘;Tó)b /'klInt˘n, 'Eıt˘n, 'bÈ:≤t˘n/ Clin-
ton˚ Elton˚ Burton˘
Besides, we find (¤n) for the grammeme >-i«≥: ('<x¨l¤n, '<OU-)m;b ('>‘¨l¤˙)b /'<OU-
lI˙/ rolli«÷ this example also shows the oscillation for /OU/ before /lé/, under the in-
2. english 105
2.4.3.7. Our examples have already shown that one of the most evident charac-
teristics, which is socially stigmatized, is the çzeroÇ realization of /h/ (= (`)):
('[h]A;Ù)m;b ('ha;Ù)b /'haE/ high. Consequently, even hypercorrections are frequent,
as happens with the name of the letter h (not without a certain çinternalÇ logic):
('[h]ÄIöc)m;b ('™Ic)b /'EIc/. On the other hand, the reduced form of him is less fre-
quent: ('çhÉP[§]h¤m)m;b ('Th™l¤m)b /'tElIm/ tell him÷ equally for the reduced form
of a\ (ÄI'mE;n)m;b (È'mπ;n)b /È'mπn/ a man˘ Also for /0˘nò/, a less reduced form is
more frequent: (ph<È&nansi'ÄISÈn, 'GEöksÈn)m;b (ph>È&nånsi'™ISó, 'Gπksó)b /p<Èn√n-
si'EIS˘n, 'Gπks˘n/ pronunciation˚ Ja$son˘
˛e sequences /tˆ, dˆ, nˆ/ have the peculiarity of typically corresponding to /c,
G/ and (~): ('chT;%n)m;b ('Thj¯;un)b /'tˆUun/ tune˚ ('GT;%n)m;b ('Dj¯;un)b /'dˆUun/
dune˚ ('~T;%)m;b ('nj¯;u)b /'nˆUu/ n[ (in Cockney we actually find ('nÈ;%) = /'nUu/).
Occasionally, /†, ∑/ can become /f/ and /òd, v/ respectively (which is a typical
Cockney pronunciation, and can be heard even on the borders of the çEstuaryÇ
area {and in further –mostly metropolitan– areas, which have been influenced by
this accent}): ('†<Ù;I, 'f<-)m;b ('†>I;i)b /'†<Ii/ `ree˚ ('∑¤s, 'D-)m;b ('∑¤s)b /'∑Is/ `is˚ ('ma∑a,
-va)m;b ('må∑å)b /'m√∑È≤/ mo`er.
2.4.3.8. ˛ere is nothing to say about /l/, while for /ı/ there is a typical çvocal-
izationÇ of (ı, ®) which become (P) (û 2.19), after V or C˚ as many examples have
already shown (particularly those of the neutralization before /ı/). ˛e lateralized
velar rounded approximant (§) (in brackets in the figure) indicates a less marked
pronunciation, where the lateralized contoid is added to the vocalization, (P), in
order to try to attenuate this characteristic, which is often socially stigmatized. So,
we can find (ù), without lip-rounding, as a kind of halfway compromise.
Here are some examples, to complete the survey: ('l¤öçP[§])m;b ('l¤T®)b /'lIt®/ lit-
tle˚ ('phÙIöpP[§])m;b ('phIip®)b /'pIip®/ people˚ ('D™vP[§]Ω)m;b ('D™v®Ω)b /'dEv®z/ devils˚
('m¤P[§]k)m;b ('m¤ık)b /'mIık/ milk˚ ('woP[§]öça)m;b ('wø;ıTå)b /'wO:ıùÈ≤/ Walter˚
('G™nöçP[§]mÈn)m;b ('G™nT®mÈn)b /'GEnù®mÈn/ gentleman. ≈nally, (&a˙ökPl'È‘-
106 a handbook of pronunciation
b‘öç, -P'h-)m;b (&å˙k®'h‘;bÈT)b /'√˙k® 'hÈ:≤bÈ≤t/ uncle Herbert˚ where we can see the
normal kind of prevocalic l ((l)), and the intense one, again prevocalic: ('nOvͤsç,
-ßç)m;b ('nØvͤsT)b /'nØvÍIst/ noveli®˘
2.4.3.9. In cases like the following, we often find secondary stresses: ('D¤kSÈ&n™-
<ÙI, 's™mÈ&ç™<ÙI, 'DoomÈ&ço<ÙI)m;b ('D¤kSó>i, 's™m¤˛>i, 'Dø;m¤˛>i)b /'dIkSÈnŒ<i, 'sEm¢tŒ-
<i, 'dO:≤m¢tO:<i/ dictionary˚ cemetery˚ dormitory.
Regarding intonation, we have to say that the pitch of the first stressed syllable
and of the following internal unstressed syllables in a preintoneme is less high than
in the neutral pronunciation, as can be seen from û 2.20. ˛e suspensive into-
neme is more similar to the American one; and, often, the interrogative intoneme,
besides being as in neutral British English, can be rising-falling (again in û 2.20,
where we show only the di‡erent type): (¿GÈÌspÙIök '¤˙gl¤S 'wÉ;P[§]12)m;b (¿DjÈ5spIik
'¤˙gl¤S 'w™ı:21, ¿GÈ-)b Do you speak E«li´ well?
Text
2.5.0. ˛e story †e North Wind and the Sun (by Aesop) follows. It is given in
five di‡erent (çnormalizedÇ) versions. In fact, they systematically and coherently
present the most typical characteristics, which are acknowledged as peculiar. We
start with the American and British versions in (neutral) English, which is the first
step of the phonetic method. ˛e international version follows together with the
two mediatic versions of American and British English.
In the other chapters of the book, for each language dealt with, at least two kinds
of foreign pronunciations are given: first the foreign pronunciation of English, and
lastly the British English pronunciation of the foreign language in question, accord-
ing to the same principles. ˛e speakers are supposed to be neutral speakers of their
own language, fluent in English (after prolonged contact with native speakers, but
with no help from the phonetic method), who have adequately learned the rela-
tive prominences, but who substantially use segments (vowels “ consonants) and
intonation elements, which are typical of neutral English (although, of course, a
neutral accent is not so common). Obviously, the same principle is valid for the
foreign pronunciations of English, given first. Sometimes further accents have been
added as can be seen in the correspondent chapters.
2. english 107
Graphemic text
2.5.2.0. †e Nor` Wind and `e Sun were disputi« whi> was `e ®ro«er,
when a travel(l)er came alo« wrapped in a warm cloak. †\ agreed `at `e one
who fir® succeeded in maki« `e travel(l)er take his cloak o‡ ´ould be considered
®ro«er `an `e o`er.
†en `e Nor` Wind bl[ as hard as he could, but `e more he bl[ `e more
closeI did `e travel(l)er fold his cloak around him; and at la® `e Nor` Wind gave
up `e attempt. †en `e Sun ´one out warmI, and immediateI `e travel(l)er took
o‡ his cloak. And so `e Nor` Wind was obliged to confe˛ `at `e Sun was `e ®ro«-
er of `e ‘o.
Did you like `e ®ory? Do you want to hear it again?
2.6.1. ˛ere are many peculiar unifications of vowels, with the merging of /O:,
O;, Ø, Ø;/ into (Ø:) and of /O:≤, O:<, Ø<, Ø;</ into (ø:<) (('sø;<i) /'sØ<i/ sorry). In addition,
there are peculiar diphthongs with narrow taxophones of /EI, OU/ (eI, oU), and of
those of /aE, aO/ followed by a voiceless C within the word, (åÛ, √¯): ('nåÛT, '√¯T)
/iò, ié/ (i), /’I˙/ (I˙) >-ing≥ /ué/ (¯), /UÈ≤/ (U≥)
/IÈ≤/ (I≥), /I˘</ (I<), /I[ı]/ (¤, ¤ı) /U˘</ (U<), /U[ı]/ (¨, Uı)
f
/aO/ (π¨, ÅÖ), /aO=/ (x¯) /Ii, EI, aE, aO, OE, OU, Uu/ + /ı/ = /éé®/
/'naEt, 'aOt/ night˚ out. ˛ere are neutral and non-neutral taxophones for many vow-
els and diphthongs + /ı/. ˛e neutral ones which change phones are: /πı, √ı, Uı÷
Iiı, EIı, aEı, OEı, Uuı/; the others are /Iı, Eı/ and /√ı/ with other variants; in addi-
tion to a possible insertion of (û) before (ı). As far as phoneme distribution is con-
cerned, certain words are pronounced with çBritishÇ vowel elements, others with
çAmericanÇ ones. For the consonants, the use is similar to the American one.
In current and mediatic pronunciations (given in the second set of vocograms),
we find that hypercorrection can produce /jUu/ for /Uu/ [noon˚ too˚ do…), due to
the fact that, for /ˆUu/, careful speakers prefer /jUu/, after /n, t, d/. In addition, we
find the nasalization of /éöò, éö˘/ (–öò, –ö˘), even in /é˘ö/ (above all for /π/);
lastly, for /l/, we have (]é), in these kinds of pronunciations.
‡
/π[ı]/ (E, Äı)
/Ø[ı]/ (Ø, Øı)
‡
/√[ı]/ (å, √ı), /A:/ (a:)
neme /≤/ follows British use (although some young Australians, especially females,
who live abroad, present a fluctuating and non-neutral use of (>)). Let us notice
(and very well too) the various taxophones + /ı/.
We systematically find ç/OUlé/Ç (whereas in Cockney and in mediatic British
English minimal pairs occur such as ç/√Ulé/Ç – ç/OUlé/Ç, as polar]˘ Occasionally, we
can have ç/O:È≤/Ç, for /O:≤/ (either ç/Or/Ç, or ç/or/Ç). Except in neutral pronunciation,
for the /Ii, EI, aE, OE/ diphthongs, we have (ééı, éé®); besides, /{é}éÈı/ ({é}éûı,
{é}é®), without vocalizations (with (ı, ®), after phones with labial component).
˛e diaphoneme /ù/ has ([), as in American English. For /0®ò, 0Èıò/, neutral
pronunciation has (0®ò, 0ûıò); thus, it has (léı), without vocalizations (which are
present, though, in broad pronunciations). It is possible to hear a non-neutral pro-
nunciation with /l/ (]), for l + -y˚ -ie˚ -i«˚ -er… (grammemes {or even pseudo-gram-
memes} added to /ıò/).
2.6.3. In addition to the neutral accent, with three vocograms (the first three),
we present the mediatic accent (whose centering diphthongs {fourth and fifth vo-
cograms} correspond to the neutral ones) and the broad accent (whose monoph-
thongs {sixth and seventh vocograms} correspond to the mediatic ones, whereas
the centering diphthongs are peculiar, including the unification of /EÈ≤, E˘</ with
/IÈ≤, I˘</, and the possible unification of /UÈ≤, U˘</ with /O:≤, O:</). ˛us, for /E˘</
(e>, e;>, e‘>) (even in less neutral speech, the timbre is always (e): (e>, e˛)). For /I</
we have (Ù>÷ ¢>, ¢˛); whereas, for /I˘</, (I>, iI˛); for idea\ /aE'dIiÈ/.
/π</ (E>÷ ™>) is always distinct (in all accents). Occasionally, we can find ç/O:È≤/Ç,
for /O:≤/ (both ç/Or/Ç, and ç/or/Ç). ˛e triphthongs are not attenuated into diph-
thongs. Even in neutral pronunciation, we have /I˙/ (¤˙) (not ç(Ù˙, ¢˙)Ç). In New
Zealand, /π;, A;/ = /A:/, /Ø;, O;/ = /Ø/, /¢/ = /È, ±I/ (¤ /I/ is only an intentional choice);
fini´ /'fInIS/ ('fÙnÙS÷ 'f¢n¢S), visit /'vIz¢t/ ('vÙzÈT, 'v¢zÈT).
˛e most typical characteristic consists in (Ù, e) for /I, E/, in neutral pronunci-
ation (but (¢, ¤), in the other accents), in addition to the pronunciation of the diph-
thongs /Ii, EI, aE, OE, aO, OU, Uu/ (which is similar to the Australian, mediatic Brit-
ish, and Cockney accents), as can be seen from the respective vocograms. Besides,
/È:<, È:≤/ is rounded, (ê:) (or only partially so, (+È, +‘|), in the other pronunciation
kinds).
/i/ (i) /u/ (¯)
/I[ı], I˙/ (Ù, ¢ı, ¤˙) /U[ı]/ (¨, Uı)
/O:/ (o:)
/E[ı]/ (e, ™ı)
/È:≤/ (ê:), /È/ (È, û*, å|)
/π[ı]/ (E, Äı) /√[ı]/ (å, √ı)
/A:/ (a:) /Ø[ı]/ (O, Øı)
Mediatic-accent variants
/i/ (Ii) {“ /Ii/ (¤i, Ùi)} /u/ (%)
/U[ı]/ (w, Uı)
/I[ı]/ (¢, ûı)
/È/ (È, û*, å|÷ ‘)
/E[ı]/ (¤, ¤[Ù]0ò, Ùı) /O:[ı]/ (oø, oUı)
/È:≤/ (+È, +‘|)
/Ø[ı]/ (O,
° Oı)
#
/π[ı]/ (™, ™[É]0ò, Éı) /√[ı]/ (å, √ı, ∏ı)
/A:/ (Åa)
Broad-accent variants
/Ii[ı]/ (Ùi, i¤ı) /Uu[ı]/ (¢%, ¨uı)
/OE[ı]/ (PI, PÙı)
/OU[ı]/ (x%, ÖUı, ÖUl)
/aO/ (Ex)
/EI[ı]/ (a¤, aÙı) /aE[ı]/ (ùÙ, ùÉı)
/’mÅÙ, ’mI, ’mÈ/÷ besides, /’EI, ’OU/ are (™, P) before ('): vacation˚ november; /-OU/ is
(P) in compounds when it is at the end of the first lexeme (even if separated): win-
dow sill. For ar, ç/A√/Ç (A;√) is possible; even ç/OÈ≤/Ç for /O:≤/. ≈nally, it has (é‘>é)
for /éÈ<é, é˘<é/, and (¤) for /iò, ié/: ('khA:, 'khA;√÷ 'mO;√÷ 'mE‘>¤, -R¤) /'kA:≤, 'mO:≤,
'mE˘<i/ car˚ more˚ Mary.
Triphthong attenuation is very frequent (but less than in a‡ected pronuncia-
tion or in Cockney), even between word, also for /ééò’I/: ('fÅ;√, 'fÅ:÷ ÅÙ¤~'GO¤T, Ť~-)
/'faEÈ≤, aEIn'GOEIt/ fire˚ I enjoy it. Substantially, its intonation patterns correspond
to the neutral ones; with the conclusive and suspensive intonemes, a creaky phona-
tion type is very frequent.
2.6.5. It can be flaunted by aristocrats and people of high social, religious, and
cultural standings. But it is generally thought of as too a‡ected. It is characterized
2. english 117
by more peripheral /I, U/, ¤ (I, U) (instead of (¤, ¨)); but they are more centralized
in /IÈ≤, I˘<÷ EÈ≤, E˘<÷ UÈ≤, U˘</ (Ùx, Ù√|÷ Äx, Ä√|÷ Px, P√|); it presents /éÈ<é/ (éx>é),
and (Ù) for /’iò|/, as well: ('mÄx>Ù) Mary.
‹en the phoneme /I/ is fully unstressed, it is (Ù), as /iò/ too (instead of (i)),
and, before pauses, it even becomes (É): (&v¤zÙ'b¤lÙTÙ, -TÉ); it has /-I˙/ (-Ùn, Ù˙)÷ also
/¢/ is (Ù), as many /È/ are, as well.
Besides, /E, π/ are closer (e, E); when /I, E, π/ are final, in stressed checked sylla-
bles ending in voiced C before a pause, they diphthongize as (IÙ, eÉ, EÄ): ('bIÙâ,
'weÉÊ, 'bEÄfl) big˚ web˚ bad.
/È:≤, È:<÷ √÷ A:/ are backer, (x:, √, å:) (and the last one occurs more frequently
than in neutral pronunciation); for ar, ç/A√/Ç (å;√) is possible. ˛e diphthongs are
considerably narrow and have particular timbres, manly /Ii, Uu/ (ii, uu) and /aO,
OE, OU/ (Ax, @Ù, É+) (and (‘Pı), for (Ii, ¯u, aÖ, øÙ, ‘¨) and (Ö¨ı)). For short vow-
els and for diphthongs with front first elements, there are taxophones requiring
di‡erent symbols.
f
/U[ı], ué/ (U, Uı)
/I[ı], ié/ (I[ÙÊ|], ¤ı, Ié), /’I/ (Ù)
/iò/ (Ù, É|) /O:/ (o:)
/È/ (È, û*, √|), /È≤:/ (x:)
/E[ı]/ (e[ÉÊ|], ™ı) /È≤z˙d/ (È, x|), /È≤/ (È, √|), /¢/ (Ù)
/π[ı]/ (E[ÄÊ|], Äı) /√[ı]/ (√, ∏ı)
/A:/ (å:), /Ø/ (Ø)
/Ii[ı]/ /[j]Uu[ı]/ /[j]UÈ[≤], -U˘</
([j]uu, /IÈ[≤], I˘</
(ii, iiı) (Ùx, Ù√|) ([j]Px, [j]P√|)
f ‡
F
[j]uuı)
f
(“ (jx:) (“ /[j]O:≤/ “
/’OU+/ (P) = /jÈ:≤/} (jx:) = /jÈ:≤/}
/EI[ı]/
(™Ù, ™Éı) /OU[ı]/ (É+, ‘Pı) /O:≤˙z˙d|/
/OE[ı]/ /EÈ≤, E˘</ = /O:[È]≤/
/aE[ı]/ (@Ù, @Éı) (Ä‘, Ä√|) (o:|, o:√|,
(ÅÙ, ÅÉı) /aO/ (Ax) o:[x]z˙D|)
/IiÈ≤/ /UuÈ≤/ (u:x,
(i:x, i:√|) u:√|) (“ /UÈ≤/,
(“ /IÈ≤/
(Ix, I√|)} not /O:≤/! “
(Ux, U√|)}
/EIÈ≤/ /êUÈ≤/ (É‘,
(™‘, ™√|) É√|)+/È:≤/ /OEÈ≤/
(“ /EÈ≤/} (@x, @√|)
/aEÈ≤/ /aOÈ≤/
(Å:, Å√) (A:, A√)
It has no glottalization of /p, t, k, c/ (¤ neither (ö0), nor (ô); with the only pos-
sible occurrence of (ö) for /t/, before sonants: ('skØTlÈnfl, 'skØölÈnfl) Scotland]÷ ç/tj,
dj, sj, zj, †j/Ç are fairly frequent, as in ('khwesTjÈn, -scÈn÷ 'sjuup√, 'suu-), que®ion˚
super (with (S Z, Sj Zj) only in colloquial or informal pronunciation, for /'kwEscÈn,
'sUupÈ≤/ ('khw™scÈn, 's¯upå)); /Ø;f, -†, -s, -ft, -st/ = /O:/ (o:), in addition to /Ø/ (Ø) in
some words, today, but o‡ maintains /O:/.
/</ (>, 'éRé, †R, ∑R, /</ (>, 'éRé, †R, ∑R, pR, bR, kR, gR) (also ('éVé), and even ('éeé,
118 a handbook of pronunciation
'é=é) in some frequent words: very˚ terrible˚ sorry˚ tomorrow); non-written and
non-etymological /<òé/ is frequent. Often, the çaspirationÇ of /p, t, k, c/ is very
weak (contrary to Cockney usage). In addition, it has (`) for /ò’h/ hotel÷ /·w/ (w),
('weÉn) when÷ unstressed my is /’mI, ’mÈ/÷ /-OU/ is (P) in compounds when it is at
the end of the first lexeme (even if separated): window sill.
Triphthong attenuation is extremely frequent (as in Cockney), even between
words, also for /ééò’I/: ('fÅ;√, 'fÅ:÷ ÅÙ¤~'G@¤T, Ť~-) /'faEÈ≤, aEIn'GOEIt/ fire˚ I enjoy it.
Substantially, its intonation patterns correspond to the neutral ones; with the con-
clusive and suspensive intonemes, the creaky phonation type is very frequent; para-
phonic pitch expansion is typical; in intonemes, syllables are lengthened.
/ié/ (i), /i|/ = /Ii/ (ÈÛ) /ué/ (%), /U[ı]/ (¨, UP, ’¨)
/I[ı]/ (¤, ¤P, ’¤), /¢/ (¤) /È/ (È, û*, Å|)
/E[ı]/ (™[¤0], ™P) {(™)»} /O:/ (oU) {(o:)»} {(oUÈ +
/È:≤/ (È:) “ (È‘) z/D, oUÅ|, oU>é) = /UÈ/}
/π[ı]/ (E[Ù0], ÄP) {(E)»} /Ø/ (O[ø0]) {(O)»}
/√[ı]/ (Å, aP) /A:/ (å:) {(Ø:)*]
/aOÈ≤/
/aEÈ≤/
(πX‘, πXÅ|)»
(åÙ‘, åÙÅ|)»
“ (π‘, πÅ|)»
“ (å‘, åÅ|)»
/IÈ≤/
(¤‘, ¤:ò, ¤Å|, ¤:>é)» /[j]UÈ≤/
([j]o‘, [j]o:ò, [j]oÅ|,
/EÈ≤/ [j]o:>é)» = /[j]O:≤/
(™‘, ™:ò, ™Å|, ™:>é)»
thong /aO/, quite often independently from context). For the grammeme /I˙/ we
have (In, ’n, ó); and, for -`i«˚ (-f¤˙ök) is frequent: ('mÕ3n, 'samf¤˙ök, '™n'I&f¤˙ök)
120 a handbook of pronunciation
As far as C are concerned, the most typical characteristic refers to /p, t, k, c/,
which are typically preglottalized, (ö=), even (öÒ), in all cases where in the British
accent synglottalization is possible (Ô § 2.2.6.1-2, § 2.2.7.1), or where in mediatic
British English preglottalization occurs (Ô § 2.4.3.6). Even for the phonetic reali-
zations we find some di‡erences. In fact, in the most typical and broadest pronun-
ciations, /p, t, k/ are realized as the corresponding stopstrictives: (p, ç, w), even
çaspiratedÇ (in the normal contexts expected for neutral pronunciation, too): (ph,
çh, wh) (which can give the impression of stronger çaspirationÇ). However, the
most typical and broad element is the substitution of /t, ù/ with (ö) in all the cases
seen in § 2.2.5.2, but with further typifying contexts (¤ except before a tautosyl-
labic stressed nucleus, or after pauses, or after /s/, ('çh, |çh, 'sç, sç)).
Examples: ('bÅöÅ) /'b√ùÈ</ butter˚ ('woUöÅ) /'wO;ùÈ≤/ water˚ ('whOöó) /'kØtó/ cot-
ton˚ ('s¤öó) /'sItó/ sitti«˚ ('bOöõ) /'bØùÈm/ bottom˚ ('bÈ‘öó) /'bÈ:≤tó/ Burton˚ ('ÉPöó)
2. english 121
/'Eıtó/ Elton˚ ('whl¤nöó) /'klInt˘n/ Clinton˚ ('l¤öP) /'lIù®/ little˚ ('phå;öna, -Ø;ö-)
/'pA:≤tnÈ≤/ partner˚ ('s™öp>¤ölÈÛ) /'sEp˘<Ètli/ separateI˚ ('lOös) /'lØts/ lo_˘
More examples: ('pha’nöÅ, -n,Å, -nÅ, -n[Å, -nöçÅ) /'pEInùÈ≤/ painter˚ ('pha’n-
ö¤ö, -n,¤ö, -n¤ö, -n[¤ö, -nöç¤ö) /'pEInù¢t/ paint it˚ ('woUöÅ) /'wO:ıùÈ≤/ Walter˚ ('oUö¤ö,
-,¤ö, -[¤ö, -öç¤ö) /'hO:ıù¢t/ halt it˚ ('ßçå;ö¤ö, -,¤ö, -[¤ö, -öç¤ö) /'stA:≤ù¢t/ ®art it˚ ('¤ö¤ö, '¤,¤ö,
'¤[¤ö, '¤öç¤ö) /'hIù¢t/ hit it˚ (È'l¤öP 'b¤ö È'bÅöÅ) /È'lIù® 'bIù Èv'b√ùÈ≤/ a little bit of butter˚
('ph¨ö 'Åöp) /'pUù '√p/ put up˚ ('ph¨ö¤ö 'Åöp) /'pUù¢ù '√p/ put it up˘
In less broad pronunciations, an incomplete, attenuated stop is possible: (,),
which is less çinvasiveÇ; the vocoid preceding (ö) can even be laryngealized, where-
as (ö) can become çzeroÇ, especially before another vocoid (adding, however, the
creaky phonation type), (éöé = é,é = ü,é = üé): (È'l¤P 'b¤ È'bÅÅ). Generally, forms
such as lill ('l¤P) /'lIù®/ and little ('l¤P, 'l¤P;, 'l¤PU) /'lIù®/ maintain some di‡erences
even if the latter is actually pronounced in this way; as a matter of fact, in addi-
tion to the creaky phonation type, /®/ is often lengthened (at least in an intoneme).
Before V (even if derived from /®/, and even between words), also a less broad
variant, ([), is possible (or even (öç) in çelegantÇ speech, which we do not indi-
cate). It is also possible for (n[) to become (n); here we will report the relevant ex-
amples, without spelling, following the order in which they are given above (in-
cluding water and Walter): ('bÅ[Å, 'woU[Å, 's¤[’n, 'bØ[’m, 'l¤[P), ('pha’n[Å, -nÅ÷
'pha’n[¤ö, -n¤ö), ('woU[Å, 'oU[¤ö, 'ßçå;[¤ö, '¤[¤ö, È'l¤[P 'b¤[ È'bÅ[Å, 'ph¨[ 'Åöp, 'ph¨[-
¤[ 'Åöp).
Other consonants can become (ö), especially /p, k/: ('ßçOöõ) /'stØpI˙/ ®oppi«˚
('f¤l¤ö 'lØÉöç¤ö) /'fIlIp 'laEktIt/ Philip liked it˚ ('whOönÈÛ) /'kØkni/ Co$n\˘ In a previ-
ous example, we have seen that typically /d/ becomes stopstrictive, (Ç); besides,
commonly, /éd/ is realized as (éö), when it is word-final and followed by C or V˚
and in the grammeme sequence >-dn't≥ /-dót/, as well: ('b>™öõ 'bÅöÅ) /'b<Edó 'b√ùÈ≤/
bread and butter˚ ('g¨ö 'bo;Ù) /'gUd 'bOE/ good boy˚ (ØÉ'Ǥöó{öç}) /aE'dIdót/ I didn't˘
For /st, st<, sc/, broad pronunciations have (ßç, ߲>, Sc): ('ßçÅ;Ù) /'stEI/ ®Z˚
('߲>O;˙) /'st<Ø;˙/ ®ro«˚ ('wwh™ScÈn, -có) /'kwEsc˘n/ que®ion˘ In broad pronuncia-
tions, /†, ∑/ become /f, v/; however, there are many intermediate nuances, includ-
ing the realizations of normal pronunciation: (f, v÷ ƒ, √÷ Ï, ƒ÷ fl, ∂÷ †, ∑). More of-
ten, /ò∑/ can be realized as (`, Ã, d, D): (¤s'EXs ¤z'mÕ3n, äs-, d¤s-, D¤s-) /∑Is'haOs Iz-
'maEn/ `is house is mine. As we have seen, the typical realization of /h/ is (`),
which is a stigmatized pronunciation, and therefore can lead many speakers to hy-
percorrecting: ('hÈÛö{ç}) /'Iit/ eat. For /nˆ, tˆ, dˆ/, the typical Cockney pronuncia-
tion has no /ˆ/, but, in less broad pronunciations, mediatic-like types are possible,
as well: ('nÈ;%, 'n¢;%, '~¢;%, '~T;%) /'nˆUu/ n[˚ ('çh';ßn, 'çhK;ßn, '⁄hK;ßn, '⁄hfi;ßn,
'chK;ßn, 'chfi;ßn) /'tˆUun/ tune˚ ('ÇÈ%öw, 'Ç¢%öw, 'Á¢%ök, 'ÁT%ök, 'G¢%ök, 'GT%ök)
/'dˆUuk/ duke.
Triphthong attenuation is extremely frequent, even between words, also for
/ééò’I/: ('fØ;Å, 'få;Å÷ ؤ~'Go¤T, å¤~-) /'faEÈ≤, aEIn'GOEIt/ fire˚ I enjoy it. Substantially,
its intonation patterns correspond to the neutral ones. For [n]ei`er we generally
find /Ii/.
3. Italian
3.0.1. ˛is chapter will analyze modern neutral Italian pronunciation, keeping
it separate from traditional pronunciation, which will be looked at later, in order
to demonstrate the accepted and shared di‡erence between them. However, this
di‡erence is not evident in dictionaries and grammar books, which is hardly sur-
prising given the inertia and lack of dynamism shown by schools and obviously
dictionaries and grammar books. ˛ey continue to perpetuate çconventional con-
victionsÇ, a legacy from grammar books and dictionaries from the 1800's, without
even doubting that things may have changed in the meantime… Endless plagiariz-
ing, with very few innovations as far as orthoepy (¤ pronunciation) is concerned.
˛is can be seen in the way çcàsaÇ for casa, ¤ /'kasa/ is still given. ˛is is the tradi-
tional pronunciation, while the modern one has /'kaza/. Evidently, this shows a
lack of attention for the phonic aspect of the language. It is also true that this kind
of notation can, almost systematically, escape non-attentive dictionary-users, who
can thus presume that when a dictionary reads >-s-≥ it means /z/. ˛is case is not so
serious as it coincides exactly with modern pronunciation; but what would hap-
pen for -asi in qualsiasi?
In the Zingarelli dictionary (çcarrying the yearÇ 1997), the present writer has
given two variants, indicated as cà{a (= /'kaza, -sa/), and many other words, increas-
ing the double possibilities for /e, E÷ o, O÷ q, Q/ as well, though not reaching the
variety of pronunciation in DiPI, with its modern, traditional, acceptable, tolerat-
ed, slovenly, intentional and lofty variants (explained in MaPI, as well).
However, before we even begin to deal with these çsubtletiesÇ, we will have to
discredit a good few beliefs which are part of çcommon cultureÇ, especially upheld
by teachers (even language teachers, abroad too), perpetuated by society and
schools worthy of the first millennium of our era…
Earlier we mentioned the pronunciation shown in dictionaries and grammar
books, which, instead of depicting the actual reality, as can be perceived by mere-
ly paying attention with çopen earsÇ, they are çeasilyÇ satisfied with using and reus-
ing what has always been printed, without checking to verify if it is still in current
use.
Unfortunately, the same happens in various brief guides to diction and pronun-
ciation, that continue to proliferate, regardless of objective change (that one
should be able to perceive or, above all, want to perceive). Even in books about lin-
guistics, glottology, dialectology and in historical grammar books or text books or
–even– descriptive grammars, one can find a continual perpetration, as if it were
real, of what is unproposable as far as both vowels and consonants, or word-stress
and co-gemination are concerned.
3. italian 123
resse (i compiti) çshe corrected the homeworkÇ, with /E/ (çèÇ), or in (se) fosse (vero)
çif it were trueÇ with /o/ (çóÇ), which di‡ers from (le) fosse (&avate) çthe dug
gravesÇ, with /O/ (çòÇ). ˛erefore, there are seven Italian vowel phonemes: /i, e, E,
a, O, o, u/.
Let us continue our çsearchÇ for the number of phonemes, looking at consonant
phonemes. ˛e automatic answer (from a person with normal schooling, without
appropriate reading or consideration) is sixteen: b˚ c˚ d˚ f˚ g˚ h˚ l˚ m˚ n˚ p˚ q˚ r˚ s, t˚
v˚ z. ˛e very order of letters inevitably betrays the fact that the incredible figure
of sixteen was reached by means of an arithmetical operation, using the alphabet,
instead of actual sounds.
Hence, if you take the çLatinÇ alphabet, take away the five çItalian vowelsÇ and
the five çnon-Italian consonantsÇ [j˚ k˚ w˚ x˚ y), it seems logical to get to the çsadÇ
list given above.
˛e number of phonemes in a language, however, cannot be obtained through
spelling, but through the distinctive sounds of that language. Just to begin with,
the list contains two letters which have nothing to do with Italian sounds: q is
practically useless; and h is only a diacritic, it is necessary to distinguish ha from
a, ho from o, hanno from anno, and also º /ki*/ from ci /ci»/, ªro /'giro/ from
giro /'Giro/; there is even a vowel di‡erence between ho /O*/ and o /o*/. Incidental-
ly, we can but disapprove of the çmethodÇ used by those teachers who, worried
only and exclusively about spelling (and not really understanding anything about
the pronunciation of the language!) dictate things such as */has'kritto, hannostu-
'djato/ to students, for ha &ritto˚ hanno studiato çhe/she has written, they have
studiedÇ. ˛e poor (selfish and dishonest) result of managing to obtain çcorrectÇ
written tests –to show them o‡ shamelessly– produces, instead, a much worse (and
absurd) error in making pupils believe that above all, in formal situations, you re-
ally should say such çobscenitiesÇ as */hOs'kritto/ ho &ritto, instead of the only pos-
sible and admissible version: /Os'kritto/.
3.0.4. For the moment, let us say that there are 23 consonant phonemes in Ital-
ian and they are certainly not in alphabetical order, but in a phonic order, accord-
ing to places and manners of articulation and phonation type, after having çdis-
coveredÇ them, not by graphic deduction, but by looking for opposites in mini-
mal pairs, that contain two similar words, but with one di‡erent phoneme, which
makes the meaning of the two words change, as seen with /ki*, ci»/ and /'giro, 'Gi-
ro/ (respectively º and ci, ªro and giro).
Furthermore, in the number of phonemes, we must bear in mind that the two
graphemes s and z each have (as do e˚ o) two di‡erent phonemic values, as in pre-
sento [una persona çI introduce a personÇ, /pre'zEnto/, from presentare] and presen-
to [un avvenimento çI have a presentiment of an eventÇ, /pre'sEnto/, from presen-
tire˚ ¤ pre-sentire) – in the same way we have razza (çray, spokeÇ: /'raQQa/) and
razza (çrace, breedÇ: /'raqqa/)…
We must not forget other spelling çproblemsÇ that derive from the fact that Ital-
ian represents a transformation and evolution of Latin, which had a certain num-
ber of phonemes (obviously di‡erent from Italian, both phonically and numerical-
3. italian 125
Vowels
also occur for /'e-oò/: ('ve:do, -dø) /'vedo/ vedo (but not for /'o-eò/: ('do:ve) /'dove/
dove). ‹at remains to be said, as can be guessed (and confirmation is always pre-
cious), is that this happens with a final C too: ('ri:d™R, 'ri:døn) /'rider/ rider(e)˚ ri-
don(o), ('ip:siløn, 'su:tøR) /'ipsilon, 'sutor/ ipsilon, sutor (Lat.). ˛ere are however,
other less regular cases which are dealt with in § 3.3 of MaPI.
Let us take a look at some examples of the seven vowel phonemes: ('vi:ni) /'vini/
vini˚ ('se:te) /'sete/ sete˚ ('sEt:te) /'sEtte/ sette˚ ('ra:na) /'rana/ rana˚ ('Ot:to) /'Otto/ otto˚
('sot:to) /'sotto/ sotto˚ (kul'tu:Ra) /kul'tura/ cultura. Clearly, /j, w/ are not vowels, but
(approximant) consonants, as in: ('pju) /'pju*/ più˚ ('kjE:do) /'kjEdo/ ºedo˚ ('kwa)
/'kwa*/ qua˚ ('bwO:no) /'bwOno/ buono. It is equally true that /ju, jE, wa, wO/ Â are
not çdiphthongsÇ at all (Ô § 3.1.2) but simply sequences of CV˚ like /su, tE, va, nO/,
Â.
û 3.1. Italian monophthongs.
/i/ (i) /u/ (u)
Diphthongs
3.1.2. Italian grammar books put a lot of e‡ort into complicating what is, in
fact, quite simple. Indeed, instead of the three very common structures, ¤ the re-
al diphthong (('éé, &éé, ’éé)), the hiatus ((é'é, é&é)) and the heterophonic sequence
((0é)), ™, (jé), (wé), and the like), they continue to consider only two of them:
çdiphthongÇ (with fusion: çsyneresisÇ) and çhiatusÇ (with separation: çdieresisÇ),
but with strained interpretations of medieval origin, of a graphic-grammatical and
graphic-metric nature. In fact, çsemi-vowelsÇ or çsemi-consonantsÇ do not exist:
they are merely an çincredibly successfulÇ magic trick!
As a matter of fact (unless one expects to do çmagicÇ in phonetics using graphic-
-grammatical categories), it is phonetically absurd to speak about a çdiphthongÇ
for (—jé, —wé) (('pjE:no) /'pjEno/ pieno˚ ('gwan:to) /'gwanto/ guanto]˚ as only (—éi,
—éu) (('fai) /'fai/ fai˚ ('pa;uza) /'pauza/ pausa] are real diphthongs, as any sequence
of ('éé, &éé, ’éé) (('a;uto) /'auto/ auto˚ (au'tEn:tiko) /au'tEntiko/ autentico]˘
It is equally absurd to speak about çhiatusÇ for ('ié, 'ué), as only (i'é, u'é) are re-
al hiatuses, as any other sequence like (é'é, é&é) ((pa'u:Ra) /pa'ura/ paura]˚ compared
to (pau'ro:zo) /pau'rozo/ pauroso˚ a real diphthong. Hence, one cannot believe liter-
ature that only uses two categories (¤ diphthong and hiatus) and, what is more,
they dangerously mix them up to a point where they include –in çdiphthongsÇ–
the heterophonic sequences (/jé, wé/), and –in çhiatusesÇ– the real diphthongs (as
in ('ma;i, 'ca;o, 'bO;a) /'mai, 'cao, 'bOa/ mai˚ ciao˚ boa… – Ô § 1.4.4).
3. italian 127
3.1.3. ˛us, in languages like Italian, diphthongs are vowel sequences, which
are more or less common, formed by a combination of the seven vowel phonemes
and nine taxophones (/i, e, E, a, O, o, u/ (i, e, ™, E, a, O, ø, o, u)), that writing tidi-
ly reproposes without the problems presented –for example– by Germanic lan-
guages. ˛ese really do have monophonemic diphthongs, because, in di‡erent ac-
cents they vary as to their realizations, independently from monophthongs (and
more than monophthongs do), and also because they have varied historical spell-
ings.
˛erefore for Italian, it seems pointless –in this chapter– to make a list of diph-
thongs, of very di‡erent frequency; it would be appropriate however, to look at §
8.26 of NPT/HPh (as well as § 5.1.2-3 of MaPI]˘
By forcing (and violating) reality, grammar and metrics continue to call, above
all, /jE, wO/ çascendant diphthongsÇ, because they derive from the Latin /E, O/ (™,
ø), and they find they need to have to çinventÇ (as the best lawyer for the worst
criminals) fictitious realities, like çsemi-consonantsÇ and çsemi-vowelsÇ.
˛ese seem to be nearly cinematographic special e‡ects, that make one believe
one is facing something real, but which is completely invented! If phantaphonet-
ics –or virtual phonetics– is not our aim, then sequences like (jé, wé) are –natural-
ly– part of /{0}0é/ groups, as in /'fjanko, 'franko÷ 'gwado, 'grado÷ 'twOno, 'trOno÷
'gjande, 'grande, 'glande/ fianco˚ franco÷ guado˚ grado÷ tuono˚ trono÷ ªande˚ gran-
de˚ ∫ande. Paradigmatically, /j, w/ are in opposition with /0/, clearly not with /é/
and can under no circumstance belong to vowel groups.
Even çsemi-vowelsÇ are a truly distorted reality and have been invented to try to
explain (but they are only deceived into trying to explain) what has no need of ex-
planations. Indeed, what need do normal diphthongs, like /ai, au/, have for alibis
to defend themselves from grammatical and metric fanaticism, so as to demon-
strate that they constitute one syllable and not two: ('ma;i, 'ka;u{to}) /'mai, 'kau{to}/
mai˚ cau(to)?
Even /ia/ is a –monosyllabic– diphthong, as in ('mi;a) /'mia/ mia. It is evident,
in the same way as the Earth is round and rotates around the Sun; and yet, it has
not been at all easy for it to be accepted…! ˛e diphthong, hiatus, and /CV/-se-
quence matter, is much simpler: one does not have to be a scientist; all it takes is
observation and thought (using both ears)! Yet… there is no change! Let us trust-
ingly refer to § 5.1.2-3 of MaPI as well.
Consonants
3.2.0. û 3.2 shows the table of the neutral Italian consonant articulations, in-
cluding their taxophones ((M, ~, ˙÷ R÷ ¬)), which are necessary for satisfactory pro-
nunciation.
Instead, û 1.9-15 gives orograms, grouped by manners of articulation of all con-
toids given in the chapters of this volume, even as secondary, occasional or region-
al variants of the 12 languages dealt with.
128 a handbook of pronunciation
atal protruded
velar rounded
postalveo-pal-
labiodental
postalveo-
-palatal
alveolar
bilabial
palatal
dental
velar
ö m (M) (n) n (~) N (˙)
F pb t d k g
Ô qQ c G
ƒ f v
_ s z S {Z}
ß j w
ó r|(R)
‹ (l) l (¬) L
Nasals
3.2.1. ˛ere are three nasal phonemes, /m, n, N/ (m, n, N), and four taxophones,
for /n/, that can be rendered by three supplementary symbols (M, ~, ˙) (a fourth
symbol, ((˙)), in front of /t, d÷ q, Q÷ s/, can be useful to bring attention to the as-
similation, from alveolar to dental, even if the di‡erence, though real, is less im-
portant than in other cases): ('mam:ma) /'mamma/ mamma˚ ('ma:ma) /'mama/ m'a-
ma˚ ('nOn:no) /'nOnno/ nonno˚ ('nO:no) /'nOno/ nono˚ ('baN:No) /'baNNo/ ba√o˚ (peR-
'NOk:ki) /per'NOkki/ per √ocº˚ ('toM:fo) /'tonfo/ tonfo˚ ('dEn:te) /'dEnte/ dente˚
('fra~:Ga) /'franGa/ frangia˚ ('fa˙:go) /'fango/ fango.
In neutral Italian, /N/ is self-geminant (Ô § 3.3.1.5.): ('soN:No, loN'NO:mo) /'soNNo,
loN'NOmo/ so√o˚ lo √omo. We can observe that the self-geminant /N/ is in opposi-
tion to geminates, as in ('soN:No) /'soNNo/ so√o and ('son:no) /'sonno/ sonno, but
not simple C (('so:no) /'sono/ sono); the minimal pair is between the first two exam-
ples, not the third.
In order to closely examine the manifestation of nasal assimilation in neutral
Italian, a whole series of nouns could be shown (beginning with all possible C, Ô
§ 3.4 of MaPI]˚ preceded by (kon) /kon/ con çwithÇ, but we will limit ourselves to
(kom'mar:ko) /kom'marko/ con Marco and to the improbable (koN'NE;o) /koN'NEo/
con Gneo (a rare male name, even in ancient Rome).
Various problems (not only with consonants, but with vowels and intonation,
too) deriving from various regional pronunciations are examined in some chap-
ters of MaPI, as well. We can now briefly mention the fact that, very often, in the
north of Italy, the sequences of /n0/ are not homorganic by assimilation, but have
a velar articulation, (˙), or semi-prevelar (without full contact), («), as in (&u˙ba˙-
'bi:nø, &u«ba«'bi:nø) (concentrating, here, only on the nasals, because the actual re-
gional di‡erences, doubtlessly, also regard other C, many V˚ the type of phonation,
the syllabic structure and the intonation), for neutral (&umbam'bi:nø) /umbam'bi-
no/ un bambino.
In the same way, in the north, /N/ is very often not self-geminant: ('sO:No, 'so:No÷
3. italian 129
lo'NO:mo); but, generally, /N, nj/ are confused with something intermediate, which
we show quite generally here: (ka˙'pa;~ja, ka«-) which stands both for /kam'panja/
Campania and for /kam'paNNa/ campa√a. In central areas, /nj/ becomes (N) (but
in this case not self-geminant): (an'tO:No) /an'tOnjo/ Antonio˚ (aN'NEn:to) /an'njEnto/
anniento˘
Stops
3.2.2. Italian has three diphonic pairs of stops, /p, b÷ t, d÷ k, g/ (p, b÷ t, d÷ k, g):
(peR'bE:ne) /per'bEne/ per bene˚ (ti'dO) /ti'dO*/ ti do˚ (ko˙'gwaL:Lo) /kon'gwaLLo/
congua˙o. Although before a front V –and in front of /j/– /k, g/ are realized as pre-
velar, it is not necessary to systematically use the special symbols ((´, Ò)): (&kjakkje-
'ra:Re) /kjakkje'rare/ ºacºerare˚ ('kik:kø) /'kikko/ ºcco˚ ('gE˙:ga) /'gEnga/ ghen-
ga.
˛e greatest regional problem concerning stops is provided in the çgorgia tosca-
naÇ (¤ çTuscan throatÇ) and in the çcentral-southern voicingÇ, which will be exam-
ined in general terms (also because there are di‡erences between areas; but for
more precise details, various chapters of MaPI can be consulted). ˛erefore, the
following gives just a general idea: (&ihaFi'Ïa:ni) /ikapi'tani/ i capitani and ('ci˙âwe
im'Êun:∂ø) /'cinkwe im'punto/ cinque in punto˘
Stopstrictives
3.2.3. ˛ere are two diphonic pairs, dental and postalveopalatal protruded, /q,
Q÷ c, G/ (q, Q÷ c, G): ('dan:qa, 'pjaq:qa) /'danqa, 'pjaqqa/ danza˚ piazza˚ ('QO:na,
aQ'QO:to) /'QOna, aQ'QOto/ zona˚ azoto˚ ('ce:na, 'fac:co) /'cena, 'facco/ cena, fac-
cio˚ ('GEn:te, 'OG:Gi) /'GEnte, 'OGGi/ gente˚ oggi.
˛e main regional problems regarding the north concern the sequential realiza-
tions of /q, Q/, as: ('da˙tsa, -fia, -†a, 'da«-÷ 'pja{t}tsa, -ßa, -†a÷ d'zO:na, d'fi-, d'∑-÷ ad'zO:-
to, ad'fi-, ad'∑-) (or, in broader accents, simply as constrictive: ('da˙sa, -ßa, -†a, 'da«-÷
'pjassa÷ -ßßa, -††a÷ 'zO:na, 'fi-, '∑-÷ a'zO:to, a'fi-, a'∑-)), instead of real stopstrictive
phones, (q, Q). For almost the whole center and many parts of the south, with-
out going into too much detail, the most obvious characteristic is given by the
change of /écé/ to its corresponding constrictive, (éSé): ('pe:Se) /'pece/ pece˚ (&diSe-
'Si:lja) /dice'cilja/ di Cecilia˚ ('ka:So) /'kaco/ cacio. In Tuscany the same goes for /é-
Gé/ = (éZé): ('a:Zile, la'Zak:ka) /'aGile, la'Gakka/ agile˚ la giacca.
It is important to focus on the fact that we are dealing with single postvocalic
/c, G/ (and hence, in this case, intervocalic too), thus avoiding ridiculous imita-
tions which have, unfortunately ended up amongst examples shown in books
about linguistics, dialectology, language history, language teaching and sometimes
even in… phonetics and phonology!
˛erefore, it is legitimate to give (la'Zi:†a) /la'Gita/ la gita, as an example (al-
though the o‚cial symbol (†) is hardly recommendable), but not simply *('Zi:†a)
130 a handbook of pronunciation
/'Gita/ gita, because after silence or after C, it stays for (G), ('Gi:†a), as does, obvi-
ously, (aG'Gi:na) /aG'Gina/ a Gina (because of co-gemination, Ô § 3.3.2.1-3; cer-
tainly not *(aZ'Zi:na)). Naturally, the same is true of (la'Se:na) /la'cena/ la cena˚
while we can only have: (peR'ce:na, ac'ce:na) /per'cena, ac'cena/ per cena˚ a cena
(absolutely not *(peR'Se:na, aS'Se:na))!
Apart from Tuscany (and some northern parts of Umbria and the Marches), in
the center and south (still generalizing the area and the precision in transcription,
too), we have /éGé/ = (éGGé): ('aGGile, laG'Gi:∂a)…
Constrictives
3.2.4. Neutral Italian has two diphonic pairs of constrictives, /f, v ÷ s, z/ (f, v÷ s,
z) as well as a voiceless postalveopalatal protruded self-geminant /S/ (S) (Ô §
3.3.1.5.), with the voiced correspondent, but non-geminant (Ô the beginning of
the same paragraph), in foreign words, above all French: ('fa:va) /'fava/ fava˚ ('sO:-
zja) /'sOzja/ sosia˚ ('peS:Se, loS'Sa:me, {laS}'SE:na) /'peSSe, loS'Same, {laS}'SEna/ pe&e, lo
ßame, (la) &ena˚ (&aba'Zu:r, &aZi'go) /aba'Zur, aZi'go/ abat-jour˚ à gigot˘
In front of any voiced C there is always /z/ in Italian: (biz'bE:tiko, zve'ni:R™, zle-
'ga:Re, zden'ta:to) /biz'bEtiko, zve'nire, zle'gare, zden'tato/ bisbetico˚ svenire˚ slegare˚
sdentato (Ô (sten'ta:to) /sten'tato/ stentato]˘
As far as VsV is concerned, modern neutral pronunciation resolves the problem
of the traditional one (Ô § 3.3.5.2.); actually, every postvocalic intralexemic -s- (¤
in simple words, not compound] is voiced, /z/ (z): ('ka:za, 'ri:zø, 'pre:za, ci'ne:ze, Ge-
'lo:zo) /'kaza, 'rizo, 'preza, ci'neze, Ge'lozo/ casa˚ riso˚ presa˚ cinese˚ geloso˘ Only in
compounding, is the initial -s- of lexeme or grammeme /s/ (s): (&pR™ndi'so:le, &sotto-
&segRe'ta:Rjo, &pResa'la:Rjo, kwal'si;asi, Ri'sal:to) /prEndi'sole, sottosegre'tarjo, presa'la-
rjo, kwal'siasi, ri'salto/ prendisole˚ sottosegretario˚ presalario˚ qualsiasi˚ risalto˘ ˛e
last example means çI jump againÇ; whereas, the noun means çto stand outÇ and
is (Ri'zal:to) /ri'zalto/… Forms such as prosieguo, whose composition –nowadays–
is heard in a less definite way, vary: (pRo'sjE:gwo, -z-) /pro'sjEgwo, -z-/.
For /s, z/, in the north, an alveolar instead of dental articulation is frequent; ('ßO:-
fija) /'sOzja/ sosia (be warned that, too often, neutral Italian articulation, which is
dental {with lowered tip}, is defined çalveolarÇ, even in phonetic books!).
In central-southern regional pronunciations (except in Tuscany), for VsV (even
if with an approximant inserted after s), we do not have /z/, but /s/: ('rO:se, 'a:sja)
/'rOze, 'azja/ rose˚ Asia. However, the prestige of voiced pronunciation, with /z/, is
such that, many speakers try to go from (s) to (z); but from a regional situation
without /z/ (and with no help from the written word), the substitution often gen-
eralizes (z) (or (Ω)), even in words like: *(&laza'lu:t™, &laΩa-) /lasa'lute/ la salute, that
is (&lasa'lu:t™).
In the center (except ∫orence and Prato) we find the typical change /s/ = (q) af-
ter /n, r, l/: /'pEnso, 'pErso, 'falso/ ('pEn:qo, 'pEr:qo, 'fal:qo) penso, perso, falso. The
same happens in Tessin and in northern and easter Lombardy. It is very frequent
also in some areas of southern Italy, including its partial or complete voicing: (ë,
3. italian 131
Approximants
3.2.5. ˛e two Italian approximants are /j, w/ (j, w): ('jE:Ri) /'jEri/ ieri˚ (gjac'ca:-
jo) /gjac'cajo/ ªacciaio˚ ('wO:vo) /'wOvo/ uovo˚ (kwa'lu˙:kw™) /kwa'lunkwe/ qua-
lunque˘
In central-southern pronunciation, we sometimes find /j/ = (jj): ('paj:jo, 'pa;jjo)
for ('pa:jo) /'pajo/ paio˘ In the south we often have even (i'e:Ri, u'o:vø) for /'jEri, 'wO-
vo/ ieri˚ uovo (simplifying a little).
Trills
3.2.6. ˛ere is only one trill phoneme, /r/, with two important taxophones, (r,
R), which depend on the strength of the syllables; in stressed syllables, before or af-
ter the syllabic nucleus we find a trill, (r) (lengthened, (r:), if in a checked syllable
of an intoneme); whereas in unstressed (or half-stressed) syllables we have a tap:
('ra:Ro) /'raro/ raro˚ ('pOr:ta) /'pOrta/ porta˚ (Ri'prE:ndeRe) /ri'prEndere/ riprendere˚
(&pRepa'rar:si) /prepa'rarsi/ prepararsi˚ ('kar:Ro) /'karro/ carro˚ (aR'ri:vø) /ar'rivo/ arri-
vo˚ (&aRRi'va:Re) /arri'vare/ arrivare˘ As one can see, even for /rr/, the choice between
(r, R), depends on the position of the stress.
˛ere are various types of accents with a çFrench rÇ, in Italian, that range from
uvular, like (K, º, ˜, r), to labiodental (with or without uvularization), (V, ◊).
For regional pronunciations (as shown in MaPI), we could have a generalized
132 a handbook of pronunciation
(R) or (r), independently of the stress; we could also have uvularization, (5, R); or
in Venice, (¸, Í, ®).
Let us take a detailed look at the çSicilian typeÇ (in Sicily, of course, and in
southern Calabria and the Salentina Peninsula as well) for /òr, rr, tr, dr, str, sdr/.
Indeed, simplifying a little (even the transcriptions) for /òr, rr/, we have the voiced
slit alveolar constrictive, (z), self-geminant if postvocalic: ({laz}'za:djo) /{la}'radjo/
(la) radio˚ ('kaz:zø) /'karro/ carro˘ It is clearly not the postalveolar trill phone ((#),
çcacuminalÇ, çó¶), as it has been çdescribedÇ for generations, without proof.
Relative to this, we also have the çbizarreÇ pair ç=ó, ƒóÇ which is persistently pre-
sented as ç(˛#, Ã#)Ç, without listening carefully. In fact the most normal articula-
tion is given, simply, by the stopstrictives which correspond to (z), ¤ (., …) (slit
alveolar): ('.a:v™) /'trave/ trave˚ ('…a:gø) /'drago/ drago˘
Lastly, the /str/ sequence (lexically quite common), is more often rendered by
the voiceless postalveo-prevelar constrictive (self-geminant, if postvocalic) (.): ('na.:-
.ø) /'nastro/ nastro˚ ({la.}'.a:da) /las'trada/ (la) strada; instead, the (less common)
sequence /zdr/ is more often rendered by a voiced (grooved + slit) alveolar se-
quence, (fi…): ({la}fi'…a:jø) /{la}z'drajo/ (la) sdraio (much less commonly by (la{÷}-
'÷a:jø)).
We will conclude by indicating the assimilation of /r0/, which is more typical
of eastern Sicily: ('pat:tø) /'parto/ parto.
Laterals
3.2.7. Italian has two lateral phonemes, /l, L/, alveolar and palatal (this last is
quite rare in languages, and – in neutral Italian – it is self-geminant, as well): ('la:la)
/'lala/ l'ala˚ ('lal:la) /'lalla/ Lalla˚ ('dir:Li) /'dirLi/ dir˙˚ ('fOL:La) /'fOLLa/ fo˙a.
We can observe how the self-geminant /L/ opposes to geminates, as in ('paL:La)
/'paLLa/ pa˙a and ('pal:la) /'palla/ palla, not to (('pa:la) /'pala/ pala); the minimum
pair is between the first two examples not the third, which is a single consonant.
In some central areas, /L/ becomes /j/, or /lj/ becomes (L) (in this case it is not
self-geminant): ('paj:ja, 'pa;jja) /'paLLa/ pa˙a˚ ('O:Lo) /'Oljo/ olio˚ (&paLLa'ti:vø) /pal-
lja'tivo/ palliativo˘ On the other hand, in the north, generally, /L, lj/ are mixed in-
to something intermediate that we can render quite generally here: (li'ta;¬-ja) that
stands for /li'talja/ l'Italia and for /li'taLLa/ li ta˙a. Neutral Italian has /l/ = (¬) +
/c, G, S/: ('a¬:ce) alce˘
Structures
3.3. We will now look at the characteristics which go beyond simple segments,
those with syllable, length, stress, and intonation relevance.
˛e intermediate vocoids, (™, ø), were mentioned in § 3.1.1; for assimilations
and consonant taxophones, the relevant characteristics are shown, including the
most common regional peculiarities, in their respective paragraphs.
3. italian 133
It has already been stated that in Italian, diphthongs are biphonemic and are
formed by joining two vocalic phonemes/phones, retaining their normal typical
realizations, with no modification, in particular of the second element, that is al-
ways distinct, even in the case of /ai, au/ (ai, au), contrarily to many languages,
above all Germanic (Ô § 8.26-8 of NPT/HPh).
Taxophonics
3.3.1.1. It is necessary and useful to speak out against the widespread practice,
led by journalists and unrefined publishers, of yielding to laziness of various forms
regarding the spelling of grammemes, ™: la˚ una˚ della…, -re (in infinitives) –we
refer to the linguistic convention, that places an asterisk (*) in front of undesir-
able forms, which report a decidedly inaccurate use– like in *la esattezza, *una op-
portunista, *della università, *voltare pagina… for the more usual and definitely
more harmonious: (&lezat'teq:qa, &unop&poRtu'nis:ta, del&luni&veRsi'ta, &voltaR'pa:Gina)
/lezat'teqqa, unopportu'nista, delluniversi'ta*, voltar'paGina/ l'esattezza, un'oppor-
tunista, dell'università, voltar pagina…
As a matter of fact, Italian is a real native language only in central Italy (where
local dialects present substantially, the same structure as the Italian language, on-
ly with di‡erences of register); ¤ in Tuscany, Umbria, the Marches and Latium (al-
though their linguistic borders do not coincide exactly with administrative bor-
ders, but are a little less widespread, for the four regions and above all for the
Marches, as even the whole province of Pesaro and Urbino do not come into the
çlinguistic CenterÇ).
In the center, the dropping of grammeme endings [-i˚ -e˚ -a˚ -o {in clear phonic
order}), even for nouns, adjectives, conjunctions, adverbs and verbs, is a normal
and natural procedure, as seen in: ancor indietro˚ sempr'avanti˚ er'andato˚ quant'al-
tri mai˚ cinqu'anni˚ or'ott'e trenta (o&Røttet'tren:ta), Dant'Aliªeri˚ Â.
For words in connected speech (unless there are particular reasons, such as high-
lighting, or emphasis, which should be present or risk lacking expressiveness), neu-
tral pronunciation resorts to elision, ¤ the dropping of one of two identical V (in-
cluding the dropping of the final V of the articles and some pronouns {considered
by grammar, in front of even di‡erent V}): (lat'te:za, &unat&tivi'ta, le&zeRcitaq'qjo:ni,
aL&Lita'lja:ni, su&naRgo'men:to÷ la'mi:ka, u'nal:tRa, u'nal:tRo, lin'seN:No) /lat'teza, unat-
tivi'ta*, lezercitaq'qjoni, aLLita'ljani, sunargo'mento÷ la'mika, u'naltra, lin'seNNo/
l'attesa˚ un'attività˚ l'esercitazioni˚ a∫'italiani˚ s'un argomento÷ l'amica˚ un'altra˚
l'inse√o.
If one is not influenced by spelling without elision, and one is, at the same time,
free from non-central and non-neutral sounds, one can easily see that normal pro-
nunciation really is not: *(laat'te:za, &unaat&tivi'ta, le&ezeRcitaq'qjo:ni, &aLLi&ita'lja:ni,
su&unaRgo'men:to÷ laa'mi:ka, &una'al:tRa, liin'seN:No), even for the words written as:
le esercitazioni, a˙ italiani, su un argomento, li inse√o.
For this reason, it is more advisable to use writing which accurately considers
these facts, explicitly showing how closely linked genuine pronunciation and spell-
134 a handbook of pronunciation
3.3.1.2. Amongst the bad habits we have just examined, unfortunately we can
find, propagated by schools, the (humiliatingly mechanical, ¤ without the slight-
est consideration, resulting in a damaging conditioned reflex) use of the notorious
çeuphonic dÇ –¤ çwell-sounding dÇ, however, the only euphonic part of it is its
high-sounding name– as in: *ad Emilia, *ed an>e, *od altro, (not to mention: *ed
educazione, *ad Adele, *od odore!).
Once more, the central way of speaking, which is not ruined by schools or jour-
nalism, spontaneously and correctly, gives: (ae'mi:lja, e'a˙:ke, o'al:tRo) /ae'milja, e-
'anke, o'altro/ a Emilia, e an>e, o altro˚ and even: e educazione, a Adele, o odore
(not including the really çcacophonous dÇ, forming sequences such as ç/édédé/Ç).
Contrarily, neutral language, prefers cases such as: (&ada˙'ko:na, e&deve'li:na, &o-
dol'fat:to) /adan'kona, edeve'lina, odol'fatto/ ad Ancona˚ ed Evelina˚ od olfatto.
However, we obviously have: (e'E:va, o'Ot:to) /e'Eva, o'Otto/ e Eva˚ o otto, as the V˚
in these cases are not the same at all (phonically speaking)!
˛e cases of the preposition da and of pronouns una and uno as well followed
by an adjective are clearly di‡erent. ˛ey must never be elided (or çtruncatedÇ for
uno), as they are necessary to maintain an important distinction between the pre-
position di and the articles (written una˚ uno as well): (dan'da:Re) /dan'dare/ d'an-
dare is only çdi andareÇ, not çda andareÇ; in the same way, (u&nita'ljana, &unan'ti:kø)
/unita'ljana, unan'tiko/ un'italiana˚ un antico are nouns; hence, di‡erent from: (&u-
na&ita'ljana, &unoan'ti:kø) /unaita'ljana, unoan'tiko/ una italiana˚ uno antico, as in:
ne cerco… (for instance: canzone, mobile) çI'm looking for an Italian song/piece of
furnitureÇ.
A last reflection, linked to spelling and school wrongdoings, has to be cast re-
garding the absurd complication of insisting on teaching çthe exception to the
ruleÇ, where the pronoun sé is written with an acute (accent) in order to di‡er
from se (conjunction and also allotrope, or variant, of sé in front of other weak pro-
nouns or ne): (&sevveR'ra;i, &sela'ri:d™, &sene'va) /sevver'rai, sela'ride, sene'va*/ se verrai˚
se la ride˚ se ne va˘
It is not at all çclearÇ why, once the spelling rule is formulated, to distinguish sé
from the other se, one must rack one's brains to excogitate the çexceptionÇ to sé
stesso, sé medesimo… which some grammars –and many pedants– would rather we
wrote *se stesso and *se medesimo, with no good reason to do so. Indeed, we can-
not exclude the possibility of finding (even in an ambiguous initial position) sen-
tences such as: Se stessi male, non potrei venire çIf I were ill, I couldn't comeÇ, which
is di‡erent from: Sé stessi malediranno ˙ azzeccagarbu˙ della grammatica ç˛e
pettifogging lawyers of grammar will curse themselvesÇ; or: Se medesime compa√e
3. italian 135
di gioco perdono, vengono e&luse… çIf equal playmates lose, they are excludedÇ,
di‡erent from: Sé medesime comparano a tutte le altre ç˛ey compare themselves
to all the othersÇ.
˛erefore, eliminating the accent from sé is not only utterly pointless, but caus-
es awkward ambiguity, as well as undue analogies that lead to many people writ-
ing *a se stante, instead of the only form possible: a sé stante çseparate, apartÇ.
Let us briefly turn our attention to the çimproperÇ use (decidedly incorrect –
proudly lead by a fair number of çintellectualsÇ) of *e non˚ *o non, as in: *italiani
e non. In Italian, the negative non /non/ çnotÇ has to be followed by the term
which is denied, as in: giallo, non verde; cotto, non crudo; un ºlo, non due (ºli);
coßa, non petto; essere, o non essere? Instead, the absolute negation (hence, final in
the sentence) is only no /'nO*/ çnoÇ (Eng. /'nOU/ ('n‘;¨, 'nø;¨)) – even if the given
term is not expressed or repeated. ˛erefore, logically, we must even have: italiani
e no. Is it possible to imagine an Italic Hamlet saying *Essere, o non? – or, *Io ti a-
mo, ma tu non!? (for correct English çTo be, or not to be?Ç, of course, and çI love
you, but you don'tÇ). Such usage is much worse than English cases like: *I don't
want no bread˚ *We ain't going (no more)˚ *She don't love you (no more)˘
˛is aspect can be hard to grasp, especially if accurate transcriptions are not
availed of (and obviously, if no keen listening is made); so it is quite a common
problem for many actors, dubbers and presenters of northern descent, who –as
çvoice professionalsÇ– are somewhat lacking; unfortunately they are more like
çsemi-professionalsÇ…
As seen, phonically we have (bas-'ta:-Re) /'bas-ta-re/ bastare, but also (las-'tO:-Rja)
/las-'tO-rja/ la storia÷ therefore, likewise (s'tO:-Rja) /s'tO:rja/ storia˚ with (s't) /s't/, as
–when there is a V in front– the phono-syllabification is (s-'t) /s-'t/, as in the second
example [la storia]˘ Even acoustic data confirm the fact that (|s'té) /|s'té/ (after a
pause, or çsilenceÇ) is part of the same syllable (a little particular, possibly, on the
scale of syllabicity, but nothing really surprising) whereas, obviously, (és'té) /és'té/
constitute two phono-syllables bordering between two C (Ô § 12.2-6 of NPT/HPh]˘
3.3.1.5. However, in neutral Italian, not all C are geminable; in fact, /z, j, w/
are always single (¤ non-geminable, as also the xenophoneme /Z/, which has been
integrated into the Italian phonological system for centuries now): ('pO:za) /'pOza/
posa˚ ('kwO:jo) /'kwOjo/ cuoio˚ (a'Zu:r) /a'Zur/ à jour˘ Nevertheless, phonetically, in
the appropriate contexts, /z/ is lengthened: ('riz:ma) /'rizma/ risma.
˛ere are also five self-geminant C (/N, S, L÷ q, Q/), which, in a postvocalic posi-
tion, are necessarily geminated; but with no chance of phonological opposition
with a single C: ('baN:No) /'baNNo/ ba√o˚ ('laS:So) /'laSSo/ laßo˚ ('vOL:Lo) /'vOLLo/ vo-
˙o˚ ('viq:qi) /'viqqi/ vizi˚ ('viq:qjø) /'viqqjo/ vizio˚ (oQ'QO:no) /oQ'QOno/ ozono˘
˛e reason for this is to be found in Latin; as a matter of fact, these five phonemes
were not part of that language, and generally derive from two or more C (or from
other languages); therefore, by assimilation, the result is a geminate C (Ô § 5.6.1-
7 of MaPI “ § 1.6 of DiPI]˘
We can observe that (only) phonetics can supply explanations about the gram-
mar rules (which are çtroublesomeÇ for foreigners and for northern Italians), as the
articles (&uno, lo, Li) /&uno, lo, Li/ uno˚ lo˚ ˙ must be used in cases such as: (&unoS-
'Se:mø, loQ'Qa;ino, LiQ'Qi;i, LiN'NOk:ki) /unoS'Semo, loQ'Qaino, LiQ'Qii, LiN'NOkki/
uno &emo˚ lo zaino˚ ˙ zii˚ ˙ √ocº (as also (&unos'trac:co, los'kOp:pjo, Lis'kOL:Li)
/unos'tracco, los'kOppjo, Lis'kOLLi/ uno straccio˚ lo &oppio˚ ˙ &o˙). In fact, it is a
question of two –heterosyllabic– consonants, whereby the neutral Italian structure
could not tolerate more complicated sequences, as they would make these sylla-
bles too heavy to pronounce.
˛e remaining 15 C, /m, n÷ p, b, t, d, k, g÷ c, G÷ f, v, s÷ r÷ l/, in a postvocalic po-
sition, can be single or geminate, distinctively, as shown in the example chosen:
('va:no) /'vano/ vano and ('van:no) /'vanno/ vanno˚ ('fa:to) /'fato/ fato and ('fat:to)
/'fatto/ fatto˚ ('mO:Go) /'mOGo/ mogio and ('mOG:Go) /'mOGGo/ moggio˚ ('be:ve)
/'beve/ beve and ('bev:ve) /'bevve/ bevve˚ ('ka:Ro) /'karo/ caro and ('kar:Ro) /'karro/ car-
ro˚ ('pa:la) /'pala/ pala and ('pal:la) /'palla/ palla˘ We can see that ('ka:za) /'kaza/ casa
and ('kas:sa) /'kassa/ cassa do not constitute a minimal pair, in modern neutral pro-
nunciation (but only in traditional pronunciation or… as far as spelling is con-
cerned).
/'pane/ pane˚ ('ve;Ido, 've;edo) /'vedo/ vedo˚ ('do;Uve, 'do;ove) /'dove/ dove˚ for the
neutral ('pa:ne, 've:do, 'do:ve)…
In the north, geminates, do not often have su‚cient length, even where vowel
length does not present the lengthening mentioned at the beginning of this sec-
tion: (af'fatto, af'fa;tto) /af'fatto/ a‡atto.
In the central-southern areas, there can be other self-geminant consonants (Ô §
3.3.1.5.); which takes place more widely for /b, G/ – in parts of the centre for /j/
too: ('rOb:ba, 'rO;bba) /'rOba/ roba˚ (lab'bar:ka, lab'ba;rka) /la'barka/ la barca˚ ('viG:-
Gil™, 'vi;GGil™) /'viGile/ vigile˚ (laG'Gak:ka, laG'Ga;kka) /la'Gakka/ la giacca˚
('nOj:ja, 'nO;jja) /'nOja/ noia˚ (daj'jE:Ri) /da'jEri/ da ieri˚ for the neutral ('rO:ba, la'bar:-
ka, 'vi:Gil™, la'Gak:ka, 'nO:ja, da'jE:Ri).
In Rome (and other central areas), instead, we can find the degemination of /r/,
in broad and typical accents: ('ko:ReRe) /'korrere/ correre˚ (e'ro:Re) /er'rore/ errore˚ (a-
'ro:ma) /ar'roma/ a Roma (with co-gemination, Ô § 3.3.2.1.), for the neutral ('kor:-
ReRe, eR'ro:Re, aR'ro:ma).
Co-gemination
Stress
3.3.3.2. Going back to the examples of the rhythm groups given above, we can
observe that unstressed syllables (or rather, with weak stress), and those with half-
-stress (¤ with mid or secondary stress), alternate in such a way as to have one or
two weak phono-syllables amongst others with secondary (or primary) stress.
Secondary stress is assigned rhythmically, bearing in mind –as far as possible–
the position of the stress in the original forms (¤ fundamental forms, from which
they derived {certainly not in an evolutive sense, from Latin}), as occurs in lexical
compounds: (&pøRtasa'po:ne) /pOrtasa'pone/ portasapone˚ (pu&liSSis'kar:pe) /puliSSis-
'karpe/ puliß&arpe.
In derivatives, instead, the origin counts for less than rhythmic reasons, even if,
in case of a double possibility (rhythmic or original), the stress of the original form
can exert a decisive influence: (&pata'ti:n™) /pata'tine/ patatine (notwithstanding
(pa'ta:ta) /pa'tata/ patata), (&atten'qjo:ne) /atten'qjone/ attenzione (notwithstand-
ing (at'tEn:to) /at'tEnto/ attento]˘
For syllables that follow a stressed syllable of a word, furthermore, they behave
as follows: ('fab:bRo) fabbro, ('fab:bRika) fabbrica, ('fab:bRika&no, 'fab:bRika&mi) fab-
bricano˚ fabbricami, ('fab:bRikame&lo) fabbricamelo (let us also add a rather improb-
able ('fab:bRika&mice&lo) fabbricamicelo – Ô the end of § 6.4.2 in NPT/HPh]˘
3.3.3.3. As regards the syllables of a word that precedes a stressed syllable, the
spontaneous formation of a rhythmic alternation of weak and half-strong stresses
is formed, going back towards the beginning of the word, starting from the strong
syllable, of this kind: (ò{&à}à&àà&àà'à). However, given that polysyllabic words are
generally derivatives or compounds, there is a clear and agreed tendency to place
secondary stresses on the same syllables of the simple forms of the lexeme which
are originally stressed.
˛e only exception, so to speak, is made up of the rhythmic tendency that inter-
rupts both sequences of weak syllables that are too long, introducing some second-
ary stresses, and sequences of stressed syllables (¤ strong and half-strong), suppress-
ing some secondary stresses or shifting them one or two syllables.
Some examples: (&soce'ta) società, (ka&paci'ta) capacità, (pRo&babili'ta) probabili-
tà, (be&ati'tu:din™) beatitudine, (&Raqqjo&naliQ&Qabili'ta) razionalizzabilità, (aR&tifi-
&cali'ta) artificialità, (u&tilita&Ristika'men:te) utilitaristicamente, (e&zeRcita'to:Re) eser-
citatore, (aR&tikolaq'qjo:ne) articolazione, (as&socaq'qjo:ne) associazione, (o&ceano-
'gra:fiko) oceanografico, (ka&RatteRiQ'Qa:bile) caratterizzabile, (in&telliGen'tis:sima)
intelligentissima, (oS&Sillo'gram:ma) oßllogramma, (ef&feRveSSen'tis:simø) e‡erve&en-
tissimo˘
Other examples: (do&loRo&zissima'men:te) dolorosissimamente, (as&tuta'men:te) a-
stutamente, (im&maGinaq'qjo:ne) immaginazione, (mo&difikaq'qjo:ne) modificazio-
ne, (kom&poziq'qjo:ne) composizione, (ak&kjappafaR'fal:le) acºappafarfalle, (e~&c™-
142 a handbook of pronunciation
3.3.3.5. Neutral Italian does not in any way accept strong stress on two contigu-
ous syllables (in an intoneme), but attenuates the first, even by shifting it or, if nec-
essary, it shifts it (without attenuation), as in the third (and fourth) variants giv-
en, which are possible, but less common: /fa'rOk 'kwesto = farOk'kwesto/ (&faRøk-
'kwes:to), (fa&Røk'kwes:to), ('faRøk 'kwes:to), ('fa;Røk 'kwes:to) farò questo, /skoper'kjO
il'tetto = skoperkjOil'tetto/ (s&kopeRkjøil'tet:to), (s&kopeR&kjøil'tet:to), (s'kopeRkjø il-
'tet:to), (s'ko;peRkjø il'tet:to) &operºò il tetto, /nonnepo'tep 'pju* = nonnepotep-
'pju*/ (&nonne&potep'pju), (&nonnepo&tep'pju), (&nonne'potep 'pju), (&nonne'po;tep
'pju) non ne poté più, /bEn'kek k'reda = bEnkek'kreda/ (&b™˙kek'kre:da), (b™˙&kek-
'kre:da), ('bE˙kek 'kre:da) ben>é creda, /me'tad 'dOze = metad'dOze/ (&metad'dO:ze),
(me&tad'dO:ze), ('metad 'dO:ze), ('me;tad 'dO:ze) metà dose˘
One aspect of attenuation, owing to rhythmic reasons, can concern the moving
back of the reduced stress too, as seen in the following examples: /'vEngo 'dopo di-
'te* = 'vEngo dopodi'te*/ ('vE˙go &dopodi'te), ('vE˙go do&podi'te) vengo dopo di te˚
/'karo pa'pa* = karopa'pa*/ ('ka;Ro pa'pa), (&kaRopa'pa), (ka&Ropa'pa) caro papà˚ /lo-
'facco 'anke per'te* = lo'facco ankeper'te*/ (lo'facco &a˙kepeR'te), (lo'facco∞a˙&ke-
peR'te) lo faccio an>e per te˚ /si'kjama 'forse ko'zi = si'kjama forseko'zi*/ (si'kja;ma
&foRseko'zi), (si'kja;ma foR&seko'zi) si ºama forse così?
Intonation
3.3.4. û 3.3 shows the preintonemes and intonemes in neutral Italian (both
modern and traditional), which we will illustrate with simple examples:
/./: (Ri'partono 'tutti 'sa:bato23) /ri'partono 'tutti 'sabato./ Ripartono tutti sabato.
/?/: (¿Ri'partono 'tutti 'sa:bato21) /¿ri'partono 'tutti 'sabato?/ Ripartono tutti saba-
to?
/÷/: (&seRRi'partono 'tutti 5sa:bato12| &mene'va;do a˙'ki;ø23) /serri'partono 'tutti 'saba-
to÷ mene'vado an'kio./ Se ripartono tutti sabato, me ne vado an>'io.
In the north, regional intonation has consistent and varied shifts; in the south
they are rather repetitive; while, in the center, they are more contained (and, gen-
erally, more similar to neutral patterns). ˛e major di‡erences are found in the sus-
pensive intoneme, /÷/, as can also be seen in the phonosyntheses of é 16 of NPT/
HPh (and of é 10-15 of MaPI).
which, at that time were defined as of çTuscan languageÇ, more often than of çItal-
ianÇ.
˛e peculiarities of traditional pronunciation, therefore, are more rigid than
those of modern pronunciation (having less variations {and here, we relate above
all the çstrangestÇ} and –at the same time– most çcapriciousÇ). The forms followed
by * are found only in some dictionaries or specific lists.
As regards the vowel phonemes /e, E÷ o, O/, some examples of words and endings
follow. ˛e more bizarre forms are marked with simple (çorthoepicalÇ) written ac-
cents: A√èse˚ annèsso˚ auróra*˚ Bertóldo*˚ cè‡o˚ cócca çcorner, notchÇ, collètta˚ con-
nètto˚ crèsima˚ enórme˚ esòso˚ fólla˚ Giórgio˚ gótta˚ gróppo˚ lèttera˚ mòccolo˚ nèsso˚
nórma*˚ órgano*˚ rócca çdista‡Ç, Rómolo˚ sgómino˚ siète˚ stòrpio˚ strènna˚ tè&hio÷
as well as the endings: amaró√olo˚ stètti˚ stèttero…
As far as the distribution of /q, Q/ is concerned, the major traditional character-
istic regards initial /q/ for many words like: zampa˚ zio˚ zoppo˚ zuc>ero˚ zucca÷
and internal /Q/, in cases such as: aguzzino˚ amazzone˚ brezza˚ ªribizzo˚ intiriz-
zito˚ lazzi˚ lezzo˚ olezzo˚ pettegolezzo˚ ribrezzo˚ rubizzo˚ &orza˚ Ωarzo*˚ sozzo*˘
3.3.5.2. ˛e greatest di‡erence, however, concerns the use of single /s/, within
a word, between V (even if with /j, w/ before the second). ˛e most significant cas-
es (including derivatives) are: asino˚ casa˚ ºesi˚ ºuso˚ cosa˚ così˚ desidero˚ deside-
rio˚ (il) fuso˚ mese˚ naso˚ peso˚ Pisa˚ posa˚ raso˚ riposo˚ riso, and adjective endings
-ese˚ -oso˚ and verb endings -esi˚ -osi\ /in'glese/ in∫ese (but /fran'ceze/ francese!) /ci-
nese'ria/ cineseria (but /borge'zia/ bor!esia!), /go'loso/ goloso˚ /cellu'losa/ cellulosa˚
/kurjosi'ta*/ curiosità÷ /'presi, -sero, -so/ presi˚ presero˚ preso (but /'lezi, -zero, -zo/ lesi˚
lesero˚ leso!), /impre'sarjo/ impresario˚ /'rosi, -sero, -so/ rosi˚ rosero˚ roso˚ /ris'posi, -se-
ro/ risposi˚ risposero˚ /ro'sikkjo/ rosicºo (but /ero'zjone/ erosione!).
However, in traditional pronunciation, we find /z/ in cases such as: biso√o˚ ca-
so˚ ºesa˚ muso˚ paese˚ quasi˚ sposa˚ viso˚ misi˚ misero˚ (io) fusi˚ fusero˚ (ho) fuso˚ in-
cisi˚ incisero˚ inciso…
3.3.5.3. Traditional co-gemination is more systematic and more widespread
than modern co-gemination; indeed, it has a higher number of activating forms
(as da\ da Milano /dammi'lano/; for the modern: /dami'lano/), and does not pro-
vide for de-gemination è la mia /Ella'mia/; modern: /Ela'mia/); what is more, the
following are always co-geminant, too, come˚ dove˚ qual>e˚ sopra (unless they are
used as nouns): come si fa? dove vai? qual>e volta˚ sopra quel tavolo /komessi'fa, do-
vev'vai, kwalkev'vOlta, soprakkwel'tavolo/ (but: il sopra del tavolo /il'sopra del'tavo-
lo/); modern: /komesi'fa*, dove'vai, kwalke'vOlta, soprakwel'tavolo/).
As far as accentuation is concerned, there are no particular peculiarities; any
di‡erences are due to the fact that generations have passed, and, as one knows, the
accentuation of learned terms and classical nouns are subject to trends of prefer-
ence, which can be agreed upon to a greater or lesser extent, but are not under dis-
cussion: they are merely documentable (Ô the end of § 3.3.3.4).
Intonation does not change, except for the possible adoption (by some) of Tus-
can-type preintonemes (Ô the relative phonosyntheses in é 16).
3. italian 145
Text
3.4.0. ˛e story †e North Wind and the Sun follows, given in eight di‡erent
çnormalizedÇ versions. We start with the (neutral) Italian pronunciations of (neu-
tral British) English – this is the first step of the phonetic method (the written text
is given in § 2.5.2.0). ˛e Italian translation follows, in its neutral modern and tra-
ditional versions. ˛ree further accents are given: two are mediatic (respectively
northern and central – of Milan and Rome), which, for segmental and intonation-
al aspects, are not as broad as those presented –with vocograms and tonograms–
in é 10 “ é 12 of MaPI (in which there are 23 regional versions, in é10-14).
˛e respective (dialectal) phonosyntheses in é 16 of NPT/HPh (looking at Mi-
lanese, § 16.15, and Roman, § 16.42) can also provide useful preliminary infor-
mation. In northern mediatic pronunciation, stopstrictives often occur as se-
quences: (ts, dz÷ TS, DZ) (special symbols would be better: ((tS, dZ))).
Lastly, the a‡ected accent (snobbish, § 3.4.9) is examined, as well, with a çFrench
rÇ (generally, an approximant: labiodental, (V), or uvular, (˜), or uvularized labio-
dental, (◊), which has been chosen as a representative type, having both compo-
nents) and other characteristics linked to that kind of pronunciation, as çtenserÇ
but also çlaxerÇ V, at the same time (Ô û 3.4).
˛ey are realized –basically– by higher and more advanced vocoids than nor-
mal, which, in stressed syllables (and in preintonemes as well), are diphthongized,
but with second elements rather central in their vocogram boxes, therefore, end-
ing in di‡erent –opposite– positions from the initial ones.
Furthermore, /c, G÷ S, {Z}/ are postalveo-palatal over-rounded, (&, 1, «, {»});
there is also çdrawlingÇ and splitting of the tonic –in the real tonetic sense– and
posttonic syllables (sometimes even of the previous ones), in particular before a
pause, and above all for intonemes when they are di‡erent from the conclusive
ones.
Lastly, we note a bigger pitch extension (which we will indicate paraphonical-
ly, with §(ˇ)@, at the beginning of each paragraph) and creaky voice in intonemes
(¤ tonic and posttonic syllables), which is represented by the specific diacritic: (ü).
146 a handbook of pronunciation
3.4.1. At the end, as always, there is the version which gives the English pronun-
ciation of Italian, by neutral British speakers, fluent in Italian (after prolonged con-
tact with native speakers, but with no help from the phonetic method), who have
adequately learned the relative prominences, but who substantially use segmental
and intonation elements which are typical of neutral British English (for reference
purposes, although, of course, a neutral accent is not so common). Obviously, the
same principle is valid for the foreign pronunciation of English, given first.
Speakers of American English could prepare their own version of the Italian pro-
nunciation of American English, with /a/ even for English /Ø/ (¤ ('Sat:$) instead of
('SOt:$), for shot) and so on, as an excellent exercise, by listening to native speakers,
best of all after recording them. Of course, speakers of other languages could pre-
pare both their own pronunciation of Italian and the foreign Italian accent of their
language, as well. ˛e author would be happy to receive their transcriptions and
recordings, both in case of help –should they need it– and to make their contribu-
tion known to others (possibly in our website on canIPA Natural Phonetics – Ô §
0.12).
Italian text
4.0. As well as the çmodern neutralÇ accent, we will be presenting the çinter-
nationalÇ accent, as an alternative teaching proposal.
Lastly, we will also consider the çmediaticÇ accent (¤ from television), based on
current Parisian, and two types of accents which are very di‡erent from these, in
so far as they are structurally very di‡erent: southern pronunciation, represented
by Marseilles, and Canadian, from Québec.
Vowels
4.1.1.2. û 4.1 shows the French vowel articulations, which have to be compared
to those of the other languages to show the similarities and, above all, the di‡eren-
ces. In our transcriptions, we use twenty-three elements (plus another less impor-
152 a handbook of pronunciation
tant one). ˛e front series presents five vocoids: (i, I, e, ™, E), for three phonemes:
/i, e, E/.
Although these three, and even (™), have the same symbols as in other languages,
they di‡er slightly: (mi'_i) /mi'di/ midi˚ (e'te) /e'te/ été˚ ('fEt) /'fEt/ fête˚ ('E) /'E/ e®÷
note: (s™'m°Å) /sE'mwa/ c'e® moi˚ (v™Æ'+Û) /vEK'ty/ vertu˚ (_I˜'lP) /diK'lo/ dirlo (with
unstressed /E/, and with unstressed /i/ in a checked syllable in /K/). ˛e French /a/
is definitely fronter than in Spanish, Italian, German, Â: (bÅ'áÅ:\) /ba'gaZ/ bagage˚
('Åm) /'am/ âme (traditional ('A:m), mediatic/Parisian ('å:m)); furthermore: (påÆ-
'+i:˜) /paK'tiK/ partir˚ (&påÆ+i'ºÚ) /paKti'KÚ/ partirons˚ (på˜'lÅ) /paK'la/ par là˘
˛e phoneme /A/ of neutral traditional pronunciation was mentioned, which
is no longer up-to-date; for this reason it is not in û 4.1 (however, its tongue posi-
tion is slightly fronter than that of /Õ/ (ú), in the vocogram of the çinternationalÇ
accent, Ô § 4.4.1).
˛e çbackÇ (and rounded; actually, back-central as can easily be seen in û 4.1)
series is evidently di‡erent from that in most languages, although the phonemic
symbols are the same ((¯, ¨, P, Ö, ∏) for three phonemes /u, o, O/): (t¯'\¯:˜) /tu'ZuK/
toujours˚ (dP'dP) /do'do/ dodo˚ ('k∏m) /'kOm/ comme÷ note: (kÖ'm∏d) /kO'mOd/
commode˚ (\Ö'li) /ZO'li/ joli˚ (k¨Æ's¿e) /kuK'sje/ coursier˚ (p¨˜'v¯) /puK'vu/ pour vo¨
(with unstressed /O/, and with unstressed /u/ in checked syllable in /K/).
˛e real di‡erence lies in the fact that normally, forms with çunstable eÇ are ex-
pressed with the çzeroÇ phone and possible voicing assimilations, as seen in the sec-
ond transcriptions. ˛e solution of the çzeroÇ phone might even leave natives in
doubt (for rarer words), who are frequently undecided, about the proper spelling
too, between /§, °/ e˚ S˚ even if /°/ S cannot be dropped and should not give
rise to confusion. For the moment, having avoided the unrealistic çsolutionsÇ, ™
/È/, and /ê, +/ too, which are less improbable diaphonemically, to play safe we will
adopt /ù/ (#), before transforming it definitively into /§/, with the possibility of
indicating its presence or absence with /[§]/ – or /ù/, in opposition to an un-
stressed çstable eÇ, represented by /’§/.
‹en /ù/ is stressed, as in d¤-le˚ le çeÇ, it actually becomes the phoneme /°/ (re-
inforcing further what has just been said): (_i'l°) /di'l°/, (l#'°) /lù'°/ (it is the writ-
ing that is capricious, not the phonic structure). In stressed checked syllables in
/K/, the articulation of /E, §, O/ can be more or less back and lowered (((E/, §/, ∏/)),
but they are transcribed in the same way): ('mE:˜) /'mEK/ mère˚ ('s§:˜) /'s§K/ sœur˚
('f∏:˜) /'fOK/ fort˘
4.1.1.4. Modern neutral pronunciation keeps the opposition /'eò/ >-é˚ -ée˚ -er˚ -ez≥
V /'Eò/ >-ès˚ -et˚ -Pe˚ -a¤˚ -P(en)t≥, which, however, is rather tiring, structurally, giv-
en that the others are neutralized: actually, there is no (longer) di‡erence between
p› and pot˚ they are both ('pP) /'po/, nor between pS and pSx\ ('p°) /'p°/, and
now, not even between bat and bø\ ('bÅ) /'ba/. We have, for example, ('kúe) /'kle/
clé V ('kúE) /'klE/ clPe˚ ('püe) /'pKe/ pré V ('püE) /'pKE/ près˚ ('fe) /'fe/ fée V ('fE) /'fE/
fPt˚ (vÅ'le) /va'le/ vallée V (vÅ'lE) /va'lE/ vallet˚ (på˜'le) /paK'le/ parler V (på˜'lE)
/paK'lE/ parla¤˚ (sÅ've) /sa've/ savez V (sÅ'vE) /sa'vE/ savPt˘
It is prevalently through school and writing that such oppositions persist, as it
is above all useful to distinguish certain verb forms: (&\#på˜'le, /p-) /ZùpaK'le, Sp-/
je parlP (past: çI spokeÇ) V (&\#på˜'lE, /p-) /ZùpaK'lE, Sp-/ je parla¤ (imperfect: çI
spokeÇ) and (\#&på˜l#'ºe, /&p-) /ZùpaKlù'Ke, Sp-/ je parlerP (future: çI will speakÇ)
V (\#&på˜l#'ºE, /&p-) /ZùpaKlù'KE, Sp-/ je parlera¤ (conditional: çI would speakÇ).
As far as -ai is concerned, we have: ('\e) /'Ze/ j'P (but ('\E) /'ZE/ j'Pe]÷ ('áe, 'áE) /'ge,
'gE/ gP˚ ('©e, '©E) /'ke, 'kE/ quP÷ ('bE÷ -e) /'bE÷ -e/ bP˚ ('mE÷ -e) /'mE÷ -e/ mP˚ ('vºE÷ -e)
/'vKE÷ -e/ vrP÷ furthermore, we find ('mE) /'mE/ ma¤ (with (m™, me) /mE, me/ in a
preintoneme), as also for ('E) /'E/ tu π [il e®]˚ ('fE) /'fE/ je fa¤ [il fPt]˚ ('sE) /'sE/ je/tu
sa¤ [il sPt]˚ ('vE) /'vE/ je va¤˚ with (™, e÷ f™, fe÷ s™, se÷ v™, ve) /-E, -e/ in a preinto-
neme: (&il™'lÅ, -e-) /ilE'la, -e/ il e® là˚ (m™'sÅ2 ‘Å'l∏:˜2’) {(me-)} /mE'sa, ‘a'lOK,’/ {/me-/}
ma¤ ça, alors!
Nasalized vowels
4.1.1.5. Obviously, the four çnasalÇ vowels remain (which are more precisely de-
fined as nasalized]˚ with their six taxophones, ('ì, ’3) /í/, ('^, ’}) /^/, (Œ) /Õ/, (Ú)
/Ú/: (&3s™Æ'tì) /ísEK'tí/ incertPn˚ (}'bº^) /^'bK^/ un brun˚ (pŒ'dŒ) /pÕ'dÕ/ pen-
dant˚ (mÚ'nÚ) /mÚ'nÚ/ mon nom÷ in stressed and checked syllables, we have:
154 a handbook of pronunciation
('pì:dX) /'pídK/ pRndre˚ ('^:ba) /'^bl/ humble˚ ('lŒ:p) /'lÕp/ lampe˚ ('nÚ:bX) /'nÚbK/
nombre˘ A çfunnyÇ example: (}'bÚ 'vì 'blŒ) /^'bÚ 'ví 'blÕ/ un bon vin blanc˘ ˛ese
phonemes can even be followed by N\ (Œ'~¥i) /Õ'n¥i/ ennui˚ (nÚ'nEtX) /nÚ'nEtK/
non-être˚ (Œm'ne) /Õm'ne/ emmener.
It must be pointed out that, in current dictionaries and manuals, only four sym-
bols are used (the same, for both phonemic or phonetic use): ç/í, ^, ˙, Ó/Ç; of
these, for modern neutral pronunciation, only /^/ can be used; the others reflect
a pronunciation which is over a century old (exactly when the International Pho-
netic Association was founded), and lasted until the 1950's, as the neutral pronun-
ciation.
As the neutral pronunciation is di‡erent today, it can still be found in various
regional pronunciations, even if, as we will see, ç/í, ^, ˙, Ú/Ç can be useful as
representatives of an çinternationalÇ pronunciation, which is less tied to Paris and
to the neutral pronunciation of Parisian origin (Ô 4.2). Certain texts (and some
dictionaries), for the first three, even use ç/e, J, Å/Ç. In the transcriptions in the
French Dictionary (by R. Boch: Zanichelli, 1995’), the present writer put /í, ^,
Õ, Ú/, as here. In books which only use one kind of transcription (often a hybrid
of phonetic and phonemic), it is probably more suitable to give ç/π, ^, Õ, Ú/Ç.
Modern neutral pronunciation firmly keeps these four nasalized vowel pho-
nemes, as do most regional pronunciations (even if with di‡erent, more traditional
timbres). Instead, for Paris (for the center and the west) /^/ merges with /í/, so,
there, /'bKí/ corresponds to brin and even to brun (neutral ('bºì, 'bº^), respective-
ly). Consequently, even the çmediaticÇ accent loses a phoneme, merging both these
forms and other similar ones. ˛is will all be dealt with later (§ 4.4.2.1-2 “ §
4.5.2.3).
4.1.2.1. As far as unstressed V are concerned, it would be best to give some indi-
cations to follow, in order to obtain the most fitting results. Transcriptions in dic-
tionaries and in manuals do not actually tally completely, as there are various fac-
tors to be considered. In order to simplify the description, let us say that –indepen-
dently of writing and dictionary transcriptions– there are the following çvowel ad-
justmentsÇ: for /’E/ (¤ the archiphoneme of /e, E/), we have (e) /e/, in an un-
checked pretonic syllable, if it is followed by a çclose(r)Ç V (¤ /i, y, u÷ e, °, o÷ Ú/}:
(e'te) /e'te/ été˚ (me'zÚ) /me'zÚ/ ma¤on˚ (le'z,°) /le'zj°/ lπ ySx˚ (&ºepe'te) /Kepe'te/ ré-
péter˘
On the other hand, we have (™) /E/, in an unchecked pretonic syllable, when it
is followed by an çopen(er)Ç V (¤ /E, §, O÷ a÷ í, ^, Õ/) and in checked syllables (in-
dependently of the context) too: (™'tE) /E'tE/ éta¤˚ (™'tŒ) /E'tÕ/ étant˚ (l™'z∏m) /lE-
'zOm/ lπ hommπ˚ (&ºep™'tE) /KepE'tE/ répéta¤÷ and (&ev™n'mŒ) /evEn'mÕ/ événement˚
(m™t'sì) /mEt'sí/ médecin˚ (p™ú'te) /pEl'te/ pelleter˚ (p™˜'_y) /pEK'dy/ perdu˚ (™d'mi)
/Ed'mi/ et demi˚ (t™Æ'p∏:˜) /tEK'pOK/ tπ reports˚ (s™ú'sÚ) /sEú'sÚ/ sπ leçons˚ (&l™sp™©'tÅka)
/lEspEk'takl/ lπ spectaclπ˚ (d™p'n°) /dEp'n°/ dπ pne¨˘ For (initial) ex-˚ esC- there is
4. french 155
a strong tendency to have /e/: (™á'zŒ:pa, e-) /Eg'zÕpl, e-/ exemple˚ (&™s©Å'¬,e, &e-) /Eska-
'lje, e-/ πcaliers˘
˛e examples of checked syllables, (intentionally) show, in these two cases, that
the timbre of the V that follows is of no importance (as it is, here, quite the oppo-
site). One must always recall that spelling (which is not in any way a true transcrip-
tion of phonic structure) can play dirty tricks… For -err-˚ we have /’EK/, but we can
also find the adjustment: (t™'ºiba, te-) /tE'Kibl, te-/ terrible˚ (s™'ºe, se-) /sE'Ke, se-/ ser-
rer˚ but (&p™˜Ö'©E) /pEKO'kE/ perroquet÷ furthermore, (&Ppü#&m,™ ˜™'tÅ:\) /opKù'mjE
KE'taZ/ au premier étage˚ (&}l™'\™ ˜Œ'~¥i) /^lE'ZE KÕ'n¥i/ un léger ennui (despite (pü#-
'm,e, le'\e) /pKù'mje, le'Ze/ premier˚ léger]˘ ˛ere are possible fluctuations for -er +
V- and for the spelling é (towards /e/): (&}le'\e ˜Œ'~¥i, &Ppü#&m,e ˜e'tÅ:\)÷ and for P˚
Z (towards /E/): (púe'zi:˜, pú™-) /ple'ziK/ pla¤ir˘
4.1.2.2. On the other hand, the isolated forms often influence the contextual-
ized ones, even in checked syllables, as in (&\evizi't™l nÅ'vi:˜, -tel) /Zevizi'tEl na'viK,
-tel/ j'P v¤ité le navire˚ also for -ez: (&v¯lÅ'v™Æ kÖ'ny) /vula'vEK kO'ny, -eK/ vo¨ l'avez
reconnu, and even for the proclitic -es monosyllables: (t™Æ'p∏:˜, teÆ-) /tEK'pOK, teK-/
tπ reports˚ (s™ú'sÚ, seú-) /sEl'sÚ, sel-/ sπ leçons˚ (&l™sp™©'tÅka, les-) /lEspEk'takl, les-/ lπ
spectaclπ˚ (d™p'n°, dep-) /dEp'n°, dep-/ dπ pne¨÷ also (™d'mi, ed-) /Ed'mi, ed-/ et de-
mi˘
Lastly, it is evident that the e‡ects of vowel adjustment, in unchecked syllables,
can go back further than the pretonic syllable, as long as syllables with di‡erent V
timbres do not intervene: (&ºepe'te) /Kepe'te/ répété˚ (&º™p™'tÅ) /KEpE'ta/ répéta÷ (&be-
áe',e) /bege'je/ bégZer˚ (&b™á™'mŒ) /bEgE'mÕ/ bégPement˘
For /’◊/ too, (not to be confused with (`), çzeroÇ), in unchecked syllables (phon-
ically of course, and always independently of transcriptions easily found), we have
(°) /°/ + a çclose(er)Ç V and (#) /§/ + an çopen(er)Ç V\ (pú#'v°Å:˜) /pl§'vwaK/ plS-
voir˚ (pú°'ºe) /pl°'Ke/ plSrer; whereas for /’O/ in unchecked syllables, we general-
ly have (Ö) /O/, unless it is immediately followed by /z/, or by a syllable with /o/,
or it does not derive from /o/ and in -otion: (&mÖnÖ't∏n) /mOnO'tOn/ monotone˚ (&sÅ-
lÖ'pEt) /salO'pEt/ salopette˚ (bÖ'se) /bO'se/ bosser˚ but (\P'zEf) /Zo'zEf/ Josè#e˚ (bP'bP)
/bo'bo/ bobo˚ (ºP'ze) /Ko'ze/ rosée˚ (&gºPs¿™Æ'te) /gKosjEK'te/ grossièreté˚ (&emP's¿Ú) /e-
mo'sjÚ/ émotion˘.
Even for the (unstressed) V spelt ô, au, the more frequent pronunciation is /O/:
(&Öpi'tÅl) /Opi'tal/ hôpital˚ (Ö'º∏:˜) /O'KOK/ aurore˚ (mÖ'ºis) /mO'Kis/ Maurice˚ (Ö˜-
'v°Å:˜) /OK'vwaK/ au revoir˘ In aujourd'hui˚ the preposition combined with the
definite article [au˚ èà+le¶] holds; instead, it is the second syllable that gives more,
due to /’uK˘/, as well: (&P\¨˜'_¥i, -\Ö˜-) /oZuK'd¥i, -OK-/; but we can hear (&Ö\Ö˜-, &¯-
\¨˜-) ç/OZOK-, uZuK-/Ç too.
For b›coup [b› + coup]˚ obviously, we have (bP'k¯) /bo'ku/, but the tenden-
cy to have /’O/, is so strong that, as it would be quite absurd to have */bO'ku/, we
very often end up with having, (b¯'k¯) ç/bu'ku/Ç. Even surtout and au fur et à
mesure present us with the frequent colloquial pronunciation (s¨˜'tU, P&f#˜Åm-
'zy:˜), for (sY˜'tU, P&fy˜eÅm'zy:˜) /syK'tu, ofyKeam[ù]'zyK/.
Likewise, for /°, §/, the base form is quite important: (d°'z,Em) /d°'zjEm/ dS-
156 a handbook of pronunciation
xième˚ (b#'ºe) /b§'Ke/ bSrrer˘ In both cases, in checked syllables, we find (#) /§/,
(Ö) /O/: (s#l'mŒ) /s§l'mÕ/ sSlement˚ (pÖs'te) /pOs'te/ po®er˘
Above, we have already dealt with /’^, ’Õ/ and /’iK˘, ’yK˘, ’uK˘/ (§ 4.1.1: see giv-
en examples). In all other cases, in modern neutral pronunciation, in unstressed
syllables, we have, (i, y, ¯÷ Å÷ Œ, Ú) /i, y, u÷ a÷ Õ, Ú/.
Consonants
4.2.0. ˛e table of û 4.2 shows the French consonant articulations, which are
necessary for the right pronunciation of that language.
û 1.9-15, instead, give the orograms, grouped for manners of articulation, of all
the contoids given in the chapters of this volume (even as secondary, occasional,
or regional variants) for the 12 languages treated. ˛is exposition makes the neces-
sary comparisons between di‡erent languages more immediate.
postpalatal rounded
provelar rounded
protuded
labiodental
alveo-velar
prepalatal
alveolar
bilabial
palatal
uvular
dental
velar
Nasals
4.2.1. In French there are three traditional phonemes: /m, n, N/. It is worth
keeping the last phoneme, even though it has lost most of its phonemicity, merg-
ing with /nj/ (as did /L/, which does not exist in French anymore, having merged
with /j/): (mÅ'mŒ) /ma'mÕ/ maman˚ (nÅ'nŒ) /na'nÕ/ nanan˚ (pÅ'~,e) /pa'nje/ pa-
nier˚ (Å'NP, Å'~,P) /a'No/ agn›˚ (&Œs™N'mŒ÷ -~,#-) /ÕsEN'mÕ/ ensRgnement˚ (mÚ'tÅN÷
-Å~,) /mÚ'taN/ montagne˘ ˛e transformation into /nj/ is more frequent (and for
many, by now, quite normal) in front of a V˘
If /N/ is losing ground, there is a xenophoneme (for English -ing]˚ which is tak-
ing root (and, in theory –possibly, given its limited use– the balance of the phono-
logical system could be restored, with the substitution between these two N]˚ how-
ever, at the moment, the pronunciation of the English -ing, as in camping, varies
substantially; the most recommendable pronunciations are velar, (-i˙) (in an En-
glish-like way), or prevelar, (-i”) (by partial assimilation): (kŒ'pi˙, -”); or even pal-
atal, (-iN) (above all for the elderly); we also find (-i˙g, -i”Ò, -iNá) (a more autoch-
tonous version).
For (tautosyllabic) /nj, n¥/, we have (~,, ~¥): (pÅ'~,e) /pa'nje/ panier˚ (y'~,∏l) /y-
'njOl/ une yole˚ (Œ'~¥i) /Õ'n¥i/ ennui. (Popularly, we find that /nj/ and /N/ tend to
merge into (N): (mÅ'~,E:˜) /ma'njEK/ manière becomes (må'NE:‰, mÅ-), but should
not be copied.)
Normally, /n/ does not assimilate a heterosyllabic C following it (as happens, in-
stead, in most other languages), so we have (including ((yn$-)), nearly (&yn#-), with
quite an evident break): (yn'p∏m) /yn'pOm/ une pomme˚ (yn'bŒ:©) /yn'bÕk/ une ban-
que˚ (yn'mE:˜) /yn'mEK/ une mère˚ (&ynp™Æ's∏n pÅ's¿Œ:t) /ynpEK'sOn pa'sjÕt/ une per-
sonne patiente˚ (yn'fi,) /yn'fij/ une fille˚ (&ynvÅ'li:z) /ynva'liz/ une val¤e˚ (yn'/E:z) /yn-
'SEz/ une >a¤e˚ (yn'No:l) /yn'Nol/ une gnôle˚ (mÅn'©ì) /man'kí/ mannequin˚ (&yn©Å-
'ºEs) /ynka'KEs/ une carπse˚ (yn'g¯t) /yn'gut/ une goutte÷ and (&yn+y'lip) /ynty'lip/ u-
ne tulipe˚ (yn&_,ÅgÖ'nÅl) /yndjagO'nal/ une diagonale˚ (&ynºe'pÚ:s) /ynKe'pÚs/ une ré-
ponse.
Only in (what is often considered to be) non-neutral pronunciation, or nearly,
is it possible to have coarticulations for /n/ before a dorsal C: (~) (with raised tip
of the tongue) + (N, ©, á) and (n) + (k, g÷ º, ˜): (y~'No:l, mÅ8'©ì, &y8©Å'ºEs, yn'g¯t,
&ynºe'pÚ:s).
Near a voiceless C, as already seen in some examples, the N undergo devoicing,
and they reach total voicelessness before a pause: (}p'n°) /^p'n°/ un pnS˚ ('püis))
/'pKism/ pr¤me˘
Stops
4.2.2. ˛ere are three (diphonic) pairs of phonemes: /p, b÷ t, d÷ k, g/, with im-
portant pairs of taxophones: prepalatal, /t, d/ (+, _), before /i, y÷ j, ¥/ (and, less sys-
tematically, before /e, °/ too); a further (less important) one, is alveolar, for /t, d/
(T, D), before /S, Z/ (whereas, before /s, z/ they remain dental); and one palatal, or
158 a handbook of pronunciation
rather, postpalatal, for /k, g/ (©, á) (more appropriate symbols are ((£, 8)), but it is
not absolutely necessary to use them), before front V (including /a, ù/), before /j,
¥/ and even at the end of syllables or rhythm groups, before a pause. In the other
cases, the çvelarÇ articulation, (k, g) can also be prevelar, ((´, Ò)), but, again, it is
not necessary to use special symbols.
Let us take a look at some examples: (p¯'pe) /pu'pe/ poupée˚ (be'be) /be'be/ bébé˚
('t¯t) /'tut/ toute˚ (d¯'d¯n) /du'dun/ doudoune˚ (+y'_i) /ty'di/ tu d¤˚ ('+¿ì) /'tjí/ tiens˚
(kÚ'_¥i:˜) /kÚ'd¥iK/ conduire˚ (e'te÷ e'+e) /e'te/ été˚ ('d°÷ '_°) /'d°/ dSx˚ (©Å¯T'/U)
/kaut'Su/ caout>ouc˚ (&ÅD\™©'+if) /adZEk'tif/ adjectif˚ (&s3mÖ'ºits) /símO'Kits/ SPnt-
-Moritz˚ (pid'zÅ) /pid'za/ pizza˚ (pi©'ni©) /pik'nik/ pique-nique˚ ('k∏©) /'kOk/ coq˚ (áÅ-
'áÅ) /ga'ga/ gaga˚ ('lŒ:á) /'lÕg/ langue˚ ('gºP) /'gKo/ gros˚ ('gºi) /'gKi/ gr¤˚ ('kúÅ©) /'klak/
clac! Before tautosyllabic C˚ (k, g) remain unchanged ({pro}velar), as can be seen
in some of these examples.
We find complete voicing assimilation to the second element (if diphonic), in
cases such as: (&Ån™á'd∏t) /anEk'dOt/ anecdote˚ (&º™T/P'se) /KEdSo'se/ rez-de->a¨sée˚
(m™t'sì) /mEd'sí/ médecin˚ (&ÅpsÖ'ly) /absO'ly/ absolu˚ (&s™d'dÅm) /sEt'dam/ cette da-
me˚ ('k¯b d#/Œ'pÅN) /'kup dùSÕ'paN/ coupe de >ampagne˚ (/Åg'\¯:˜) /Sak'ZuK/ >a-
que jour˚ (&Åv™á'v¯) /avEk'vu/ avec vo¨.
In these cases, a slowing down of the rate of speech can lead to only partial assi-
milation: (&Ån™ˆ'd∏t, &º™fl/P'se, m™∂'sì, &ÅÊsÖ'ly, &s™∂'dÅm, 'k¯Ê d#/Œ'pÅN, &Åv™ˆ'v¯).
˛is is the reason why, in phonemic transcriptions, we keep etymologic pho-
nemes, whereas for V we indicate the actual timbres, as we use more precise sym-
bols (despite dictionary –even pronunciation-dictionary– transcriptions which do
not, however, consider intermediate articulations!).
On the other hand, if the second segment is not diphonic (including /j/), the
aforementioned voicing assimilation does not occur: (&Åv™©'n¯) /avEk'vu/ avec no¨˚
(&Åv™©'1¥i) /avEk'l¥i/ avec lui˘
Constrictives
4.2.3.1. ˛ere are three (diphonic) pairs, /f, v÷ s, z÷ S, Z/, as well as two isolated
voiced phonemes, /j, K/ (instead of ç/J, º/Ç, the more çlegitimateÇ symbols), which
we will see presently. We can observe that, usually, /s, z/ are articulated with the
tip of the tongue raised, therefore, if we wanted to highlight this aspect, for discus-
sion and teaching, we could resort to supplementary symbols, ((s, z)); but, the most
important characteristic concerns /S, Z/, which, are generally, prevelarized postalve-
olar protruded, (/, \) (with a deeper timbre, caused by the lowering of the back of
the tongue between the two articulatory strictures, postalveolar and prevelar). Ex-
amples: ('fE:˜) /'fEK/ fPre˚ ('vif) /'vif/ vif˚ ('sŒ) /'sÕ/ cent˚ ('vÅ:z) /'vaz/ vøe˚ ('/Å) /'Sa/
>at˚ ('pÅ:\) /'paZ/ page˘
Voicing assimilation (to the second element) also concerns diphonic pairs of
constrictives: (n¯v'zÚ) /nuf'zÚ/ no¨ fa¤ons˚ ('ºP;s påÆfy'me) /'Koz paKfy'me/ rose
parfumée˚ ('vÅ\ 'v,E,) /'vaS 'vjEj/ va>e viRlle˚ (/'sE) /Z'sE/ je sa¤÷ but we have (/f) /Sv/:
(/'fÅl) /S'val/ >eval˚ (Å/'fe) /aS've/ a>evé˘ Slowing down, we can even have (n¯Ñ'zÚ,
4. french 159
'ºP;Ω påÆfy'me, 'vÅ? 'v,E,, ?'sE÷ /'ÑÅl, Å/'Ñe). On the other hand, speaking quickly, we
can find cases such as: (/™'pÅ) /Z{ùn}sE'pa/ je (ne) sa¤ pø.
About /j/, it must be said immediately that, more than a real constrictive, it is
a çsemi-constrictiveÇ, (,), indeed, it is placed half-way between the approximant,
(j), and the truly (voiced) constrictive, (J) (definitely rarer in world languages):
(',E:˜, i',E:˜) /'jEK, i'[j]EK/ hier˚ (©Å',e) /ka'je/ cahier˚ ('p¿e) /'pje/ pied˚ (,P',P) /jo'jo/ yo-
-yo˚ (&Åt#'¬,e) /at§'lje/ atelier˚ (fÅ'mi,) /fa'mij/ famille˚ (sÖ'lE,) /sO'lEj/ solRl˚ (f#¿'tÚ)
/f§j'tÚ/ fSilleton˘ Before a pause, we often have (¿): (fÅ'mi¿, sÖ'lE¿). French phono-
tactics, contrary to that of other languages, also has /Sj, Zj/: ('/¿ì) >ien˚ (ºe'\,Ú) ré-
gion˘
Sequences such as //0Ki'é, 0li'é// are given with (i',é), and therefore even the
most appropriate and most modern phonemic transcription gives /i'jé/: (püi',e)
prier˚ (püi',§:˜) priSr˚ (púi',Å) plia˚ (&sÅbli',e) sablier˘ Hence, (bºi',Œ) stands for
both brillant and Briand÷ but we could –possibly– have (bºi'Œ) /bKi'Õ/ for Briand˚
in controlled pronunciation. Furthermore, we have: (pe'i, pe',i) /pe'[j]i/ pZs˚ (&Åbe'i,
-e',i) /abe'[j]i/ abbZe˘
Approximants
4.2.4. Apart from the taxophone (˜) of /K/ (just treated with the constrictives),
we have two central approximant phonemes, (¥) /¥/ (postpalatal rounded) and
(°) /w/ (provelar rounded, for which the symbol (w) of velar rounded phone
could quite easily be used, as in other kinds of pronunciation, treated at the end
of this chapter; but the opportunity to demonstrate not negligible gradation
would be lost): ('s¥i) /'s¥i/ su¤˚ ('~¥i) /'n¥i/ nuit˚ ('¬¥i) /'l¥i/ lui˚ ('l°i) /'lwi/ Lou¤˚
('m°Å) /'mwa/ mo¤˚ ('p#Å) /'pwa/ po¤˚ ('s#Å:˜) /'swaK/ soir˘ ˛ese examples also
show assimilation of voicing and place of articulation, as well as the important fact
that /¥/ and /w/ are two di‡erent phonemes, di‡erent from /j/ too. Regarding /lw/,
we can have by assimilation, (]°), with /l/ realized as semi-velar. ˛e /0K, 0l/ + /y,
u/ sequences remain the same, with no insertion of approximants (contrary to
what happens for //é'i, i'é//, which normally become /é'ji, i'jé/, § 4.2.3): (&Öpstüy'e)
ob®ruer˚ (gly'Œ) gluant˚ (kú¯'e) clouée˚ (tü¯'Œ) trouant˘
Laterals
4.2.5. ˛ere is only one lateral phoneme (today, Ô § 4.2.1), (l) /l/, which assimi-
lates for voicing (and, before /j, ¥/, for place of articulation): ('lyn) /'lyn/ lune˚
('p#Ål) /'pwal/ poil˚ (Å'le) /a'le/ aller˚ ('bl°) /'bl°/ blS˚ ('kúe) /'kle/ clef˚ ('fúy) /'fly/ flux˚
(Åú'pì) /al'pí/ alpin˚ (™á'zŒ:pa) /Eg'zÕpl/ exemple˚ ('Ú:ka) /'Úkl/ oncle˚ ('Ú:ga) /'Úgl/ on-
gle˚ ('s¯fa) /'sufl/ sou·e˘ Slowing down, or enunciating, we can also have (lï): ('Ú:ga÷
-lï), between a voiced C and a pause. Sometimes, we can hear something in be-
tween, with (Í): ('ÚkÍ, 'Ú:gÍ). In /lj, l¥/ the articulation is prepalatal: (s¯'¬,e) /su'lje/
soulier˚ ('¬,°) /'lj°/ liS˚ ('¬¥i) /'l¥i/ lui. Often, in non-neutral pronunciation, /lj/ and
/j/ tend to merge into /j/: (mi'¬,Ú÷ mi',Ú) million˚ (mi'¬,°÷ mi',°) miliS˚ therefore (s¯-
'¬,e) soulier and (&fyzi'¬,e) f¨ilier can correspond to (s¯',e) souiller˚ (&fyzi',e) f¨iller˘
Structures
4.3.0. ˛e greatest problems for segments are caused by the unstressed vowel
phoneme graphically expressed by e (apart from some exceptional cases, as mon-
siSr˚ fa¤ons); whereas, with regard to phone groups –or connected speech– the
phenomenon of lia¤on is typical (§ 4.3.3.1-3).
debatable. More often than not, it seems that /ù/ (#) is introduced, in pronuncia-
tion, to avoid long sequences of C which turn out to be di‚cult to pronounce.
˛erefore from a phonemic transcription such as //msj°, ddÕ, at'lje//, Â, we could
actually have: (m#'s¿°, d#'dŒ, &Åt#'¬,e) monsiSr˚ dedans˚ atelier˚ as isolated forms;
in connected speech, the most common forms are, for instance: (}≈'s¿°) /^m'sj°/
un monsiSr˚ (lÅd'dŒ) /lad'dÕ/ là dedans˚ whereas atelier remains the same.
We could, on the other hand, start from a çfullÇ or çisolatedÇ form, that retains
all these unstable e (which is the most common in the traditional reading of verse,
and is even stronger in the way of speaking in the south of France, the Midi]˚ drop-
ping all the possible /ù/, without complicating the pronunciation with di‚cult
or impossible groups: //bOnù'tKi// bonnetrie˚ //ZùtùlùKù'di// je te le red¤˚ for the
normal (bÖn'tüi, &/t#lº#'_i÷ &\#tú#˜'_i). Generally, in phonemic transcriptions in
dictionaries, the /ù/ that do not drop in isolated forms are retained, omitting the
others; hence, we can now see when, even those which are normally indicated, can
drop.
4.3.1.2. In practice, the dropping of one or more /ù/ can occur if the conso-
nant groups, which come in contact, can occur within the word, ™ /lst, ksj, kskl,
kspK, ksplw, Kkw, Ks¥, KstK, Kmn/, Â, as in: (sÖús'tis) /sOls'tis/ sol®ice˚ (&™ázÅ©'s¿Ú)
/Egzak'sjÚ/ exaction˚ (&™kskúÅ'me) /Ekskla'me/ exclamer˚ (&™kspüi'me) /EkspKi'me/ ex-
primer˚ (™ks'pú°Å) /Eks'plwa/ exploit˚ (p¨Æ'k#Å) /puK'kwa/ pourquoi˚ (p¨Æ'sÎit)
/puK's¥it/ poursuite˚ (&syp™Æs'tüÅ) /sypEKs'tKa/ super®rat˚ (&ip™˜mne'zi) /ipEKmne'zi/
hypermnésie.
It is generally possible to increase the number of C in contact, if when dropping
/ù/, constrictives, approximants, laterals and nasals (but also stops) are added –
before or after: (©is'sE;Æt 'sÅ÷ &©is#'s-) qui se sert de ça?˚ (\#n&l#˜_i'pÅ, \&n#lº#-) je ne le
red¤ pø˚ (&ynp+it'fi,) une petite fille˚ (&+ynsüÅ'pÅ) tu ne serø pø˚ ([&i¬],Å&bPk¯d'mÚ:d)
il y a b›coup de monde˚ (&Únpå˜'lE ©#t'sÅ) on ne parlPt que de ça˚ (&™s©#&\#úsÅ'vE,
™zá&\#ú-) e®-ce que je le sava¤?˚ (\mŒ'fi/) je m'en fi>e!
Furthermore: (/&kü°Å'b,ì) je cro¤ bien˚ (s&n™pÅ'sy:˜) ce n'e® pø sûr˚ (s&©i[ú]t#'fP,
s©it'fP) ce qu'il te faut˚ (s&tü3'lÅ) ce trPn là˚ (/&sÎikÚ'tŒd lÅ'v°Å:˜) je su¤ content de la
voir [de l'avoir]˚ (/t#ld'mŒ:d, /&t#ld#'m-) je te le demande˚ (&s™á\#n&t#ldv™'pÅ, -d#v™-,
s™©/&t#l) c'e® que je ne te le deva¤ pø˚ (lve'v¯, &l#-) levez-vo¨!˚ (m&nem°Å'lÅ, m#&ne-)
menez-moi là!˚ (\'_i kúemÖ'+iv dlŒ'fŒn sÚpÅ'bÚ, -+if) je d¤ que lπ motifs de l'enfant
ne sont pø bons˚ (&sY˜l#'bŒ, syl-) sur le banc˚ (si&\#nt#l_i'pÅ, &si/t#l-) si je ne te le d¤
pø˚ (il&m#ldmŒt'pÅ, in&m-, &inm#l&d#-) il ne me le demande pø.
©thin a word, we can find: (bÖn'tüi) bonneterie˚ (©Ås'ü∏l) cøserole˚ (Åm'ne)
amener˚ (Å'púe) appeler˚ (\¯'ºe) jouerP˚ (pú™n'mŒ) plRnement÷ but: (&Œgl#'tE:˜) An-
gleterre˚ (&m™Ækü#'_i) mercredi˚ (&på˜l#'mŒ) parlement˚ (&fÖÆt#'mŒ) fortement˚ (&Œ-
pú#'mŒ) amplement˘ For parle-m'en, we often have (på˜l'mŒ), which we can also
hear for parlement˚ in fast or slow speech.
4.3.1.3. It is important to bear in mind the di‡erence that exists in French, for
the /0/ + /m, n, K, l/ + /j/ sequences, that would be too heavy, hence, a stable /ù/
is inserted: (&sÖm#'¬,e) sommelier˚ (&n¯s#'m,Ú) no¨ semions˚ (&sŒt#'~,e) centenier˚
162 a handbook of pronunciation
(&v¯t#'~,e) vo¨ teniez˚ (&n¯s#'º,Ú) no¨ serions˚ (&v¯/Œt#'º,e) vo¨ >anteriez˚ (&ºi/#-
'¬,°) Ri>eliS˚ (&n¯zÅp#'¬,Ú) no¨ appelions˚ but (©Ås'p¿e) cøse-pieds˚ (s™t'p¿Es) cette
pièce˚ (bÖn'+¿e) bonnetier˚ (/åÆ'+¿e) >arretier˚ (p™ú'+¿e) pelletier˘ ˛e same happens
for /¥, w/: (/e&s#¬¥i'si) >ez celui-ci˚ (}&b¯d#'l°Å) un bout de loi÷ even if it is not unu-
sual to hear: (&/es¥i'si) >ez celui-ci˚ (&}b¯'dl°Å), (&l#\°'dl°Å) le jS de l'oie˚ and simi-
lar phrases.
˛e /ù/ does not generally drop in the first syllable of names: ('l°i ˜#'nP) /'lwi
Kù'no/ Lou¤ Renault˚ (&Ř#'nÅ:˜) /aKù'naK/ à Renard˚ not even for de\ (&d#lÅ'ºi:v)
/dùla'Kiv/ De la Rive˚ (m#&s¿°d#'gP:l) /mùsj°dù'gol/ M. De Gaulle÷ but if it is pos-
sible to simplify, it is done, on the radio and television too, even if it is a little stig-
matized, as in: de De Gaulle which is nearly always (d#d'gP:l). Surnames usually
resist better than first names, indeed, Renaud and Den¤, in appropriate contexts
are often: (˜'nP, d'ni).
However, we regularly find: (}˜'nÅ:˜) /^K'naK/ un renard˚ (snÚ'lÅ2 &v,ìd'gP:l)
{(s#n-)} /sùnÚ'la, vjíd'gol/ ce nom-là vient de "Gaule&˚ (\&nepÅd'gP:l) /Znepad'gol/
je n'P pø de gaulπ (in a real or figurative sense), (P&bÖ˜dlÅ'ºi:v) /obOKdla'Kiv/ au
bord de la rive÷ but: (ºi/'lE) /KiS'lE/ Ri>elet (obviously in the case of (&ºi/#'¬,°) /Ki-
Sù'lj°/ Ri>eliS, it is di‡erent). ˛e same happens before /*é/ (¤ çdisjunctiveÇ ini-
tial V˚ usually represented by the so called çaspirateÇ h and by the names of num-
bers): (l#e'ºP) /lùe'Ko/ le héro˚ (l#'Ú:z) /lù'Úz/ le onze˘
Taxophonics
4.3.2.2. ‹en, in connected speech, a word ends in /0Kò, 0lò/ and is then fol-
lowed by another which begins with /ò0/, in slow and wary pronunciation, /ù/ is
inserted; but normally, even /K, l/ are dropped: ('©Åt 'fÅm÷ '©Åtü#) /'katK 'fam/ qua-
tre femmπ˚ (&ynPt'f#Å÷ y&nPtü#-) /ynotK'fwa/ une autre fo¤˚ (&m™ddP'tEl÷ 'mEtü# d-)
/mEtKdo'tEl/ maître d'hôtel˚ (l#'pP;v bÖ'n∏m, ú'p-÷ l#'pP;vº#) /lù'povK bO'nOm/ le
pauvre bonhomme˚ (i≈'sŒ;p ©#'nÚ÷ &ilm#'sŒ;bl#) /ilmù'sÕbl kù'nÚ/ il me semble que
non˚ (&3pÖ'sib d#ú'fE:˜÷ -ibl#) /ípO'sibl dùl'fEK/ impossible de le fPre˚ (ú'p§b _itÅ'li÷
l#'p§pú#) /lù'p§pl dita'li/ le pSple d'Italie˘
In cases such as /vOtKùp'n°/ votre pnS˚ apart from a slow and wary (&vÖtü#p-
'n°), we also have (vÖp'n°, -t'n°, vÖtp'n°), and (&vÖtp#'n°), considered rather un-
educated because of the di‡erence with its spelling. Decidedly uneducated (and
intentionally joky) is (&™©s#'püE, -e) for (™ks'püE) exprès˚ whereas (™s'püE, -e) is rather
fast colloquial.
As seen (in the previous section), we have to have /0ùùj/ in a word (/ù/ refers
to sonants: /m, n, K, l/), but it is not valid in sentences: (s™l'm,ì) c'e® le mien˚ (&i[ú]-
fPl'~,e, fP-) il faut le nier˚ (&i[l]zŒ'+¿En '¬,°) ils en tiennent liS˚ (in'vÅl 'º,ì, &iln#-) ils
ne valent rien˘
However, one can definitely also have: (&d#n#'º,ì 'fE:˜) de ne rien fPre˚ (&\#nd#-
'mŒ;d# 'º,ì) je ne demande rien˚ and even: (&s™l#'m,ì), (&i[ú]fPl#'~,e, fP-), (&i[l]zŒ'+¿En#
'¬,°), (in'vÅl# 'º,ì, &iln#-). Let us also consider: (&bÖnÅpåÆt#'mŒ '/P) bon appartement
>aud˚ (&bÖnÅ'pÅÆt ≈Œ'/P, -t#) Bonaparte man>ot˘ If the context does not create
any ambiguity, we can easily have bon appartement >aud (&bÖnÅpåÆt'mŒ '/P).
4.3.2.3. For sequences of monosyllabic words with /ùò/ [le˚ je˚ me˚ te˚ se˚ ce˚ de˚
ne]˚ there are often clear, general preferences, but not absolute, such as: (\#n÷ \n#)
je ne˚ (\#m, \m#) je me˚ (\#l, \l#) je le˚ (d#l) de le˚ (d#n) de ne˚ (d#m) de me˚ (d#t)
de te˚ (d#s) de se˚ (d#s, ts#) de ce˚ (©#l) que le˚ (©#n) que ne˚ (©#m) que me˚ (©#s) que
se/ce˚ (©#t) que te˚ (s©#) ce que\ (&d#nº#v'ni;Æ ©#ú's#Å:˜) de ne revenir que le soir˚
(&m°Å\#n&s™©#d'_i:˜) moi je ne sa¤ que te dire˚ (&\#ns™'º,ìt s#tüy'kúÅ) je ne sa¤ rien
de ce truc là˘
For /Z/ in je me le demande˚ we find (&\#ml#d'mŒ:d) and (\&m#ld#'mŒ:d)÷ the lat-
164 a handbook of pronunciation
ter is at times considered less advisable; similarly, for the second forms in: (\l#'v°,
\#l'v°) je le vSx˚ (\#≈'+¿ì, \m#-) je me tiens˚ (/&tÅv™'_id v#'ni;˜, -'_i d#v'ni:˜) je t'a-
va¤ dit de venir (spelling influences the çchoiceÇ of preferably retaining word-in-
ternal /ù/).
˛e second forms here are, thus, also çuneducatedÇ: (/'kü°Å÷ #/'kü°Å) je cro¤˚
(\mŒ'fU÷ &#\-) je m'en fo¨˚ (l™'mEá d#lÅ'ºy, l™'mE ©#dlÅ'ºy) lπ mecs de la rue˚ (&dek¯-
'vºi;˜ l#'mÚ:d, &dek¯'vºi; ˜#l'mÚ:d) découvrir le monde˚ ('püŒd l#me'tüP, 'püŒ d[º]#l-)
prendre le métro˚ (i'mEt ú#'püi, i'mEt [ü]#ú'püi) y mettre le prix (if it could be confused
with ils mettent le prix (i'mEt ú#'püi), the /K/ does not drop), (i[ú]'/EÆ/ s#©i[l]'v°, i[ú]-
'/EÆ/#s ©i[l]'v°) il >er>e ce qu'il vSt÷ actually, the second forms are often more
çnaturalÇ, but –alas– di‡erent from the spelling!
In the case of forms such as (&Åb°Å'mŒ) aboiement˚ (\n™'t#Å) je nettoie˚ (i[l]'v°Å)
ils voient˚ (/pe'ºe) je pZerP [pPerP]˚ (©i[ú]'s#Å) qu'ils soient˚ (&©#+y'E, ©#'+ÎE, ©#'tE)
que tu Pπ÷ forms such as (&Åb°Å,'mŒ, \n™'t#Å,, i[l]'v°Å,, /p™,'ºe, ©i[ú]'s#Å,) and (&©#-
+y'E,, ©#'+ÎE,, ©#'tE,) are certainly çuneducatedÇ.
ing) is considerable and typical: (}n&mi©i'lP) /^dmiki'lo/ un demi kilo˚ (Ån'mì) /ad-
'mí/ à demPn˚ (t¯n'mEm) /tud'mEm/ tout de même˚ (Œ&tü3nmŒ'\e) /ÕtKídmÕ'ZE/ en
trPn de manger˚ (yn'gºŒ;n me'zÚ) /yn'gKÕd me'zÚ/ une grande ma¤on˚ (lŒn'mì)
/lÕd'mí/ lendemPn˚ (d™'gºŒ;n 'dÅm) /dE'gKÕd 'dam/ dπ grandπ damπ˚ (lÅ'/Œ;m dÅ-
'mi) /la'SÕb[K] da'mi/ la >ambre d'am¤˚ (kÚ'm,ì) /kÚ'bjí/ combien˚ (&intÚ;≈'pÅ, &iút-,
il&n#-) /ilnùtÚb'pa/ il ne tombe pø˚ (l™'lŒ;˙ mÖ'dE˜n) /lE'lÕg mO'dEKn/ lπ languπ
modernπ˚ (yn'lÚ;˙ 'áE:˜) /yn'lÚg 'gEK/ une longue guerre˘
For voiceless stops, we have assimilation of places of articulation in the same
way, while we can have voicing, devoicing, or voicelessness, for the type of phona-
tion (again, as well as a slower or more controlled possibility, which corresponds
to the phonemic transcription ): (&}≈+i'k¯, &})-) /^pti'ku/ un petit coup˚ (m3n'nŒ,
-n'nŒ, -£'nŒ, m3'nŒ) /mít'nÕ/ mPntenant˚ (v3n'd°, -n'd°, -£'d°) /vít'd°/ vingt-dSx˚
('bŒ;~ pe&˜ife'ºi©, -) p-) /'bÕk peKife'Kik/ banque péri#érique˚ (\#n&m#tüÚm'pÅ, -≈-
'pÅ, -)'pÅ) /ZùnmùtKÚp'pa/ je ne me trompe pø˘
Lastly, this assimilation can occur even before a V\ (pŒ'nŒ) /pÕ'dÕ/ pendant˚ (t¯l-
&mÚn™'lÅ) /tulmÚdE'la/ tout le monde e® là˚ as even before a pause: (t¯l'mÚn) /tul-
'mÚd/ tout le monde˚ (mÅ'lŒ:˙) /ma'lÕg/ ma langue˘ It can also be found between a
non-nasalized V and N\ (&mÅnm°Å'zEl) /madmwa'zEl/ mademo¤elle˚ (&m™nm°Å'zEl)
/mEdmwa'zEl/ mπdemo¤ellπ˚ (&º#nmŒ'de) /KùdmÕ'de/ redemander˚ (&Ånmi'ºe) /ad-
mi'Ke/ admirer˚ (&_,ÅNnÖs'+i©, -˙n-) /djagnOs'tik/ diagno®ic˚ (&Œ\Œm'mŒ) /ÕZÕb'mÕ/
enjambement˚ and adverbial -ment: (fü°Ån'mŒ) /fKwad'mÕ/ froidement˚ (vÅ˙'mŒ)
/vag'mÕ/ vaguement˚ (&kÚpú™n'mŒ, -n'mŒ, -£'mŒ) /kÚplEt'mÕ/ complètement˘
Let us also note cases such as (l°i'©Em püÖ'/ì, -mp p-) /lùwi'kEnd pKO'Sí/ le week-
-end pro>Pn˘
4.3.2.6. In fluent, familiar speech, there are certain reductions (even for stress),
which simplify speech, without compromising communication; instead, by slow-
ing down speech, pronunciation can correspond to the phonemic transcription:
(s't∏m) /sE'tOm/ cet homme˚ (st#'fÅm) /sEt'fam/ cette femme˚ (Ås't§:˜) /asE't§K/ à cet-
te hSre˚ (stÅ'_i:˜) /sEta'diK/ c'e®-à-dire˚ ({m™}p&t™t, -tX) /{mE}p°'tEtK/ [ma¤] pSt-être˚
(Å&°Å:˜) /a'vwaK/ avoir˚ (°Å'lÅ, v'lÅ) /vwa'la/ voilà˚ (vlÅPT'/P:z, &°ÅlÅ-) /vwalaotK'Soz/
voilà autre >ose˚ (≈&s¿°, )'s-, p's-) /mù'sj°/ monsiSr˚ (&k¯te'm°Å) /ekute'mwa/ écou-
tez-moi˚ (s©yz™'m°Å) /Ekskyze'mwa/ exc¨ez-moi˚ (s¿¯&púE) /silvu'plE/ s'il vo¨ plaît˘
In a preintoneme, /swa'sÕt/ soixante is currently reduced to ('s#Œ;t), especially
in compounds: (s#Œt'sis, &s#Œd_iz'n§f, -n_-) 66˚ 79˘
Furthermore: (pÅs'©#, &pÅs©#, &pÅs©, &Ås©, s©#, s©, sk) /paKs[ù]kù/ parce que˚ (&™s-
©#, s©#) /Eskù, skù/ e®-ce que {(p¨Æ'k#Ås ©#&+y_i'sÅ) pourquoi e®-ce que tu d¤ ça?]˚
(sÎi'si) /sùl¥i'si/ celui-ci˚ (&pi) /'p¥i/ pu¤˚ (&pis©#, -s©, -sk) /'p¥isk/ pu¤que˚ ([™]'bì)
ç/[E]'bí/Ç [eh] bien (bin˚ bi'n would be better), ('pÛ) /'ply/ pl¨ (possibly: p'¨]˚ (mŒ-
&fì) /mEÕ'fí/ ma¤ enfin [m'enfin]˚ (mÅ&l∏;˜) /mEa'lOK/ ma¤ alors [m'alors]˚ (&bŒ) ç/'bÕ/Ç
bon! (ban! would be better), (&nŒ) ç/'nÕ/Ç non! (nan! would be better), ('°™, -E, -e, -É)
/*'wE, -e/ oua¤!˚ (&k∏;˜) /Õ'kOK/ encore˚ (st#&mŒ, /t#-) /Zystù'mÕ/ ju®ement˘
Other examples: (dmÅ&~,E;˜, t≈-) /dùtutma'njEK/ de toute manière˚ (tfÅ&sÚ) /dù-
tutfa'sÚ/ de toute façon˚ (Æ&k#Å) /puK'kwa/ pourquoi˚ (+y˜™l&mŒ) /natyKEl'mÕ/ natu-
rellement˚ (s&≈Œ) /s§l'mÕ/ sSlement˚ (D&\¯:˜) /tu'ZuK/ toujours˚ (s&pÅ, &pÅ) /nEs'pa/
166 a handbook of pronunciation
n'e®-ce pø?˚ (D&\Å) /de'Za/ déjà˚ (D\°'ne) /deZ°'ne/ déjSner˚ (™s&púi©Å's¿Ú, s&p-) /Ekspli-
ka'sjÚ/ explication˚ (\™'s¿Ú) /ZEs'tjÚ/ ge®ion˚ (&©Åt) /'katK/ quatre˚ (©™©'f#Å) /kElkù'fwa/
quelquefo¤˚ (©™©'/P:z) /kElkù'Soz/ quelque >ose˚ (†tÅ'kU) /tuta'ku/ tout à coup˚ (†tÅ-
'l§:˜) /tuta'l§K/ tout à l'hSre˚ (†tP'púy[s]) /tuto'ply[s]/ tout au pl¨˚ (Å™á'v¯) /avEk'vu/
avec vo¨˘
More still: (/Îi'lÅ, /sÎi-, //i-) /Zùs¥i'la/ je su¤ là˚ (\¥ie'_i÷ \,e-) /Zùl¥ie'di/ je lui P
dit˚ (&tŘe'zÚ) /tyaKe'zÚ, t¥a-/ tu ø ra¤on˚ (t™'f¯, te-) /tyE'fu, t¥E-/ tu π fou˚ (t&s™) /ty-
'sE/ tu sa¤˚ (&+yÅv™'_i, &+¥Å-, &tÅ-) /tyavE'di, t¥a-, ta-/ tu ava¤ dit˚ (l™'b,ì) /ilE'bjí/ il e®
bien˚ (v&zÅve'vy) /vuzave'vy/ vo¨ avez vu˚ (v*&zŘe'te) /vuvuzaKe'te/ vo¨ vo¨ arrê-
tez?˚ (nó&zŒnÅ'lÚ) /nunuzÕna'lÚ/ no¨ no¨ en allons˚ (&isÚ'bÚ) /ilsÚ'bÚ/ ils sont bons˚
(&izÚ'püi) /ilzÚ'pKi/ ils ont pr¤˚ (,Å) /ilja, ja/ il y a˚ (&,ŒnÅ'vE) /iliÕna'vE, ilj-, j-/ il y en
avPt (uneducated even (&,ÅnÅ'vE)). In fast speech, y˚ si˚ ni˚ tu˚ ou, oú˚ before a V˚
often and usually, have consonant variants (often condemned by schools, all be-
cause of disguising spelling): (si™l'v°, s¿™-) /siEl'v°, sjE-/ si elle vSt˚ (niÅ'n¯, ~,Å-) /ni-
a'nu, nja-/ ni à no¨˚ (¯Å'¬,Ú, °Å-) /ua'ljÚ, wa-/ ou à Lyon˚ (¯™'tEl, °™-) /uE'tEl, wE-/
oú-e®-elle?˚ (i[ú]&fPiÅ'le, -P,Å-, -P+-) /ilfoia'le, -oja-, -ot-/ il faut y aller˘
In rapid speech, between nasalized V, continuous C can be nasalized a little, but
it is not worth transcribing as it is barely perceptible: (&mÚ\Œ'bÚ) /mÚZÕ'bÚ/ mon
jambon˚ (Œ'vì) /Õ'ví/ en vPn; the same occurs for oral V preceded and followed by
N\ (nÖ'nEt) /nO'nEt/ nonnette˚ (y&nÅnimi'te) /ynanimi'te/ unanimité˘ On the other
hand, in other languages, such as English, Spanish, Italian, the nasalization –of V
between N– is more obvious, but this too, is hardly worth mentioning as it occurs
quite automatically.
An alveolar semi-lateral, /l/ (¬), is possible for the grammemes (articles or pronouns)
le˚ la˚ les˚ lui (non-final) preceded by the grammemes par˚ pour˚ sur˚ vers: (&på˜l™f-
'nEtX, -˜¬™-÷ &p¨˜¬¥i'_i:˜, -˜¬¥i-÷ &sY˜lÅ'tÅba, -˜¬Å-÷ &v™˜l#'s°Å:˜, -˜¬#-) /paKlef'nEtK, puK-
l¥i'diK, syKla'tabl, vEKlù'swaK/ par les fenêtres˚ pour lui dire˚ sur la table˚ vers le soir˘
çLiaisonÇ
4.3.3.1. For the all-important lia¤on, it must be said that it concerns, to di‡er-
ent degrees, all types of pronunciation: from a minimum of linking in familiar
speech, to a maximum which can be found in classic poetry (a certainly more elab-
orate language). More or less in between, we can find current conversation (real
language). Lia¤on only occurs within rhythm groups, between words which are
linked from a morphosyntactic and semantic point of view. Some are obligatory,
others impossible, some optional, depending on the style of diction and the choic-
es of the speaker.
˛e most normal and frequent linking occurs with: /z/ [s˚ x˚ z]˚ /t/ [t˚ d]˚ /n/ [n]\
(&lezÅ'mi) lπ am¤ (“ mπ˚ tπ˚ sπ˚ dπ˚ cπ]˚ (&d°'z§:˜) dSx hSrπ˚ (™&+ilÅ'le) e®-il al-
lé?˚ (&}gºŒ't∏m) un grand homme˚ (&p¿etÅ'tE:˜) pied-à-terre˚ (&Œne'te) en été˚ (&mÚnÅ-
'mi) mon ami (“ ton˚ son÷ once (&mÖnÅ-) was neutral, but today it is no longer so,
although this pronunciation is still widely used), (&ÚnÅ'tŒ) on attend˚ (&}nP'tEl) un
hôtel (“ aucun hôtel]˚ (&º,3nÅ'fE:˜) rien à fPre˚ (&b,3nÅ'se) bien øsez˘
4. french 167
Other forms with nasalized V do not link, except bon and adjectives such as
plRn˚ vPn˚ ancien˚ certPn˚ pro>Pn˚ soudPn˚ vilPn (but they lose their nasality):
(&bÖnÅ'mi) bon ami˚ (&Œpú™'nE:˜) en plRn Pr˚ (l#&m°Å,™'nÅ:\) le Moyen-Âge˘
We never find lia¤on after et, or before the names of numbers or a çdisjunctiveÇ
h, (or çaspiratedÇ h, because it was pronounced like that… centuries ago!) and gen-
erally, before w˚ y\ (eÅ'l∏:˜) et alors˚ (¬¥ie'El) lui et elle˚ (&iúsÚ'Ú:z) ils sont onze˚ (l#'^)
le un˚ (l#'¥it) le huit˚ (dee'ºP) dπ héros˚ (Œ'P) en haut˚ (&}°is'©i) un wh¤ky˚ (}',P:t,
}',∏t, }',Å©) un ya>t÷ and (}'°i) un oui, as well.
Even in current conversation, it is obligatory to link a noun with the çdetermi-
nantsÇ that precede it: (le'z∏m) lπ hommπ˚ (&sez°Å'zP) cπ o¤›x˚ (te'z,°) tπ ySx˚
(le&zPtü#'z∏m, &lezPd'z∏m) lπ autrπ hommπ˚ (&l#˜zÅ'mi) lSrs amiπ˚ (&d°zÅ'mi) dSx
am¤˚ (°ºŒ'z∏m) dπ grands hommπ˚ (}&gºŒtÅ'mi) un grand ami˘
Subject pronouns and verbs are linked: (&n¯zÅ'vÚ) no¨ avons˚ (&v¯zÅ'le) vo¨ al-
lez˚ (il'zEm, i'zEm) ils Pment (Ô (i'lEm) il Pme]˚ (&ÚnÅ'vE) on avPt˚ (\Œ&nepå˜'le) j'en
P parlé˚ (™&+ilv#'ny, &™+iv'ny) e®-il venu?˚ (&Ú+il'vy, &Ú+i'vy) ont-ils vu?˚ (p°'tÚ) pSt-on?
4.3.3.2. ˛ere is a case in which pronunciation does influence spelling, and this
should be remembered. It occurs in nearly all imperatives, without -s when isolat-
ed, but with /z/ for the pronouns y and en: vø-y (vÅ'zi), pensπ-y (&pŒs#'zi), mangπ-en
(&mŒ\#'zŒ)…
As well as forms such as va-t'en! (with elision of the pronoun te) we find analog-
ical çeuphonicÇ t's in questions whith subject-auxiliary inversion: e®-il?˚ but P-
me-t-on?˚ viendra-t-elle?˚ convPnc-t-il?…
Similarly, the adverbs très˚ tout˚ bien are linked to adjectives (or adverbs) modi-
fied by them: (&tü™zy'+il) très utile˚ (&t¯tŒ'+¿e) tout entier˚ (b,3&nÅtŒ'+if) bien attentif˘
Instead, pø˚ pl¨˚ moins˚ trop˚ fort˚ øsez˚ jama¤ can link or not; but in normal
conversation they do not usually do so: (&pÅ[z]Œ'k∏:˜) pø encore˚ (&tüPetü°Åt'≈Œ,
tüÖ&pe-) trop étroitement˘ In fixed expressions they obviously link; for instance,
pl¨˚ as in (&púyz¯'m°ì) pl¨ ou moins˚ which is always so.
(Monosyllabic) prepositions and conjunctions link to the forms that follow
them: (&Œni'vE:˜) en hiver˚ (&dŒz}'m°Å) dans un mo¤˚ (sŒ'zEl) sans elle˚ (/e'z°) >ez
Sx˚ (/e&z}nÅ'mi) >ez un ami÷ (kŒ&+il™v'ny) quand il e® venu (but linking is only
possible with the interrogative adverb, as in (kŒ&[t]™s©il™v'ny) quand e®-ce qu'il e®
venu?÷ but not in ('kŒ ™+ilv#'ny, ™+iv'ny) quand e®-il venu?˚ so as to avoid /tété/).
If the prepositions and conjunctions are polysyllabic, a lia¤on is equally only possi-
ble: (&Åpü™Å&v°ÅÆ/Œ'te, Å&pü™zÅ&v°ÅÆ/Œ'te) après avoir >anté˚ (d#&vŒynme'zÚ, -Œ+y-)
devant une ma¤on˘ ˛ere is linking with dont˚ en, too: (l#/'fÅl dÚ+i&lÅpå˜'le) le >e-
val dont il a parlé÷ and en˚ y have a çpre-liaisonÇ: (&pü#nÚ'zŒ) prenons-en˚ (&n¯zi's∏m)
no¨ y sommπ˘
©th auxiliaries and semiauxiliaries, lia¤on, though possible, is becoming more
and more unusual in current conversation. It nearly always occurs between e® (and
often sont˚ ont] and a following adjective or past participle, especially with allé\
(s™&t3pÖ'siba) c'e® impossible˚ (i[ú]&sÚtÅ'le) ils sont allés˚ (&i[l]lÚ'+Û) ils l'ont S (e ils l'ont
tu). ˛ere are even some fixed expressions which require linking: (le&/Œzeli'ze) lπ
≥amps-Élyséπ˚ (&leze&tÅzy'ni) lπ États-Un¤˚ (&vizÅ'vi) v¤-à-v¤˚ (d#&tŒzŒ'tŒ) de temps
en temps˘
168 a handbook of pronunciation
Considerations on stress
4.3.4.1. It is merely in theory that French only has a primary stress at the end
of each rhythm group. ˛e previous examples have quite thoroughly shown the
use of secondary stress in French rhythm groups; they generally alternate, mov-
ing back from the primary stress.
Let us now consider some other examples and some structural di‡erences. We
normally find: (&mŒ\eŒ'k∏:˜) mangez encore˚ (yn&p#+it'fi,) une petite fille˚ (&lÅmit-
'p¿E:˜) l'ami de Pierre˚ unless there are two rhythm groups (to emphasize the first
part, for a particular reason): (mŒ'\e Œ'k∏:˜, &ynp#'+it 'fi,, lÅ'mi d#'p¿E:˜). Let us al-
so note such as cases: (&lÅmÖÆ'sy:˜) la morsure˚ (lÅ&mÖÆ'sy:˜, lÅ'm∏;Æ 'sy:˜) la mort
sûre÷ (Ú&sŒde'g¯t) on s'en dégoûte˚ (Ú'sŒ de'g¯t) on sent dπ gouttπ÷ (&lÅk¯'l§:˜) la
coulSr˚ (&_yv3'º¯:\) du vin rouge˚ (&lÅk¯'l§;˜ _yv3'º¯:\) la coulSr du vin rouge˚
(\v¯'dºE _y'vì 'º¯:\) je voudra¤ du vin rouge˘
Traditional teaching describes the French stress in rhythm groups as follows:
with no emphasis, there is a primary stress on the last syllable, and, for the rest of
the rhythm group, the lexemes reduce their stress to secondary, whereas gram-
memes lose it completely.
To native speakers, this distribution sounds a bit too çpreciseÇ, çtext-bookÇ; it is
actually, a kind of mid-way between two kinds of transcriptions for the examples,
seen above, respectively with one or more intonemes: (mŒ&\eŒ'k∏:˜, ynp#&+it'fi,, lÅ-
&mit'p¿E:˜). In reality, as we have seen, there are certain rhythmic needs, for which
the use of secondary stresses is quite di‡erent.
Intonation
4.3.5. û 4.3 gives the preintonemes and the four intonemes of neutral French.
It can be noted that, compared to other languages (and variants of French itself),
instead of four preintonemes, there are five. Indeed, for partial questions, as well
as the normal (echo-like) anticipation of the interrogative rise, we find a general
falling movement, from half-high (similar to that of an imperative preintoneme,
except for the obvious and correct di‡erence of echo anticipation).
A more çstructuredÇ solution, rather than adding a fifth preintoneme, would be
a split in two (something like ç/»¿/Ç and ç/”¿/Ç, or even ç/¿/Ç and ç/¿¿/Ç, or ç/¿¡/Ç, or
ç/¡¿/Ç), leaving the task of clarifying the çoddityÇ to observation — explanations.
We prefer a more concrete approach (Ô § 11.9). A careful examination of all the
preintonemes (and intonemes), as well as of the symbols is revealing.
It is worth carefully observing the position of the pretonic syllable of the contin-
uative and conclusive intonemes: although it does not manage to be really high,
it is significantly raised, and the auditory e‡ect is noteworthy, as well (even if not
immediate), so much so that it contributes to giving the utterance a sort of promi-
nence, (too) often mistaken for stress.
As we will see, for the çmediaticÇ accent, such prominence is increased by the
half-lengthening of the V in the pretonic syllable (again, without any particular
û 4.3 Modern neutral French preintonemes and intonemes.
stress).
We only give examples of the three marked intonemes (referring the imperative
and emphatic preintonemes to § 4.3.4). As can be seen in the transcriptions of the
text (§ 4.5), there is the çmidÇ parenthesis, as well (Ô § 13.24 of NPT/HPh):
/./: (\v°Å'le Psi%neçmÅ3 3) /Zv°a'le osine'ma./ Je vSx aller au cinéma.
/?/: (¿[&™s©#]v¯&på˜%le'b,ì füŒ•sE1 1) /¿[Eskù]vupaKle'bjí fKÕ'sE?/ [E®-ce que] vo¨ par-
lez bien frança¤? – (¿på˜&le%v¯'b,ì füŒ•sE1 1) /¿paKlevu'bjí fKÕ'sE?/ Parlez-vo¨ bien
frança¤? – (&1kÖÌmŒ tÅleçv¯3 3) /&kO'mÕ tale'vu./ Comment allez-vo¨?
/÷/: (Ú&p°i%Å'le Œv°Å5+y:˜1 1 %¯Å'p¿e3 3) {(-°%,Å-, %°Å-)} /Úp°ia'le Õvwa'tyK÷ ua'pje./ On
pSt y aller en voiture, ou à pied.
Other accents
4.4.0. It will be interesting to compare what has just been said –about neutral
pronunciation– to other pronunciations which present more or less di‡erent char-
acteristics.
çInternationalÇ accent
4.4.1.1. For teaching purposes, the çinternationalÇ accent could be even more
appropriate than the neutral accent seen so far, inasmuch as it is considerably less
influenced by Paris and, therefore, undoubtedly nearer to many other varieties of
pronunciation, once any marked peculiarities have been eliminated.
˛erefore, an çinternationalÇ pronunciation might even be more advisable (and,
generally, easier to pick up and command), giving unquestionably valuable results,
as long as there is coherence within all elements and no improvisation or variation
among di‡erent types (including the usual interference, not only phonological,
but due to spelling).
˛e vocogram in this section gives the çinternationalÇ vowel articulations
which, as can be observed immediately from the comparison with û 4.1, are less
marked; in particular, this refers to (u, o, O) which are back and not back-central;
even (a) is less peculiar, as it is central, even if fronted. ˛e nasalized vowels are
still four, (í, ^, ú, Ú). In the figure there are two grey markers, for /’E, ’í, ’O/ (¤ un-
stressed – a third would be for /’§, ’^/, which, however, coincides with /ù/); but
could be removed to simplify the structure.
As a matter of fact, there are two satisfactory practical solutions, as, given the
lack of stress, either quite spontaneously we already have the lower-mid timbre ((™,
™, #, }, ø), which we prefer to use); or the higher-low timbre ((E, í, §, ^, O)), if
weakened, could be acceptable. ˛erefore, it is better not to introduce less impor-
tant taxophones, keeping (i, y, u, a) in the context /’éK˘/: (&oZu˜'d¥i) aujourd'hui÷
avoiding devoicing too.
Let us look at some examples, only for the cases in which there is a phonetic
di‡erence with the neutral (given in round brackets): ('lu) {('l¯)n} loup˚ ('o) {('P)n}
4. french 171
›˚ ('nOt) {('n∏t)n} note˚ ('la) {('lÅ)n} là˚ (d#'dú) {(d#'dŒ)n} dedans˚ ('bjí) {('b,ì)n}
bien˚ (bj™'to) {(b,3'tP)n} bientôt˚ (pa˜'ti) {(påÆ'+i)n} parti˘
çMediaticÇ accent
4.4.2.1. ˛is accent is spread by television and radio. It is based on the Parisian
accent, and shares quite a few characteristics with modern neutral, although it
shows further evolution which moves it further away from the çinternationalÇ ac-
cent (§ 4.4.1).
172 a handbook of pronunciation
It also manifests some fluctuation both towards and away from neutral; how-
ever, we show it here in its most typical form, indicating some internal di‡erences
too, for instance of an uneducated or suburban level (a peculiarity of the Parisian
banliSe).
In this section the vocogram is given of the vowels for which there are definite-
ly gradations (comparing it to fig 4.1, of neutral); here, we bring your attention
to the major di‡erences. ˛e fronting of /u, O/ (%, @) (and /’O/ (ê), ¤ unstressed)
is evident; the (anti-clockwise) rotation of /í, Õ, Ú/ (a, _, œ), (and /’Õ/ (Ô)), is just
as evident, as is the absence of /^/ which merges with /í/ (a), or varies between
the two, even with an intermediate realization (not shown explicitly in the voco-
gram) (8), with slight rounding. (In the fluctuations towards the neutral accent,
we have (£), as well; or (Õ), ¤ with a slight derounding).
˛e /EK, §K, OK/ sequences often shift back one box: (Ä[:]‰, ’ɉ÷ @[:]‰, ’ê‰÷ ∏[:]‰
’Ö‰÷). Furthermore, note in the vocogram the values for a\ (even if with individ-
ual fluctuations) the distinction is generally maintained between /a/ and /A/,
which traditional neutral had adopted. However, the timbres (of Parisian and) of
the çmediaticÇ accent are: /a/ (å), but (Ä) for /aK[0]é, wa/, and /'A[0]ò/ (å), but /’A/
(A) (in the banliSe˚ we have /wa/ (’°A, '°å), often /aKò/ (å:‰) ar(C)˚ even more
often /ajò/ (å:,), but (ås¿œò) for /AsjÚ/ -ation˚ of traditional neutral).
Some examples follow (where t indicates traditional neutral): ('t°) {('tU)n} tout˚
('n@t) {('n∏t)n} note˚ (bê'n@:‰) {(bÖ'n§:˜)n} bonhSr˚ (&P\T‰'_¥i, -ê‰-) {(&P\¨˜'_¥i)n}
aujourd'hui˚ ('b,a) {('b,ì)n} bien˚ (\Ô't_) {(\Œ'tŒ)n} j'entend˚ ('bœ) {('bÚ)n} bon˚ (P'©a,
-8) {(P'©^)n} aucun˘
Others: ('pÄ:‰) {('pE:˜)n} père˚ ('p@:‰) {('p§:˜)n} pSr˚ ('p∏:‰) {('p∏:˜)n} port˚ ('p@ú
's∏:‰) {('p∏ú 's∏:˜)n} Paul sort˚ (pÄ'‰i) {(pÅ'ºi)n} Par¤˚ (pÄ”'+i:‰) {(påÆ'+i:˜)n} partir˚
(v°Ä'lå) {(v°Å'lÅ)n} voilà˚ ('s#Ä:‰) {('s#Å:˜)n} soir˚ ('©å:‰) {('©Å:˜)n} car˘ Also: (på'på)
{(pÅ'pÅ)n} papa˚ (pA'så:\) {(pÅ'sÅ:\)n} pøsage˚ ('g‰å) {('gºÅ)n, ('gºA)t} grø˚ ('g‰å:s)
{('gºÅs)n, ('gºA:s)t} grøse˚ ('_,å:ba) {('_,Åba)n, ('_,A:ba)t} diable˘
For the banliSe accent; (v°A'lå) {(v°Å'lÅ)n} voilà˚ ('s#A:‰) {('s#Å:˜)n} soir˚ ('kå:‰)
{('©Å:˜)n} car˚ ('på:,) {('pÅ,)n} pPlle˚ (t”å'vå:,) {(tüÅ'vÅ,)n, (-A:,)t} travPl˚ (p”ê&nœs¿å-
's¿œ) {(püÖ&nÚs¿Å's¿Ú)n, (-s¿A-)t} prononciation˘
˛ere is a strong tendency to merge /eò, Eò/ into /e/ (but with many oscillations
due to hypercorrection and undecidedness; some even use the intermediate tim-
bre, (™), always or above all for -P): (p%'le) /pu'lE/ poulet˘
4.4.2.2. For the C, as well as for /K/ (as seen in some examples), which is typical-
ly (‰), but can be –as well as in neutral– even (˜), especially to give emphasis: ('‰y)
{('ºy)n} rue˚ ('t”%) {('tü¯)n} trou˚ ('©åtx) {('©ÅtX)n} quatre÷ we must add that on an
uneducated level, the çpalatalizationÇ of /t, d÷ k, g/ is much more evident, with ar-
ticulations going from stops to stopstrictives, (⁄, Á÷ ›, k): (⁄y'Ái) {(+y'_i)n} tu
d¤˚ ('mEk) {('mE©)n} mec˚ ('›id) {('áid)n} guide; /w/ is provelar rounded, (°):
('m°Ä) /'mwa/ mo¤.
˛e length in the mediatic accent, apart from what has already been said for the
neutral one, presents a typical half-lengthening of the vocoid of the unchecked pre-
tonic syllable, ¤ the syllable that precedes the stressed syllable of an intoneme (of-
4. french 173
ten this prosodic phenomenon is incorrectly described as a shift of stress from the
last syllable to the penultimate in a rhythm group): (%pÄ;'‰i) {(%pÅ'ºi)n} Par¤˚ (&pÄ-
%‰i;'z,a) {(&pÅ%˜i'z,ì)n} par¤ien (in these examples, even for neutral French, we ex-
plicitly highlight the marked height of the pretonic syllable, which is almost half-
-high in the two accents, but in the çmediaticÇ one there is also the half-lengthen-
ing, seen above). ˛e tonogram gives the intonation of the çmediaticÇ accent:
make all the necessary considerations, paying a (fair) bit of attention.
4.4.3.1. We will now look at one of the accents which is furthest from the neu-
tral accent (but also from the çinternationalÇ and çmediaticÇ ones): Marseilles
French as a good representative of the pronunciation of the Midi˘ As can be seen
from the vocogram, the V are few: seven plus /ù/. ˛ere is no phonemic opposi-
tion between /e, E÷ °, §÷ o, O/; least of all between /a, A/. For /E, ◊, O/ we always
have (e, °, o) (even in unstressed syllables), except in stressed syllables followed by
C (with or without /ù/), where we find (™, #, ø).
For /a/ there is a central timbre, (a). ˛e articulation of /ù/ is unrounded cen-
tral, (È), except in contact with /K/, where there is rounding, (+); it corresponds to
every e in the spelling (except for Ve\ amie = ami˚ (a'mi)}, and it often appears be-
tween C.
Furthermore, the nasalized V, /í, ^, Ú, Õ/, are nothing but a sequence of oral
vowels and a nasal contoid homorganic to the following C÷ but, before a pause,
174 a handbook of pronunciation
we find the semi-provelar, («). ˛e timbre of the çnasalizedÇ vowels is: (e, °, o, å)
in unstressed syllables, (™, #, ø, a) in stressed syllables, but (™e, #°, øo, aå) in ab-
solute final (stressed) syllables.
˛e only kind of normal lengthening, apart from emphasis, is the half-lengthen-
ing, in an intoneme, of the V (followed by a C, with or without /ù/), and the diph-
thongization of the çnasalizedÇ V in absolute final position before a pause. In less
broad pronunciation, the lengths can be more similar to those of neutral pronun-
ciation (as the timbres and the distribution of /'E, '§, 'O÷ ’E0, ’§0, ’O0/, as well).
Some interesting examples follow: ('tu) {('tU)n} tout˚ ('tre) {('tüE)n} très˚ (te'r™e«)
{(t™'ºì)n} terrPn˚ (s°l'maå«) {(s#l'mŒ)n} sSlement˚ (&promÈ'nadÈ) {(püÖm'nÅd)n}
promenade˚ ('v™e«) {('vì)n} vin˚ (&bjene'me) {(&b,3ne'me)n} bien-Pmé˚ ('m™ns) {('mì:s)n}
mince˚ (o'k#°«) {(P'©^)n} aucun˚ (l°n'di) {(l}'_i)n} lundi˚ ('#mblÈ) {('^:ba)n} humble˚
(mo«'nøo«) {(mÚ'nÚ)n} mon nom˚ ('nømbr+) {('nÚ:bX)n} nombre˚ (òån'taå«) {(\Œ-
'tŒ)n} j'entend˚ ('lampÈ) {('lŒ:p)n} lampe˚ ('di;r) {('_i:˜)n} dire˚ ('vwa;r) {('v°Å:˜)n}
voir˚ (pÈ'n°) {(p'n°)n} pnS˚ (&òÈtÈ&lÈdÈ'mandÈ) {(/&t#ld#'mŒ:d)n} je te le demande˚
(&òÈnÈ&tÈl+&r+dÈ'mandÈ 'pa) {(\#n&t#lº#d&mŒn'pÅ)n} je ne te le redemande pø˘
A less broad accent can have some partially nasalized V in stressed syllables:
('m™ens) {('mì:s)n} mince˚ (ëi'™e«) {('/¿ì)n} >ien˚ ('#JmblÈ) {('^:ba)n} humble˚
('br#J«) {('bº^)n} brun˚ ('laAmpÈ) {('lŒ:p)n} lampe˚ ('baA«) {('bŒ)n} banc˚ ('nøÚmbr+)
{('nÚ:bX)n} nombre˚ ('bøÚ«) {('bÚ)n} bon.
4.4.3.2. For the consonants, we can observe that all the taxophones of neutral
pronunciation are not present; furthermore: /n/+/0/ (m, M, n, ~, N, ˙), /nm, nK/
(«m, «r), /S, Z/ (ë, ò), /j/ (j), /w/ (°), /N, nj/ (~j), /Ni/ (ni), /lj, l¥/ (¬j, ¬¥), /'[0]/+/jé,
¥é, wé/ ([0])+(i'é, y'é, u'é), /K/ (r) (even (‰), especially before a consonant; and,
in less broad accent, also (K, ˜)}: (åM'v™;r) {(Œ'vE:˜)n} envers˚ ('ba˙kÈ) {('bŒ:©)n} ban-
que˚ (å«'ri) {(Œ'ºi)n} Henri˚ (òÈ'ë™rëÈ) {(\#'/EÆ/)n} je >er>e˚ (&trava'je) {(&tüÅvÅ',e)n}
travPller˘ ˛e banlieu accent has /t, d/ (+, _) + /i, j, y, ¥/, and /K/ tendentially voice-
less (3) (and (x, ', ˜)).
Furthermore: (mon'ta~jÈ) {(mÚ'tÅN)n} montagne˚ (&ma~je'rizmÈ) {(&mÅ~,e-
'ºis))n} maniér¤me˚ (¬je'zøo«) {(¬,e'zÚ)n} lia¤on˚ (k¥i'zi;nÈ) {(©Îi'zin)n} cu¤ine˚
(&òÈsy'i) {(\#'sÎi, /'sÎi)n} je su¤˚ (ly'i) {('¬¥i)n} lui˚ (lu'i) {('l°i)n} Lou¤˚ (bi'™e«)
{('b,ì)n} bien˚ ('ra;r+, '‰a;‰+) {('ºÅ:˜)n} rare˘ Hence, (j) is maintained well only be-
fore a non-final V; indeed, we generally have: ('p™i) {('pE,)n} pZe˚ (pe'i) {(pe',i)n}
pZs˘
Consonant clusters are simplified, as in: (di'r™;k) {(_i'ºEkt)n} direct˚ (di'r™;t) {(_i-
'ºEkt)n} directe˚ (os'ky;r) {(Öp's©y:˜)n} obscur˚ (&aòek'ti;f) {(&ÅD\™©'+if)n} adjectif˚ (se-
'tambr+) {(s™p'tŒ:bX)n} septembre÷ (a've) {(Å'vE©)n} avec÷ but: (sò) in words such as:
dix˚ six˚ alors˚ lors˚ gens˚ Sx˚ cSx˚ cours˚ vers˚ jad¤˚ av¤˚ moins˚ Roux˚ Poux˚ †iers˚
Arno¨˘
˛ere is no devoicing, as can be seen in the examples given, neither for V, nor
for C. Lia¤on is rare, as a matter of fact, it only occurs with /z/ in articles, pro-
nouns and plural masculine adjectives; however, there is no concept of a çdisjunc-
tive hÇ, hence, the use of linking and elision, ™: (le'zø;mÈ) {(l™'z∏m)n} lπ hommπ˚
(&meza'mi) {(&mezÅ'mi)n} mπ am¤˚ (&noza'mi) {(&nPzÅ'mi)n} nos am¤˚ (&l°rza'f™;r+)
4. french 175
/e, E/ (eò, ™0, ™0È, ’e) /o, O/ (oò, ø0, ø0È, ’o)
/°, §/ (°ò, #0, #0È, ’°) /ù/ (È, r+, +r)
/í/ (™ö0, ™e«ò, ’eö) /Ú/ (øö0, øo«ò, ’oö)
/^/ (#ö0, #°«ò, ’°ö)
/a/ (a) /Õ/ (aö0, aå«ò, ’åö)
(&pA-)t} pât¤sier÷ we also have an expansion of its frequency, to practically all cas-
es of /a|, wa|, av|, az|, as|, aZ|, aj|, aö|, aK|, ’aK/ (with possible fluctuation, especially
for /av|, az|/ and -ation˚ which have both /A/ and /a/). ˛ere is actual neutraliza-
tion too, as for: ('kå) which goes for both ('©Å) {('kA)t} cø and ('©Å) {('©Å)t} k÷ how-
ever, the contextual distribution –with ç/A/Ç in an intoneme– leads to situations
such as: ('sÅ '\E≈ 'så) ça… j'Pme ça˘
˛ere is fluctuation between (wå|) and (wÅ|) (the latter is generally considered
to be more uneducated), up to (w™), decidedly dialectal: ('mwå÷ 'mwÅ÷ 'mw™) moi˘
˛e çoldÇ opposition between /E/ and /E:/ is stable, we have for example: ('mEt5)
/'mEtK/ mettre˚ ('mE™t5) /'mEtK/ {/'mE:tK/t} maître÷ ('fEt) /'fEt/ fPte˚ ('fE™t) /'fEt/
{/'fE:t/t} fête˚ ('sEn) /'sEn/ sPne˚ ('sE™n) /'sEn/ {/'sE:n/t} scène (which now, in neutral
European French, is only: ('mEtX, 'fEt, 'sEn)n); whereas for the çnewÇ opposition
constituted by ç/e:/Ç, we have examples such as: ('neI\) ç/'ne:Z/Ç nRge˚ ('bReI©)
ç/'bKe:k/Ç bréque (or brake] (/'nEZ, 'bKEk/n ('nE:\, 'bºE©)n).
4.4.4.2. In checked stressed syllables in /v, z, Z, K, vK/, all recurring V are real-
ized as diphthongs (here are only some examples): ('lIiv5) /'livK/ livre˚ ('pYyR) /'pyK/
pur˚ ('RUu\) /'KuZ/ rouge˚ ('Rå∏\, 'RÅå\) /'KaZ/ rage (('li:vX, 'py:˜, 'º¯:\, 'ºÅ:\)n). ˛e
same occurs for /E:, e, °, o, A/ in an intoneme for any checked syllables (without
repeating examples): ('f°Yt5) /'f°tK/ fStre˚ ('\oUn) /'Zon/ jaune (('f°:tX, '\P:n)n), and
even in a preintoneme for unchecked syllables (again, without repetition): (&ÅR™e-
'te) /aKe'te/ arrêter˚ (neI'\e) /ne'Ze/ nRger˚ (f°Y'tÍe) /f°'tKe/ fStré˚ (\oU'nå∏t5) /Zo-
'natK/ jaunâtre˚ (få∏'/e) /fa'Se/ fâ>é ((&Ře'te, ne'\e, f°'tüe, \P'nÅtX, fÅ'/e)n).
In the broader or more uneducated accent, there is even the fusion of these two
characteristics, whereby for an unchecked syllable in a preintoneme, the V often
diphthongize if followed by /v, z, Z, K, vK/ (the first vocogram of § 4.6.3 shows (™e',
#°', øo')): (&ÅqIi'Re) /ati'Ke/ attirer˚ (\Yy\'mÅA) /ZyZ'mÕ/ jugement˚ (&epUu'ze) /epu'ze/
épo¨ée˚ (t™e'R™e) /tE'Kí/ terrPn˚ (b#°'Re) ç/b§'Ke/Ç bSrrée˚ (°Y'R°) /°'K°/ hSrSx˚
(&øRløo'\E™R) /OKlO'ZEK/ horlogère˚ (&ekÍÅå'ze, -å∏-) /ekKa'ze/ écrøé÷ or if they are /e, °,
o/ (even when non-derived or in syntagms): (leI'pªe) /le'pje/ lπ pieds˚ (d°Y'm™e)
/d°'mí/ dSx mPns˚ (soU'få) /so'fa/ sofa˘
Nasalized V diphthongize in an intoneme (¤ both in tonic and pretonic sylla-
bles), but they are short in a preintoneme: ('f™e) /'fí/ fin˚ (s™e'qYyR) /sí'tyK/ cRnture˚
(o'©}J) /o'k^/ aucun˚ (}J'\UuR) /^'ZuK/ un jour˚ (mÓ9'nÓ9) /mÚ'nÚ/ mon nom˚ (\ÅA-
'tÅA) /ZÕ'tÕ/ j'entends÷ (&™fi'ni) /ífi'ni/ infini˚ (&}nÅ'mi) /^na'mi/ un ami˚ (&mÓnÅ'mi)
/mÚna'mi/ mon amie˚ (Å&nÅtÅA'dÅA) /ÕnÕtÕ'dÕ/ en entendant (('fì, s3'+y:˜, P'©^, }-
'\¯:˜, mÚ'nÚ, \Œ'tŒ÷ &3fi'ni, &}nÅ'mi, &mÚnÅ'mi, Œ&nŒtŒ'dŒ)n).
k.s'qI©) /akus'tik/ acou®ique {(fi'lip, y'ni©, k¯'zin÷ &püimi'+i:v, &\y˜i'_i©÷ &minis'tE:˜, &Å-
k¯s'+i©)n). ˛e less broad accent can limit the occurrence of (I, Y, U) only in
checked syllables, or only in stressed checked syllables, as well; the decidedly less
broad accent, and more controlled, could have no (I, Y, U) at all.
Again, /i, y, u/ present two other peculiarities (except in slow or controlled
speech); indeed, when unstressed, they are devoiced between voiceless C, in
checked or unchecked syllables: (&kÓfN'qYyR) /kÚfi'tyK/ confiture˚ (&åÍqNs'qI©) /aKtis-
'tik/ arti®ique˚ (&QIspÛ'te) /dispy'te/ d¤puté˚ (&deku'på∏\) /deku'paZ/ découpage
{(&kÚfi'+y:˜, &ÅÆ+is'+i©, &_ispy'te, &dek¯'pÅ:\)n).
˛ey can be dropped between a constrictive or sonant and another C, or be-
tween a stop and a non-stop (even other V˚ above all, /e/): (&pÍøfe's§#R, pÍøf's-) /pKO-
fE's§K/ profπsSr˚ (&Yniv™Ísi'te, &Ynv™Ís'te) /ynivEKsi'te/ université˚ (&Åbili'te, &ÅbIú'te)
/abili'te/ habilité˚ (&mÅNI'fI©, mÅ˚'-, mÅJ'-, mÅ9'-) /maNi'fik/ magnifique˚ (bI'zå∏R, b'z-)
/bi'zaK/ bizarre˚ (pø&pylå∏'sªÓ9, &pøpúå∏-) /pOpyla'sjÚ/ population˚ (e&pyRå∏'sªÓ9, &e-
pÍå∏-) /epyKa'sjÚ/ épuration˚ (&bujø'ne, bjø-) /bujO'ne/ bouillonner˚ (&pinÅ'je, pnÅ-)
/pina'je/ pinPller˚ (&øRe'je, øR'je) /OKe'je/ orRller˚ (&lÅkø'mOd, lÅ©'≈Od) la commode
{(&püÖf™'s§:˜, &yni&v™Æsi'te, Å&bili'te, &mÅNi'fi©, bi'zÅ:˜, pÖ&pylÅ's¿Ú, e&py˜Å's¿Ú, &b¯,Ö'ne,
&pinÅ',e, &Ö˜e',e, &lÅkÖ'm∏d)n). Even /ù/ drops more often than in European French:
('lY© s+pÍø'mEn, 'lYks p-) Luc se promène {('ly© s#püÖ'mEn)n}, and presents even dis-
tributions which are often considered uneducated, like: ('fE™R+ú 'f¨) fPre le fou
{('fE;˜ l#'fU)n}. Furthermore, there is generally no vowel adjustment: (™'me, ™'mE)
–in a broader accent (™e-)– {(e'me, ™'mE)n} Pmé˚ Pma¤˘
4.4.4.4. For the C, besides the examples already given, we can observe that /nj,
n¥/ are (~j, ~¥) (on a broad, uneducated level /nj/ becomes /N/): (pÅ'~je) /pa'nje/
panier˚ (ÅA'~¥i) /Õ'n¥i/ ennui {(pÅ'~,e, Œ'~¥i)n); /N/ varies between (N) and (J):
(mÓ9'tå∏N, -'tå∏J) /mÚ'taN/ montagne˚ (&Ås™N'mÅA, &Ås™J'-) /ÕsEN'mÕ/ ensRgnement
178 a handbook of pronunciation
4.4.4.5. ˛e groups with final C are often simplified, above all in less controlled
language, even in the case of /0K, 0l/ (before a pause too): ('wEs) /'wEst/ oue®˚ (&åÍ-
/i'tE©) /aKSi'tEkt/ ar>itecte˚ (&søsªÅ'lis) /sOsja'lism, -ist/ social¤me˚ -i®e˚ ('mYs©)
/'myskl/ m¨cle˚ ('Ó9á, 'Ó9˙) /'Ú:gl/ ongle˚ (øÍ'©Es) /OK'kEstK/ or>e®re˚ ('Ó9b, 'Ó9m)
/'ÚbK/ ombre÷ on the other hand, especially on an uneducated level, we can have
/`/ = /t/: (\¥i'jEt, \¥I-) /Z¥i'jE/ juillet˚ (qi'©Et, qI-) /ti'kE/ ticket˚ (d+'bUt) /dù'bu/ de-
bout˚ ('pÍEt) /'pKE/ prêt˚ ('~¥It) /'n¥i/ nuit˚ ('lIt) /'li/ lit˚ (kÓ9'púEt) /kÚ'plE/ complet˚
(bRy'nEt) /bKy'nE/ Brunet˚ (&møRi'sEt) /mOKi'sE/ Mor¤set˚ (tÅl'bOt) /tal'bo/ Talbot˘
Cases such as (p'qi tÅ'mi) /pù'ti ta'mi/ petit ami and (p'qI tÅ'mi, p'qIt Å'mi) /pù'ti
ta'mi/ petite amie maintain the distinction of the V in a checked syllable (of petite].
˛e tonogram gives the intonation of Canadian French.
4. french 179
Text
4.5.0. ˛e story †e North Wind and the Sun follows, given in seven di‡erent
çnormalizedÇ versions. We start with the (neutral) French pronunciation of (neu-
tral British) English – this is the first step of the phonetic method (the written text
is given in § 2.5.2.0). ˛e French translation follows, in its neutral version; then
the çinternationalÇ accent, followed by the çmediatic/ParisianÇ one and, lastly, the
Marseilles and Canadian versions.
At the end, as always, there is the version which gives the English pronuncia-
tion of French, by neutral British speakers, fluent in French (after prolonged con-
tact with native speakers, but with no help from the phonetic method), who have
adequately learned the relative prominences, but who substantially use segmental
and intonation elements which are typical of neutral British English (for reference
purposes, although, of course, a neutral accent is not so common). Obviously, the
same principle is valid for the foreign pronunciations of English, given first.
Speakers of American English could prepare their own version of the French
pronunciation of English and of their pronunciation of French, as an excellent ex-
ercise, by listening to native speakers, best of all after recording them. Of course,
speakers of other languages could do the same thing. ˛e author would be happy
to receive their transcriptions and recordings, both in case of help –should they
need it– and to make their contribution known to others (possibly in our website
on canIPA Natural Phonetics – Ô § 0.12).
French Text
4.5.2. La b¤e et le solRl se d¤putPent˚ un jour, prétendant l'un comme l'Qtre être
le pl¨ fort, lorsqu'ils virent s'avancer un voyagSr, enveloppé dans son mant›. Lπ
dSx adversPrπ décidèrent, alors, que serPt déclaré vPnquSr celui qui, le premier,
parviendrPt à le lui fPre ôter.
180 a handbook of pronunciation
La b¤e se mit, alors, à sou·er de toutπ sπ forcπ; ma¤, pl¨ elle sou·Pt, pl¨ le
voyagSr se serrPt dans son manteQ, tant et si bien, qu'à la fin, la pQvre b¤e dut
renoncer à sπ intentions. Le solRl se montra, alors, dans le ciel, et le voyagSr, qui Q
bout d'un moment commençPt à avoir >Qd, retira son manteQ. C'e® Pnsi que la
b¤e dut reconnaître que le solRl étPt le pl¨ fort dπ dSx.
Tu ø Pmé cette hi®oire? Tu vSx la réentendre?
Canadian pronunciation
4.6.0. We would like to add some further information: about the a‡ected Pa-
risian accent, in the form of phonosynthesis; and about the Parisian variants of the
banliSπ˚ compared to mediatic Parisian (Ô § 4.4.2.1-2 “ § 4.5.2.3); lastly about
Canadian variants (Ô § 4.4.4.1-5 “ § 4.5.2.5). In NPT/HPh, you can find the pho-
nosyntheses of Acadian French (acadien˚ east Canada, § 21.10) and cadien, Cajun
(from Louisiana, § 21.16).
A‡ected Parisian
4.6.1. ˛is accent can be heard from representatives of the aristocracy and up-
per classes, but is not usually highly valued. We give the realizations of the vowels
/˚ / (˚ 2 2 ' 2 1 ' 2 1 Ì 2)
4. french 183
Canadian variants
4.6.3. As well as what has already been given in § 4.4.4.1-5 “ § 4.5.2.5, we add
both some broader variants (≠, ≠≠, ≠≠≠) and less broad variants (±): /e, °, o, A/ (±ee',
±°°', ±PP', ±AA'); /wA/ (wå, ≠wù, ±wÅ) (çdialectalÇ (≠≠≠w™)); as also /EK0/ = /≠≠≠aK0/
is çdialectalÇ; /u/ (±¯, ±¨¯, ±¨, ±P, ±P¨), /O/ (±’Ö, ±∏, ±∏Ö). We can observe the use
of (≠≠¤, ≠≠T, ≠≠¨) (considerably centralized), which is independent from the moder-
ate systematic advancement of back phonemes (of ç±Ç type). We also indicate the
uvulo-postalveolarized velar rounded realization of /Kò/ (= (<)) and its çvocaliza-
184 a handbook of pronunciation
tionÇ through (Ù, +, P÷ Ä, §, ∏÷ √) (as the last element at the end of the diphthongs
e triphthongs), in /iK, yK, uK÷ EK, §K, OK÷ aK/, given in the first vocogram (which
will be dealt with and illustrated with examples in the book in progress indicated
in the bibliography). Naturally, the vocograms will be less grouped up, and more
numerous, to properly demonstrate the various peculiarities (it was not possible
to place the variant (O÷P°) of /O(:)0ò/). In rural accents, and above all for speakers
who are not so young, /S, Z/ (/, \) have di‡erent, peculiar and broad realizations,
as well as (S, Z): (», «÷ S, q÷ ¢, Ú÷ '·, ’h, 'â, ’H). On an uneducated level, before a
rounded V˚ /f/ = (å). ˛e French may have di‚culties in understanding Canadian
French, especially if the accent is very broad and uneducated.
/Ú/ (±9Ú)
/í/ (±í™, ≠ì/ò ≠a3), /^/ (±^}) /O/ (±’Ö, ≠’ê, ≠’+, ±∏, ±∏Ö)
/Eò/ (≠Ä÷), /E:0ò/ (≠≠ÅÉ, ≠≠≠aÙ)
/O(:)0/ (≠≠åø)
/a(:)0/ (≠≠å¨)
/Õ/ (≠ÅÚ, ≠ìíò, ≠≠As0ò, ±Œ_, ±úÕ) /A(:)0/ (≠åø, ≠≠ùø, ≠≠å¨)
5. German
Vowels
5.1.1. û 5.1 shows the neutral articulations of modern neutral German, which
presents seven pairs of opposing short and long vowels in stressed syllables, with
the important addition of a timbric di‡erence too (except for /a, a:/, whose tim-
bric di‡erence is now regional). As can be seen from û 5.1, too, long vowels are
realized by closer vocoids, and a modern phonemic transcription must obviously
reflect this characteristic: /i:, I÷ e:, E÷ a:, a/, /y:, Y÷ °:, §/, /u:, U÷ o:, O/.
But we have to add another çinconvenientÇ long phoneme, /E:/, which opposes
both /e:/ (because of timbre) and /E/ (because of length). Frequently, it merges with
/e:/, mainly in the North, but this pronunciation is not recommendable as neutral,
all the more so as current spelling clearly shows it. Indeed, for a modern neutral
pronunciation, the most appropriate realization is an intermediate one, (™:), in fact
in û 5.1 the other variants are shown as separate ones.
û 5.1. German monophthongs.
/i:, ’i/ (i:, i), /y:, ’y/ (y:, y) /u:, ’u/ (u:, u)
/I/ (I), /Y/ (Y) /U/ (U)
/e:, ’e/ (e:, e), /°:, ’°/ (°:, °) /o:, ’o/ (o:, o)
/È/ (¢, È˘º, ºÈ, º¢ò)
/E:/ (™:, ’™;) {(E:÷ e:÷ ’e;)}
/O/ (O, ’ø)
/E/ (E, ’™), /§/ (§, ’#)
/Ú/ (…) {“ (√)}
/a/ (a), /a:/ (a:)
In addition, we find the typical unstressed phoneme, /È/ (¢), which is realized
di‡erently when close to /K/: (È) (/È˘K, KÈ/ (Ⱥ, ºÈ), except for word-final /KÈò/ (º¢)),
and in checked syllables with /K/ (as we will see shortly after, in the examples),
where we actually have /Ú/ (…); although it derives from //ÈK//).
186 a handbook of pronunciation
Let us now see some examples: ('mi:t¢) /'mi:tÈ/ M¤te˚ ('mIt¢) /'mItÈ/ Mitte÷ ('ve:k)
/'ve:k/ Weg˚ ('vEk) /'vEk/ weg÷ ('Sta:t) /'Sta:t/ Staat˚ ('Stat) /'Stat/ Stadt÷ ('hy:t¢) /'hy:tÈ/
Hüte˚ ('hYt¢) /'hYtÈ/ Hütte÷ ('h°:l¢) /'h°:lÈ/ Höhle˚ ('h§l¢) /'h§lÈ/ Hölle˘ In addition,
('ö™:º¢÷ 'öE:-) /'E:KÈ/ ∏hre˚ ('öe:º¢) /'e:KÈ/ Ehre and ('th™:l…÷ 'thE:-) /'tE:lÚ/ Täler˚ ('thEl…)
/'tElÚ/ Teller÷ finally, ('zOn¢) /'zOnÈ/ Sonne˚ (â¢'thüU˙kô) /gÈ'tKU˙kô/ getrunken˚ (âÈ-
'ºa:d¢) /gÈ'Ka:dÈ/ gerade˚ ('öUnzȺ¢) /'UnzÈKÈ/ unsere˚ ('öUnzȺ…) /'UnzÈKÚ/ unserer˘
˛e last example (as well as a previous one – Täler] shows (…) /Ú/ -er˚ which is
more traditionally rendered phonemically as //ÈK// (and even as //Èr//); but /Ú/ is
more convenient, since its most normal and frequent realization is as an intense
faryngealized uvular approximant (…). It is a fact that a çvocalizationÇ is also pos-
sible, as (√) (and this symbol is certainly more appropriate than ç(å)Ç, which how-
ever is found in many publications), that is represented by a broken-line white
marker (û 5.1); but it does not occur in a systematic way. As a matter of fact, more
often we find (…), or even (1, &); occasionally, even something like ((√‰, √˜, √º)) (¤
a back-central vocoid with traces of the contoids shown) is surely apter than a sim-
ple and pure vocoid.
5.1.2. We must add that, in unstressed syllables, /E, §, O/ are realized as closer
vocoids, (™, #, ø) (Ô û 5.1): (ö™nt'SUldIgU,) /Ent'SUldIgU˙/ En˛uldigu«˚ (öøk-
'tho:b…) /Ok'to:bÚ/ Oktober˚ ('ph§‰k#lt) /'p§Kk§lt/ Pörkölt˘ Again, in unstressed
syllables, also /i, e÷ y, °÷ u, o/ are to be added, even in phonemic transcriptions:
(&miküo'fo:n) /mikKo'fo:n/ Mikro#on˚ (&ºeali'th™:t÷ -E:t) /Keali'tE:t/ Realität˚ (&∂yna-
'mi:t) /dyna'mi:t/ Dynamit˚ (&ö°kolo'gi:) /°kolo'gi:/ Ökolog¤˚ (&öuni&v™‰zi'th™:t÷ -E:t)
/univEKzi'tE:t/ Universität˚ (&fonolo'gi:) /fonolo'gi:/ Phonolog¤˘ We must say at once
that, in current or familiar pronunciation, unstressed /i, e÷ y, °÷ u, o/ easily become
/I, E÷ Y, §÷ U, O/ (I, ™÷ Y, #÷ U, ø) (except when word-final): (&mIküø'fo:n, &º™alI'th™:t
{-E:t}, &∂Yna'mi:t, &ö#kølø'gi:, &öUnI&v™‰zI'th™:t {-E:t}, &fønølø'gi:), and (ö™nt'SUldI&gU,,
¢n-), but: ('âUmi) /'gUmi/ Gummi˚ ('zi:mi&le, -I-) /'zi:mile/ Simile˚ ('phúa:ty) /'pla:ty/
Platy˚ ('qhe:bu) /'qe:bu/ Zebu˚ ('mEksi&ko, -I-) /'mEksiko/ Mexiko˘
Unstressed /a, a:/ pose no problems to modern neutral pronunciation, since
they always become (a): (öa'ga:v¢) /a'ga:vÈ/ Agave˚ (öa'me:ºi&ka, -I-) /a'me:Kika/ Ame-
rika˚ ('öa:füi&ka, -I-) /'a:fKika/ Afrika˚ (&öafüi'kha:t¢, -I-) /afKi'ka:tÈ/ A‡rikate˘ Instead,
in certain transcriptions of traditional neutral pronunciation, we may find ç/A/Ç,
which somehow corresponds to ç/A:/Ç; but there is not much coherence at all a-
mong di‡erent dictionaries and textbooks: (öa'ga:v¢, öA'gA:-÷ öa'me:ºi&ka, öA'me:ºi-
&kA, -I-÷ 'öa:füi&ka, 'öA:füi&kA, -I-÷ &öafüi'kha:t¢, -A:t¢). Except for some loanwords, the
Ÿ∑∂å systematically gives ç/A/Ç (¤ /’A/) for unchecked grapho-syllables (a çstrangeÇ
criterion indeed, to be frank), whereas Langenscheidt's dictionaries still using /A:/
mark /’a/.
5.1.3. û 5.2 shows the diphthongs of neutral German: three basic and two sec-
ondary ones (the latter are marked in grey and are used in English loanwords).
˛eir neutral duration, both in stressed and unstressed syllables, is always interme-
diate between the short (('öal¢) /'alÈ/ alle) and the long (('öa:l¢) /'a:lÈ/ Ahle) ones. ˛e
three genuine diphthongs are (ae, ao, OY) /ae, ao, OY/ (often badly represented as
5. german 187
ç/ai, au, oy˙oi/Ç): ('Êaen) /'baen/ BRn˚ ('öael¢) /'aelÈ/ Eile˚ ('mae) /'mae/ MP÷
('Êaom) /'baom/ BQm˚ ('öaog¢) /'aogÈ/ Auge˚ ('thao) /'tao/ TQ÷ ('lOYtó) /'lOYtó/ lT-
ten˚ ('öOYl¢) /'OYlÈ/ Eule˚ ('nOY) /'nOY/ nS˘ ˛e two additional ones, in grey, in the
most spontaneous pronunciation type, merge into /e:, o:/ (but this depends on
words and speakers, too): ('le:di, 'l™Idi) /'leIdi/ Lady˚ ('âe:, 'â™I) /'geI/ gay˚ (&no:'hao,
&nøU'hao) /noU'hao/ Know-how˚ ('So:, 'SøU) /'SoU/ Show˘
˛e broken-line white markers show two possible realizations of /ae, ao/, in re-
duced forms (Ô § 5.3.1), (åe, åo), as in: (åen) /aen/ Rn˚ (åos) /aos/ Qs (also (ÖY)
/OY/: (ÖYÂ) /OYÂ/ S>). ˛e marker corresponding to (å) is also valid for reduced
forms with /a/, as a possible variant: (∂ås) /das/ das. We have intentionally not
placed it in the monophthong vocogram (û 5.1), also to avoid confusion with the
–quite frequent, but not recommendable– use of ç(å)Ç to indicate our (‰, …) (or,
at best, (√), see û 5.1 and the first part of § 5.1, and the second part of § 5.2).
û 5.2. German diphthongs.
Consonants
5.2. û 5.3 shows the necessary consonant articulations which are needed for a
good German pronunciation, including taxophonic variants. ˛e examples will
even show the devoicing of voiced stops, after pauses or voiceless consonants and
of voiced constrictives only after voiceless consonants.
188 a handbook of pronunciation
postalveo-palatal
pharyngealized
denti-alveolar
protruded
labiodental
prepalatal
laryngeal
uvular
alveolar
bilabial
palatal
uvular
dental
velar
ö m (M) (n) n (~) ˙ (,)
F pb 47 kg (ö)
Ô ∫ q c
ƒ (6) f v
_ s z S Z (À) Â (,) /x/ (X º)
ß /j/ (‰) h
ó ([) /K/
‹ (R) l (¬)
+(º, ¸, n) “ /õ, ó/ (õ, è, ó, «, ô, ?÷ :, !, õ, ò, ô, /, ¡, |, ˘) (Ô û 1.9.1)
Nasals
5.2.1.1. ˛ere are three phonemes for this manner of articulation, /m, n, ˙/,
with a certain number of taxophones by assimilation ((m), (m, M, n, n, ~, ˙, ,),
and (˙, ,)): ('man) /'man/ Mann˚ ('hEmt) /'hEmt/ Hemd˚ ('ve:m) /'ve:m/ wem˚
('naen) /'naen/ nRn˚ (k'ni:) /k'ni:/ Kn¤˚ ('qhe:n) /'qe:n/ zehn˚ ('thüI˙kô) /'tKI˙kó/
trinken˚ ('öam&bi;tó) /'anbi:dó/ anb¤ten˚ ('fYMf) /'fYnf/ fünf˚ ('öUns) /'Uns/ uns (strict-
ly (('öU˙s)), but (n) is su‚cient), (∂i'thant¢) /di'tantÈ/ d¤ Tante (strictly ((fli'4han4¢))),
('vYnSó) /'vYnSó/ wün(en, (khøn'ºa:d¢) /kOn'Ka:dÈ/ Konrade, ('mY~«) /'mYnÂó/
Mün>en˚ ('öa˙Ín) /'a˙Ín/ a«eln˚ ('lE˙st) /'lE˙st/ lä«®˚ ('lE˙…) /'lE˙Ú/ lä«er˚ ('zI˙ô)
/'zI˙ó/ si«en˚ ('qhU˙¢) /'qU˙È/ Zu«e˚ (Ê¢'dI˙U,) /bÈ'dI˙U˙/ Bedi«u«˚ ('qhaetU,)
/'qaetU˙/ ZRtu«˚ ('qhaetU,?) /'qaetU˙ó/ ZRtu«en˘ (For /U˙ò, U˙óò/ we have
(U,ò, U,?).)
Please, note that the sequence /nj/ –and /lj/ as well– has two di‡erent realiza-
tions in stressed syllables, for tautosyllabic (after /é:/) or heterosyllabic (after /é/)
elements: ('li:n,¢) /'li:njÈ/ Lin¤˚ (Sam'pha~,…) {((-a~-,…))} /Sam'panjÚ/ ≥ampagner.
In lexical compounding, even for the negative prefix un-, in slow — accurate
pronunciation, people try to maintain (n), whereas –currently– it is assimilated
to the place of articulation of the following C˘
As an actual compromise, here we will show that it is possible to have articula-
tions which maintain an apical contact but add secondary coarticulations (without
full contacts). ˛ese coarticulations are: bilabial (º), labiodental (¸), velar (n):
('öUm&b™ndIÂ, 'öUº-) /'UnbEndIÂ/ unbändig˚ ('öUM&vI‰kúIÂ, 'öU¸-) /'UnvIKklIÂ/ unwirk-
5. german 189
li>˚ ('öU˙g¢&ºe;gÍt, 'öUn-) /'UngÈKe:gÍt/ u«eregelt˘ Below, we will find that a couple
of these phones occur as intense too, (/, ¡).
In other cases, with less di‡erent coarticulations, assimilation (which is often
considered to be less recommendable, on pure graphical and morphological bases)
appears to be less noticeable, thus it is applied in a spontaneous though uncon-
scious way.
5.2.1.2. Including two intense nasal phonemes too, /õ, ó/ (instead of //Èm,
Èn//), we have five of them, with all their various taxophones ((õ) for /õ/, and (õ,
è, ó, «, ô, ?) for /ó/) ('öa:tõ) /'a:tõ/ Atem˚ ('zi:bõ) /'zi:bó/ s¤ben˚ ('ne:mõ) /'ne:-
mó/ nehmen˚ ('Sa‰fè) /'SaKfó/ (arfen˚ ('hO∫ès) /'hO∫ós/ Hoπens˚ ('khaqó) /'ka-
qó/ Ka˛en˚ ('∂u:dó) /'du:dó/ Duden (rigorously (('flu:7~))), ('phancó) /'pancó/ pan-
˛en˚ ('öap&ʺEnó) /'apbKEnó/ abbrennen˚ ('m™:+«÷ 'mE:+-) /'mE:tÂó/ Mäd>en˚ ('fúi:-
gô) /'fli:gó/ fl¤gen˚ ('za:gô) /'za:gó/ sagen˚ ('qhaetU,?) /'qaetU˙ó/ ZRtu«en˚ ('ʺI-
˙ô) /'bKI˙ó/ bri«en˚ ('vIº?, -º¡, -I‰,, -I‰n) /'vIKó/ wirren˚ ('qha:ló) /'qa:ló/ zahlen˚
('âe:ó) /'ge:ó/ gehen˘
Let us observe, however, that there are cases where intense nasals are less favored,
as results from such forms as ('ÊIndód¢n) /'bIndódÈn/ bindenden (with theoretical
sequences of (M), ('ÊIndódó), which are possible in faster speech, but mostly not
in intonemes).
As can be seen, it is more convenient to distinguish between /M/ and /Èö/. It is
su‚cient to know that, if pronunciation becomes slower or more controlled, it is
easy to go back to (ó), or (¢n): from ('zi:bõ, 'za:gô) to ('zi:bó, 'za:gó) and ('zi:b¢n,
'za:g¢n), or from ('qha:ló, 'âe:ó) to ('qha:l¢n, 'âe:¢n). However, generally, except in
intonemes, forms such as gehen can even become ('âe:n).
5.2.1.3. In the previous section, we have shown up to six taxophones for the
two intense nasals, /õ, ó/. At first sight, they might seem to be excessive; neverthe-
less, by concretely considering actual phonetic reality, it is clear that only in quite
approximate transcriptions could it be possible to renounce them. Besides, in or-
der not to complicate things çwithout reasonÇ, even nine could be useful, indeed:
(:, !, õ, ò, ô, /, ¡, |, ˘). (All this without considering a denti-alveolar ((~)), for
which a simple (ó) is su‚cient, as in (&öaen¡'gu:tó 'tha˙k) ((&öaen¡'gu:4~ '4ha˙k))
/aenó'gu:tó 'ta˙k/ Rnen guten Tank.)
However, we will show them only here, since these examples belong to actual
sentences taken from passages of connected speech, and we will put them in dou-
ble square brackets that indicate an objective precision, but perhaps a less pressing
one. We are dealing with sequences where /ó, õ/ and the tautosyllabic C before
them, /˘0M˘/, determine their exact place of articulation, (õ, è, ó, «, ô, ?), just
as in the examples seen above. Even the initial C of following syllables a‡ect the
result, by coarticulation, in a more or less evident way, according to the places of
articulation directly involved. ˛is happens every time two phones come into con-
tact with one another; and, of course, this is done to render pronunciation more
fluent and natural (certainly not to further complicate life!).
Among the taxophonically relevant actual cases in German, after experiencing
190 a handbook of pronunciation
all possibilities, the following instances are worthwhile mentioning and exemplify-
ing. As a matter of fact, should they not be applied, the result would be articulato-
ry straining, indeed.
5.2.1.4. For practical reasons, we will start from /0ó/ (0õ). We can see that
when they are followed by labiodentals (/∫, f, v/), the coarticulation of (õ) and
(∫, f, v) produces an intense labiodentalized bilabial ((:)): (('zi:b: '∫he:‰7¢)) /'zi:bó
'∫e:KdÈ/ s¤ben Pferde˚ (('zi:b: 'fE4…)) /'zi:bó 'fEtÚ/ s¤ben Vetter˚ (('zi:b: 'va:gô)) /'zi:bó
'va:gó/ s¤ben Wagen˘
‹en /0ó/ (0õ) is followed by laminal C, /t, d, q, z, n, l/, we have a denti-al-
veo–bilabial ((!)): (('zi:b! '4ha:g¢)) /'zi:bó 'ta:gÈ/ s¤ben Tage˚ (('zi:b! '7e:&ma‰k)) /'zi:bó
'de:maKk/ s¤ben DM˚ (('zi:b! 'qhIm…)) /'zi:bó 'qImÚ/ s¤ben Zimmer˚ (('zi:b! 'nEÂ4¢))
/'zi:bó 'nEÂtÈ/ s¤ben Nä>te˚ (('zi:b! 'lEn7…)) /'zi:bó 'lEndÚ/ s¤ben Länder˘
≈nally, for /0ó/ (0õ) + /k, g, K/, we find an intense velarized bilabial nasal ((ò)):
(('zi:bò 'khaqó)) /'zi:bó 'kaqó/ s¤ben Ka˛en˚ (('zi:bò 'gE˙¢)) /'zi:bó 'gE˙È/ s¤ben
Gä«e˚ (('zi:bò 'º™:7…÷ 'ºE:-)) /'zi:bó 'KE:dÚ/ s¤ben Räder˘
5.2.1.5. As far as /0õ/ (0õ) -em is concerned (which is not derived by assimila-
tion, but is original, for //0Èm//), three groups deserve careful attention. In fact,
for /0õ/ (0õ) + /∫, f, v/, we find a labiodental (è): (mIt&∂i;zè'∫he:‰t) /mItdi:zõ-
'∫e:Kt/ mit d¤sem Pferd˚ (öIn&di;zè'valt) /Indi:zõ'valt/ in d¤sem Wald˘
For /0õ/ (0õ) + /t, d, q, z, n, l/, once again, we have a denti-alveo–bilabial
((!)): ((qhu&7aen!'qhIm…)) /qudaenõ'qImÚ/ zu dRnem Zimmer˚ ((bae&7i;z!'lIÂt))
/baedi:zõ'lIÂt/ bR d¤sem Li>t˘ For /0õ/ (0õ) + /k, g, K/, a velar–bilabial ((õ)):
((fo‰&maenõ'khO∫)) /fo:Kmaenõ'kO∫/ vor mRnem Koπ˚ ((mI4&flaenõ'ºI˙)) /mItdae-
nõ'KI˙/ mit dRnem Ri«˘
5.2.1.6. For /0ó/ (0ó) + /p, b, m/, for the third time we find a denti-al-
veo–bilabial ((!)): (('âu:4! 'phüaes)) /'gu:tó 'pKaes/ guten PrRs˚ (('âu:4! 'bal)) /'gu:tó
'bal/ guten Ball˚ (('âu:4! 'mO‰gô)) /'gu:tó 'mOKgó/ guten Morgen˘
For /0ó/ + /∫, f, v/, we have a labiodentalized alveolar ((/)): (('âu:4/ '∫hle:g…))
/'gu:tó '∫le:gÚ/ guten Pfleger˚ (('âu:4/ 'fa˙)) /'gu:tó 'fa˙/ guten Fa«˚ (('âu:4/ 'vaen))
/'gu:tó 'vaen/ guten WRn˘ For /0ó/ + /k, g, K/, we find a velarized alveolar ((¡)):
(('âu:4¡ 'khEl…)) /'gu:tó 'kElÚ/ guten Keller˚ (('âu:4¡ 'gas4)) /'gu:tó 'gast/ guten Ga®˚
(('âu:4¡ 'ºo:4&Ñaen)) /'gu:tó 'Ko:tvaen/ guten RotwRn˘
Besides, for /0ó/ (0ô) + /p, b, m/, we again find a velar–bilabial ((õ)): ((&âe;gõ-
'pho:ló)) /ge:gô'po:ló/ gegen Polen˚ ((&âe;gõ'bElg,«, -jó)) /ge:gô'bElgjó/ gegen Belg¤n˚
((&âe;gõ'mEksi&ko)) /ge:gô'mEksiko/ gegen Mexiko˘
For /0ó/ (0ô) + /∫, f, v/, we find a labiodentalized velar ((|)): ((&âe;g|'∫han4))
/ge:gô'∫ant/ gegen Pfand˚ ((&âe;g|'füa˙k&üaeÂ)) /ge:gô'fKa˙kKaeÂ/ gegen FrankrR>˚
(('âe:g|&va‰4)) /'ge:gôvaKt/ Gegenwart˘
For /0ó/ (0ô) + /t, d, q, z, n, l/, we have a denti-alveolarized velar ((˘)): (('âe:g˘-
&4hael)) /'ge:gôtael/ GegentRl˚ ((&âe;g˘'7ºae)) /ge:gô'dKae/ gegen drR˚ ((&âe;g˘'qhñae))
/ge:gô'qvae/ gegen zwR˚ ((&âe;g˘'zEks)) /ge:gô'zEks/ gegen se7˚ ((&âe;g˘'nOYn)) /ge:gô-
'nOYn/ gegen nSn˚ (('âe:g˘&laes4U,)) /'ge:gôlaestU˙/ GegenlR®u«˘
5. german 191
5.2.1.7. In the case of /0ó/ (0è) + /∫, f, v/, obviously, things do not change:
(m¢t'Sa‰fè 'fo:tos) /mIt'SaKfó 'fo:tos/ mit (arfen Fotos÷ equally, for /0ó/ (0è) + /t,
d, q, z, n, l/: (m¢t'Sa‰fè 'qh™:nó÷ 'qhE:-) /mIt'SaKfó 'qE:nó/ mit (arfen Zähnen˘
Instead, for /0ó/ (0è) + /k, g, K/, we have a velarized labiodental ((ô)): ((m¢t'Sa‰-
fô 'khúI˙ô)) /mIt'SaKfó 'klI˙ó/ mit (arfen Kli«en˘
For /0ó/ (0«) + /p, b, m/, for the fourth time we find a denti-alveo–bilabial
((!)): (('m™:+Â!&p™nz,o&na;4, -a˙-, -Å˙-, -Å-, -s,-÷ 'mE:+-)) /'mE:tÂópenzjona:t, -a˙-, -sj-/
Mäd>enpensionat÷ besides, + /∫, f, v/, we have a labiodentalized alveolar ((/)):
(('m™:+Â/&füOYntúIÂ÷ 'mE:+-)) /'mE:tÂófKOYntlIÂ/ mäd>enfrSndli>÷ then, + /k, g/, we
have a velarized alveolar ((¡)): (('m™:+¡&khúas¢÷ 'mE:+-)) /'mE:tÂóklasÈ/ Mäd>enklasse˘
For /0ó/ (0?) + /p, b, m/, we find for the third time a velar–bilabial ((õ)):
(('qhae4U˙õ &publi'qhi:º?, -º¡, -‰,, -‰n)) /'qaetU˙ó publi'qi:Kó/ ZRtu«en publiz¤-
ren˘
For /0ó/ (0?) + /∫, f, v/, we again have a labiodentalized velar ((|)): (('qhae4U˙|
f…'khaofè)) /'qaetU˙ó fÚ'kaofó/ ZRtu«en verkQfen˘
Lastly, for /0ó/ (0?) + /t, d, q, z, n, l/, once again we find a denti-alveolarized
velar ((˘)): (('qhae4U˙˘ 'le:zó)) /'qaetU˙ó 'le:zó/ ZRtu«en lesen˘
In grammemes (and frequent lexemes), as in˚ Rn˚ von˚ ins˚ uns˚ dann˚ (on˚
Montag, in current speech, it is not at all rare that /én/ + /[0]˘, [0]˘/ becomes (◊n,
–n, –) (with progressively stronger nasalization, up to a complete dropping of /n/):
(öIns'khi:no, öIs-) /Ins'ki:no/ ins Kino…
Stops
5.2.2.1. ˛ere are three diphonic pairs, /p, b÷ t, d÷ k, g/, which are peculiar be-
cause in word-final position (and in postnuclear position, but with some çodditiesÇ
that we will see) only the voiceless elements occur. In addition, /p, t, k/, after paus-
es and in stressed syllables (not preceded by /s, S/ that belong to the same lexeme),
are çaspiratedÇ. Besides, /t, d/ are denti-alveolar (and rigorously the most appropri-
ate symbols would be ((4, 7)), but it is su‚cient to use (t, d), unless it is thought to
be important to insist on this nuance).
Here are some examples: (pha'phi:‰) /pa'pi:K/ Pap¤r˚ ('phúaq) /'plaq/ Pla˛˚
('SpüE«) /'SpKEÂó/ spre>en˚ ('halp) /'halp/ halb˚ ('öapt) /'apt/ Abt˚ (öIÂ'pha‰k¢) /IÂ-
'paKkÈ/ i> parke˚ ('öUnt) /'Unt/ und˚ ('lu:p¢) /'lu:pÈ/ Lupe˚ ('öEb¢) /'EbÈ/ Ebbe˚ ('ÊIt¢)
/'bItÈ/ bitte÷ ('tho:t) /'to:t/ Tod˚ ('thüe:tó) /'tKe:tó/ treten˚ ('Stüa:s¢) /'StKa:sÈ/ Stra•e˚
('laet) /'laet/ LRd˚ (¢s'thüI˙kt) /Ès'tKI˙kt/ es trinkt˚ ('laed…) /'laedÚ/ lRder˚ ('∂a˙kô)
/'da˙kó/ danken˚ ('khInt) /'kInt/ Kind˚ ('khüaes) /'kKaes/ KrRs˚ (∂ås'khi:no) /das'ki:-
no/ das Kino˚ ({öae˙}k'nal) /{aen}k'nal/ (Rn) Knall˚ (â¢'bI‰g¢) /gÈ'bIKgÈ/ Gebirge˚
('âI˙) /'gI˙/ gi«˘
5.2.2.2. ‹at is too often described as çaspirationÇ of stops before pauses, is ac-
tually nothing but the çexplosionÇ of the phones (that we will show only here with
($)), not çaspirationÇ, which is a di‡erent thing: ('skalp$) /'skalp/ Skalp˚ ('tha:t$)
/'ta:t/ Tat˚ ('khEk$) /'kEk/ ke$˘ Some authors go as far as to indicate something like
192 a handbook of pronunciation
ç('thaphth¢, 'ºYthl¢)Ç, for ('thapt¢) /'taptÈ/ tappte˚ ('ºYtú¢) /'KYtlÈ/ rüttle˚ when they
simply want to hint at the fact that the transition from one stop to another, or to
any other phone, can be audible, that is çexplodedÇ (('4hap$4$¢, 'ºY4$ú¢)); but there
is certainly no need to mark such things.
By assimilation, in /pf, pv/, we have (]) + /f, v/: ('öa]&fal) /'apfal/ Abfall (di‡er-
ent from (∫) /∫/)˚ ('öa]&Ñe;‰tó) /'apve:Któ/ abwerten˘ In addition, as we have seen
for /n/ (§ 5.2.1), even /t, d/ assimilate to a following C: ('mIk&âIft) /'mItgIft/ Mit-
gi⁄˚ ('mIk&kømõ) /'mItkOmó/ mitkommen˚ ('mIp&ʺI˙ô) /'mItbKI˙ó/ mitbri«en˚
('mIp&≈™nS) /'mItmEnS/ Mitmen(˚ ('ʺo:k&khaofè) /'bKo:tkaofó/ Brot kQfen˚
('ʺo:p&ʺI˙ô) /'bKo:tbKI˙ó/ Brot bri«en˘ Note that we have (+) in: ('m™:+«÷ 'mE:+-)
/'mE:tÂó/ Mäd>en˘
5.2.2.3. Let us also consider: ('öap&úaot) /'aplaot/ AblQt˚ ('öapúa&ti;f) /'aplati:f/ Ab-
lativ˚ ('öap&üu;f) /'apKu:f/ Abruf˚ (öat'ÑE‰p) /at'vEKp/ Adverb˚ (&öat≈i'ºa:l) /atmi'Ka:l/
Admiral˚ ('Sto:f&úaen) /'Sto:flaen/ StovlRn˚ ('li:p&úo;s) /'li:plo:s/ l¤blos˚ ('ºe:k&úo;s) /'Ke:k-
lo:s/ reglos˚ ('m°:kúIÂ) /'m°:klIÂ/ mögli>˚ ('ni:túIÂ) /'ni:tlIÂ/ n¤dli>˚ ('hantúU,) /'han-
tlU˙/ Handlu«˚ ('öy:púIÂ) /'y:plIÂ/ übli>˚ ('va:knIs) /'va:knIs/ Wagnis˚ but: ('laebnIq)
/'laebnIq/ LRbniz˚ ('ºe:dn…) /'Ke:dnÚ/ Redner˚ ('âe:gn…) /'ge:gnÚ/ Gegner˚ ('va:gn…)
/'va:gnÚ/ Wagner˚ ('öa:dl…) /'a:dlÚ/ Adler˚ ('öa:dlIÂ) /'a:dlIÂ/ adlig˚ ('öa:dlIg¢) /'a:dlIgÈ/
Adlige˚ ('O‰dnU,) /'OKdnU˙/ Ordnu«˚ ('ºe:gl…) /'Ke:glÚ/ Regler˚ (zIg'na:l) /zIg'na:l/
Signal˚ (&zIgna'thu:‰) /zIgna'tu:K/ Signatur˚ ('öe:bn¢) /'e:bnÈ/ ebne˚ ('öy:bºIÂ) /'y:bKIÂ/
übrig˘
Actually, in this second group of forms, the consonantal sequences are voiced
because they belong to the same lexeme (with a possible diachronic drop of /È/ -e-]\
reden˚ gegen˚ Wagen˚ Adel˚ Regel˚ Orden˚ eben˚ über˘ On the contrary, in the first
group we have combinations (which may be clear or opaque) with grammemes
(¤ prefixes or su‚xes), so we will consider only three cases: Ab-[lQt]˚ [üb]-li>˚
from üben˚ and [Wag]-nis˘
Words of lofty or foreign origin generally maintain a voiced stop before sonants:
/bl/ Double˚ mobl¤ren˚ Problem˚ publik˚ Republik˚ sublim˚ Tablett˚ Variable÷ /bK/
Algebra˚ Fabrik˚ Kabrio˚ Kobra˚ Libretti®˚ Soubrette˚ Vibration˚ Zebra÷ however
/bK, pK/ Abrasion˚ Abrasit and /pK, bK/ abrupt÷ /dK/ Adresse˚ Hydrant˚ Madrider˚
Madrigal˚ Melodrama˚ Quadrant˘
Other cases: /gm/ Dagmar˚ Dogma˚ Egmont˚ Fragment˚ Magma˚ Paradigma˚
Phlegma˚ Pigment˚ Pragmatik˚ Pygmäe˚ Segment˚ Sigma˚ Stigma÷ /gn/ Agno®iker˚
Diagnose˚ ignor¤ren˚ indign¤rt˚ kognitiv˚ Magnat˚ Magnol¤˚ Magnus˚ Physiogno-
m¤˚ prägnant˚ Stagnation˚ and /gn, ˙n/ inkognito˚ Magnesit˚ Magnesium˚ magne-
ti(˚ Signal˚ signikant÷ /gl/ eglomis¤ren˚ H¤rogly#e˚ Iglu˚ Negligé˚ and /˙l/ e«-
li(˚ Jü«li«˚ lä«li(˚ and /˙Í/ Si«le÷ /gK/ agrari(˚ Agronom¤˚ Allegro˚ Dia-
gramm˚ Emigrant˚ Migräne˚ Regress˚ Sa«ria˚ Telegraf˘
But we have: /tm/ Admiral˚ Kadmium˘ In addition: /p, b/ Abort [= Abtritt]˚ ab-
oral˚ abort¤ren˚ and /kn, gn/ ErRgnis˚ in spite of /kn/ Befugnis˚ Wagnis˚ ZSgnis˘
5.2.2.4. From the examples provided, we see that German has a çglottalÇ stop
too (or çglottal catchÇ, but laryngeal stop is better) (ö): (&öyb…'öal, &öY-) /y[:]bÚ'al/ ü-
5. german 193
berall˚ (â¢'öO‰k) /gÈ'OKk/ Georg (more frequently, ('âe:ø‰k) /'ge:OKk/), (Ê¢'öaeló) /bÈ-
'aeló/ beRlen˚ (the'öa:t…) /te'atÚ/ †eater˚ (f…'öaen) /fÚ'aen/ VerRn˚ ('vael [ö]IÂ'öaos-
&öa‰baet¢) /'vael IÂ'aosaKbaetÈ/ wRl i> QsarbRte˚ ('öIs 'öaoX &öaen[ó]'öa∫Í) /'Is 'aox
aenó'a∫Í/ iss Q> Rnen Aπel˘
˛is phone precedes every V which is syllable-initial (thus with no C before that,
always in the same syllable), either in stressed or unstressed position, either in lex-
emes or grammemes, even in sentences. It is mostly noticeable after pauses and in
stressed syllables; if pronunciation becomes more relaxed, (ö) may be missing but
only in completely unstressed syllables (and we indicate it here as ([ö])), provided
composition is not evident.
As can be seen, it also occurs within lexemes, in the context /é'é/ (é'öé), but not
if the stress is weak: ('öaf¢) /'afÈ/ A‡e˚ ('öap&öa;‰t) /'apa:Kt/ Abart˚ (f…'öaXtó) /fÚ'axtó/
vera>ten˚ (Ê¢'öaXtó) /bÈ'axtó/ bea>ten˚ (d™s&öavu'öi:º?, -º¡, -‰n, -‰,) /dEsavu'i:Kó/
desavou¤ren˚ (∂e'öEMfazIs) /de'EmfazIs/ Deem#asis˚ (öant'öa:qidUm) /ant'a:qidUm/
Antacidum˚ (&öant[ö]a'qhi:t) /anta'qi:t/ Antacid˚ ('öaosö…≤zó) /'aosÚle:zó/ Qserlesen˚
('mUnt&öa;‰tó) /'mUnta:Któ/ Mundarten˘ ˛us we have: (thea'thüa:lIS) /tea'tKa:lIS/
theatrali(˚ but (&theöa-) only in a kind of… theatrical pronunciation.
In addition, we find (ö) even if the V is çdeepÇ or assimilated, as in (…) /Ú/, for
//ÈK//, provided it is an initial syllabic nucleus: ({qhu}ö…'fa:º?, -º¡, -‰n, -‰,) /{qu}Ú-
'fa:Kó/ [zu] erfahren˘
But we have: (öae'nand…) /ae'nandÚ/ Rnander˚ with non-morphemic syllabifica-
tion, which is acknowledged by the new orthography as well.
Further useful examples: (&ö…'öa‰&baetó) /ŒK'aKbaetó/ erarbRten˚ ('öaofö…&Ste;U,)
/'aofŒKSte:U˙/ Aufer®ehu«˚ ('vi:d…ö…&öo;b…,, -…n) /'vi:dÚŒKo:bÚn/ w¤dererobern˚
('âlat&öaes) /'glataes/ GlattRs˚ (la˙'ne:z¢&öaes) /la˙'ne:zÈaes/ La«neseRs˚ ('âa˙&öa;‰t)
/'ga˙a:Kt/ Ga«art˚ ('qhOlö…&h°;U,) /'qOlÚh°:U˙/ Zollerhöhu«˚ (khlo'öa:k¢) /klo'a:kÈ/
Kloake˚ (&öide'öa:l) /ide'a:l/ ideal˘
(0é), although the phonemic transcriptions do not use ç/ö/Ç, but just for the sake
of simplicity. Actually, for teaching purposes, it could be more useful to write it,
especially if no real phonetic transcription is added to the phonemic one. How-
ever, if only one type of transcription is used, it is decidedly more recommendable
to also write (ö) or /ö/, especially for important cases.
A possible variant of (öé{:/0}, ö…) is (ü, ü:, üó, 0): (&öIÂö…'öa‰&baet¢, &öIÂö™‰'ö-),
(&IÂ0'aé&baet¢, &I™é'aé-) /IÂŒK'aKbaetÈ/ i> erarbRte˚ (öIÂ'öant&Ñø‰t¢), (IÂ'act-) /IÂ'ant-
vOKtÈ/ i> antworte˚ (&ide'a:l) /ide'a:l/ ideal˘ ˛erefore, we have (ö) which fuses with
the voiced segments of a syllable, and realizes them as creaky. For native speakers,
this phonation type sounds similar to the sequences we have seen.
(In Danish, something similar occurs, although in the opposite direction: ®@d
{Ô § 17.36 of NPT/HPh} entails a distinctive use of creaky voice for voiced pho-
nemes, (ü, ó); however, certain speakers change these phones into (éö, 0ö) sequen-
ces, with (ö) in second place.)
5.2.2.6. It is interesting to note that (ö) can also occur mainly as a substitute of
/t/ (but of /d, p, k/ as well), when preceded — followed by /ö, M/. Even variants
with creaky intense nasal segments occur ((≠, =, ±)), which represent a frequent
compromise, as we will see. ˛e cue and most examples are taken from Kohler
(1994), with integrations and changes, both for the transcriptions and contexts of
occurrence: ('öaeMf…&Standó, -nn=, -nöó) /'aenfŒKStandó/ Rnver®anden˚ ('fo:‰&han-
dó, -nn=, -nöó) /'fo:Rhandó/ vorhanden˚ ('StUndó, -nn=, -nöó) /'StUndó/ Stunden˚
('öEndó, -nn=, -nöó) /'Endó/ Enden˚ ('öEntó, -nöó, -nn=) /'Entó/ Enten˚ ('kh§ntó,
-nöó, -nn=) /'k§ntó/ könnten˚ ('∂a˙kô, -˙öô, -˙~±) /'da˙kó/ danken˘
Further examples: ('lOYÂp&pU˙któ, -˙tó, -˙öô, -˙~±) /'lOYÂtpU˙któ/ LS>tpunk-
ten˚ ('qhEntn…, -nön…) /'qEntnÚ/ Zentner˚ ('khlEmpn…, -mön…) /'klEmpnÚ/ Klemp-
ner˚ ('thaozõp '≈e;t…, -õö '≈-, -≠ 'm-) /'taozótme:tÚ/ tQsend Meter˚ (Ê¢'öamtó, -mp-
tó, -möõ, -m≈≠) /bÈ'amtó/ Beamten (not for (Ê¢'öamt¢, -mpt¢) /bÈ'amtÈ/ Beamte]˘
More: (&öInöaen'nOYÈs 'öamt 'öaeM&fy;º?, -ampt, -amp, -º¡, -‰,, -‰n) /Inaen'nOYÈs
'amt 'öaenfy:Kó/ in Rn mSes Amt Rnführen˚ ('phU˙któ, -˙tó, -˙öô, -˙~±) /'pU˙któ/
Punkten˚ ('lOYtnant, -ön-) /'lOYtnant/ LStnant˚ (mIp'≈Ind¢stós, mIö-) /mIt'mIndÈs-
tós/ mit minde®ens˚ ('hIntó, -nöó, -nn=) /'hIntó/ hinten˚ (&mItnõ, -önõ) /&mItnõ/
mit 'nem˚ ('öaX&qhe;ntó, -öó, -n=, -&qen-, -q™n-, -q¢n-, -qó-) /'axqe:ntó/ a>˛ehnten˘
Stopstrictives
5.2.3.1. German has three voiceless stopstrictives, (∫, q, c) /∫, q, c/. From a
descriptive and teaching point of view, it is not a good thing to exclude them from
the phonemic inventory and consider them as phonemic sequences such as ç/pf,
ts, tS/Ç; also because these sequences do exist in the language (as we will see)! More
often, /q/ is articulated with lowered tip of the tongue, whereas /c/ is with raised
tip (but it is not necessary to use a special symbol (c)). ˛e three of them are all
çaspiratedÇ after pauses or in stressed syllables, except after a tautomorphemic /s/:
('∫hEf…) /'∫EfÚ/ Pfe‡er˚ ('∫húanq¢) /'∫lanqÈ/ Pflanze˚ ('SIM∫è) /'SIm∫ó/ (im-
πen˚ ('khO∫) /'kO∫/ Koπ÷ ('qhaet) /'qaet/ ZRt˚ (na'qh¿o:n) /na'qjo:n/ Nation˚
('nIÂq) /'nIÂq/ ni>_˚ ('sqe:n¢) /'sqe:nÈ/ Szene÷ ('chE¢) /'cEÂÈ/ T(e>e˚ ('chElo÷ 'SE-)
/'cElo÷ 'SE-/ Cello˚ ('∂OYc) /'dOYc/ dS˛˚ ('phac¢) /'pacÈ/ Pa˛e˘
Let us now see some cases of /pf, tS/ (]f, tS): ('öa]f…&la˙ô) /'apfÚla˙ó/ abverla«en˚
(ö™nt'SpInó, ö¢n-) /Ent'SpInó/ en_pinnen˚ (ö™nt'SúUs, ö¢n-) /Ent'SlUs/ En˛luss˘
However, in fast speech they can become (∫, c): ('öa∫…&la˙ô, ö™nc'pInó, ö¢n-,
ö™nc'úUs, ö¢n-) (generally without çaspiratingÇ the following stopstrictives or stops,
as also in: (ö™nt'SUldIgU,, -n'cU-, ö¢n-) /Ent'SUldIgU˙/ En˛uldigu«˚ Ô § 5.3.2.5 as
well).
Instead, German has no ç/G/Ç, but /tZ/, which occurs in loanwords: (t'Ëi:ns)
/t'Zi:ns/ Jeans˚ (&khilimant'Ëa:ºo) /kilimant'Za:Ko/ KilimanÃaro˚ ([öIm]t'ËU˙Í)
/[Im]t'ZU˙Í/ [im] D(u«el÷ especially in current pronunciation, (tË) /tZ/ can alter-
nate with /tS, c/.
Constrictives
5.2.4.1. ˛ere are five diphonic pairs: (f, v÷ s, z÷ S, Z÷ Â, ,÷ º, X). Generally, /s, z/
are articulated with lowered tip of the tongue, whereas /S, Z/ are with raised tip (but
it is not necessary to use special symbols as (x, ç)). Strictly speaking, (Â) could be
considered not to be a phoneme (by çsimplyÇ introducing a çjuncture phonemeÇ,
as we will see shortly). In addition, (,) is actually a çsemi-constrictiveÇ phone (even
if it were indicated with (J)). On the other hand, functionally, (º) belongs to the
particular category of sonants, and to the çvibrantÇ group; as a matter of fact, one
possible realization is exactly (K) /K/.
˛us it is important to clearly know the relations among the various constric-
tive elements. ˛erefore, it is not di‚cult –in this case, thanks to spelling as well–
to see that (Â, X) are in (almost) complementary distribution. On the other hand,
even (,) has close relationships (mostly in northern speech) with (Â) and also (X,
k) (even with (g, Ÿ) as well, which in the North realize /g/ together with (,), that
belongs to the phoneme /j/ too).
196 a handbook of pronunciation
But let us see some examples: ('fYMf) /'fYnf/ fünf˚ ('vas) /'vas/ was˚ ('ºaesó) /'Kae-
só/ rR•en˚ ('ºaezó) /'Kaezó/ rRsen˚ ('zo:n) /'zo:n/ Sohn˚ ('S°:n) /'S°:n/ (ön˚ ('Staen)
/'Staen/ StRn˚ ('thIS) /'tIS/ ‰(˚ (Ze'ni:) /Ze'ni:/ Gen¤˚ (e'tha:Z¢) /e'ta:ZÈ/ Etage˚ ('Êu:X)
/'bu:x/ Bu>˚ ('Êy:Â…) /'by:ÂÚ/ Bü>er˚ ('ÊaX) /'bax/ Ba>˚ ('ÊE¢) /'bEÂÈ/ Bä>e˚ ('ºu:IÂ)
/'Ku:IÂ/ ruhig˚ ('ºaoºIs) /'KaoKIs/ RQris˘
5.2.4.2. ˛ere are three more approximant taxophones for slit constrictives,
which occur before /s, S/, and we will notate them with their çcommittedÇ variants
as well (which are more demanding), because they are not really necessary, al-
though very frequent and çtypifyingÇ. It would not be useless to acquire them to
obtain a smoother pronunciation: (&, â, ˜) /f, Â, x/: ('öao&&Sibõ, 'öaof&Sibõ)
/'aofSi:bó/ Qf(¤ben˚ ('h°:âst, 'h°:Âst, -ÂÀt) /'h°:Âst/ hö>®˚ ('ºaoX&Sp™k, 'ºao˜&Sp™k)
/'KaoxSpEk/ RQ7pe$˘
In preintonemes (¤ not in intonemes), forms such as Qf˚ i>˚ mi>˚ di>˚ si>˚
-ig˚ -i>˚ -li>˚ -zig˚ S>˚ Q>˚ do>˚ no>˚ in current speech, often have (&, â, ˜) /f,
Â, x/, irrespective of the context just seen.
Quite rightly, German pronouncing dictionaries clearly indicate both (Â) /Â/
and (X) /x/ – it is necessary to provide precise indications, not mysterious algebraic
formulae. In fact, unless through an appropriate morphonological knowledge,
there would be no way to tell the di‡erence between ('thao«) /'taoÂó/ TQ>en
[->en] çsmall ropeÇ and ('thaoX?) /'taoxó/ tQ>en çto plungeÇ, or ('khu:«) /'ku:Âó/
Kuh>en [->en] çcalfÇ and ('khu:X?) /'ku:xó/ Ku>en çcakeÇ.
5.2.4.3. ˛e su‚x -ig is (IÂ) /IÂ/, even when followed by C (except for those be-
longing to su‚xes which contain (Â)): ('kh°:nIÂ) /'k°:nIÂ/ König˚ ('öaenIâst, -Âst,
-ÂÀt) /'aenIÂst/ Rnig®˚ (f…'öaenIÂt) /fÚ'aenIÂt/ verRnigt˚ ('qhñanqIâst¢, -Âst¢, -ÂÀt¢)
/'qvanqIÂstÈ/ zwanzig®e÷ but: ('kh°:nIkúIÂ) /'k°:nIklIÂ/ königli>˚ ('kh°:nIk&üaeÂ)
/'k°:nIkKaeÂ/ KönigrR>˚ ('kh°:nIgIn) /'k°:nIgIn/ Königin˘
However, it is possible to find ('kh°:nIÂ&üaeÂ), and even –fairly systematically–
(Ik) /Ik/: ('kh°:nIk, 'öaenIkst, f…'öaenIkt, 'qhñanqIkst¢), which is the only pronun-
ciation really accepted in southern Germany, Austria, and Switzerland, in initial
position too: ('Âe:mIS, 'k[h]-) >emi(˚ ('Âi:na, 'k[h]-) ≥ina˘
˛e graphic sequence 7 has two di‡erent phonic structures: within lexemes
we find (ks) /ks/: ('vEksÍn) /'vEksÍn/ we7eln˚ ('vaeksÍ) /'vaeksÍ/ WR7el˚ ('öOks¢)
/'OksÈ/ O7e˚ ('zEks) /'zEks/ se7˚ ('fUks) /'fUks/ Fu7˚ (∂…'daks) /dÚ'daks/ der Da7
çbadgerÇ, while we have: (Xs) and (âs, Âs, ÂÀ) /xs, Âs/, when s is a grammeme: (∂¢s-
'∂aXs) /dÈs'daxs/ des Da7 çof the roofÇ, (∂¢s'phEâs, -Âs, -ÂÀ) /dÈs'pEÂs/ des Pe7
ç(black) pitchÇ.
/s, z/ are mostly articulated with lowered tip of the tongue; but, for /sÂ/, we have
(ÀÂ): ('ÊIÀ«) /'bIsÂó/ bis(en÷ and, except in slow speech, for /sS, sZ/, we have (SS,
SË): ('öaeS&Süa˙k) /'aesSKa˙k/ Eis(rank˚ ([öI]Â≈US'SnEl 'vEk) /IÂmUs'SnEl 'vEk/ i> muss
(nell weg˚ (&∂aSËe'ni:) /dasZe'ni:/ das Gen¤. /S, Z/ are with raised tip (((x, ç))).
In addition, in the /kv, qv, Sv/ sequences, /v/ is more often realized as bilabial,
(6): ('khñac) /'kvac/ Qua˛˚ ('qhñae) /'qvae/ zwR˚ ('SñEst…) /'SvEstÚ/ S>we®er˘ In
neutral German, /j/ is semi-constrictive, in all contexts: (',a:‰) /'ja:K/ Jahr˚ (&ma,o-
5. german 197
5.2.4.4. For /K/, it is necessary to add a few observations, both on our notation
and phonemic interpretation. ˛e çgenericÇ phonemic symbol seems to be useful
in indicating its normal place of articulation, which is uvular – and the most rec-
ommendable, for teaching purposes. As far as its manner (of articulation) is con-
cerned, the voiced constrictive one, (º), is normal (and the most recommendable
as well).
For emphasis, or for individual peculiarities, it is certainly possible to have a
(voiced) uvular trill: (K). On the other hand, among native speakers it is not rare
to find (voiced) alveolar articulations: a tap, (R), or a trill, (r) (mostly in the South,
and Austria and Switzerland). However, before stressed or unstressed V˚ we will
transcribe only (º) /K/: ('ºaoºIs) /'KaoKIs/ RQris˚ ('ʺI˙ô) /'bKI˙ó/ bri«en.
After tautosyllabic V˚ instead, we will always transcribe (‰) (voiced pharyngeal-
ized uvular approximant) for /K/, which is intense after C or diphthongs, (…) /Ú/:
('vi:‰) /'vi:K/ wir˚ ('vI‰) /'vIK/ wirr˚ ('mae…) /'maeÚ/ MRr (like MRer˚ Meyer]˚ ('ÊEs…)
/'bEsÚ/ besser˚ (h…'fo:‰) /hÚ'fo:K/ hervor.
Many phonemic traditions have considered //ÈK//, mostly in parallel with //Èn,
Èm, Èl//, which we prefer to transcribe and interpret as (ó, õ, Í) /ó, õ, Í/, although
we are perfectly aware that, in slower or emphatic speech, (¢n, ¢m, ¢l) are possible,
and (Ⱥ÷ ÈK÷ Ș) as well – even ç(¢R)Ç, or rather (ÙR, ÙR) (in Switzerland) and (Ù‰, Ùº,
Ù˜) (in Austria).
Contrary to what we do –and Kohler (1977, 1995”) and others as well– several
authors distinguish between (éº) and (é:‰) (¤ ç/ér/Ç and ç/é:å/Ç, above all in teach-
ing textbooks), but we do not think it is the best proposal, also because it is more
complex and far from being the most widespread pronunciation (Ô § 5.1.1 too).
For the prefixes er-˚ her-˚ ver-˚ zer-˚ we will transcribe ({0}…) /{0}Ú/, since it is the
normal and most recommendable pronunciation, although in pronouncing dic-
tionaries (except the Ÿ∂∑å) we still find ç/{0}Er, {0}Eå/Ç. A slow, careful, or em-
phatic, pronunciation –with ({0}™‰) /{0}EK/– can be safely inferred from our nota-
tion as well. Diaphonemically, though, we might use /{0}Œ‰/.
Approximants
5.2.5. For this manner of articulation, the most important German phoneme
is certainly (h) /h/, which often has the voiced variant, among voiced phones (espe-
cially in non-primary-stressed syllables): ('haos) /'haos/ HQs˚ (â¢'haem) /gÈ'haem/
gehRm˚ ('da:&ha;b/vI‰, -H-) /'da:ha:bóvi:K/ da haben wir…˚ ('öUn&hael, -H-) /'Unhael/
UnhRl˚ ('vIlh™lm, -H-) /'vIlhElm/ Wilhelm˘ Notice, however, cases like: ('âe:ó) /'ge:ó/
gehen˚ ('ºu:IÂ) /'Ku:IÂ/ ruhig˚ ('za:) /'za:/ sah˘
Graphically, h is a device often used to indicate that the preceding V is long,
mostly before m˚ n˚ l˚ r (the so-called çDehnungs-hÇ): ('ne:mõ) /'ne:mó/ nehmen˘
In addition to the variant of /v/, (6), occurring in the sequences /kv, qv, Sv/ (§
198 a handbook of pronunciation
5.2.4.3), we will briefly say something about other possible variants for /0j, yé,
u'öé, o'a/: ('fe:º,«, -ºjó) /'fe:Kjó/ Fer¤n˚ (na'qh¿o:n, -'qhªo:n, -qi'o:n) /na'qjo:n/ Na-
tion˚ (qhya'no:z¢, qh¥a-) /qya'no:zÈ/ Zyanose˚ (&manu'öEl, -u'El, -'njEl, -'nwEl) /ma-
nu'El/ manuell˚ (thoa'lEt¢, thja-, thwa-) /toa'lEtÈ/ Toilette˘
Laterals
5.2.6. In German neutral pronunciation we find only (l) /l/ (with the corre-
sponding intense one, (Í) /Í/) and the variant (¬), before (Â, ,) (the last one is het-
erosyllabic only): ('lant) /'lant/ Land˚ ('fOl) /'fOl/ voll˚ (fa'mi:l,¢) /fa'mi:ljÈ/ Famil¤˚
(va'nI¬,¢) {((-nI¬-,¢)) and (-Il¢)} /va'nIl[j]È/ Vanille˚ ('mI¬Â) /'mIlÂ/ Mil>˚ ('hElt) /'hElt/
Held˚ ('öe:zÍ) /'e:zÍ/ Esel˚ ('∂u:dÍn) /'du:dÍn/ dudeln˘
Structures
5.3.0. In this part, we will make the necessary observations for pronouncing
words in actual sentences. ˛us, we will deal with assimilation, reduction, stress,
and intonation.
Reduced forms
(∂…'le:º… &hatós'hOYt¢ 'nIÂq 'öaofâ¢≥bõ) der Lehrer hat uns hSte ni>_
Qfgegeben
(∂ó'thIS 'StElèvI‰ öIndI'öEk¢2| ∂ó'Süa˙kô dI'öandȺ¢3 3) den ‰( ®ellen wir in d¤
E$e, den S>rank in d¤ andere
(∂…'thIS ‘∂™‰öódI'öEk¢&Ste;t’) der ‰(, der in d¤ E$e ®eht…
(∂ó'thIS ‘∂¢MvI‰ódI'öEk¢ g¢'StElt&haõ’) den ‰(, den wir in d¤ E$e ge®ellt ha-
ben…
200 a handbook of pronunciation
5.3.1.2. Here is a list of reduced forms; it is fairly complete, which could even
seem to be too analytical. However, it is of paramount importance to carefully ex-
amine all the modifications given, in order to grasp the mechanism.
It is obvious that the most çnormalÇ forms, in this list, are not the first given,
but some of the succeeding ones, in suitable conditions.
aber ('öa:b…, &öa;b…, &öab…, &ab…, &a6…) 'öanó, &öanó, &anó, aó, an:, ån:)
also ('öalzo, &öalzo, &alzo, &azo) an'r˚ an der ('öand…, &öand…, &and…, 'öa-
an ('öan, &öan, an, ån), (-m, -M, -~, -˙, n…, &öan…, &an…)
-,) an die ('öandi, &öa-, &a-, -ani, -I, -¢)
am˚ an dem ('öandõ, &öandõ, &adõ, auf ('öaof, &öaof, &aof, aof, åof)
'öaõ, 'öam, &öam, am, åm) aufs˚ -f's˚ -f das ('öaof∂as, 'öaof∂ås, &öao-
ans˚ an's˚ an das ('öandås, &öandås, &an- f¢s, &öaof#, &öaofs, &aofs, öaofs, aofs,
dås, 'öans, &öans, ans, åns) åofs)
an'n˚ an den ('öandó, &öandó, &andó, aufm˚ auf'm˚ -f dem ('öaof∂õ, 'öaofõ,
5. german 201
ly after V] -gegen˚ …)
die ('∂i:, &∂i;, &∂i, ∂i, ∂I, ∂¢), (d-), ([-) er ('öe:‰, &öe;‰, &öe‰, e‰, ™‰, …) ((…) only
(only after V] when not preceded by (¢, ‰, …) /È, K,
die (rel.) ('∂i:, &∂i;, &∂i, ∂i, ∂I), (d-) Ú/÷ the most reduced forms do not
die (dim.) ('∂i:, &∂i;, &∂i, ∂i), (d-) occur in final position)
diese ('∂i:z¢, &∂i;z¢, &∂iz¢, &∂Iz¢), (d-) er- (ö™‰'ö-, ™‰'ö-, …'ö-) [-arbRten˚ -übri-
diesem ('∂i:zõ, &∂i;zõ, &∂izõ, &∂Izõ), gen˚ …)
(d-) (ö™‰'-, ™‰'-, …'-) [-klären˚ -saufen˚ …)
diesen ('∂i:zó, &∂i;zó, &∂izó, &∂Izó), (d) es˚ 's ('öEs, &ö™s, ™s, ö¢s, ¢s, #, s) (the
dieser ('∂i:z…, &∂i;z…, &∂iz…, &∂Iz…), (d-) most reduced forms occur in final
dieses ('∂i:z¢s, &∂i;z¢s, &∂iz¢s, &∂Iz¢s), (d-) position too, in many cases)
dir ('∂i:‰, &∂i;‰, &∂i‰, ∂I‰, ∂…), (d-), ([-) eu> ('öOYÂ, &öøYÂ, öøYÂ, øYÂ, øYâ,
(only after V÷ the most reduced ÖYÂ, ÖYâ)
forms do not occur in final posi- euer ('öOY‰, &öøY‰, öøY‰, øY‰, ÖY‰)
tion) eure ('öOY‰, &öøY‰, öøY‰, øY‰, ÖY‰)
do> ('∂OX, &∂øX, ∂øX, ∂¢X), (d-) eurem ('öOYºõ, &öøY-, &öøY‰m, öøY-,
du ('∂u:, &∂u;, &∂u, ∂u, ∂U, ∂¢), (d-), ([-) öøYõ, øY-, ÖY-)
(only after V÷ the most reduced euren ('öOYº?, -º¡, &öøY-, &öøY‰,, -‰n,
forms do not occur in final posi- öøY-, øY-, ÖY-)
tion) eurer ('öOYº…, &öøY-, öøY…, øY…, ÖY…)
dur> ('∂U‰Â, &∂U‰Â, ∂U‰Â), (d-) eures ('öOYº¢s, &öøY-, &öøY‰s, öøY‰s, øY-,
dur>s˚ -h das ('∂U‰Â∂ås, '∂U‰Â∂¢s, ÖY-)
'∂U‰Â#, '∂U‰Âs, &∂U‰Âs, ∂U‰Âs, -ÂÀ, für ('fy:‰, &fy;‰, &fy‰, fY‰, fY, f…)
-âs), (d-) fürs˚ -r's˚ -r das ('fy:‰dås, 'fy:‰d¢s, 'fy:‰-
dur>'n˚ -h den ('∂U‰Â∂ó, &∂U‰Â∂ó, d#, 'fy:‰s, &fy;‰s, &fy‰s, fY‰s, fYs, f…s)
'∂U‰Â«, &∂U‰Â«), (d-) für den˚ -r'n ('fy:‰dó, 'fy:‰?, 'fy:‰,,
dur> die ('∂U‰Â∂i, &∂-, -I, -¢), (d-) &fy;‰,, &fy‰,, fY‰,, f…,, fYn), (-m, -M,
eben ('öe:bõ, &öe;bõ, &öebõ, öeõ, -n, -~, -˙)
öem, em) für die ('fy:‰di, &fy;-, &fy-, &fY-, -I, -¢)
Rn˚ 'n ('öaen, &öaen, aen, ¢n, ó), (-m, gar ('âa:‰, &âa;‰, &âa‰, âa‰, âa, â…), (g-)
-M, -~, -˙, -,) gegen ('âe:gó, -gô, ≥-, &ge-, âeô), (g-)
Rne˚ 'ne ('öaen¢, &öaen¢, &aen¢, &ån¢, &¢n¢, gegens˚ -n's˚ -n das ('âe:gódås, 'âe:gód¢s,
ó¢, n¢) 'âe:gós, -gôs, ≥-, &ge-, âeôs), (g-)
Rnem˚ R'm˚ 'nem ('öaenõ, &öaenõ, gegen die ('âe:gódi, -ón-, -ó-, ≥-, &ge-,
&aenõ, &aemõ, &ånõ, &åmõ, öaem:, -I, -¢), (g-)
aem:, aem, åem, åm, n¢m, ¢m, õ) geworden (â¢'vO‰dó, â¢&vø‰dó, â¢-
Rnen˚ R'n˚ 'nen ('öaenó, &öaenó, &aenó, &vø‰,), (g-÷ -m, -M, -n, -~, -˙)
öaen:, öaen, aen, åen, ån, n¢n, nó, guten Abend! ('âu:tó 'öa:bõt, &âu;tó'ö-,
ó), (-m, -M, -~, -˙, -,) &âut-, &âUt-, &âUd-, âUn'ö-, âô'ö-, ô'ö-,
Rner˚ 'ner ('öaen…, &öaen…, &aen…, &åe- ó'ö-, [-]'öa:mt, [-]'öamt), (g-)
n…, &ån…, &¢n…, ó…, n…) guten Morgen! ('âu:tó 'mO‰gô, &âu;tó-
Rnes˚ 'nes ('öaen¢s, &öaen¢s, &aen¢s, &åe- 'm-, &âut-, &âUt-, &âUd-, -õ'm-, âUn'm-,
n¢s, &ån¢s, &¢n¢s, ó¢s, n¢s) -m'm-, âô'm-, ô'm-, ó'm-, õ'm-, 'm-,
Rnmal ('öaen&ma;l, &öaenmal, -ål, -å, [-]'mO‰,, [-]'mO,), (g-)
-mm-, &óm-, &õm-) guten Tag! ('âu:tó 'tha:k, &âu;tó'tha:k,
ent- (ö™nt-, ™nt-, ¢nt-, ót-) [-wi$eln˚ &âut-, &âUt-, &âUd-, âUn'th-, âô'th-, ô'th-,
5. german 203
so ('zo:, &zo;, &zo, zo, zø, z¢) ver- (f™‰'ö-, f…'ö-) [-Rsen˚ -arbRten˚ …)
soll ('zOl, &zøl, zøl, zø) (f™‰'-, f…'-) [-rRsen˚ -lassen˚ …)
sonst ('zOnst, &zønst, zøns, z¢ns) von ('fOn, &føn, føn, f¢n, fó), (-m, -M,
-tägigen (-&th™;gIgô, -t™g-, -g¢g-, -ggô, -~, -˙, -,)
-gô) [zwR-˚ drR-˚ …) vom˚ -n dem ('fOndõ, 'fOnõ, 'fOmõ,
über (unstressed) (&öyb…-, &öY-, -6-) 'fOm, &føm, føm, f¢m, fõ)
überm˚ -r'm˚ -r dem ('öy:b…dõ, &öy;b…- von'n˚ -n den ('fOndó, 'fOnó, &fønó,
dõ, 'öy:b…õ, 'öy:b…m, &öy;b…m, &öy- føó, føn:,), (-m, -M, -~, -˙, -,)
b…m, &Yb…m, -6…m) von'r˚ -n der ('fOnd…, 'fOn…, &føn…)
übern˚ -r'n˚ -r den ('öy:b…dó, &öy;b…dó, vor ('fo:‰, &fo;‰, &fo‰, fo‰, fø‰, f…)
'öy:b…?, 'öy:b…,, &öy;b…,, &öyb…,, &Y- vorn˚ -r den ('fo:‰dó, 'fo:‰,, &fo;‰,,
b…,, -6…,), (-m, -M, -n, -~, -˙) &fo‰,, fo‰,, fø‰,, f…,), (-m, -M, -n,
übers˚ -r's˚ -r das ('öy:b…dås, 'öy:b…d¢s, -~, -˙) (Ô adv. ('fO‰n, &fø‰n)}
'öy:b…d#, 'öy:b…s, &öy;b…s, &öyb…s, &Y- vors˚ -r das ('fo:‰dås, 'fo:‰d¢s, 'fo:‰d#,
b…s, -6…s) 'fo:‰s, &fo;‰s, &fo‰s, fo‰s, fø‰s, f…s)
über die ('öy:b…di, -I, -¢, &öy;-, &öy-, &Y-, vor'm˚ -rm˚ -r dem ('fo:‰dõ, 'fo:‰nõ,
-6…-) 'fo:‰mõ, 'fo:‰m, &fo;‰m, &fo‰m,
um ('öUm, &öUm, Um) fo‰m, fø‰m, f…m)
ums˚ um's˚ um das ('öUmdås, 'öUmd¢s, vor'r˚ -r der ('fo:‰d…, 'fo:…, &fo;…, &fo…,
'öUmd#, 'öUms, &öUms, Ums) fo…, fø…)
um'n˚ um den ('öUmdó, 'öUmõ, &öU- vor die ('fo:‰di, &fo;-, &fo-, -I, -¢)
mõ, Uõ, Um:, Um) vor- (fo'º-) [-an˚ -auf˚ -aus˚ -in˚ -über˚
um die ('öUmdi, &öU-, U-, -I, -¢) …)
und ('öUnt, &öUnt, &öUn, Un, ¢n, ó), (-m, (fo‰'ö-) [-ab˚ -Rnst÷ -arbRten˚ -Rlig˚
-M, -~, -˙, -,) …)
-und- (-&öUnt-, -&öUn-, -Unt-, -Un-, -¢n-, (fo‰'-) [-bR˚ -weg˚ …)
-ó-), (-m-, -M-, -~-, -˙-, -,-) (fo‰'-), ('fo:‰&-) [-her˚ -hin˚ -zu˚ …)
uns ('öUns, &öUns, &Uns, ¢ns, ós, ns) (the während ('v™:º?t, &v™;-, -º¡t, 'v™:‰,t,
most reduced forms do not occur in -‰nt, &v™;-, v™-÷ 'vE:-)
rhythm groups which are in into- was (non-interrogative) ('vas, &vas,
nemes, in neutral pronunciation) vas, vås)
unser ('öUnz…, &öU-, &U-) wegen ('ve:gô, &ve;gô, &vegô, veô)
unsere ('öUnzȺ¢, &öU-, &U-, -zº¢) wRl ('vael, &vael, vael, våel, vål)
unserem ('öUnzȺõ, &öU-, &U-, -zº-, -…m, wem ('ve:m, &ve;m, &vem, vem, v™m)
-zõ) wen ('ve:n, &ve;n, &ven, ven, v™n), (-m,
unseren ('öUnzȺ?, -º¡, &öU-, &U-, -zº-, -M, -~, -˙, -,)
-…,, -zó) wenn ('vEn, &v™n, v™n, v¢n), (-m, -M,
unserer ('öUnzȺ…, &öU-, &U-, -zº…, -…‰) -~, -˙, -,)
unseres ('öUnzȺ¢s, &öU-, &U-, -zº¢s, -z¢s) wer ('ve:‰, &ve;‰, &ve‰, ve‰, v™‰)
unter ('öUnt…, &öUnt…, &Unt…) werden ('ve:‰dó, &ve;‰dó, &ve‰dó,
untern˚ -r den ('öUnt…dó, 'öUnt…,, &ve:‰n, ve‰n, v™‰n, -‰,), (-m, -M, -n,
&öUnt…,, &Unt…,), (-m, -M, -n, -~, -˙) -~, -˙)
unterm˚ -r'm˚ -r dem ('öUnt…dõ, 'öUn- werde ('ve:‰d¢, &ve;‰-, &ve‰-, ve‰t)
t…m, &öUnt…m, &Unt…m) werdet ('ve:‰d¢t, &ve;‰-, &ve‰-)
unters˚ -r's˚ -r das ('öUnt…dås, 'öUnt…- wider ('vi:d…, &vi;d…, &vid…, &vId…)
d¢s, 'öUnt…d#, 'öUnt…s, &öUnt…s, &Unt…s) wider- (unstressed) (&vid…-, &vId…-)
206 a handbook of pronunciation
wie ('vi:, &vi;, &vi, vi, vI) würde ('vY‰d¢, &vY-, vY‰t, vT‰t)
wie viel ('vi:fil, vi'fi:l, &vifIl, vI&fil, &vifÍ) würdest ('vY‰d¢st, &vY-, -s)
wieder ('vi:d…, &vi;d…, &vid…, &vId…) würdet ('vY‰d¢t, &vY-)
wieder- (unstressed) (&vid…-, &vId…-) -zehn (-qe:n, -qe;n, -qen, -q™n, -q¢n,
will ('vIl, &vIl, vIl, vI) -qó), (-m, -M, -~, -˙)
wir ('vi:‰, &vi;‰, &vi‰, vI‰, v¤‰, v¢‰, v…) -zehnte (-qe:nt¢, -qe;n-, -qen-, -q™n-,
(the most reduced forms do not oc- -q¢n-, -qó-)
cur in rhythm groups which are in zer- (q™‰-, q…-) [-fahren˚ -rinnen˚ …)
intonemes, in neutral pronuncia- zu ('qhu:, &qu;, qu;, qu, qU, q¢)
tion, not even when they are encli- zum˚ zu dem ('qhu:dõ, &qu;dõ, &qu-
tic after verbs) dõ, &qUõ, 'qhUm, &qUm, qUm, q¢m,
wird ('vI‰t, &vI‰t, vI‰t, v¤‰t, v¢‰t, v…t) qõ)
wirst ('vI‰st, &vI‰st, vI‰st, v¤‰st, v¢‰st, zur˚ zu der ('qhu:d…, &qu;d…, &qud…,
v…st, -s) 'qhu:‰, &qu;‰, qu;‰, qu‰, qU‰, qw…,
wo ('vo:, &vo;, &vo, vo, vø) q…)
wo- (vo'-, vø'-, v¢'-) [-bR˚ -her˚ -hin˚ zu'n˚ zu den ('qhu:dó, 'qu;dó, 'qudó,
-vor˚ -zu˚ …) 'qhu:ó, 'qhu:n, &qu;n, qu;n, qun,
(vo'º-, vø'-, vÈ'-), ('vo:&º-) [-ran˚ -rauf˚ qUn, q¢n, qó), (-m, -M, -~, -˙, -,)
-raus˚ -rüber˚ -runter˚ …) -zu- (-qu-, -qU-, -q¢-) [ab…holen˚ …)
worden ('vO‰dó, &vø‰dó, vø‰n, -‰,), zusammen (qhu'zamõ, qu-, qU-, q¢-)
(-m, -M, -~, -˙) zwar ('qhña:‰, &qña;‰, &qña‰, qña‰,
wurden ('vU‰dó, &vU‰n, vU‰n, -‰,), qña, qñ…)
(-m, -M, -~, -˙) -ren˚ -r'n (after –short or long–
wurde ('vU‰d¢, &vU‰-, &v…-) stressed V˚ where /Kó/ (º?, º¡) be-
wurdest ('vU‰d¢st, &vU‰-, &v…-, -s) comes /Kn/ (‰,), through ç/Ún/Ç)
wurdet ('vU‰d¢t, &vU‰-, &v…-) (-'é[:]º?, -'é[:]º¡, -'é[:]‰,, -'é[:]‰n),
würden ('vY‰dó, &vY‰n, vY‰n, vT‰n, (-m, -M, -~, -˙).
-‰,), (-m, -M, -~, -˙)
Taxophonics
5.3.2.1. ˛ere are various other ways of simplifying words –and above all sen-
tences– which make speech easier and more fluent, but less çclearÇ to foreigners.
It is important, thus, to know the mechanisms implied, both to understand bet-
ter and move closer to the native speakers' genuine pronunciation, and to be un-
derstood better. Following Kohler (1977, 1995”, § 6.3.1), we will give several cas-
es, by appropriately integrating and completing them.
We have already seen that sequences of (MM) /MM/, in fast speech, are possible
only in preintonemes: (∂I'öapâ¢S&nItóó 'ºo:zó) d¤ abge(nittenen Rosen˚ but: ('nIm
dI'öapâ¢S&nItó¢n) nimm d¤ abge(nittenen˘
5.3.2.2. Often -e (¢) /È/ of the present-tense first person singular is dropped, pro-
vided this does not produce any (M) (¤ intense nasal): ([öI]Â'≈aX) /IÂ'max[È]/ i>
ma>'˚ ('maXIÂ) /'max[È]IÂ/ ma>' i>˚ ([öI]Â'hab¢s â¢'ze:ó, -Ê-) /IÂ'hab[È] ÈsgÈ'ze:ó/ i>
5. german 207
hab' es gesehen˚ ('ha:bIÂ, -ÊIÂ) /'ha:b[È]IÂ/ hab' i>˚ ('le:zIÂ, -ΩIÂ) /'le:z[È]IÂ/ les' i>˚ and
('ö§fnIÂ) /'§fn[È]IÂ/ ö‡n' i>, ('öa:t≈IÂ) /'a:tm[È]IÂ/ atm' i>˚ but: ([öI]Â'ö§fn¢) /IÂ'§f-
nÈ/ i> ö‡ne, ([öI]Â'öa:t≈¢) /IÂ'a:tmÈ/ i> atme; in the past tense, -e (¢) /È/ may be
dropped, before an enclitic i>, provided the verb does not end in -ete (¢t¢) /ÈtÈ/:
('maXtIÂ) /'maxt[È]IÂ/ ma>t' i>˚ ('∂U‰ftIÂ) /'dUKft[È]IÂ/ dur⁄' i>˚ but: ('ö§fn¢t¢&öIÂ)
/'§fnÈtÈIÂ/ ö‡nete i>˚ ('öa‰&baet¢t¢&öIÂ) /'aKbaetÈtÈIÂ/ arbRtete i>˘ Even -e (¢) /È/ be-
longing to nouns may often be dropped while maintaining the C voiced or half-
-voiced: ('li:bUn 'laet, -Ê-) /'li:b[È] Unt'laet/ L¤b' und LRd˘
We have already dealt with the assimilation of place of articulation to a follow-
ing C for /t, d, n, ó, s/ (§ 5.2.1-2 “ 5.2.4). We have also mentioned some alterna-
tions, especially for approximants and /K/ (§ 5.2.4-5).
5.3.2.3. As already seen, /ó/ assimilates to a preceding C÷ but /n/ remains un-
changed when -e- /È/ (¢) is dropped (and a voiced syllable-final –but lexeme-inter-
nal– C does not become voiceless): ('khOmõ) /'kOmó/ kommen˚ ('le:bõt) /'le:bót/
lebend˚ ('öe:bõ) /'e:bó/ eben (but: ('öe:bn¢) /'e:bnÈ/ ebne˚ unchanged); ('fa˙ô) /'fa˙ó/
fa«en˚ (',u:gôt) /'ju:gót/ Jugend˚ (f…'lo:gô) /fÚ'lo:gó/ verlogen (but: (f…'lo:gn¢) /fÚ-
'lo:gnÈ/ verlogne˚ unchanged). Even double assimilation occurs (but not in into-
nemes): (öaof'öe:bõõ 'Stüa:só) /aof'e:bóÈn 'StKa:só/ Qf ebenen Stra•en˚ (∂I'öae˙g¢-
&Súa:gôô 've:g¢) /di'ae˙geSla:góÈn/ d¤ R«e(lagenen Wege˘
Even after nasals (which may be intense or not), /t, d/ can assimilate, if they are
unexploded: ('hEmt, -mpæ) /'hEmt/ Hemd˚ ('hEmdó, -mbõ) /'hEmdó/ Hemden˚ (∂I-
',u:gót 'öaM&fOY…n, ∂I',u:gôkæ) /di'ju:gót 'anfOYÚn/ d¤ Jugend anfSern˚ (∂I'ge:gót
'öan&ze:ó, ∂I'ge:gôkæ) /di'ge:gót 'anze:ó/ d¤ Gegend ansehen˚ (∂I'ge:gódó 'öan&ze:ó, ∂I-
'ge:gôgô) /di'ge:gódÈn 'anze:ó/ d¤ Gegenden ansehen˚ (Ê¢'öamtó, -mpõ) /be'amtó/
Beamten˚ (Ê¢'haoptó, -ppõ, -põ) /bÈ'haoptó/ behQpten˚ (∂I'le:bódó 'laedó, -bõ-
bõ) /di'le:bódÈn 'laedó/ d¤ Lebenden lRden˘
5.3.2.4. Frequently, the sequence /gn/ assimilates as well (and also /bn, dn/,
even if preceded by N˚ with a subsequent reduction): (#'üe:gn¢t, #'üe:˙n¢t) /Ès'Ke:g-
nÈt/ es regnet˚ ('öagn™s, 'öa˙n™s) /'agnEs/ Agnes˚ (mag&nifi'qhEnq, ma˙&n-, -IfI-) /mag-
nifi'qEnq/ Magnifizenz˚ (zIg'na:l, zI˙'na:l) /zIg'na:l/ Signal, ('va:gn…, 'va:˙n…) /'va:g-
nÚ/ Wagner÷ ('öe:bn¢, 'öe:mn¢) /'e:bnÈ/ ebne˚ ('le:bót, -mõt) /'le:bót/ lebend˚ ('ºe:dn…,
'ºe:nn…) /'Ke:dnÚ/ Redner˚ ('ve:‰dó, -nó) /'ve:Kdó/ werden˚ ('fIndó, 'fInó) /'fIndó/ fin-
den˚ ('hEmdó, -mbõ, -mõ) /'hEmdó/ Hemden (Ô above).
Even sequences of N + a voiced stop can undergo nasal assimilation: (no'vEmb…,
-mm…) /no'vEmbÚ/ November˚ ('öUmb¢&haltó, 'öUmm¢-) /'UmbÈhaltó/ umbehalten˚
(qhUm'bae&Spi;l, -m'm-) /qUm'baeSpi:l/ zum BRsp¤l˚ ('ÊUnd¢s&Êa˙k, 'ÊUnn¢s-) /'bUn-
dÈsba˙k/ Bundesbank˚ ('vandȺ…, 'vannÈ-) /'vandÈKÚ/ Wanderer˚ ('vUnd…&ba;‰ 'vUn-
n…-) /'vUndÚba:K/ wunderbar˚ (qhU'mInd¢st, -Inn¢st) /qu'mIndÈst/ zuminde®˚ ('öU˙-
g¢&vIs, 'öU˙˙¢-) /'UngÈvIs/ u«ewiss˚ ('öa˙g¢≥bõ, 'öa˙˙¢-) /'a˙gÈge:bó/ a«egeben˚
('öa˙≥bõ, 'öa˙&˙e;-) /'a˙ge:bó/ a«eben˚ ('öU˙ga‰, 'öU˙˙a‰) /'öU˙gaK/ U«ar˘
5.3.2.6. In fast speech, between vowels and in unstressed syllables, /b, g/ can be
realized as (6, Ÿ): ('ha:b¢, -6¢) /'ha:bÈ/ habe˚ ('le:g¢, -Ÿ¢) /'le:gÈ/ lege÷ /t/ can become
([): ('ÊIt¢, 'ÊI[¢) /'bItÈ/ bitte˘ Under the same conditions, voiceless stops and con-
strictives may become half-voiced or totally voiced: (∂ås&hat…'dOX â¢'maXt, -∂-, -d-)
/dashatÚ'dOx gÈ'maxt/ das hat er do> gema>t˚ (∂åS'Sa:fI '∂OX&nIÂt, -Ñ-, -v-, ∂å'S-) /das-
'SafI 'dOxnIÂt/ das (a‡' i> do> ni>t˚ (∂ås&≈UsIÂ'≈aX?, -Ω-, -z-) /dasmUsIÂ'maxó/
das muss i> ma>en˚ (∂ås'≈aXI '∂OX&nIÂt, -ü-, -º-) /das'maxI 'dOxnIÂt/ das ma>' i>
do> ni>t (obviously, because of ambiguity for native speakers –which, however,
does not coincide with foreigners' ambiguity!– this typical fast-speech neutraliza-
tion is avoided).
After pauses, voiced constrictive phonemes remain unchanged; whereas, after
voiceless C they become half-voiced (but more often /z/ becomes voiceless): ('vi:n)
/'vi:n/ W¤n˚ (∂ås'Ñas…) /das'vasÚ/ das Wasser˚ ('qhñanqIÂ) /'qvanqIÂ/ zwanzig˚ ('zi:-
bõ) /'zi:bó/ s¤ben˚ ('ºa:t&sa;m, -t'Ω-) /'Ka:tza:m/ ra_am˚ (&ZU‰na'lIst) /ZUKna'lIst/ Jour-
nali®˚ (t'ËU˙Í) /t'ZU˙Í/ D(u«el˚ (',a:) /'ja:/ ja˚ (∂ås'¿a:‰) /das'ja:K/ das Jahr˚ ('ºo:t)
/'Ko:t/ rot˚ ('Süa˙k) /'SKa˙k/ 6rank˘ After voiceless stops, /K/ may become (X)
(while phonotactically remaining /FK/): ('phüaes, 'phXaes) /'pKaes/ PrRs˚ ('thüe:tó,
'thXe:tó) /'tKe:tó/ treten˚ ('khüaes, 'khXaes) /'kKaes/ KrRs˘
Voiced stops are realized as half-voiced, both after pauses and after voiceless C:
('ÊU‰k) /'bUKk/ Burg˚ ('mEs&Êa;‰) /'mEsba:K/ messbar˚ ('∂e:nU,) /'de:nU˙/ Dehnu«˚
(&∂as∂u'öEt) /dasdu'Et/ das Duett˚ ('âu:t) /'gu:t/ gut˚ ('öaos&âe:ó) /'aosge:ó/ Qsgehen˘
-m:, -m) /'e:bó/ eben˚ ('nEnó, -n:, -n) /'nEnó/ nennen˚ ('hE˙ô, -˙:, -˙) /'hE˙ó/ hä«en˚
('le:gô, -˙ô, -˙:, -˙) /'le:gó/ legen˘ Please, note that kommen and nennen can become
identical with komm! and nenn! – however, in case of ambiguity, obviously, dis-
tinct forms are used.
Geminates derived by assimilation can be reduced when they occur between
stressed and unstressed syllables: ('li:bõs&Ñe;‰t, -mõ-, -õ-, -m-) /'li:bósve:Kt/ l¤bens-
wert˚ ('ÊUnd¢s≈inIs&te;º,Um, -nn-, -n-, -mI-) /'bUndÈsminIste:KjUm/ Bundesmini®e-
rium˚ ('öand…s, -nn-, -n-) /'and…s/ anders˚ ('öa˙g¢≥bõ, -˙˙-, -˙-) /'angÈge:bó/ a«e-
geben˘ In addition to ('Stüo:&m™˙¢) /'StKo:mE˙È/ Strohme«e˚ we can also have
('Stüo:m&m™˙¢, -&m-) /'StKo:mmE˙È/ Stromme«e˚ with possible reduction; however,
in case of ambiguity, no reduction occurs.
Geminate stops can be reduced to their second element, when they are very sim-
ilar: ('öap&ÊIlt, 'öa&ÊIlt) /'apbIlt/ Abbild˚ ('ºa:t&∂aM∫…, 'ºa:&∂-) /'Ka:tdam∫Ú/ Raddam-
πer˚ ('mIt&thaeló, 'mI&th-) /'mIttaeló/ mittRlen˚ ('mIt&qh™;ló, 'mI&qh-) /'mItqE:ló/
mi˛ählen˚ ('vEk&khømõ, 'vE&kh-) /'vEkkOmó/ wegkommen˚ ('vEk&âe:ó, 'vE&â-) /'vEk-
ge:ó/ weggehen˘ ˛e same holds for grooved constrictives: (∂ås'sElb¢, ∂ås'Ω-, ∂å's-)
/das'zElbÈ/ dasselbe˚ (∂ås'SIf, ∂åS'S-, ∂å'S-) /das'SIf/ das 6i‡˚ ('öaes&Süa˙k, -S&S-, -&S-) /'aes-
SKa˙k/ Eis(rank˘ Instead, the /Âs/ sequence, as we have already seen, has a couple
of possible variants: ('h°:Âst, -âst, -ÂÀt) /'h°:Âst/ hö>®˚ whereas /sÂ/ is only slightly
modified: ('ÊIÀ«) /'bIsÂó/ bis(en˘
5.3.2.8. From the above, a phonation-type di‡erence can become almost func-
tional in three ways: ('öap&ÊIlt, 'öa&ÊIlt) /'apbIlt/ Abbild˚ in comparison with ('öa:pÍ)
/'a:pÍ/ Apel and ('öa:bÍ) /'a:bÍ/ Abel˘ Let us also consider: ('mEndÍ&zo;n) /'mEndÍzo:n/
Mendelsohn˚ ('mEndÍs&so;n, -Ís&Ω-, -Í&s-) /'mEndÍszo:n/ Mendelssohn˘
If the geminates are produced by assimilation of place (of articulation), they can
be reduced (provided the syllable of the first element is unstressed): (&öImmu'ze:ó,
&öImu-) /Immu'ze:ó/ in Museen˚ (mIk'kha‰l, mI'kh-) /mIt'kaKl/ mit Karl˚ (&magnifi-
'qhEnq, &ma˙n-, &ma˙˙-, &ma˙-, -IfI-) /magnifi'qEnq/ Magnifizenz˚ (qhUm'bae&Spi;l,
-m'm-, -'m-) /qUm'baeSpi:l/ zum BRsp¤l˚ (qhu&mInd¢s'∂as&∂a;, -Inn-, -In-) /qu'mIndÈst
'dasda:/ zuminde® das da˘
We have, though: ('öam&m™ldó) /'anmEldó/ anmelden˚ ('mIk&khømõ) /'mItkO-
mó/ mitkommen˚ ('ʺo:k&khaofè) /'bKo:tkaofó/ Brot kQfen˚ ('ºa:kâ¢&fa:º?, -º¡,
-&fa;‰,, -‰n) /'Ka:tgÈfa:Kó/ Rad gefahren˚ ('öam&bIndó, 'öam&m-) /'anbIndó/ anbinden˚
('öagn™s, -˙n-, -˙˙-) /'agnEs/ Agnes˚ ('öa˙≥bõ, 'öa˙&˙-, -e;mõ, -e;õ, -e;m) /'ange:bó/
a«eben˚ ('ÊUnd¢s, -nn-) /'bUndÈs/ Bundes˚ ('vandȺ…, -nn-) /'vandÈKÚ/ Wanderer˘
5.3.2.10. Certainly, the transcriptions seen thus far are useful to conveniently
separate German spelling from its phonic structure, which might seem rather
strange to foreigners. But this is reality. Here are some examples for reflection:
('öap&üaez¢) AbrRse˚ ('th™:kúIÂ÷ 'thE:k-) tägli>˚ ('mUnt&öa‰tó) Mundarten˚ ('öUn&öaof-
üIÂtIÂ, -f&üIÂ-) unQfri>tig˚ (f…'öap&üe;dU,) Verabredu«˚ (Ê¢'öa:p≈õ, -t≈õ) beat-
men…
It is obvious then that German has very complex consonant sequences; here we
will give just one example: ('hE‰ps[t]&Êl™t…) Herb®blätter˘ Please, note carefully even
the spelling sh /s˘h/: ('land¢s&haos÷halt) /'landÈshaoshalt/ LandeshQshalt.
Stress
5.3.3.1. In long words and in sentences, there are unstressed syllables that alter-
nate with stressed or half-stressed ones. Even rhythm groups follow the same prin-
ciple; and all this is already clear from the transcriptions we have given.
German compounds have particular patterns. ˛e most normal and widespread
one shows a primary stress on the prominent syllable of the first lexeme and re-
duced stresses on the ex-prominent syllables of succeeding lexemes.
Here are some examples of two-lexeme compounds: ('SIf&fa;‰t) 6i‡fahrt˚ ('∂aM∫-
&SIf) Damπ(i‡˚ ('Êa:n&ho;f) Bahnhof˚ ('öaezó&ba;n) Eisenbahn˚ ('ho:X&∂OYc) Ho>-
dS˛˚ ('füy:&StYk) Früh®ü$˚ ('öaoto&ba;n) Autobahn˚ (fa'mi:l,ó&na;m¢) Famil¤nna-
me˚ ('mUnt-ha‰&mo;ni&ka) Mundharmonika˚ ('Su:&maX…) 6uma>er˘
It is useful to note that phonemic transcriptions, which do not mark secondary
stresses, naturally imply such a reality. ˛erefore, pronunciations as the following
for four of the compounds just seen are typically foreign-like (in this case Italian):
ç('Sif:faRt, 'dam:Sif, 'ba:nof, 'Ok:døic)Ç. We cannot help ending with an observation
on the compound 6uhma>er and its derived family name 6uma>er˚ which
is generally pronounced by Spanish-, Portuguese-, and Italian-speaking people with
peculiar stress patterns that we will illustrate here using the Italian ones: current-
ly ç(Su'ma:keR), up to an ugly hybrid, which is neither German nor Romance: ç('Su:-
maxeR)Ç.
5.3.3.4. ˛ere are even cases where secondary stresses actually determine the
meaning of certain compounds: ('∂aM∫÷SIf&fa;‰t) Damπ(i‡fahrt çsteamboat voy-
ageÇ, ('∂aM∫&SIf÷fa;‰t) Damπ(i‡fahrt çsteam navigationÇ. In cases such as these,
of course, rhythmic fluctuations are less frequent, although context must (and can)
help very much.
We also find some compounds which do not bear primary stress on their first
lexeme, as: (&,a;‰'qhe:nt) Jahrzehnt˚ (&öo;st…'mo:n&ta;k) O®ermontag˚ (&Sña‰q'Ñaes)
(warzwR•˚ (&ºo;t'khüOYq&Sñ™st…) RotkrSz(we®er˚ (&za;‰'bºYkô) Saarbrü$en˘
Others have more than one primary stress: ('Êlao 'gºy:n) blQ-grün, ('∂Um 'StOlq)
dumm-®olz˚ ('Súe:sñIÂ 'hOlStaen) 6leswig-Hol®Rn÷ however, they can become
(&Êlao'gºy:n, &∂Um'StOlq, &Súe;sñIÂ'hOlStaen), mostly in preintonemes.
Let us observe: ('Êlao 'gºy:n) blQ-grün çblue and green (¤ two colors put along-
side, as in a flag or sports †-shirt)Ç and ('Êlao&gºy:n) blQgrün çbluish green, water
green (¤ two colors blended together, as painters do)Ç.
5.3.3.5. ˛e faster speech becomes, the more non-primary stresses can be re-
duced — dropped, mostly in preintonemes. On the contrary, if speech becomes
slower, the stresses can be restored. In particular in intonemes, the last non-strong
stress can often be strengthened (we will mark it only here with (“)): ('öaoto“ba;n÷
'öan“Stalt÷ 'öalt&ho;X“∂OYc÷ 'haopt&öaof“âa;b¢)÷ besides, ('öaezó÷ba;M&fa;‰“phúa;n, -“fa;‰-
÷phúa;n÷ &Spi;l&va;º?'öaosó&handÍsâ¢&z™l“Saft, -⢓z™l÷Saft÷ &ho;X&ÑakuUm'lIÂt&Êo;gô&S≈™lq-
“öo;fè, -“S≈™lq÷öo;fè).
˛ere are frequent cases of (homographic) compounds with (stressed or un-
stressed) prefix grammemes, which (in their unitary –non-inflected– forms) is the
most important element of semantic di‡erentiation: überse˛en »('öy:b…&z™qó) çto
pass overÇ, ”(&öyb…'zEqó, &öY-) çto translateÇ, umfahren »('öUM&fa;º?, -º¡, -a;‰,, -a;‰n)
çto run downÇ, ”(öUM'fa:º?, -º¡, -a:‰,, -a:‰n) çto drive/sail roundÇ.
Generally, compounds with unstressed monosyllabic su‚xes have /I, È, ó, õ, Í,
212 a handbook of pronunciation
Ú/, even without secondary stresses (except for rhythmic reasons, when several un-
stressed syllables are arranged in sequence, as happens for (-laen) -lRn˚ (-U,) -u«
as well): ('v§Â«túIÂ) wö>entli> (('v§Â«t&úI«, -÷úI«) wö>entli>en].
5.3.3.8. Cases such as the following are interesting too: 'ma>en S¤ d¤ 'Re>-
nu« &fertig˚ 'halten s¤ d¤ 'Pässe be&rRt˚ where the non-conjugated form of the verb
is an indeclinated adjective; the old orthography used to give: fertigma>en˚ berRt-
halten˚ instead of: fertig ma>en˚ berRt halten˘
For subordinate clauses, we have conjugated forms in the last position, infinitive
5. german 213
ones with um … zu: s¤ 'kann 'ni>t &kommen, &wRl s¤ 'morgen &sehr 'früh 'Qf&®ehen
&muss (except when dependent infinitives are more than two: s¤ i® &sehr 'müde, &wRl
s¤, 'hSte 'Morgen, 'sehr &früh hat 'Qf&®ehen &müssen÷ i> 'ho‡e, &dass er d¤ 'Re>nu«
&hat be'zahlen &können÷ i> 'ho‡e, &dass s¤ &Rne 'nSe 'Wohnu« hat 'finden &können]÷
i> 'für>te, &dass er 'Re>t &hat÷ der 'Arzt &wollte 'wissen, &ob er '‡ber ge&habt &habe÷ s¤
'sagt, dass &ihre 'Kinder im 'Garten &sp¤lten÷ i> 'glQbe, &dass s¤ 'ni>t 'h¤r &i®÷ i>
'wR•, dass er vor &zwR 'Jahren &ope'r¤rt &worden &i®÷ i> 'brQ>e 'ni>t zu &antworten÷
es i® 'immer 'besser, 'pünktli> &anzu&kommen÷ um 'pünktli> &anzu&kommen, &müssen
wir d¤ 'U-&Bahn &nehmen˘ ˛e constructions with modals and infinitives in the last
position are to be noticed too: du 'sollte® zum 'Arzt &gehen˘
With inflected separable verbs, postponed particles generally bear primary
stresses ((')); however, it is often weaker (up to a secondary degree), thus we will
indicate it here with (“): s¤ 'kommt 'morgen {an˚ i> 'lege es {bR˚ du 'ruf® {an˚ er 'gab
den 'Kamπ {Qf˚ das 'Da> 'spri«t {über˘
Let us add some typical examples, since the concepts expressed by some words
can (or must) come out: 'Anna darf 'ni>t &mitkommen÷ der 'Br¤f &wurde von 'mir
ge&(r¤ben. More: &kann man 'h¤r 's>wimmen? (two concepts), however: 'können
S¤ Kla'v¤r &sp¤len? something like… çto pianoÇ; 'morgen˚ &muss i> 'fl¤gen÷ i> muss
'morgen &fl¤gen [¤ fl¤gen is already known).
‹en no complement or predicate is present, destressing certainly does not oc-
cur, as in the following sentences, where verbs are preceded only by words (and
concepts) with very low information power: 'kann i> 'mal telefo'n¤ren?÷ er 'fuhr
in d¤ 'Fer¤n, um &si> zu er'holen÷ er 'l¤• das Re®Q'rant, 'ohne zu be'zahlen÷ 'bitte,
'blRben S¤ &do> 'si˛en˘ Let us further consider the following examples: das i® be-
'®immt 'unab&si>tli> ge&(ehen÷ er hat ge'sagt, dass es 'unab&si>tli> ge&(ehen &i®÷
er hat ge'sagt, dass es 'unab&si>tli> ge&ma>t &worden &i®÷ er hat ge'sagt, dass es 'unab-
&si>tli> ge&ma>t &worden sRn &kann÷ er hat ge'sagt, dass es 'unab&si>tli> ge&ma>t
&worden sRn &könnte˘ Here, the particular syntactic structure of German subordi-
nate clauses, with verb phrases at the end of sentences, makes it possible for the
last primary stress to be followed by a considerable number of syllables.
Intonation
5.3.4. û 5.4 shows neutral German intonation. Each tonogram must be ana-
lyzed very carefully. Here we will simply provide examples for the three marked
intonemes. In § 5.3.1.1, the third and fourth utterance (of the second group of ex-
amples, before the reduced-form list) show mid-pitch parentheses (Ô § 13.24 of
NPT/HPh), which is also typical of German and French. Here is another example:
… und, &wenn es 'mögli> zu 'ma>en &wäre, … Others may be found in the transcrip-
tions of the text (§ 5.5):
/./: (öIÂ'su:X¢ öaem'm°:bÍg¢&S™ft3 3) /IÂ'zu:xÈ aen'm°:bÍgÈSEft./ I> su>e Rn Möbelge-
(ä⁄.
/?/: (¿'khOmõzi öaos'ö°:st…&ºaeÂ1 1) /¿'kOmózi aos'°:stÚKaeÂ?/ Kommen S¤ Qs Ö®er-
rR>?
214 a handbook of pronunciation
Other accents
North-eastern-Germany German
5.4.1.1. ˛e first vocogram shows the vowels of the ex-∂∂® mediatic (and neu-
tral) accent. It is not the local pronunciation of Berlin or of Saxony. As will be seen
shortly, by comparing this vocogram with that of û 5.1, long stressed V are actual-
ly monotimbric diphthongs, or vocalic doublings, with a slight upward move-
ment. Besides, /E, E:, §, O/ are slightly closer (su‚ciently so to use di‡erent phonet-
ic symbols: (™, ™™, #, ø)).
We just give some examples to directly show the di‡erence, thanks to phonetic
transcriptions: ('fi;il) /'fi:l/ v¤l˚ ('ze;eó) /'ze:ó/ sehen˚ ('Sp™;™t) /'SpE:t/ spät˚ and:
('Êa;an) /'ba:n/ Bahn˚ ('zo;o) /'zo:/ so˚ ('âu;ut) /'gu:t/ gut˚ ('thy;y‰) /'ty:K/ Tür˚ ('S°;°n)
/'S°:n/ (ön˘
5. german 215
˛e two realizations of /È/ are slightly lower: ('öUnz‘ºÈ, -ndz-, -nQ-) /'UnzÈKÈ/ un-
sere÷ whereas, the possible vocalization of /K, Ú/ is closer: ('vas…, -x‰, -x˜, -xº, -x) /'va-
sÚ/ Wasser˘
5.4.1.2. By comparing the vocogram provided, it can be seen that the second
elements of /ae, ao, OY/ are more central (and even lower for /OY/): (aÙ, aP, ø+). In
addition, especially the first element of /ao/ is slightly backer and the first of /OY/
is slightly higher, and indeed it crosses the edge towards the upper box: ('öaÙs) /'aes/
Eis˚ ('ÊlaP) /'blao/ blQ˚ ('nø+) /'nOY/ nS˘ ˛ere is little di‡erence for the two xeno-
phonemes /eI, oU/; while, for the possible first-element variants of /’ae, ’ao, ’a÷ ’OY/
they coincide.
We add for C that there is quite a frequent change of voiceless constrictives (and
/z/), after /n, l/, into homorganic sequences of stops + constrictives, or even into
real stopstrictives (Ô § 5.3.2.9). ≈nally, before pauses, /kò/ can be realized as a (pre)-
velar stopstrictive (w, ∞): ('khüi;ik, -w, -∞) /'kKi:k/ Kr¤g˚ ('tha;ak, -w, -∞) /'ta:k/
Tag˚ ('qhu;uk, -w, -∞) /'qu:k/ Zug˘ However, all these realizations are not at all rec-
ommendable.
˛e tonogram shows intonation, which sounds slightly less peculiar than the
çwesternÇ neutral one, approaching what may be called çinternationalÇ.
/i:, ’i/ (i;i, i), /y:, ’y/ (y;y, y) /u:, ’u/ (u;u, u)
/I/ (I), /Y/ (Y) /U/ (U)
/e:, ’e/ (e;e, e), /°:, ’°/ (°;°, °) /o:, ’o/ (o;o, o)
/È/ (È, ‘˘º, º‘, ºÈò)
/E/ (™, ’™), /§/ (#, ’#)÷ /E:/ (™;™, ’™;) /O/ (ø, ’ø)
/Ú/ (…) {“ (x)}
/a/ (a), /a:/ (a;a)
Austrian German
5.4.2.1. ˛is accent is the çmediaticÇ one, which prevails in radio and television
broadcastings and is half-way between the slightest and broadest accents. ˛ese ob-
servations are drawn from our (growing) sound archives.
çLongÇ monophthongs are realized as geminate or doubled vocoids and have a
shorter duration although gemination gives an impression of greater fullness,
which almost equalizes true length: not (é:), but (éé).
It is at once clear even that the articulations corresponding to /I, Y, U÷ E[:], §, O/
are closer than in neutral German, so much so that they are represented as (i, y, u÷
™[™], #, ø) (in less broad accents we find (I°, Y°, U°÷ E[E]°, §°, O°), with the unstressed
variants (™, #, ø), slightly closer than in neutral pronunciation): ('fiS) /'fIS/ „(˚
('fyMf) /'fYnf/ fünf˚ ('hunt) /'hUnt/ Hund˚ ('™lf) /'Elf/ elf˚ ('Sp™™t) /'SpE:t/ spät˚ ('q6#lf)
/'qv§lf/ zwölf˚ ('øft) /'Oft/ o⁄˘ Certain words have /e:/ (ee) for /E:/.
5.4.2.2. Both /a/ and /a:/ are back-central: ('StAt) /'Stat/ Stadt˚ ('StAAt) /'Stat/ Staat
(in less broad accents we find (a, aA) (and (Å, ÅÅ) in refined, or a‡ected, ones),
whereas in broader ones we have: (å, åå) and (Ø, ØØ) as well (which will certain-
ly be noted at once).
Austrian traditional dialects generally present phonemic opposition between a
back a and a fronter one, both short and long. It is possible to find this timbre dis-
tinction in language too, where the back timbre is used in traditional words,
whereas the fronter one is used in loanwords and neologisms, as happens for in-
stance in Bank\ /'bA˙k/ ('ÊA˙k) çbenchÇ, /'ba˙k/ ('Êa˙k) çbank, gambling tableÇ.
A very broad accent has /e:, °:, o:/ (eÙ, #°, øo): ('vee, ≠'veÙ) /'ve:/ weh˚ ('Ê°°,
≠'Ê#°) /'b°:/ Bö˚ ('soo, ≠'søo) /'zo:/ so˘
5.4.2.3. /È/ is realized as (Ù, È), with the possible vocalization of /K, Ú/ (å) (with
consonantal accompaniments too); thus we have: ('unsȺÙ) /'UnzÈKÈ/ unsere and
('vAs…, -å‰, -å˜, -åº, -å) /'vasÚ/ Wasser˘ ˛e symbol (å) corresponds to the one current-
ly used, though less rigorously, in neutral German as well, for (√), which on the
contrary is appropriate for less broad Austrian accents that have (Õ) (and even
(x÷)); whereas the broadest accents have (a[í]). It is to be noted that (Ù) /È/ is front-
-central, ¤ considerably fronter than in neutral German. ˛e most refined or a‡ec-
ted accents may reach (Ùº, Ù√) for /ÈK, Ú/ (Ô the last vocogram but one).
Moving to diphthongs, we can see that /OY/ has just a higher first element, while
/ae, ao/ are decidedly di‡erent, (ÄÙ, åø): ('ÄÙs) /'aes/ Eis˚ ('Êlåø) /'blao/ blQ˚ ('nøY)
/'nOY/ nS˘ Less broad accents have, (ÅÙ, Aø, O°), respectively, while the broadest
ones have (EÙ, ∏ø, √Y) (as in Vienna, where even (πÉ, Øø, ∏ê) occur, and even (O,),
as a compromise, and (@°, §°) as well). Xenophonemes are slightly closer than in
neutral German: /eI, oU/ (ee, ™i÷ oo, øu).
5.4.2.4. In the second vocogram, two further typical diphthongs are evident (al-
though marked in grey, since they only occur in local family and place names), (iÙ,
uÙ) /iÈ, uÈ/: ('∂iÙks, '∂iiks) D¤x˚ ('ºuÙq, 'ºuu™q) Rue˛˘ No variant of /ae, ao, OY/ in
5. german 217
reduced forms are given, because they tend to remain as indicated; nevertheless,
we have shown /’a/ (√), which is possible in reduced forms.
In comparison with neutral German, there are some di‡erences in phonemic
length too, as in: ('Alt…&tum) /'altÚtu:m/ Altertum˚ ('Spiil&Êa‰) /'Spi:lba:K/ sp¤lbar˚
('hA‰q) /'ha:Kq/ Harz˚ ('ÊA‰S) /'ba:KS/ Bar(˚ ('Sust…) /'Su:stÚ/ 6u®er÷ (âÙ'Sloos) /gÈ-
'SlOs/ Ge(loss/-o•˘
/ / (2 2 Ç 2 2 Ç 2 2 Ç 2) /./ (2 ' 2 3)
/¿ / (¿ 2 2 Ç 2 2 Ç 2 2 Ç 2) /?/ (2 Ç 1 2)
/¡ / (¡ 2 2 Ç 2 2 Ç 2 2 Ç 2) /÷/ (2 ' 2 1)
/˚ / (˚ 2 2 Ç 2 2 Ç 2 2 Ç 2) /,/ (2 ' 2)
/i:, ’i/ (±ii, ±i), /y:, ’y/ (yy, y) /u:, ’u/ (±uu, ±u)
/I/ (±I), /Y/ (Y) /U/ (±U)
/e:, ’e/ (±ee, ±e), /°:, ’°/ (±°°, ±°) /o:, ’o/ (±oo, ±o)
/È/ (±Ù, ±ºÙò, ±Èº, ±ºÈ)
/E/ (±E, ±’™), /§/ (±§, ±’#) /O/ (±O, ±’ø)
/E:/ (±EE, ±’™;) /Ú/ {(…) “) (±√)
/a/ (±a), /a:/ (±aA)
/ÈK-/ (±±Ùº-)
/e:/ (≠≠eÙ, ee), /°:/ (≠≠#°) /o:/ (≠≠øo)
/OY/ (O°, O,, ≠√Y, ≠≠∏ê, ≠≠@°, ≠≠§°)
/ae/ (±ÅÙ, ≠EÙ, ≠≠πÉ) /ao/ (±Aø, ≠∏ø, ≠≠Øø)
/…/ (±±Ù√, ±x, ≠a) /a[:]/ (±±Å[Å], ≠å[å], ≠≠Ø[Ø])
218 a handbook of pronunciation
˛e last two vocograms show both less broad ((±)) vowel realizations (up to
some a‡ected ones, (±±)), and –on the contrary– some of the broadest realizations
((≠)).
5.4.2.5. Moving to C˚ we find that often /˙ò, ˙0/ become /˙kò, ˙k0/ (but they
are not recommendable): ('qÄÙtu˙, -˙k) /'qaetU˙/ ZRtu« (however, even without
(k), the nasal is always velar, not uvular, in /Kó/ as well: (SpA'qiiºô, -ii‰˙, -ii‰n)).
/t, d/ are more dental than denti-alveolar: (∂e't™kto‰) /de'tEktoK/ Detektor÷ of-
ten /kò, kl, gl/ can be prevelar or retracted palatal (and /kò/ may even become stop-
strictive, but these variants are best avoided): ('tAAk, -´, -∞, -©, -k) /'ta:k/ Tag˚ ('∂u˙-
kl…, -”´l-, -N©-) /'dU˙klÚ/ dunkler˚ ('ºeeâl…, -"l-, -ˆl-) /'Ke:glÚ/ Regler˘
Voiceless stops and stopstrictives are not çaspiratedÇ (except in less broad pro-
nunciations), as we have just seen for /q/: ('pool) /'po:l/ Pol˚ ('toon) /'to:n/ Ton˚
('kint) /'kInt/ Kind˚ ('∫unt) /'∫Unt/ Pfund˚ ('cAko) /'cako/ T(ako˘
Voiced stops are half-voiced, (Ê, ∂, â); but they can also be (b, d, g), between
voiced phones, again in less broad accents; however, in the broadest ones, we gen-
erally find (p, t, k): ('ÊuuÊÙ, -6Ù) /'bu:bÈ/ Bube (often we have (é6é) /ébé/), ('∂uu-
∂ó) /'du:dó/ Duden˚ ('âeeâô) /'ge:gó/ gegen.
On the other hand, within words, b˚ d˚ g followed by heterosyllabic /n, l/ can
become /b, d, g/ (instead of /p, t, k/, as in neutral German pronunciation, which
has neutralization): ('ø‰∂nu˙) /'OKdnU˙/ Ordnu«˚ ('m°°âliÂ) /'m°:klIÂ/ mögli>.
No glottal stop is used, except in less broad accents: (iÂ'™sÙ) /IÂ'EsÈ/ i> esse˚ (te'AA-
t…) /te'a:tÚ/ †eater˘
est and most typical speakers. In the broadest accents (as in the traditional local
dialect – see the cennese phonosynthesis, § 17.28 of NPT/HPh]˚ we find velar (ı,
”), in the sequences /kl, gl÷ kÍ, gÍ/. In cennese pronunciation again, in addition
to (“) (and (”)), we often find it vocalized in (,) (as in the dialect): ('Spiiâ”, -â,, 'ee-
s“, -s,) /'Spi:gÍ, 'e:zÍ/ Sp¤gel˚ Esel.
Swiss German
5.4.3.1. Even for this accent we will add, when relevant but without further
figures, some more non-neutral characteristics, taken from our sound archives. ˛e
main accent illustrated is the mediatic one – something in between the broadest
“ most local ones and a neutralized, unmarked, accent – expressly acquired.
˛e first vocogram shows the short and long çmonophthongsÇ; actually, in this
accent çlongÇ vowels are diphthongs with narrow movements: ('fiil) /'fi:l/ v¤l˚
('f5yy) /'fKy:/ früh˚ ('âuut) /'gu:t/ gut˚ ('Êeet) /'be:t/ Beet˚ ('°°l) /'°:l/ Öl˚ ('Ñoo) /'vo:/
wo˚ ('SpEEt) /'SpE:t/ spät (which is very stable and typical) ('StaAt) /'Sta:t/ Staat˘ ˛e
last two have non-neutral broad variants: (πE) and (åå) (or (åA) as well, which is
typical of Bern, and (ù∏), typical of Zurich). (˛e phonosynthesis in § 17.20 of
NPT/HPh shows the characteristics of Zurich 6wyzertü˛ {which is Aleman-
nic}, with still more di‡erent realizations.)
Among short V, /I, Y, U/ have closer realizations (i, y, u): ('fiS) /'fIS/ „(˚ ('fyMf)
/'fYnf/ fünf˚ ('hun∂) /'hUnt/ Hund÷ /E, §, O/ are relatively open: ('Elf) /'Elf/ elf˚
('q6§lf) /'qv§lf/ zwölf˚ ('Oft) /'Oft/ o⁄ (with unstressed taxophones (™, #, ø)). For
the spelling ä˚ we can also find a broad non-neutral realization (π): ('fElt, ≠'fπlt)
/'fElt/ fällt÷ /a/ is back-central (A) (which, in broad marked pronunciation, is
rounded too, (ù), or fully back, (å)): ('StAt) /'Stat/ Stadt˘
220 a handbook of pronunciation
5.4.3.2. For /È/ we have (Ù), in every context, since /K/ is not uvular: (âÙ'5aA∂Ù)
/gÈ'Ka:dÈ/ gerade˚ ('unsÙRÙ) /'UnzÈKÈ/ unsere˚ ('unsÙR) /'Unz…/ unser˘ Besides, we gen-
erally find (-Ùn, -Ùm, -Ùl) for /ó, õ, Í/: ('faARÙn) /'fa:Kó/ fahren˚ ('looΩÙm) /'lo:zõ/ lo-
sem˚ ('eeΩÙl) /'e:zÍ/ Esel˘ Only in very careful pronunciation, can we find (ó, õ, Í),
and (…, √‰, √˜, √º) as well, even for speakers lacking /K/ (º).
Neutral Swiss diphthongs are /ae, ao, OY/ (aÙ, Ao, OI): ('aÙΩ) /'aes/ Eis˚ ('ÊlAo)
/'blao/ blQ˚ ('nOI) /'nOY/ nS÷ in the broadest accents we find: /ae/ (EI, ÄI), /ao/ (ÄU,
∏U), /OY/ (êI, oÙ). Xenophonemes are decidedly closer than in neutral German: /eI,
oU/ (ee, ei÷ oo, ou).
/ / (2 2 ç 2 2 ç 2 2 ç 2) /./ (2 ç 2 3)
/¿ / (¿ 2 2 ç 2 2 ç 2 2 ç 2) /?/ (2 Ì 1 2)
/¡ / (¡ 2 2 ç 2 2 ç 2 2 ç 2) /÷/ (2 ç 2 2)
/˚ / (˚ 2 2 ç 2 2 ç 2 2 ç 2) /,/ (2 ' 2)
In the second vocogram again, three further typical diphthongs are evident (al-
though marked in grey, since they only occur in local family and place names), (iÙ,
yÙ, uÙ) /iÈ, yÈ, uÈ/: ('∂iÙt) D¤th˚ ('5yÙâ) Rüegg˚ ('huÙp) Huep˘ Except in less marked
5. german 221
–or a‡ected– accents, often /y:, y, yÈ/ become /i:, i, iÈ/. In the same figure, possible real-
izations of reduced forms of /ae, ao, OY, a/ are given: (åÙ, ao, Ö¤, √).
In comparison with neutral German, there are some words with short V, as in:
('A5t) /'a:Kt/ Art˚ ('flOs) /'flo:s/ √o• [√oss\ in Switzerland • is not used], ('jAâ∂) /'ja:kt/
Jagd˚ ('nun) /'nu:n/ nun˚ ('∫E5∂) /'∫e:Kt/ Pferd˚ ('fO5) /'fo:K/ vor˚ ('ÑE5∂Ùn) /'ve:Kdó/
werden˚ ('E5∂Ù) /'e:KdÈ/ Erde˚ ('Ñuks) /'vu:ks/ Wu7˚ ('qE5tliX, -á) /'qhE:‰tlIÂ/ zärt-
li>˘
˛e last vocogram shows broad realizations of /E, E:, a, a:÷ ae, ao, OY/ (and less
broad ones of /I, Y, U/).
5.4.3.3. Moving to C˚ we have to say that /t, d/ are decidedly dental: (∂e'tEktoR)
/de'tEktoK/ Detektor÷ whereas, /k/ is (k, w, ∑, k): ('kin∂, 'w, '∑-, 'k-) /'kInt/ Kind˚
('sAk, -w, -∑, -k) /'zak/ Sa$; please, note that in the transcription of § 5.5.2.4
(and § 5.5.1.2), we write (∑), to insist on this typical realization (although it tends
to be avoided in neutral-like pronunciations, perhaps resorting to a velar stop-sem-
i(con)strictive, (˜), which is half-way between (k) and the other stopstrictives in-
dicated).
/p, t, k÷ ∫, q, c/ are not çaspiratedÇ (unless çaspirationÇ is voluntarily used):
('pool) /'po:l/ Pol˚ ('toon) /'to:n/ Ton˚ ('kuu, 'w, '∑-, 'k-) /'ku:/ Kuh˚ ('∫un∂) /'∫Unt/
Pfund˚ ('qeen) /'qe:n/ zehn˚ ('cAkko) /'cako/ T(ako.
Furthermore, (ö) is lacking, except in less marked and less typical accents: (te-
'aAtÙR) /te'a:tÚ/ †eater˚ (iX'EssÙ, iá-) /IÂ'EsÈ/ i> esse÷ as can be seen, though, lexeme
and grammeme boundaries are respected; only in very broad pronunciations resyl-
labification is possible: (i'XEssÙ, i'á-).
Voiced C of diphonic pairs are actually half-voiced, (Ê, ∂, â÷ Ñ, Ω, Ë), even before
pauses or when syllable-final (where, in neutral German, they are neutralized and
change to /p, t, k÷ f, s, S/). ˛erefore, we generally find /s/ for /z/, after pauses or C\
('ÊuuÊÙ) /'bu:bÈ/ Bube˚ ('ÊuuÊ) /'bu:p/ Bub˚ ('∂uu) /'du:/ du˚ ('un∂) /'Unt/ und˚ ('5ee-
∂Ùn) /'Ke:dó/ reden˚ ('ÊooâÙn) /'bo:gó/ Bogen˚ ('taAâ) /'ta:k/ Tag˚ (âÙ'Ñin) /gÈ'vIn/ Ge-
winn˚ ('leeΩÙn) /'le:zó/ lesen˚ ('saAâÙn) /'za:gó/ sagen˚ ('ÊinsÙ) /'bInzÈ/ Binse˚ ('paAËÙ)
/'pa:ZÈ/ Page˘
5.4.3.4. Often, /f/ occurs instead of (neutral) /v/: (no'fEmÊÙR) /no'vEmbÚ/ No-
vember˚ (f™n'tiil) /vEn'ti:l/ Ventil˚ (ful'kaAn, -w, -∑-, -k-) /vUl'ka:n/ Vulkan˘
For sp˚ ®˚ internal or final we have /sp, st/, but in broad accents we find /Sp, St/:
(k'nOspÙ, ≠-SpÙ) /k'nOspÈ/ Knospe˚ (Êist, ≠-iSt) /bIst/ bi®˚ ('âEstÙRn, ≠-St-) /'gEstÚn/
ge®ern˘ Generally, for internal /S/ (after stressed short V, like for the other C] we
have (SS): ('muSSÙl) /'mUSÍ/ Mu(el÷ for 7 /ks/, in marked pronunciation it is fair-
ly possible to have ç/xs/Ç: ('fuks, ≠-Xs, ≠-ás) /'fUks/ Fu7˘
Only in less broad speech, can we find /Â/ (but articulated as postpalatal (…), or
prevelar (¢)); whereas, normally, we have /x/ (X, á): ('milX, -á÷ ±-…, -±-¢) /'mIlÂ/
Mil>˚ ('AoX, -á) /'aox/ Q>˘ -ig is regularly /Ik/ (¤ ç/Ig/Ç; while /IÂ/ is decidedly
rare and only voluntarily used): ('q6Anqiâ) /'qvanqIÂ/ zwanzig÷ and /òÂ/ is normal-
ly /k/: ('kiinA, 'w-, '∑-, 'k-) /'Âi:na/ ≥ina÷ /j/ is an approximant, /j/: ('jaA) /'ja:/ ja˘
222 a handbook of pronunciation
South-Tyrol/Alto-Adige German
5.4.4.1. ˛e first vocogram shows the short and long monophthongs of South-
-Tyrolese mediatic pronunciation. Even for this accent it is immediately clear that
the articulations corresponding to /I, Y, U÷ E[:], §, O/ are closer than in neutral Ger-
man – so they are represented with (i, y, u÷ ™[:], #, ø) (in less broad accents we find
(I°, Y°, U°÷ E[:]°, §°, O°), and unstressed (™, #, ø) only slightly closer than in neutral
pronunciation): ('fiS) /'fIS/ „(˚ ('fyMf) /'fYnf/ fünf˚ ('hunt) /'hUnt/ Hund˚ ('™lf) /'Elf/
elf˚ ('Sp™:t) /'SpE:t/ spät˚ ('qV#lf) /'qv§lf/ zwölf˚ ('øft) /'Oft/ o⁄˘
5.4.4.2. /a:/ is diphthongized: ('Stat) /'Stat/ Stadt˚ ('StaAt) /'Stat/ Staat (in less
broad accents /a, a:/ are more similar to one another: (a, a:); whereas, in the broad-
est accents, near the dialects, we can also have (ù, Ø:) (which will certainly be not-
ed at once; Ô § 16.15 of NPT/HPh, where the phonosynthesis of the South-Tyrolese
5. german 223
5.4.4.3. /È/ is realized as (Ù, É), with the vocalization of /K, Ú/ (√) (with conso-
nantal accompaniments too), and in the most broad pronunciations even (∏), as
in the dialects (Ô the phonosynthesis); thus we have: ('unsɺÙ) /'UnzÈKÈ/ unsere and
('vas…, -√‰, -√˜, -√º, -√÷ 'vùss∏, -∏‰, -∏˜, -∏º) /'vasÚ/ Wasser˘ It is important to note that
(Ù, É) /È/, in less broad accents, given in the last vocogram, are realized slightly
di‡erently (as other V are too) although they are represented by the same symbols.
≈nal -e (not only of verbs) and the first one in -ere(C)ò tend to be dropped (except
in Val Pusteria, or Pustertal), especially in non-controlled pronunciation: ('ty:pÙ,
'ty:p) /'ty:pÈ/ Type (which can coincide thus with ('ty:p) /'ty:p/ Typ]˘
˛e second vocogram shows the diphthongs of South Tyrolese German. ‹ile
/OY/, in comparison with neutral German, di‡ers mostly in its second element that
is unrounded, (O¤); /ae, ao/ are decidedly less wide, (aÙ, aP): ('aÙs) /'aes/ Eis˚ ('plaP)
/'blao/ blQ˚ ('nO¤) /'nOY/ nS˘ In the dialect, for /ae, ao/, we have (aÙ, AP), but also
(aP), which are even less wide; only in Bolzano the same is true for /OY/ (ø¤), since
in traditional dialects it has merged into /ae/ (aÙ), or it is represented by /ui, oi/:
('nui) /'nui/ nui {= nS}, ('foiR) /'foiK) foir {= FSer}. ˛e xenophonemes are quite
rarely used (and, if they are, it is in a completely intentional way): /eI, oU/ (e:, ±ei÷
o:, ±ou).
5.4.4.4. In the second vocogram, two further typical diphthongs are evident (al-
though marked in grey, since they only occur in local family and place names), (iÙ,
uX) /iÈ, uÈ/: ('tiÙtÍ, ±'∂iitÍ) D¤tl˚ ('wuXns, ±'ku:ns) Kuens˘ In the same diagram, the
possible realizations of /ae, ao, OY, a/ (åÙ, åP, Ö¤, å), in reduced forms, have been
marked.
As already said, the last vocogram shows less broad vowel realizations ((±)),
which are rather similar to the neutral German ones; however, this pronunciation
is still recognizable, since the prosodic characteristics have typical di‡erences.
˛ere are di‡erent distributions too, as for instance /E/ in ('™‰st, 'E-÷ ≠'™∏St) /'e:Kst/
er®˚ ('∫™‰t, 'E-÷ ≠'∫™∏t) /'∫e:Kt/ Pferd (which in neutral German have /e:/: ('öe:‰st,
'∫he:‰t)).
224 a handbook of pronunciation
/ / (2 2 Ì 2 2 Ç 2 2 Ç 2) /./ (2 ' 2 3)
/¿ / (¿ 2 2 Ì 2 2 Ç 2 2 Ç 2) /?/ (2 ' 2 1)
/¡ / (¡ 2 2 Ì 2 2 Ç 2 2 Ç 2) /÷/ (2 ' 2 2)
/˚ / (˚ 2 2 Ì 2 2 Ç 2 2 Ç 2) /,/ (2 ' 2)
/i:, ’i/ (±i:, ±i), /y:, ’y/ (±y:, ±y) /u:, ’u/ (±u:, ±u)
/I/ (±I), /Y/ (±Y) /U/ (±U)
/e:, ’e/ (±e:, ±e), /°:, ’°/ (±°:, ±°) /o:, ’o/ (±o:, ±o)
/E/ (±E, ±’™), /§/ (±§, ±’#) /È/ (Ù, ºÙò, ±Éº-, ±ºÉ-)
/O/ (±O, ±’ø)
/E:/ (±E:, ±’™;)
/Ú/ (…) {“ (±Éº, ±√)}
/a/ (±a)÷
ƒ /a:/ (±a:)‡
5.4.4.5. Moving to C˚ we find that generally internal /˙/ becomes /˙g/ (but it
is not so in less broad accents): ('pºi˙âÙn, ±'ʺI˙ô) /'bKI˙ô/ bri«en (however, even
without (g), as in less broad accents, the nasal is always velar, not uvular). Espe-
cially in checked syllables with N (above all in stressed positions), V are nasalized
in broad accents: ('vi:n, 'vi:n) /'vi:n/ W¤n˚ ('fa˙, 'fÅ˙) /'fa˙/ Fa«˘
/t, d/ are dental: (te't™kto‰) /de'tEktoK/ Detektor÷ often /òk, kò, kô/ are stopstric-
tives, or even stop-semi(con)strictives, velar (or postpalatal, near front V]\ ('kømõ,
'w-, '˜-) /'kOmõ/ kommen˚ ('søkô, 'søwÙn, -˜-) /'zOkô/ So$en˚ ('tik, -w, -˜, -k,
-%) /'dIk/ di$˘
Voiceless stops and stopstrictives are not çaspiratedÇ (generally even in less broad
accents) as we have just seen for /k/: ('po:l) /'po:l/ Pol˚ ('to:n) /'to:n/ Ton˚ ('kint, 'w-,
'k-) /'kInt/ Kind˚ ('∫unt) /'∫Unt/ Pfund˚ ('cako) /'cako/ T(ako˘
Phonemic voiced stops are voiceless in typical pronunciation, but half-voiced,
5. german 225
(Ê, ∂, â) in less broad accents. However, between voiced phones, they are half-
-voiced, (Ê, ∂, â) (except, usually, in the dialects which –mostly– have (p, t, k/w),
as in the broadest accents. Instead, in less broad accents they are voiced, (b, d, g)):
('pu:ÊÙ, ±-bÈ) /'bu:bÈ/ Bube˘ Half-voiced realizations are possible before internal /n,
l/, too: ('O‰∂nu˙, 'm°:âliÂ) /'OKdnU˙, 'm°:klIÂ/ Ordnu«˚ mögli>. (In Bolzano, we
can find (éBé, éƒé, éŸé) /ébé, édé, égé/: ('Êu:BÙ).)
˛ere is no (ö), except in less broad accents, where however it is less evident than
in neutral German (and, in this case, we could mark it with a special symbol, (,)):
(iÂ'™sÙ, ±,IÂ',EsÈ) /IÂ'EsÈ/ i> esse˚ (te'a;At…, ±te',a:-) /te'a:tÚ/ †eater˘
5.4.4.6. For constrictives, it is to be noted that /z/ is normally /s/ (s) (generally,
even in less broad accents, in Bolzano as well): ('sa;Aâô) /'za:gó/ sagen˚ ('also) /'al-
zo/ also˚ ('ºaÙsó) /'Kaezó/ rRsen (which becomes identical to ('ºaÙsó) /'Kaesó/ rR-
•en]˘ In initial position, before C˚ /s/ practically never occurs (substituted by /S/);
and, as far as internal and final /st/ is concerned, in broad accents, as in the dia-
lects, we typically find /St/: ('li:pstó, ≠-Stó) /'li:pstó/ l¤b®en˚ ('sønst, ≠-nSt) /'zOnst/
son®˘ However, in case they are heterolexemic, we have /st/: ('sams&taAw) /'zams-
ta:k/ Sam®ag (as in the dialects: ('sams&tiw, 'sùms&tik) /'samstik/ sam®ig]˘
For (tautosyllabic) /˘0v/, we have (0V): ('SV™st…) /'SvEstÚ/ 6we®er˚ ('qVaÈ)
/'qvae/ zwR˚ ('kV™l) /'kvEl/ Quell÷ for /f/, it is possible to have /v/, besides in forms
such as Vize (which, even in neutral German, may have the variant with /v/, al-
though not prevailing), even for vò˚ against neutral phonemic rules, in cases such
as positiv\ ('vi:qÙ, &posi'tiÑ). In addition, /j/ is approximant (j): ('ja;A) /'ja:/ ja; gener-
ally, /x/ is velar (x): ('na;Ax) /'na:x/ na>; /Â/ is maintained (even before /K/, contrary
to what happens in Austria): ('tu‰Â) /'dUKÂ/ dur>; but, for /-IÂ/ we have /-Ik/ (ex-
cept in less broad accents) ('qVanqik, -w, -k) /'qvanqIÂ/ zwanzig÷ for /òÂ/ we syste-
matically find /k/: ('ki:na, 'w-, 'k-) /'Âi:na/ ≥ina˘ However, in the dialects, except
in Bolzano, (Â) /Â/ does not exist at all.
˛e most widely used articulation for /K/ is a uvular constrictive, (º), which co-
exists with a fairly widespread uvular trill, (K), with the possibility of postnuclear
vocalization (that, as we have already seen, is back-central (√) (or back, (∏), in the
dialects and in the broadest accents). In absolute final position, after stressed V, in
the dialects and in the broadest accents we have (R): ('ºa;A‰÷ -a;A√÷ ≠-a;AR÷ ≠-Ø:R)
/'Ka:K/ rar˚ ('hi:‰÷ 'hi:√÷ ≠'hi:R) /'hi:K/ h¤r˘
In lofty or foreign words, there is a frequent use of Italian –or Italian-like– struc-
tures: /gn/ = /nj/: (pºo~'jo:sÙ) /pKog'no:zÈ/ Prognose˚ (si~'ja;Al) /zIg'na:l/ Signal˚ (&ºe-
si~'ji:ºô, -º¡, -'ji:‰n, -‰n) /KezIg'ni:Kó/ resign¤ren÷ /t/ = /q/: (ti&ploma'qi:) /diploma-
'ti:/ Diplomat¤˚ (&kaºan'qi:) /gaKan'ti:/ Garant¤÷ /q/ = /c/: (m™‰'ce:∂™s) /mEK'qe:-
dEs/ Mercedes˚ (&pa‰ce'lo:na) /baKqe'lo:na/ Barcelona÷ /S/ = /sk/: ('ske:ma) /'Se:ma/
S>ema˚ (&skiqo'fºe:n) /Siqo'fKe:n, sÂi-/ S>izo#ren÷ /Z/ = /dZ/: (&tËu‰na'list) /ZUKna-
'lIst/ Journali®÷ /kv/ = /kw/ (kj) (in addition to (kV)): (kja‰'t™t) /kvaK'tEt/ Quar-
tett˘ For V˚ we find: /y/ = /u/: (pu&ºokºa'qi:) /byKokKa'ti:/ Bürokrat¤÷ /Oy/ = /eu/:
(eu'ºo:pa) /OY'Ko:pa/ Europa˚ (neu'tºa;Al) /nOY'tKa:l/ nStral˘
South-Tyrolese intonation is easily recognized because of its half-low stressed sylla-
bles, except the first one which is half-high; mostly the interrogative intoneme (and the
suspensive one too) have particular movements, which can be seen in the tonogram.
Text
5.5.0. ˛e story †e North Wind and the Sun follows, given in some di‡erent
(çnormalizedÇ) versions. We start with the (neutral) German pronunciation of
(neutral British) English – this is the first step of the phonetic method (the writ-
ten text is given in § 2.5.2.0). ˛e German translation follows, in its neutral ver-
sion; then the north-eastern accent, followed by the Austrian, Swiss, and South-
-Tyrolese ones. ˛ese transcriptions would have been (visually) more di‡erent, if
we had used the çspecialÇ (denti-alveolar) symbols ((4, 7, fl)), and instead of (‰, …)
(for /K, Ú/) respectively the following possible variants: (√) (neutral), (x) (north-
-eastern), (å, a) (Austrian), (5, R) (Swiss), (√, ∏) (South Tyrolese).
At the end, as always, there is the version which gives the English pronuncia-
tion of German, by neutral British speakers, fluent in German (after prolonged
contact with native speakers, but with no help from the phonetic method), who
have adequately learned the relative prominences, but who substantially use seg-
mental and intonation elements which are typical of neutral British English (for
reference purposes, although, of course, a neutral accent is not so common). Ob-
viously, the same principle is valid for the foreign pronunciation of English, giv-
en first.
Speakers of American English could prepare their own version both of the Ger-
man pronunciation of English and of their pronunciation of German, as an excel-
lent exercise, by listening to native speakers, best of all after recording them. Of
course, speakers of other languages could do the same thing. ˛e author would be
happy to receive their transcriptions and recordings, both in case of help –should
they need it– and to make their contribution known to others (possibly in our
website on canIPA Natural Phonetics – Ô § 0.12).
German text
5.5.2. Ein® ®ritten si> Nordwind und Sonne, wer von ihnen bRden wohl der
Stärkere wäre, als Rn Wanderer, der in Rnen warmen Mantel gehüllt war, des Weges
daherkam. S¤ wurden Rnig, dass derjenige für den Stärkeren gelten sollte, der den
Wanderer zwi«en würde, sRnen Mantel abzunehmen.
Der Nordwind bl¤s mit aller Ma>t, aber je mehr er bl¤s, de®o fe®er hüllte si>
der Wanderer in sRnen Mantel Rn. Endli> gab der Nordwind den Kamπ Qf. Nun
erwärmte d¤ Sonne d¤ Lu⁄ mit ihren frSndli>en Strahlen, und (on na> weni-
gen Augenbli$en zog der Wanderer sRnen Mantel Qs. Da mus®e der Nordwind zu-
geben, dass d¤ Sonne von ihnen bRden der Stärkere war.
Hat dir d¤ Ge(i>te gefallen? Wollen wir s¤ w¤derholen?
6.0. ˛e Spanish accents that will be dealt with are the neutral Iberian and the
(central-southern) American versions. ˛ere are really very few di‡erences in neu-
tral pronunciation; therefore a diaphonemic transcription is not necessary (where-
as for Portuguese and English it is); the American variant simply has two pho-
nemes less: /†, L/, which become /s, J/. ˛e former two symbols, however, are used
diaphonemically, as they distinguish the two accents.
Requiring informative symbols, to recall this phenomenon more clearly, one
could resort to /†, L/, or –if inclined to a more international type of transcription–
to /s, J/. On the other hand, again, to help foreigners, it is worth indicating the
phonemes /b, d, g/ with /b, ò, g/, in contexts where they are realized as (B, ƒ, Ÿ).
Placing /B, ∑, Ÿ/ (as seen in some text-books and dictionaries with transcriptions)
is no good usage at all: because they are not phonemes, and because in the inflec-
tion of lemmata things do not remain constant.
Although, even for Spanish, the number of speakers with çAmericanÇ accents is
decidedly more consistent than for the Iberian ones, we prefer to consider the lat-
ter çprimaryÇ, because it is closer to the written word; therefore it is more advis-
able, for teaching purposes too, also for more coherent spelling, which, on the oth-
er hand, is a safer guide to pronunciation.
Obviously, even the American accent has many internal variants, which we here
consider to be non-neutral, even if, they are often more widespread than neutral
pronunciation, which (as happens in every language) is decidedly minimal.
Generally, these variants are not stigmatized in the way easily localized pronun-
ciations are, nor do they have negative connotations, as each nation has its own
çnationalÇ accent as well as regional variants. A more in-depth study of Spanish
pronunciation will give the macro-koinés (at least seven for America), to be consid-
ered çneutralÇ, as well as more specific, regional variants.
On the other hand, the non-neutral characteristics of Spanish are shared by most
Hispanic American nations, and in Europe as well, so as to make the non-neutral
accents more alike, even if they remain recognizable. For this reason, regarding
Spanish, other accents will not be presented, other than the two neutral ones (as
done, instead for German, which has more distinct accents, above all in Switzer-
land and Austria).
However, recordings are being collected for a systematic description of the actu-
al Spanish accents. In the meantime, the phonosynthesis of eastern Andalusian can
be seen (in § 17.9 of NPT/HPh).
6. spanish 231
Vowels
6.1.1.1. Spanish has only five vowel phonemes realized as shown in the first vo-
cogram in û 6.1 (i÷ '™, ’e÷ a÷ 'ø, ’o÷ u) /i, e, a, o, u/. For e, o in the neutral Iberian-
-American accent, in stressed syllables, (™, ø) are normal, but, in unstressed sylla-
bles, we find (e, o).
We therefore have examples such as: ('ßi)i ('si)a /'si/ sì˚ ('kRi;ßiß)i (-sis)a /'kRisis/
cr¤¤˚ ('mil) /'mil/ mil˚ (†eR'n™R)i (s-)a /†eR'neR/ cerner˚ (pa'p™l) /pa'pel/ papel˚ ('tj™R-
r:a) /'tjeRr:a/ tierra˚ ('ß™iß)i ('s™is)a /'seis/ se¤˚ ('r:™ßto)i ('r:™s-)a /'r:esto/ rπto˚ ('t™;Xa)i
(-xa)a /'texa/ teja˚ (de'XaR)i (-x-)a /òe'xaR/ dejar˚ ('ßj™mpRe)i ('s-)a /'sjempRe/ siempre˚
(a't™nto) /a'tento/ atento˚ (uß't™‡)i (us-)a /us'teò/ ¨ted˚ ('p™†)i (-s)a /'pe†/ pez˚ ('d™fi-
ƒe)i (-z-)a /'òesòe/ dπde˚ ('k™;ßo)i (-so)a /'keso/ ¢πo˘
And: ('p™;Co) /'peco/ pe>o˚ (kom'pR™) /kom'pRe/ compré˚ ('anda) /'anda/ anda˚
(pa'ta;ta) /pa'tata/ patata˚ ('gøRr:a) /'goRr:a/ gorra˚ ('øi) /'oi/ hoy˚ ('r:ø;ßa)i (-sa)a /'r:o-
sa/ rosa˚ ('ø;Xa)i (-xa)a /'oxa/ hoja˚ (mo'XaR)i (-x-)a /mo'xaR/ mojar˚ (fa'BøR) /fa'boR/
favor˚ ('ßøl)i ('s-)a /'sol/ sol˚ (r:a'†øn)i (-s-)a /r:a'†on/ razón˚ ('gølpe) /'golpe/ golpe˚
('pø;Lo)i (-,o)a /'poLo/ poıo˚ (La'mø)i (,a-)a /La'mo/ ıamó˚ (de'kø;Ro) /òe'koRo/ deco-
ro˚ ('tu;Bo) /'tubo/ tuvo˚ (r:u'møR) /r:u'moR/ rumor˘
tion (for speakers — words; Ô the second vocogram of û 6.1); indeed, sometimes
(E°÷ O°) ((E, O)) can be found, especially in checked syllables, or in the /ei, oi÷ r:e, r:o÷
ex, ox/ sequences; or, (™°/e§) ((e)), (ø°/o§) ((o)), above all in unchecked syllables; and
also for /e/ in checked syllables in /m, n÷ d÷ †, s/. However, it is not necessary to
point these out, as their regular timbre, in a stressed position, (™, ø), is normal and
safe, used by real natives (despite the variations). In unstressed syllables, in the
same contexts, instead of ((E, O)), we obviously find (™, ø).
For the sake of curiosity, and only here, we give some adapted cases which are
pertinent to the aforementioned examples: ((†™R'nER)) /†eR'neR/ cerner˚ ((pa'pEl)) /pa-
'pel/ papel˚ (('tjERr:a)) /'tjeRr:a/ tierra˚ (('ßEiß))i (('sEis))a /'seis/ se¤˚ (('r:Eßto))i (('r:Es-))a /'r:es-
to/ rπto˚ (('tE;Xa))i (-xa)a /'texa/ teja˚ ((d™'XaR))i (-x-)a /òe'xaR/ dejar˚ (('ßjempRe))i ('s-)a
/'sjempRe/ siempre˚ ((a'tento)) /a'tento/ atento˚ ((uß'te[‡]))i (us-)a /us'teò/ ¨ted˚ (('pe†))i
(-s)a /'pe†/ pez˚ (('defiƒe))i (-z-)a /'òesòe/ dπde˚ (('ke;ßo))i (-so)a /'keso/ ¢πo˘
And: (('pe;Co)) /'peco/ pe>o˚ ((kOm'pRe)) /kom'pRe/ compré˚ (('gORr:a)) /'goRr:a/ gor-
ra˚ (('Oi)) /'oi/ hoy˚ (('r:O;ßa))i (-sa)a /'r:osa/ rosa˚ (('O;Xa))i (-xa)a /'oxa/ hoja˚ (mø'XaR)i
(-x-)a /mo'xaR/ mojar˚ ((fa'BOR)) /fa'boR/ favor˚ (('ßOl))i ('s-)a /'sol/ sol˚ ((r:a'†On))i (-s-)a
/r:a'†on/ razón˚ (('gOlpe)) /'golpe/ golpe˚ (('po;Lo))i (-,o)a /'poLo/ poıo˚ ((La'mo))i (,a-)a
/La'mo/ ıamó˚ ((de'ko;Ro)) /òe'koRo/ decoro˚ ((r:u'mOR)) /r:u'moR/ rumor˘
Also for /i, a, u/: ((ßen'tIR)) (ßen'tiR)i (s-)a /sen'tiR/ sentir, ((&aßIŸ'naR)) (&aßiŸ'naR)i (-s-)a
/asig'naR/ øignar, (('I;Xo)) ('i;Xo)i (-xo)a /'ixo/ hijo, (('mIRr:å)) ('miRr:a) /'miRr:a/ mirra,
(('r:I;ko)) ('r:i;ko) /'r:iko/ rico; (('ka;Nå)) ('ka;Na) /'kaNa/ caña, (('ma;Co)) ('ma;Co) /'ma-
co/ ma>o, (('ma;,o)) ('ma;,o) /'maJo/ mayo, (('ka;Le)) ('ka;Le)i (-,e)a /'kaLe/ caıe, (('bai-
le)) ('baile) /'baile/ baile; (('qlto)) ('alto) /'alto/ alto, (('mq;Xå)) ('ma;Xa)i (-xa)a /'maxa/
maja, ((kq'XOn)) (ka'Xøn) /ka'xon/ cajón, (('kqußå)) ('kaußa)i (-sa)a /'kausa/ ca¨a, ((q-
'un)) (a'un) /a'un/ aùn, ((bIl'Bqo)) (bil'Bao) /bil'bao/ Bilbao, ((q'O;Rå)) (a'ø;Ra) /a'oRa/ a-
hora; (('tuRko)) ('tuRko) /'tuRko/ turco, (('Xuntå)) ('Xunta)i ('x-)a /'xunta/ junta, (('lu;Xo))
('lu;Xo)i (-xo)a /'luxo/ lujo, ((tuR'r:On)) (tuR'r:øn) /tuR'r:on/ turrón, ((r:u'mOR)) (r:u'møR)
/r:u'moR/ rumor.
6.1.1.3. After /'i, 'u/, /eò|, oò|/ remain (e, o), even if at times we can find (’™, ’ø)
(in neutral pronunciation too): (fe'li;†eß)i (-ses)a /fe'li†es/ felicπ˚ ('r:i;ko) /'r:iko/ ri-
co˘ Even in an unstressed syllable, above all next to /x, r:/, we can have the (™, ø)
timbres, but, for neutral pronunciation, (e, o) are regular and adequate: (&Xela'ti;-
na)i (&x-)a /xela'tina/ gelatina˚ (koR'r:™o) /koR'r:eo/ correo.
We also indicate a pronunciation which can be heard, above all in words of fre-
quent usage, in not slow speech, the sequence /we/ can be pronounced (wÉ) (real-
izing a front-central vocoid); whereas in faster or less controlled speech, the se-
quence can be reduced to a single vocoid (¤ a central rounded (ê)), through pro-
gressive shifts: ('pw™;Blo, 'pwÉ;-÷ 'pw‘;-÷ 'pjê;-÷ 'pê;-) /'pweblo/ pueblo˚ ('nw™;Be, 'nwÉ;-÷
'nw‘;-÷ 'njê;-÷ 'nê;-) /'nwebe/ nueve˚ (&aßta'lw™;Ÿo, -wÉ;-÷ -w‘;-÷ -jê;-÷ -ê;-)i (&as-)a /asta-
'lwego/ høta luego˚ (kweß'tjøn, -wÙ-÷ -wÈ-÷ -j+-÷ -+-)i (-s-)a /kwes'tjon/ cuestión. How-
ever, it is not appropriate to actively adopt this particular kind of pronunciation
(which is shown in the central part of the second vocogram in û 6.1).
6. spanish 233
Diphthongs
Consonants
6.2.0. û 6.2 gives the consonant articulations, of the two neutral accents, which
are necessary for satisfactory Spanish pronunciation.
Instead, û 1.9-15, give orograms, grouped by manners of articulation of all the
contoids given in the chapters of this volume, even as secondary, occasional or re-
gional variants of the 12 languages dealt with.
û 6.2. Table of Spanish consonants.
velar round.
labiodental
postalveo-
prevelar
palatal
alveolar
bilabial
palatal
uvular
dental
velar
ö m (M) (n) n (~) N (˙) (,)i
F pb t d k g
Ô C {‚} (›)
ƒ f †i (∑)i /J/|(,) xa (Ÿ) (X)i
_ sa (z)a (ß)i (fi)i (ò)
ß (B) (ƒ) j|(ã) (F) w|(j)
ó R|r:
‹ (l) l (¬) Li
+ (N, M, M, º, ∫) Ô text
Nasals
6.2.1.1. ˛ere are three nasal phonemes, /m, n, N/, with various taxophones for
/n/ (m, M, n, ~, N, «, ˙, ,) (û 1.9.1-2; («) is semi-provelar without full contact; we
could add dental ((˙)), before /t, d÷ †/, for which, however, (n) is su‚cient; below,
we add five more taxophones, (N, M, º, M, ∫), for nasal heterorganic sequences):
('ma;no) /'mano/ mano˚ ('ni;No) /'niNo/ niño˚ (um'p™Rr:o) /um'peRr:o/ un perro˚ (im-
'bj™Rno) /im'bjeRno/ invierno˚ (iM'fj™l) /in'fjel/ infiel˚ (&konten'd™R) ((-˙te˙-)) /konten-
'deR/ contender˚ (kon'†™ñto) ((-˙'†-))i (-s-) ((-˙'s-))a /kon'†ebto/ concepto˚ (kon'ß™;Xo)i
(-s™;xo) ((-˙'s-))a /kon'sexo/ consejo˚ ('ønr:a) /'onr:a/ honra˚ ('a~Co) ((-nC-)) /'anco/ an-
>o˚ ('køN›uXe)i (-xe)a /'konJuxe/ cónyuge˚ (uN'›u˙ke) /un'Junke/ un yun¢e˚ (&koN-
Le'BaR)i (-N›-, -NJ-)a /konLe'baR/ conıevar˚ (u«'w™;Bo) /un'webo/ un huevo˚ ('ba˙ko)
/'banko/ banco˚ ('t™˙go) /'tengo/ tengo˚ ('fRa,Xa)i (-˙xa)a /'fRanxa/ franja˘
˛e articulation of /N/ is palatal, as in Italian, but short (not self-geminant, as in
neutral Italian; even if, at times, it can geminate a little (NN), after a stressed V˚
which is then short): ('ba;No÷ 'baNNo) /'baNo/ baño (Ô Italian ('baN:No) /'baNNo/ ba-
gno]˘
As for /nòw/, even with the prefixes cons-, ins-, trans-, the most normal and suit-
able articulation is with («): (&ko«ßtRu◊'†jøn)i (-«stRu◊'sj-)a /konstRug'†jon/ construc-
ción˚ (i«ß'tante)i (-s-)a /ins'tante/ instante˚ (&tRa«ßfoR'maR)i (-s-)a /tRansfoR'maR/ trans-
formar÷ obviously forms such as (&†iRku«ß'tan†ja)i (&siRku«s'tansja)a /†iRkuns'tan-
†ja/ circunstancia are included in this case.
For the first elements of /mn, nm/ sequences, we find some further taxophones,
6. spanish 235
Stops
6.2.2.2. In word-final position, (‡) /ò/ is weak (¤ articulated with less tension,
((d)), as well as devoiced, ((D)), unless it is followed by voiced phones): ('ß™‡)i ('s-)a
/'seò/ sed˚ (a'Bla‡) /a'blaò/ hablad÷ often it drops in: (uß't™[‡])i (us-)a /us'te[ò]/ ¨ted˚
(ma'ƒRi[‡]) /ma'òRi[ò]/ Madrid, and in nouns with /éòò/ (not monosyllabic nouns):
(beR'ƒa[‡]) /beR'òaò/ verdad˚ (biR'tu[‡]) /biR'tuò/ virtud˘
Furthermore, in the masculine ending -ado(s)˚ the articulation is just as attenuat-
ed (up to (`), çzeroÇ, in familiar pronunciation, above all the Iberian one, but not
systematically; generally, in American pronunciation, the drop is considered to be
non-neutral): (Le'Ÿa[;ƒ]o)i (,e'Ÿa;ƒo)a /Le'gaòo/ ıegado˚ (ßol'da[;ƒ]oß)i (sol'da;ƒos)a
/sol'daòos/ soldados˘
As seen, before a voiceless C (or before a possible pause), /b, ò, g/ are devoiced:
(oñ'taR, &a‡Xe'ti;Bo, a◊'tøR) (from the previous section). Due to an excessive influ-
ence of writing, above all in the American accent, some articulate /b, ò, g/ as (p,
b÷ t, d÷ k, g) (with voicing in relation to the spelling), before heterosyllabic C
(which, instead, represent a completely normal neutralization): (oñ'taR, op-),
(&ßuB-le'BaR, -b-)i (&s-)a, (&aƒmi'RaR, &ad-), (aƒ'møßfeRa, at-)i (-s-)a, (&a‡Xe'ti;Bo, &ad-)i
(-xe-)a÷ ('diŸno, 'dig-), (a◊'tøR, ak-).
Again, for the same reason, with a further (and more serious) removal from the
real phonic structure, due to improper spelling influence (or because of regional
accents, ™ from Valencia, (v), and from Paraguay, (V)), especially in American pro-
nunciation, some introduce the ç/v/ phonemeÇ in Spanish, which has not existed
for centuries: (&embi'aR÷ ≠&eMvi'aR) /embi'aR/ enviar˘
It is evident that the use we make of /b, ò, g/ is not diaphonemic, as the two ac-
cents substantially coincide; it is moreover, interphonemic, because it is supposed
to show where their realization is not a stop, to help foreigners use it correctly,
without strained deductions (and, often, incorrect – and therefore, the source of
endless problems).
Stopstrictives
Constrictives
6.2.4.1. ˛ere are five constrictive phonemes; four are voiceless: (f) /f/, (†i, sa)
/†/, (ßi, sa) /s/ and (Xi, xa) /x/; whereas (,) /J/ is voiced (palatal) and, if the truth be
told, only semi-constrictives, as it is half-way between an approximant, (j), and the
real constrictive ((J); but rarer in various languages).
(f) /f/ does not pose any problems, even if, often, American and Iberian speak-
ers realize it as a bilabial (constrictive, (å), or approximant, (F)): ([&uM]fa'BøR÷
[&um]åa-÷ [&um]Fa-) /[un]fa'boR/ (un) favor˘
6.2.4.2. ˛e others need further explanations. Indeed, (†) /†/i is neutral only
in the Iberian accent, whereas in the American accent it becomes /s/: (†a'pa;to)i
(sa-)a /†a'pato/ zapato˚ ('†j™;lo)i ('sj-)a /'†jelo/ cielo˚ ('di;†e)i (-se)a /'òi†e/ dice˚ ('lu†)i
('lus)a /'lu†/ luz˘ Starting from an American, or international type of transcription,
it could be more appropriate to use the diaphoneme /s/: /sa'pato, 'sjelo, 'òise, 'lus/.
Before voiced C˚ the articulation becomes voiced: (Xu∑'ŸaR)i (xuz'ŸaR)a /xu†'gaR/
juzgar˚ ('lu∑ ƒo'Ra;ƒa)i ('luz)a /'lu† òo'Raòa/ luz dorada÷ naturally before sonants, in
the American accent, (s) is preferred (as for /s/; Ô the following section): ('dj™∑mo)i
(-smo)a /'òje†mo/ diezmo˘
6.2.4.3. For /s/ the place of articulation changes, from one accent to another,
as /s/ is (apico-)alveolar in Iberian Spanish, (ß)i, but, (lamino-)dental in American
Spanish, (s)a: (eß'ta;ƒoß)i (es'ta;ƒos)a /es'taòos/ πtados˚ ('ßøl)i ('s-)a /'sol/ sol˚ ('pi;ßo)i
(-so)a /'piso/ p¤o˘ (After a stressed V˚ as well as the normal –and more appropriate–
('é;ßé)i ('é;sé)a, one can also have ('éßßé)i ('éssé)a: ('pißßo)i ('pisso)a.) Before voiced
diphonic heterosyllabic C (/b, d, g/), /s/ becomes voiced, (fi)i (z)a: (lofi'BuRr:oß)i
(loz'BuRr:os)a /los'buRr:os/ los burros˚ ('d™fiƒe)i ('d™zƒe)a /'òesòe/ dπde˚ (difi'Ÿußto)i
(-z'Ÿus-)a /òis'gusto/ d¤g¨to˘ In the /st, sò/ sequences, in neutral Iberian pronunci-
ation, (ß, fi) remain, as can be seen in the given examples; only the denti-alveolar
articulation (dental with a raised tip, ((s, z))) is possible, while it is necessary in the
(still Iberian) pronunciation of the sequence /s†/: (es'†™;na) /es'†ena/ πcena (which
in American pronunciation is (e's™;na) /e'sena/, from //es'sena//).
However, before heterosyllabic non-diphonic C (/m, n, N÷ J÷ w÷ l, L/; for /s/ + /r:/,
see the end of this section), the behavior pattern is di‡erent, even if complementa-
ry: in the Iberian accent voicing prevails, (fi, Ü, ß); whereas in the American accent
voicelessness prevails, (s, Ω, z) (even if we transcribe only the first realization for
each accent): ('mifimo)i (-s-)a /'mismo/ m¤mo˚ ('ifila)i (-s-)a /'isla/ ¤la˚ (lofi'w™;ßoß)i
(-s'w™;sos, -'sw-)a /los'wesos/ los huπos˚ ('døò ',™Rr:oß)i ('døë)a /'òos 'JeRr:os/ dos hierros˘
˛e last example shows the normal articulation of the postalveo-palatal kind, (ò)i
(ë)a, before /J/, as also before /L, N/: (laò'Lu;Bjaß)i (laë',u;Bjas)a /las'Lubjas/ lø ıuviø˚
(loò'Nø;Noß)i (loë-, -os)a /los'NoNos/ los ñoños.
If the voiced C that follow are tautosyllabic (¤ when they are part of the same
syllable), /s/ remains voiceless, as even before V˚ /sòé/ (being initial in the phono-
-syllable): ('ßj™ßta)i ('sj™s-)a /'sjesta/ siπta˚ ('ßw™;lo)i ('s-)a /'swelo/ suelo˚ (&loßa'mi;Ÿoß)i
(-sa'mi;Ÿos)a /losa'migos/ los amigos˚ (&mißeR'ma;noß)i (-seR'ma;nos)a /miseR'manos/
238 a handbook of pronunciation
6.2.4.4. ˛e criterion adopted here, only considers two kinds of neutral pro-
nunciation; however, while describing the pronunciations of each single nation,
we will inevitably broaden our criteria so as to adequately include the characteris-
tics of every single country (though, with all the non-neutral variants, which
obviously each area presents). ˛erefore, let us mention a non-neutral variant, for
(American, Canary, and Andalusian Spanish) /s˘/, ¤ a very weak laryngeal approxi-
mant ((h), voiceless; and (H), voiced before voiced phones), which is generally
found together with what is traditionally (but incorrectly), defined as çaspirate(d)Ç
/s/ (which, from a strictly phonetic point of view would mean (sh), as (th)…), be-
fore a pause, or a C˚ or even before a V˘ It is realized as if it were ç/h/Ç. For exam-
ple: (lah'kw™htah) /las'kwestas/ lø cuπtø˚ ('d™Hƒe) /'òesòe/ dπde˚ ('ihlah, 'iH-) /'is-
las/ ¤lø˚ (&loha'mi;Ÿoh) /losa'migos/ los amigos (in neutral pronunciation: (laß'kw™ß-
taß)i (las'kw™stas)a, ('d™fiƒe)i (-z-)a, ('ifilaß)i ('islas)a, (&loßa'mi;Ÿoß)i (-sa'mi;Ÿos)a]˘
Such pronunciation can even get to transform (h, H) into (`), çzeroÇ.
les/ trπ bailπ˚ (laz'Bø;tas÷ laH'Bø;tah÷ la´'Bø-÷ la'åø-÷ la'Fø-÷ -a∆) /las'botas/ lø botø˚
(loz'Ÿa;,os÷ loH'Ÿa;,oh÷ lo≠'Ÿa-÷ lo'xa-÷ lo'∆a-÷ -o∆) /los'gaJos/ los gaıos˚ (tuz'Ÿa;tos÷
tuH'Ÿa;toh÷ tu±'Ÿa-÷ tu'xa-÷ tu'∆a-÷ -o∆) /tus'gatos/ t¨ gatos ((=) and derivatives have
an intermediate type of phonation between (h) and (H)). Neutral pronunciation
is: (mifi'ƒ™;ƒoß, &tRefi'Baileß, lafi'Bø;taß, lofi'Ÿa;,oß, tufi'Ÿa;toß)i (-z-, -s)a˘
Otherwise, before son(or)ants (/m, n÷ l÷ r:/), /s/ can change into (Ò0, =0): ('miz-
mo, 'miΩ-, 'mis-÷ 'miH-, 'mi+-÷ 'mi≈-, 'mi)-) /'mismo/ m¤mo˚ ('azno, 'aΩ-, 'as-÷ 'aH-,
'a´-÷ 'an-÷ 'a£-) /'asno/ øno˚ ('izla÷ 'iΩ-, 'is-÷ 'iH-÷ 'i+-÷ 'iú-÷ 'ia-) /'isla/ ¤la˚ (&iÍr:a'™l, &i-5) /is-
r:a'el/ Israel˘ Neutral pronunciation: ('mifimo, 'afino, 'ifila)i (-s-)a (&iÍr:a'™l)˘
6.2.4.7. For /x/ too, the place of articulation changes from one accent to anoth-
er (even within the neutral accent), as /x/ is, respectively, uvular, (X)i, and velar,
(x)a: (Xa'møn)i (xa-)a /xa'mon/ jamón˚ (Xe'miR)i (xe-)a /xe'miR/ gemir˚ ('di;Xe)i (-xe)a
/'òixe/ dije÷ when word-final it is weaker, as it becomes an approximant of the same
place of articulation, or even laryngeal: ('bø˜, -h)i ('bø∆, -h)a /'box/ boj÷ it is cur-
rently lost in: (r:e'lø[˜], -ø[h])i (r:e'lø[∆], -ø[h])a /r:e'lox/ reloj˘
In the two accents, there can be some variants for /x/, which are included in neu-
tral pronunciation: respectively, a more vigorous articulation ((º)i, voiceless uvu-
lar constrictive trill), or less vigorous, ((∆)a, voiceless velar approximant), which
we refer to here only. On the other hand, its transformation into (h) (laryngeal),
which is very common in America and Andalusia, cannot be considered neutral,
according to the criteria adopted here.
240 a handbook of pronunciation
Approximants
6.2.5.1. ˛e Spanish approximants are /j, w/: ('†j™;lo)i ('sj™-)a /'†jelo/ cielo˚ ('w™;-
Bo) /'webo/ huevo˚ (&awe'kaR) /awe'kaR/ ahuecar÷ devoiced realizations after voice-
less C are dialectal (or foreign): ('tj™;ne÷ ≠'tª-) /'tjene/ tiene˚ ('kwa;tRo÷ ≠'k‹-) /'kwatRo/
cuatro˘ Generally, /j/ only occurs after a tautosyllabic C˚ as in the previous exam-
ples, and in ('r:j™;Ÿo) /'r:jego/ riego. Instead, at the beginning of a syllable, we have
/J/ exclusively (§ 6.2.4.6, 6.2.4.1, 6.2.3), except in certain areas of America, such as
Argentina, where it is found for hiV-: ('›™;lo, ',-)i ('›-, ',-÷ 'j-)a /'Jelo/ hielo˚ ('›™RBa,
',-)i ('›-, ',-÷ 'j-)a /'JeRba/ hierba˚ against ('›™RBa, ',-)i ('›-, ',-)a /'JeRba/ yerba (independ-
ently of the –non-neutral– reduction of /J/ to (j), in certain areas).
For /é[ò]bwé, é[ò]gwé/, in familiar pronunciation, there can be a simplification,
through a velarized bilabial approximant, (ñ), for the first case; or a constrictive,
()) (or (Ÿ), Ô § 9.14 of NPT/HPh), or a semi-constrictive, (m), up to the approxi-
mant (w) (both velar rounded): (a'Bw™;lo, a'm™-, a'ñ™-, a'w™-) /a'bwelo/ abuelo˚ (la-
'Bw™lta, la'm™-, la'ñ™-, la'w™-) /la'bwelta/ la vuelta˚ ('a;Ÿwa, 'a;)a, 'a;ma, 'a;wa) /'agwa/
a%a˚ (la&Ÿwape'tø;na, la&)a-, la&ma-, la&wa-) /lagwape'tona/ la %apetona˘
Instead, for /òw, éw/, in familiar pronunciation, a more vigorous realization is
more frequent (compared to the phonemic transcription): (')™;Bo, 'm™-, 'ñ™-) /'we-
bo/, (&a)e'kaR, &ame-, &añe-) /awe'kaR/. Even in the /nòw/ combination (Ô § 6.2.1.1),
we can have this more vigorous articulation: (u«'w™;Bo, u«')™-, u«'m™-, u«'ñ™-) /un-
'webo/ un huevo˘ In all these cases, we can hear even syntagmatically divided reali-
zations, (Ÿw, Bw) (and even, (˙gw, mbw), for the last case), however it is doubt-
lessly better to avoid them.
6.2.5.2. As well as (j, w) and (B, ƒ) (/j, w/, /b, ò/), there are three more approxi-
mant (taxo)phones, which, in –normal, not at all slovenly– spontaneous conversa-
tion, are the realization of /e, o, a/, in the sequences /0eé, 0oé, 0aé/; therefore we
have, (semi-palatal) /e/ (ã), (semi-velar rounded) /o/ (j) and (semi-prevelar) /a/ (F):
('pãøR) /pe'oR/ peor˚ ('tãa;tRo) /te'atRo/ teatro˚ ('pj™;ta) /po'eta/ poeta˚ (&kjaŸu'laR)
/koagu'laR/ coa%lar˚ (u'nFø;Xa)i (-xa)a /una'oxa/ una hoja˚ ('lFø;tRa pa'Ra;ƒa) /la'otRa
pa'Raòa/ la otra parada˘ In slower, or more solemn speech, we doubtlessly have
(pe'øR, te'a;tRo, po'™;ta, &koaŸu'laR, &una'ø;Xa, la'ø;tRa pa'Ra;ƒa)i (-xa)a, as the phone-
mic transcription indicates.
Furthermore, there are pronunciations which are currently considered to be un-
educated: ('pjøR, 'tja;tRo, 'pw™;ta, &kwaŸu'laR); there is a di‡erence, and it is more
than enough to distinguish the three di‡erent realizations, even if the di‡erence
between (ã, j) and (j, w) might seem negligible (not only for foreigners, but also
for natives who write articles and books, limiting themselves to only two extreme
possibilities, also because of the lack of adequate symbols).
6. spanish 241
Trills
6.2.6.1. Spanish has an alveolar trill /r:/ (r:), with three rapid tappings of the tip
of the tongue against the alveolar ridge (so, (r;) would be enough, but experience
prefers (r:)), and a tap /R/ (R), with a single rapid alveolar contact: ('r:a;Ro) /'r:aRo/ ra-
ro˘ In interphonemic transcriptions (concerning several languages), it is important
to mark the chron(em)e, because otherwise its nature could be hidden, and Span-
ish /r:/ might seem to be a simple trill (with two tappings, as Italian /r/ in stressed
syllables: ('ra:Ro)it /'raro/ raro]˘
˛e two Spanish types ((R, r:)) join together too, giving, for examples: ('tj™Rr:a)
/'tjeRr:a/ tierra (Ô Italian ('tEr:Ra) /'tErra/ terra]˚ (laR'r:a;ƒjo) /laR'r:aòjo/ la radio (Ô Ital-
ian (la'ra:djo)it /la'radjo/ la radio]˘ Furthermore, in Spanish, the two types are dis-
tinctive, between V (even if with length di‡erences, for vowels too): ('ka;Ro) /'kaRo/
caro˚ ('kaRr:o) /'kaRr:o/ carro÷ ('p™;Ro) /'peRo/ pero˚ ('p™Rr:o) /'peRr:o/ perro÷ (&en†e'RaR)i
(-s-)a /en†e'RaR/ encerar˚ (&en†eR'r:aR)i (-s-)a /en†eR'r:aR/ encerrar˘
In our phone(ma)tic analysis, (Rr:) /Rr:/ are heterosyllabic sequences, constitut-
ed by (syllable-final) (R˘) /R˘/ or (word-final) (Rò) /Rò/ + (r:) /r:/, with one + three (or
four, to give more emphasis) tappings. ˛ey are not simple segments such as ç/R,
r/Ç (or çr˚ rÇ in the Hispanic tradition).
Laterals
6.2.7. ˛ere are two lateral phonemes in neutral Iberian Spanish, (l, L)i /l, L/;
whilst in neutral American Spanish, the second merges with /J/ – (l, ,)a /l, L/; more-
over, /l/ has taxophones which are appropriately used, (¬, L) (it is not necessary to
242 a handbook of pronunciation
mark explicitly ((l)), as long as the articulation is dental): ('la;ƒo) /'laòo/ lado˚ (ka-
'løR) /ka'loR/ calor˚ (kol'ŸaR) /kol'gaR/ colgar˚ ('alto) {((-lto))} /'alto/ alto˚ (el'†i;ne)i {((el-
'†-))} (el'si;ne)a {((el's-))} /el'†ine/ el cine˚ (ko¬'Cøn) ((-L'C-)) /kol'con/ col>ón˚ (eL'Nø;-
No) /el'NoNo/ el ñoño˚ (eL'›u˙ke, eL'J-) /el'Junke/ el yunke˚ ('kla;ße)i (-se)a /'klase/
cløe˚ ('dø;Ble) /'òoble/ doble˚ (e'la;Ÿwa) /e'lagwa/ el a%a˚ (La'maR)i (,a-)a /La'maR/
ıamar˚ ('ba;Le)i (-,e)a /'baLe/ vaıe˚ (&koNLe'Bando)i (-N›-, NJ-)a /konLe'bando/ con-
ıevando˘ Phonetically, before palatals, even the American accent has (L), although
it does not have the phoneme /L/. In the same context, /J/ is articulated as a real
constrictive, (J), or as a stopstrictive, (›).
Before a pause, even for /l/ (as for /n, R/ and for /ò/), the vibrations of the vocal
folds stop at the same time as the tip of the tongue is removed from the alveolar
ridge (therefore, the following are not at all acceptable (-lÈ, -l…, -lO)): ('ßøl)i ('s-)a
/'sol/ sol˚ (lau'R™l) /lau'Rel/ laurel˘
Structures
Taxophonics
6.3.1.1. ©th regard to consonant assimilations, they have been dealt with in
the sections of the respective parts. ˛e combination of vowels within words, and
between words in sentences, will be dealt with above all here.
Within a word, two same vowels tend to be reduced to only one, except in for-
mal, slow or controlled speech: (&alBa'a;ka, al'Ba;ka) /alba'aka/ albahaca˚ (a†a'aR,
-'†aR)i (-s-)a /a†a'aR/ azahar˚ (&akRee'ƒø;Reß, &akRe'-) /akRee'òoRes/ acreedorπ˚ (bee'm™n-
†ja, be'-)i (-sja)a /bee'men†ja/ vehemencia˚ (nii'lißta, ni'-) /nii'lista/ nihil¤ta˚ (&alko-
'øl, al'køl) /alko'ol/ alcohol˚ (&koope'RaR, ko&o-, &kope-) /koope'RaR/ cooperar˚ (&†oolo-
'Xia, &†olo-)i (&so-, -'xia)a /†oolo'xia/ zoologìa˘
However, certain words, that otherwise would not be easily recognized, are not
reduced:(kRe'™n†ja)i (-sja)a /kRe'en†ja/ creencia˚ (mo'ø;ßo)i (-so)a /mo'oso/ mohoso˚
(lo'øR) /lo'oR/ loor÷ other words can be reduced except in an intoneme: (le'™R, 'l™R)
/le'eR/ leer˚ ('l™e, 'l™) /'lee/ lee˚ (kRe'™, 'kR™) /kRe'e/ creé˚ (paße'™;moß, pa'ßã™;-, pa'ß™;-) /pa-
se'emos/ pøeemos˘
©th di‡erent V˚ we have: (ko&aŸu'la;ƒo, &kja-) /koagu'laòo/ coa%lado˚ (aR'X™n-
teo, -tão)i (-x-)a /aR'xenteo/ argénteo˚ (leo'paRƒo, lão-) /leo'paRòo/ leopardo˘
ruido÷ (al'dãa;no÷ &alde'a-) /alde'ano/ aldeano˚ (lãal'ta‡÷ leal-) /leal'taò/ lealtad˚ ('li;nãa÷
-nea) /'linea/ lìnea˚ ('tja;La÷ to'a-)i (-,a)a /to'aLa/ toaıa˚ (&kjaŸu'laR÷ ko&a-÷ &koa-) /koa-
gu'laR/ coa%lar˚ ('r:ãal÷ r:e'al) /r:e'al/ real˘
It must be remembered that there is a di‡erence between normal pronunciation
(not slow): ('tãa;tRo, &empão'RaR, 'pj™;ta, &kjaŸu'laR) {/te'atRo, empeo'RaR, po'eta, koa-
gu'laR/ teatro˚ empeorar˚ poeta˚ coa%lar] and çuneducatedÇ pronunciation: ('tja;tRo,
&empjo'RaR, 'pw™;ta, &kwaŸu'laR), in addition to slow pronunciation: (te'a;tRo, &empeo-
'RaR, po'™;ta, &koaŸu'laR, ko&a-).
Except in particular cases, for rhythmic reasons (as will soon be seen), the pro-
nunciation with çunauthorizedÇ diphthongs is decidedly uneducated (although it
is very widespread, especially in Latin America): (pa'iß, ≠'paiß)i (-s)a /pa'is/ paìs˚ (ma-
'i†, ≠'mai†)i (-s)a /ma'i†/ maìz˚ (ba'ul, ≠'baul) /ba'ul/ baùl˘ ˛e transformation from
/ae, ao/ to /ai, au/ is equally uneducated: ('kaen÷ ≠'kain) /'kaen/ caen˚ (tRae'Ran÷
≠tRai-) /tRae'Ran/ traerán˚ (bil'Bao÷ ≠-au÷ ≠-aU) /bil'bao/ Bilbao˚ (&baka'lao÷ ≠-au÷ ≠-aU)
/baka'lao/ bacalao÷ in cases such as (&e[◊]ßtRa&oRƒi'na;Rjo, e[◊]ß&tRaoR-)i (-s-)a /e[k]stRaoR-
òi'naRjo/ extraordinario˚ we also have the possibility of: (&e[◊]ßtRFoR-, e[◊]ß&tRFoR-)i
(-s-)a.
6.3.1.3. It is not easy to assess the vocalic (/i, u/) or consonantal value (/j, w/)
of i˚ u in CiV˚ CuV sequences; /i, u/ are definitely more probable near a stress and
at the end of a word: (r:eß'fRio)i (-s-)a /r:es'fRio/ rπfrìo˚ (&r:eßfRi'aR)i (-s-)a /r:esfRi'aR/ rπ-
friar˚ but: (&eMfRja'm™nto) /enfRja'mento/ enfriamento (and also: (r:eß'fRja;ƒo)i (-s-)a
/r:es'fRjaòo/ rπfriado]˚ (flu◊'tuo) /fluk'tuo/ fluctùo˚ (&flu◊tu'aR) /fluktu'aR/ fluctuar
(and also: (&flu◊tu'ø;ßo)i (-so)a /fluktu'oso/ fluctuoso as well as suntuoso˚ virtuoso˚
which have only (-'twø;ßo)i (-so)a /-'twoso/) but: (&flu◊twa'†jøn)i (-'sj-)a /fluktwa-
'†jon/ fluctuación˘
˛e infinitives with /i, u/ (which are prevalently short, and their derivatives,
even when there is no longer the simple, original form) are: (re-, mal-)criar˚ fiar˚
rπfriar˚ enfriar˚ liar˚ (ex-)piar˚ (π-, ¤-)triar˚ %iar˚ (dπ)viar˚ enviar˚ reenviar÷ puar˚
ruar˚ fluctuar÷ concluir˚ excluir˚ incluir˚ ocluir˚ recluir˚ (a-, in-, re-)fluir˚ diluir˚
(re)huir˚ (π)muir˚ (re)construir˚ instruir˚ πtatuir˚ (re)constituir˚ dπtituir˚ instituir˚
rπtituir˚ s¨tituir˚ intuir˘ For ui˚ the pronunciation with /'wi/ is frequent, as well.
Other cases of /i, u/ appear for short or compound forms: (&aBi'øn) /abi'on/ a-
vión˚ (bi'™;njo) /bi'enjo/ bienio˚ (biu'ni;Boko) /biu'niboko/ biunìvoco˚ (tRi'™;njo) /tRi-
'enjo/ trienio˚ (tRi'a˙gulo) /tRi'angulo/ trián%lo˚ (&dje†i'ø;Co, &dj™-)i (-si-)a /òje†i'o-
co/ diecio>o˚ (&beinti'ø;Co, &b™i-) /beinti'oco/ veintio>o˚ (gi'øn) /gi'on/ %ión˚ (i'a;-
to) /i'ato/ hiato˚ (pRi'øR) /pRi'oR/ prior˚ (kRi'an†a)i (-sa)a /kRi'an†a/ crianza˚ (fi'an†a)i
(-sa)a /fi'an†a/ fianza˚ (fi'a;ƒo) /fi'aòo/ fiado˚ (fi'ambRe) /fi'ambRe/ fiambre˚ (pi'a;no)
/pi'ano/ piano˚ but: (&tRja˙gu'laR) /tRjangu'laR/ trian%lar˚ (kRja'tu;Ra) /kRja'tuRa/ cria-
tura˚ (fja'ƒøR) /fja'òoR/ fiador˚ (koM'fjan†a)i (-sa)a /kon'fjan†a/ confianza˚ (fjam-
'bR™;Ra) /fjam'bReRa/ fiambrera˚ (pja'nißta)i (-sta)a /pja'nista/ pian¤ta˘ Let us note,
logically: ('Cja;paß)i (-s)a /'cjapas/ ≥iapø (with no dropping of /j/ after /c/).
We also have: (u'i;ƒa) /u'iòa/ huida (as huir]˚ (di'uRno, 'dju-) /òi'uRno, 'òju-/ diur-
no (but: (dju'tuRno) /òju'tuRno/ diuturno]˚ (Xu'i;†jo, 'Xwi-)i (xu'i;sjo, 'xwi-)a /xu'i†jo,
'xwi-/ juicio (but: (Xwi'†jø;ßo, 'Xw™†)i (xwi'sjø;so, 'xw™s)a /xwi'†joso, 'xwe†/ juicio-
so˚ juez]˚ (tRi'uMfo) /tRi'unfo/ triunfo˚ (tRiuM'fal, tRju-) /tRiun'fal, tRju-/ triunfal˘ For
244 a handbook of pronunciation
muy˚ we normally have ('mwi) /'mwi/, but also ('mui) /'mui/ is frequent (even if it
is often considered as dialectal).
Preferably, words like the following have /'wi/, but pronunciation with /u'i/ is
common, too: circuito˚ fortuito˚ gratuito˚ suizo˚ cøu¤ta˚ ruido˚ ruin˚ ruina˚ arruino˘
However, in fast speech, forms with /u'é, i'é/ easily become /'wé, 'jé/: (&deßtRu'iR,
deß'tRwiR)i (-s-)a /òestRu'iR/ dπtruir˚ (&ko«ßtRu'i;ƒo, ko«ß'tRwi;ƒo)i (-s-)a /konstRu'iòo/
construido˚ (fi'a;moß, 'fja;moß)i (-s)a /fi'amos/ fiamos˚ (&a◊tu'aR, a◊'twaR) /agtu'aR/ ac-
tuar˚ (&kaRi'a~Co, ka'Rja~Co) /kaRi'anco/ carian>o˚ (&beinti'u;no, bein'tju;no) /bein-
ti'uno/ veintiuno˘
6.3.1.4. Between words˚ clusters of V are more varied and more numerous, but
the criteria are the same. ©th identical V˚ reduction is greatly favored: (laaR'r:™;Ÿlo,
lFaR-, laR-) /laaR'r:eglo/ la arreglo˚ (lee[◊]ß'pli;ko, lãeß-, leß-)i (-s-)a /lee[k]s'pliko/ le expli-
co˚ (&loolBi'ƒ™, &ljo-, &lo-) /loolbi'òe/ lo olvidé˚ ('a˙gulo oñ'tußo) /'angulo ob'tuso/ án-
%lo obt¨o˚ (&impla'ka;Ble e˙'kø;no, &impla'ka; Ble˙-) /impla'kable en'kono/ implaca-
ble encono÷ (la'pR™;ßa 'a;†e u'na~CoR r:e'manßo, la'pR™; 'ßa;†e)i (-sa, -se, -so)a /la'pResa
'a†e u'nancoR r:e'manso/ la prπa hace un an>o remanso˚ (e'laiRe 'entRa ßil'Bando, e-
'lai 'Ren-)i (sil-)a /e'laiRe 'entRa sil'bando/ el aire entra silbando˚ (&mafi'B™˙2 'kwa;tRo 'ø;-
Xoß2 ke'ƒøß, &kwa'tRø;-)i (&maz-, -xos, -øs)a /mas'ben 'kwatRo 'oxos ke'òos/ más ven cua-
tro ojos ¢e dos˘
If the V are di‡erent, we have: (loaR'r:™;Ÿlo, ljaR-) /loaR'r:eglo/ lo arreglo˚ (la&oñ-
ßeR'B™, &lFo-)i (-s-)a /laobseR'be/ la observé˚ (loe[◊]ß'pli;ko, ljeß-)i (-s-)a /loe[k]s'pliko/ lo
explico˚ (lae[◊]ß'pli;ko, lFeß-)i (-s-)a /lae[k]s'pliko/ la explico˚ (loim'pli;ko, ljim-)
/loim'pliko/ lo implico˚ (laim'pli;ko, lFim-) /laim'pliko/ la implico˚ ('tø;ƒo a'k™;Lo,
'tø; ƒja-)i (-,o)a /'toòo a'keLo/ todo a¢eıo˚ ('tRißte o'ka;ßo, 'tRiß tão-)i (-is-, -so)a /'tRiste
o'kaso/ tr¤te ocøo˚ (&entReilu'ßjø;neß, &entRi-)i (-sjø;nes)a /entReilu'sjones/ entre il¨io-
nπ˚ ('pu;ƒo außen'taRße, 'pu; ƒjau-)i (-sen'taRse)a /'puòo ausen'taRse/ pudo a¨entarse˚
('™;Roe iNmoR'tal, '™;Rje, iº-, i∫-, '™; Rjãi-) /'eRoe inmoR'tal/ héroe inmortal˚ (pa'la;†jo
au'Ÿußto, pa'la; †jjau-)i (-sjo, -sto)a /pa'la†jo au'gusto/ palacio au%sto÷ (ße'Ÿun ße&a-
no'ta;ƒo, ßãano-)i (s-, s-)a /se'gun seano'taòo/ segùn se ha notado˚ (de'ambofi 'mø;ƒoß,
'dãam-)i (-s)a /òe'ambos 'moòos/ de ambos modos˘
Obviously, the function of the vocalic elements that come into contact also
count. In fast speech, a grammeme-final V seems rather redundant; thus, it can be
dropped, even where its morphological function may seem important (however,
there are all the other elements which compensate adequately).
˛us, one could quite easily achieve even çzeroÇ, even with di‡erent V\ (la&aBRi-
'R™, &lFa-, &laB-) /laabRi'Re/ la abriré˚ (lo'ø;ƒjo, 'ljø;-, 'lø;-) /lo'oòjo/ lo odio˚ (mi'i;Xo,
'mãi;-, 'mi;-)i (-xo)a /mi'ixo/ mi hijo˚ (laeß'pø;ßa, lFeß-, leß-)i (-s'pø;sa)a /laes'posa/ la
πposa˚ (la'ø;Ra, 'lFø;-, 'lø;-) /la'oRa/ la hora˚ (lai'Xi;ta, lFi-, li-)i (-x-)a /lai'xita/ la hijita˚
(lo'u;niko, 'lju;-, 'lu;-) /lo'uniko/ lo ùnico˚ (lou'nj™;Ron, lju-, lu-) /lou'njeRon/ lo unie-
ron˚ (me'i;Ba, 'mãi;-, 'mi;-) /me'iba/ me iba˚ (me&ima'Xi;no, &mei-, &mãi-, &mi-)i (-x-)a
/meima'xino/ me imagino˘
6.3.1.5. Even clusters of various V are frequent in these examples (taken from
Navarro Tomás, but retranscribed and completed): (&eßka'l™;Ra aR'r:i;Ba, -'l™; RaR'r:i-)i
6. spanish 245
(&eska-)a /eska'leRa aR'r:iba/ πcalera arriba˚ (a'ma;ƒa eß'pø;ßa, a'ma; ƒFeß-, ƒeß-)i (-s-)a
/a'maòa es'posa/ amada πposa˚ (o'f™Rta i,'Xußta, o'f™R tFi,-, -ti,-)i (-˙'x-)a /o'feRta in-
'xusta/ oferta inj¨ta˚ (pa'la;BRa o'ßa;ƒa, pa'la; BRFo-, BRo-)i (-s-)a /pa'labRa o'saòa/ pala-
bra osada˚ ('ka;ßa u'milde, 'ka; ßFu-, ßu-)i (-s-)a /'kasa u'milde/ cøa humilde˚ ('kj™;Re a-
'BlaR, 'kj™; Rãa-, -Ra-) /'kjeRe a'blaR/ ¢iere hablar˚ ('pw™;ƒe eßkRi'BiR, 'pw™; ƒãeß-, ƒeß-)i
(-s-)a /'pweòe eskRi'biR/ puede πcribir˘
Also: ('nømbRe i'lußtRe, 'nøm bRãi-, bRi-) /'nombRe i'lustRe/ nombre il¨tre˚ ('tj™;ne
oR'Ÿu;Lo, 'tj™; não-, no-)i (-,o)a /'tjene or'guLo/ tiene or%ıo˚ ('X™nte u'milde, 'X™n tãu-÷
tu-)i ('x-)a /'xente u'milde/ gente humilde˚ ('ka;ßi apa'Ÿa;ƒo, 'ka; ßja-)i (-si, sj-)a /'kasi a-
pa'gaòo/ cøi apagado˚ (mi&eßpe'Ran†a, &mieß-, &mjeß-)i (-es-, -sa)a /miespe'Ran†a/ mi
πperanza˚ ('ka;ßi impo'ßi;Ble, 'ka; ßim-)i (-si, si-)a /'kasi impo'sible/ cøi imposible˚
(mio&BliŸa'†jøn, mjo-)i (-'sj-)a /miobliga'†jon/ mi obligación˚ (&niuna'B™†, ni&u-,
&nju-)i (-™s)a /niuna'be†/ ni una vez˘
And also: ('gRi;to a'Ÿu;ƒo, 'gRi; tja-) /'gRito a'guòo/ grito a%do˚ ('pø;ko eß'fw™R†o,
'pø; kjeß-÷ keß-)i (es-, -so)a /'poko es'fweR†o/ poco πfuerzo˚ ('n™;ŸRo iM'fj™Rno, 'n™;
ŸRjiM-, ŸRiM-) /'negRo in'fjeRno/ negro infierno˚ ('kwaRto oß'ku;Ro, 'kwaR tjoß-, toß-)i
(-s-)a /'kwarto os'kuRo/ cuarto oscuro˚ (e˙'ga;No u'ma;no, e˙'ga; Nju-, Nu-) /en'gaNo
u'mano/ engaño humano˚ (&ßuamiß'ta‡, &ßwa-)i (&s-, &s-)a /suamis'taò/ su am¤tad˚ ('im-
petu &eßpan'tø;ßo, -&tu eß-, -pe &tweß-)i (-es-, -so)a /'impetu espan'toso/ ìmpetu πpanto-
so˚ ('tRi;Bu i˙'gRa;ta, 'tRi; Bwi˙-) /'tRibu in'gRata/ tribu ingrata˚ (&poRßuo'nøR, -ßwo-)i
(-s-)a /poRsuo'noR/ por su honor˚ (eß'pi;Ritu u'ma;no, -Ri tu-)i (-s-)a /es'piRitu u'mano/
πpìritu humano˚ ('L™;Ÿa aaƒo'RaR, 'L™;Ÿa aƒo-, 'L™; Ÿaƒo-)i (',™-)a /'Lega aaòo'RaR/ ıega
a adorar˚ ('i;Ba aen†en'd™R, 'i;Ba en-)i (-s-)a /'iba aen†en'deR/ iba a encender˘
Further examples still: (be˙'gan†a ai'Ra;ƒa, -an †Fai-, -an †ai-)i (-sa, s-)a /ben'gan-
†a ai'Raòa/ venganza airada˚ (eß'ta;Ba ao'Ÿa;ƒa, -eß'ta; BFao-, Bao-)i (es-)a /es'taba ao-
'gaòa/ πtaba ahogada˚ (r:o'ßa;ƒa au'Rø;Ra, r:o'ßa; ƒFau-, ƒau-)i (-s-)a /r:o'saòa au'RoRa/
rosada aurora˚ ('kulta eu'Rø;pa, 'kul tFeu-, teu-) /'kulta eu'Ropa/ culta Europa˚ (a'pR™n-
de aa'BlaR, a'pR™n dãaa-, dãa-) /a'pRende aa'blaR/ aprende a hablar˚ ('auRea eß'pa;ƒa, 'au-
Rãa, &au RãFeß'pa;ƒa)i (-s-)a /'auRea es'paòa/ áurea πpada˚ ('mw™Rte ai'Ra;ƒa, 'mw™R tãai-,
tai-) /'mweRte ai'Raòa/ muerte airada˚ (pRe'ßu;me aon'daR, pRe'ßu; mãaon-)i (-s-)a /pRe-
'sume aon'daR/ prπume ahondar˚ ('fR™nte au'Ÿußta, 'fR™n tãau-, tau-)i (-s-)a /'fRente au-
'gusta/ frente au%sta˚ (biR'Xi;neo e˙'kanto, -nãjo, -não)i (-x-)a /biR'xineo en'kanto/
virgìneo encanto˘
'taRlo, -Bo a't-, 'bw™l Bja't) /'bwelbo aa'taRlo/ vuelvo a atarlo˚ ('b™˙go aempe'†aR,
'b™˙ gjaem-, gjFem-, gjem-)i (-saR)a /'bengo aempe'†aR/ vengo a empezar˚ ('kw™Rpo
ai'Rø;ßo, 'kw™R pjFi-, pjai-) /'kweRpo ai'Roso/ cuerpo airoso˚ (diß'pw™ßto ao&Beƒe'†™R,
diß'pw™ß tjFo-, tjao-)i (-s'pw™s-, -s™R)a /òis'pwesto aobeòe'†eR/ d¤puπto a obedecer˚
('faußto auß'pi;†jo, 'fauß tjFuß-, tjauß-)i (-st-, -s'pi;sjo)a /'fausto aus'pi†jo/ fa¨to a¨-
picio˘
And also: (an'ti;Ÿwa alti'B™†, an'ti; Ÿwal-)i (-™s)a /an'tigwa alti'be†/ anti%a alti-
vez˚ ('l™˙gwa e[◊]ß'tRa;Na, 'l™˙ gwFe-)i (-s-)a /'lengwa e[k]s'tRaNa/ len%a extraña˚ (eß-
'ta;twa iN'mø;Bil, iº-, i∫-, eß'ta; 'twFi-)i (es-)a /es'tatwa in'mobil/ πtatua inmóvil˚ ('a;-
Ÿwa olo'Rø;ßa, 'a; ŸwFo-) /'agwa olo'Rosa/ a%a olorosa˚ ('fa;twa u&mani'ƒa‡, 'fa; twFu-)
/'fatwa umani'òaò/ fatua humanidad˚ ('mu;two a'møR, 'mu; twja-) /'mutwo a'moR/
mutuo amor˚ ('aRƒwo em'p™;No, 'aR ƒwjem-) /'aRdwo em'peNo/ arduo empeño˚ (peR-
'p™;two im'p™;Rjo, peR'p™; twjim-) /peR'petwo im'peRjo/ perpetuo imperio˚ (kon'ti;-
nwo e'lø;Xjo, kon'ti; nwje-)i (-xjo)a /kon'tinwo e'loxjo/ continuo elogio˘
Lastly: ('fa;two oR'Ÿu;Lo, 'fa; twjoR-, twoR-)i (-,o)a /'fatwo oR'guLo/ fatuo or%ıo˚
('mønßtRwo u'ma;no, 'mønß tRwju-, tRwu-)i (-s-)a /'monstRwo u'mano/ monstruo
humano˚ ('r:™;Xja au&toRi'ƒa‡, 'r:™; XjFu-, Xjau-)i (-x-)a /'r:exja autoRi'òaò/ regia autori-
dad˚ (koR'r:jø aeßpe'RaRlo, Feß-, eß-)i (-s-)a /koR'r:jo aespe'RaRlo/ corrió a πperarlo˚ (pa-
'la;†jo au'Ÿußto, pa'la; †jjau-, †jjFu-)i (-sjo, -us-)a /pa'la†jo au'gusto/ palacio au%s-
to˚ (i'ni;kwo au'Ÿu;Rjo, i'ni; kwjau-, kwjFu-) /i'nikwo au'guRjo/ inicuo au%rio˚ (em-
'bi;ƒjo aeu'ß™;Bjo, em'bi; ƒjjaeu-, ƒjaeu-, ƒjFeu-, ƒjjeu-)i (-s-)a /em'biòjo aeu'sebjo/
envidio a E¨ebio˘
do and el ejido (&ele'Xi;ƒo)i (-x-)a /ele'xiòo/, hele>o and el he>o (e'l™;Co) /e'leco/, he-
leno and el heno (e'l™;no) /e'leno/, la sabπ and lø avπ (la'ßa;Beß)i (la'sa;Bes)a /la's-
abes/, ena%ø and en a%ø (e'na;Ÿwaß)i (-s)a /e'nagwas/, enojo and en ojo (e'nø;Xo)i
(-xo)a /e'noxo/˘
In emphatic or vigorous pronunciation, or in singing, after stressed V˚ the con-
tinuous non-diphonic C (voiced: /m, n, N÷ l, L/, and voiceless: /f, †, s, x/ – but not
/R, Rr:/, for which length is essential), instead of the normal structure, with ('é;0é),
often have ('é00é): ('u ;mo, "ummo) /'umo/ humo˚ ('a;na, "anna) /'ana/ Ana˚ ('ba;le,
"balle) /'bale/ vale˚ ('X™;fe, "X™ffe)i (x™-)a /'xefe/ jefe˚ ('di;†e, "di††e)i (-se, -sse)a /'òi†e/
dice˚ ('ka;ßa, "kaßßa)i (-sa, -ssa)a /'kasa/ cøa˚ ('i;Xo, "iXXo)i (-xo, -xxo)a /'ixo/ hijo˘
Stress
6.3.2.1. Spanish spelling is quite satisfactory regarding the indication of word
stress; however, some doubts remain concerning clusters of graphic VV˚ above all
with i˚ u÷ furthermore, there are possible oscillations, and di‡erences with respect
to Portuguese or Italian.
˛e position of stress is distinctive (and we can also find threefold phonemic
contrasts, the first of which are nouns, indicated graphically too), as in: ('a;nimo)
/'animo/ ánimo˚ (a'ni;mo) /a'nimo/ animo and (&ani'mø) /ani'mo/ animó; (kon'ti;-
nwo) /kon'tinwo/ continuo˚ (&konti'nuo) /konti'nuo/ continùo and (&konti'nwø)
/konti'nwo/ continuó; ('li;mite) /'limite/ lìmite˚ (li'mi;te) /li'mite/ limite˚ (&limi't™)
/limi'te/ limité÷ (de'pø;ßito)i (-s-)a /òe'posito/ depósito˚ (&depo'ßi;to)i (-s-)a /òepo'sito/
deposito˚ (de&poßi'tø)i (-s-)a /òeposi'to/ depositó˘
For stress oscillations, some examples follow: acné˚ acne÷ aeróstato˚ aerostato÷
ambrosìa˚ ambrosia÷ atmósfera˚ atmosfera÷ a¨triaco˚ a¨trìaco÷ bala¨tre˚ balaùstre÷
cartomancia˚ cartomancìa (but only farmacia]÷ cìclope˚ ciclope÷ conclave˚ cónclave÷
dinamo˚ dìnamo÷ elixir˚ elìxir÷ etìope˚ etiope (but only miope]÷ gladìolo˚ gladiolo÷
ìbero˚ ibero÷ m¤il˚ mìsil÷ olimpiada˚ olimpìada÷ orgìa˚ orgia÷ ósmos¤˚ osmos¤÷ pelì-
cano˚ pelicano÷ pensil˚ pénsil÷ pentagrama˚ pentágrama÷ perìodo˚ periodo÷ policro-
mo˚ polìcromo÷ polìgloto˚ poligloto÷ radar˚ rádar÷ reptil˚ réptil÷ reuma˚ reùma÷ ter-
mostato˚ termóstato÷ utopìa˚ utopia÷ varicπ˚ váricπ˘
6.3.2.2. Lexical compounds and adverbs ending in -mente˚ keep quite a strong
stress even on the first element (according to the scale: ('), (“), (&)): (“køRta'Bølßaß)i
(-sas)a /'koRta'bolsas/ cortabolsø (Ô (&koRta'ƒø;Ra) /koRta'òoRa/ cortadora), (“i˙ka'pj™)
/'inka'pje/ hincapié˚ (“entRe'ƒøß)i (-s)a /'entRe'òos/ entredós (Ô (&entRe'ƒøfi mu'Ca;Coß)i
(-os)a /entRe'òos mu'cacos/ entre dos mu>a>os), (r:e&al'm™nte) /r:e'al'mente/ real-
mente˚ (&beRƒa“ƒ™Ra'm™nte) /beRòa'òeRa'mente/ verdaderamente˚ (“fa†il'm™nte)i (-s-)a
/'fa†il'mente/ fácilmente˘ Also: (“tøƒa'Bia) /'toòa'bia/ todavìa˚ (a&ßi'mifimo)i (a&si-
'mis-)a /a'si'mismo/ øim¤mo (as can be seen in the last example, if a first /'/ is imme-
diately followed by another /'/, it becomes (&); on the other hand, however, if the
form were with only one /'/, ç/asi'mismo/Ç, we should have ç(&aßi'mifimo)i (&asi-
'mis-)aÇ).
To begin to acknowledge the di‡erent stress patterns, compared –for instance–
to Italian, we give some useful examples: (final stressed) acentùo˚ amplìo˚ varìo˚
(stressed on the last but one syllable) abdico˚ altero˚ animo˚ celebro˚ certifico˚ com-
puto˚ convoco˚ denomino˚ deposito˚ d¤puto˚ edifico˚ indico˚ integro˚ interrogo˚ lim-
ito˚ medito˚ modifico˚ modulo˚ penetro˚ significo /signi'fiko/, termino˚ venero˚ violo
/'bjolo/, v¤ito /bi'sito/, limpio /'limpjo/. Furthermore: (a'n™;mja) /a'nemja/ anemia˚
(bi'Ÿa;mja) /bi'gamja/ bigamia˚ (&kaRam'bø;la) /kaRam'bola/ carambola˚ (diß'pu;ta)i
6. spanish 249
6.3.2.3. Even in Spanish, in connected speech, there are normally words which
are destressed; the articles are among these: (e'lømbRe) /e'lombRe/ el hombre˚ (&elal-
'kalde) /elal'kalde/ el alcalde˚ (la'ka;ßa)i (-sa)a /la'kasa/ la cøa˚ (laß'Ci;kaß)i (-s'Ci;kas)a
/las'cikas/ lø >icø˚ (&una'mi;Ÿo) /una'migo/ un amigo˚ (&unaß'tR™inta peR'ßø;naß)i
(&unas-, -'sø;nas)a /unas'tReinta peR'sonas/ unø treinta personø˘
We then have forms (plurisyllabic too), such as salvo˚ excepto˚ mediante˚ duran-
te÷ and phrases such as rπpecto a˚ junto a˚ encima de˚ delante de\ (a'ma;laŸa) /a'ma-
laga/ a Málaga˚ (ko'n™;La)i (-,a)a /ko'neLa/ con eıa˚ (&empa'Riß)i (-s)a /empa'Ris/ en
Parìs˚ (&ßinßom'bR™;Ro)i (&sinsom-)a /sinsom'bReRo/ sin sombrero˚ (&ante'tø;ƒo) /ante'to-
òo/ ante todo˚ (&baXolo'ßaRBoleß)i (-xolo's-, -es)a /baxolo'saRboles/ bajo los árbolπ˚
(&kontRa&mißa'mi;Ÿoß)i (-sa'mi;Ÿos)a /kontRamisa'migos/ contra m¤ amigos˚ (&defiƒe-
'øi)i (-z-)a /òesòe'oi/ dπde hoy˚ (&entRelo'ßø;Xoß)i (-'sø;xos)a /entRelo'soxos/ entre los o-
jos˘
Furthermore: (&aßta'ki, -Fa-, -aa-)i (&as-)a /astaa'ki/ høta a¢ì˚ (&a†jael'pw™Rto,
-†jFel-)i (-sj-)a /a†jael'pweRto/ hacia el puerto˚ (&paRa&nweßtRofi'w™ßpeƒeß)i (-stRos'w™s-
peƒes)a /paRanwestRos'wespeòes/ para nuπtros huéspedπ˚ (&ßoBRel'ßø;BRe)i (&soBRel's-)a
/sobReel'sobRe/ sobre el sobre˚ (e◊&†eñto'ƒøß)i (-&señto'ƒøs)a /eg†ebto'òos, ek†epto-/
excepto dos˚ (&Xuntoala'pw™Rta, -tja-)i (&xu-)a /xuntoala'pweRta/ junto a la puerta˘
Also: (en&†ima&ƒela'ka;ßa)i (-si-, -sa)a /en†imaòela'kasa/ encima de la cøa˚ (de&lan-
teƒe'mi) /òelanteòe'mi/ delante de mì˘ Even in more or less long series: (&paRa&entRe-
no'ßø;tRoß, pa&RFen-)i (-'sø;tRos)a /paRaentReno'sotRos/ para entre nosotros˚ (&poRen&†i-
maƒe'tø;ƒo)i (-'si-)a /poRen†imaòe'toòo/ por encima de todo˚ (&poRƒe&lante&ƒelXaR-
'ƒin)i (-x-)a /poRòelanteòelxaR'òin/ por delante del jardìn˚ (&defiƒepo&RentRelo'ßaRBo-
leß)i (&dez-, -'saRBoles)a /òesòepoRentRelo'saRboles/ dπde por entre los árbolπ˘
Generally, conjunctions too are destressed (even in phrases, such as en cuanto
(¢e)˚ puπto ¢e˚ supuπto ¢e), except with ora˚ ya˚ bien (disjunctive), øì (consec-
utive), no obstante˚ con todo˚ fuera de (adversative), en efecto˚ por tanto˚ por consi-
%iente˚ øì ¢e (consecutive), apenø˚ aùn no˚ no bien˚ ya ¢e˚ luego ¢e˚ dπpués
¢e˚ en tanto ¢e (temporal), a no ser ¢e˚ dado ¢e˚ con tal ¢e (conditional), por
más ¢e˚ a pπar de ¢e˚ mal ¢e˚ ya ¢e (concessive), (&pweßtoke&noloßa'Bia)i (-sto-,
-sa-)a /pwestokenolosa'bia/ puπto ¢e no lo sabìa˘ At the beginning of elliptic ques-
tions y is stressed: (¿'i tuçpa;ƒRe21)i (-'pa;-)a /¿'i tu'paòRe?/ ¿Y tu padre?
ko so'noRo/ eco sonoro˚ (&aLe'Ÿa;ƒo)i (&a,-)a /aLe'gaòo/ ha ıegado and ('a;No 'ma;lo) /'a-
No 'malo/ año malo˚ (&eßto'ma;ƒo)i (&es-)a /esto'maòo/ π tomado and ('™ßto 'pi;ƒo)i
('™s-)a /'esto 'piòo/ πto pido˚ (ßo&naƒmi'Ra;ƒoß)i (s-, -s)a /sonaòmi'Raòos/ son admira-
dos and ('ßø naRmo'njø;ßo)i ('s-, -so)a /'so naRmo'njoso/ son armonioso˘
Object pronouns are destressed, la˚ le˚ lo˚ lø˚ lπ˚ los˚ me˚ nos˚ os˚ se˚ te\ (me'pa;Ro
ami'RaR, me'pa; Rja-) /me'paRo ami'RaR/ me paro a mirar˚ (ofi'Bi;mofi Be'niR)i (oz'Bi;-
moz)a /os'bimos be'niR/ os vimos venir˚ (ße&leßpeR'ƒjø)i (se&les-)a /selespeR'òjo/ se lπ
perdió÷ instead, subject pronouns and indirect pronouns are stressed:('›ø 'ß™)i ('s™)a
/'Jo 'se/ yo sé˚ ('tu 'pw™;ƒeß)i (-s)a /'tu 'pweòes/ tù puedπ˚ ('™l 'di;†e)i (-se)a /'el 'di†e/ él
dice˚ (poR'mi) /poR'mi/ por mì˚ (&paRa'ti) /paRa'ti/ para ti˘
Possessive adjectives are destressed: (&mißeR'ma;noß)i (-se-, -os)a /miseR'manos/ m¤
hermanos˚ (tu'ßø;Xoß)i (-'sø;xos)a /tu'soxos/ t¨ ojos˚ (ßu'ma;ƒRe)i (s-)a /su'maòRe/ su
madre˚ (&nweßtRo'tj™mpo)i (-s-)a /nwestRo'tjempo/ nuπtro tiempo˘
Generally demonstrative adjectives, are destressed (but, from a pragmatic point
of view, they can be considered potentially stressable): (&eßte'li;BRo, “™ß-)i (-s-)a /este-
'libRo/ πte libro˚ (&eße'p™Rr:o, “™-) /ese'peRr:o/ πe perro˚ (a&keLamu'X™R, a“k™-)i (-,amu-
'x-)a /akeLamu'xeR/ a¢eıa mujer˘
6.3.2.5. Relative pronouns [¢e˚ ¢ien{π}˚ cual{π}˚ cuyo{s}] are destressed (con-
trary to interrogative and exclamatory ones): (e'lømbRe ke'Bi;moß)i (-s)a /e'lombRe
ke'bimos/ el hombre ¢e vimos˚ (e˙&ku,o'ka;ßo)i (-so)a /enkuJo'kaso/ en cuyo cøo÷
cual˚ cualπ (with the article) and tal˚ talπ are stressed: ('t™˙go uneR'ma;no2 el'kwaL
'L™;Ÿa 'øi2 ke)i (-L ',-)a /'tengo u'neRmano, el'kwal 'Lega 'oi/ tengo un hermano, el cual
ıega hoy, ¢e…˚ (le'ƒi;Xo 'tal 'kø;ßa)i (-xo, -sa)a /le'òixo 'tal 'kosa/ le dijo tal cosa˘
Relative adverbs [como˚ cuando˚ cuanto˚ donde] are also destressed, contrary to
interrogative and exclamatory ones: (&kwandoe'laiRe ße'kalma, -dje-) /kwandoe'lai-
Re se'kalma/ cuando el aire se calma˚ (la'pla;†a &ƒondeß&tatu'ka;ßa)i (-sa, -s&t-, -sa)a /la-
'pla†a òondeestatu'kasa/ la plaza donde πtá tu cøa˘
‹ereas the indefinites [algo˚ algùn˚ al%no˚ al%ien˚ nadie˚ ningùn˚ nin%no˚ o-
tro] are stressed, cada generally is not: (&kaƒa'ƒia) /kaòa'òia/ cada dìa˘ Even tan is
not stressed, contrary to tanto˚ tanta˘
‹en used as conjunctions, the following adverbs are destressed: luego˚ mien-
trø˚ aùn (which becomes /'aun/), más˚ menos˚ cøi÷ the adverb medio is too (con-
trary to its corresponding adjective).
6.3.2.6. ˛e forms of çtratamientoÇ, don˚ doña˚ fray˚ sor˚ san˚ santo˚ santa˚ are al-
ways destressed before a name: (do˙'kaRloß)i (-s)a /òon'kaRlos/ don Carlos˚ (&doNaƒo-
'lø;Reß)i (-s)a /òoNaòo'loRes/ doña Dolorπ˚ (&ßantoƒo'mi˙go)i (&s-)a /santoòo'mingo/
santo Domingo÷ vocative forms are destressed too, señor˚ señora˚ señorita˚ padre˚ ma-
dre˚ hermano˚ hermana˚ tìo˚ tìa\ (ße&NoRmaR'ti;ne†)i (s-, -s)a /seNoRmaR'tine†/ señor
Martìnez˚ (ße&NoRama'Ria)i (s-)a /seNoRama'Ria/ señora Marìa˚ (&paƒRean'dR™ß, -ƒRãan-)i
(-s)a /paòRean'dRes/ padre Andrés˚ (tio'Xwan)i (-x-)a /tio'xwan/ tìo Juan÷ however,
when not used in forms of çtratamientoÇ, we have the regular: (&una'ßanta mu'X™R)i
(-s-, -x-)a /una'santa mu'xeR/ una santa mujer˚ ('bi;no elße'NøR maR'ti;ne†)i (-s-, -s)a /'bi-
no else'NoR maR'tine†/ vino el señor Martìnez˚ (aeß'kRi;to el'tio 'Xwan)i (aes-, 'x-)a /aes-
6. spanish 251
6.3.2.7. Certain common phrases are also destressed: (&boka'BaXo)i (-xo)a /boka-
'baxo/ boca abajo˚ (&pataßaR'r:i;Ba) /patasaR'r:iba/ patø arriba˘
In numbers formed with y˚ the first element is destressed: (&tReintai'†i˙ko)i (-'s-)a
/tReintai'†inko/ treinta y cinco˚ (kwa&Rentai'ßj™;te)i (-'s-)a /kwaRentai'sjete/ cuarenta y
siete÷ also cien(to) before mil is destressed: (†jeN'mil, -M'm-, -º'm-, -M'm-, -∫'m-)i
('s-)a /†jen'mil/ cien mil˚ (&oCo&†jentafi'mil pe'ß™;taß)i (-&sj-, -as-, -'s™;tas)a /oco†jen-
tas'mil pe'setas/ o>ocientø mil pπetø˘ Simple numbers are stressed, even if mono-
syllabic and near a stress: ('døfi 'Ÿa;toß)i (-z 'Ÿa;tos)a /'òos 'gatos/ dos gatos˚ ('tR™ß
ka'Ba;Loß)i (-s ka'Ba;,os)a /'tRes ka'baLos/ trπ cabaıos˚ ('un 'li;BRo) /'un 'libRo/ un libro
(compared to the article: (un'li;BRo) /un'libRo/ un libro).
6.3.2.8. Obviously in long words, as in sentences, some secondary stresses are
introduced, for rhythmic reasons: ('kømpRame&lo) /'kompRamelo/ cómpramelo˚ (a-
'Na;ƒaße&le)i (-s-)a /a'Naòasele/ añádøele˚ (&entRe'Ÿandoße&lo)i (-s-)a /entRe'gandoselo/
entregándoselo˘
Words such as aun¢e˚ con¢e˚ por¢e˚ sino˚ are normally /'aunke, a'unke÷ 'kon-
ke÷ 'poRke÷ 'sino/, however, in sentences, their composition can also bring to stress
patterns such as: (&au˙ke, a&u˙ke, au˙&ke, au˙'k™÷ &ko˙ke, ko˙&ke, ko˙'k™÷ &poRke,
poR&ke, poR'k™÷ &ßino, ßi&no, ßi'nø)i (s-)a, although, they are often considered incor-
rect, for spelling reasons.
Emphasis, particularly in imperatives with enclitic pronouns, can modify the
structure considerably (above all, for stresses), as in: ('da;melo, "da:melo, "da;me&lo,
"da;me'lø, 'da;me'lø, &dame'lø) /'òamelo/ ¡dámelo!
6.3.2.9. ˛e words (a'ø;Ra) /a'oRa/ ahora˚ (a'i) /a'i/ ahì˚ (a'un) /a'un/ aùn/aun˚
when they are not in an intoneme and are linked to what follows, have the fre-
quent variants ('aoRa, 'ai, 'aun): (a'ø;Ra 'Bj™;ne, 'aoRa) /a'oRa 'bjene/ ahora viene˚ (a'i
eß'ta, 'ai)i (-s-)a /a'i es'ta/ ahì πtá˚ (a'um mi 'pa;ƒRe, 'aum) /a'un mi'paòRe/ aun mi pa-
dre˚ (a'un no'ßale, 'au)i (-s-)a /a'un no'sale/ aùn no sale˘
In the case of /'jé[s]ò/, in an intoneme (more often when followed by a pause), it
is quite frequent to find, even in the neutral accent, a pronunciation which, often,
seems as if it were ç/'ié[s]ò/Ç, while, it is generally only /i'é[s]ò/: (paR'tjø, &paRti'ø)
/par'tjo/ partió˚ (a'ƒjøß, &aƒi'øß)i (-s)a /a'òjos/ adiós˚ (lim'pj™, &limpi'™) /lim'pje/
limpié˚ (em'pj™, &empi'™) /em'pje/ en pie÷ however, occasionally, we can doubtless-
ly have the structure ('i&é), even in neutral pronunciation: (paR'ti&ø, a'ƒi&øß, lim'pi&™,
em'pi&™)i (-s)a.
252 a handbook of pronunciation
6.3.2.10. In fast and familiar speech, common words and proper names in a pre-
intoneme with ('ia) /'ia/ ìa frequently become ('ja) /'ja/: (al'dia ßi'Ÿj™nte, al'dja, al-
&djaßi'-) /al'dia si'gjente/ al dìa si%iente˚ ('pø;kofi 'ƒiafi ƒeß'pw™ß, 'ƒjafi, 'pø;kofi&ƒjafi)i
(-z, -z, -s-, -s)a /'pokos 'òias òes'pwes/ pocos dìø dπpués˚ (ßu'tia ƒo'lø;Reß, ßu'tja, ßu&tja-
ƒo'-)i (s-, -s)a /su'tia òo'loRes/ su tìa Dolorπ˚ (&doNama'Ria maR'ti;ne†, &doNama'Rja, &do-
Nama&RjamaR'-)i (-s)a /doNama'Ria maR'tine†/ doña Marìa Martìnez˚ (gaR'†ia Ÿu'tj™R-
r:e†, gaR'†ja, gaR&†jaŸu'-)i (-s-, -s)a /gaR'†ia gu'tjeRr:e†/ Garcìa Gutiérrez˚ (&noßeR'Bia
paRa'na;ƒa, -'Bja, -&BjapaRa'-)i (-s-)a /noseR'bia paRa'naòa/ no servìa para nada˘
Other examples: (ßea&Bia'pw™ßto ƒe'pj™, ßea&Bja'-, ßãa-)i (s-)a /seabia'pwesto òe'pje/
se habìa puπto de pie˚ (&eßta'Ria˙ kan'ßaƒoß, -'Rja˙, -&Rja˙kan'-)i (&es-, 'sa;ƒos)a /esta'Rian
kan'saòos/ πtarìan cansados˚ (&nopo'ƒRiaN Le'Ÿa; Ra'tj™mpo, -'ƒRjaN, -&ƒRjaNLe'-)i (,-)a
/nopo'òRian Le'gaR a'tjempo/ no podrìan ıegar a tiempo˘
Intonation
6.3.3. û 6.3 shows the preintonemes and intonemes of the neutral Iberian and
American Spanish language. ˛erefore, let us simply look at the fundamental ex-
amples. It is important to make comparisons both between them and with those
of other languages:
/./: (me'Bøi ma'Na;na poR&lama'Na;na3 3)i (me'Bøi ma'Na;na poR&lamaÇNa;na3 3)a /me-
'boi ma'Nana poRlama'Nana./ Me voy mañana por la mañana.
Text
6.4.0. ˛e transcribed passage, †e North Wind and the Sun˚ is given in four
çnormalizedÇ versions. We start with the (neutral) Iberian pronunciation of (neu-
tral British) English – this is the first step of the phonetic method (the written text
is given in § 2.5.2.0). ˛e Spanish translation follows, in its neutral version.
At the end, as always, there is the version which gives the English pronuncia-
tion of Spanish, by neutral British speakers, fluent in Spanish (after prolonged con-
tact with native speakers, but with no help from the phonetic method), who have
adequately learned the relative prominences, but who substantially use segmental
and intonation elements which are typical of neutral British English (for reference
purposes, although, of course, a neutral accent is not so common). Obviously, the
same principle is valid for the foreign pronunciation of English, given first.
Speakers of American English could prepare their own version both of the Span-
ish pronunciation of English and of their pronunciation of Spanish, as an excel-
lent exercise, by listening to native speakers, best of all after recording them. Of
course, speakers of other languages could do the same thing. ˛e author would be
happy to receive their transcriptions and recordings, both in case of help –should
they need it– and to make their contribution known to others (possibly in our
website on canIPA Natural Phonetics – Ô § 0.12).
Spanish Text
6.4.2. El viento norte y el sol porfiaban sobre cuál de eıos era el más fuerte, cuan-
do acertó a pøar un viajero envuelto en an>a capa. Convinieron en ¢e ¢ien antπ
lograra obligar al viajero a ¢itarse la capa serìa considerado más poderoso.
El viento norte sopló con gran furia, pero cuanto más soplaba, más se arrebujaba
en su capa el viajero; por fin el viento norte abandonó la emprπa. Entoncπ briıó el
sol con ardor, e inmediatamente se dπpojó de su capa el viajero; por lo ¢e el viento
norte hubo de reconocer la superioridad del sol.
¿Te ha %stado el cuento? ¿Vamos a repetirlo?
7.0.3. As can be seen from phonemic transcriptions, our analysis does not con-
sider V nasalization distinctive, although it must be marked since it is clearly per-
ceptible (above all in Brazilian pronunciation) and typical of Portuguese (but, in
7. portuguese 257
Vowels
7.1.1.2. Here are some first examples (for the time being, let us ignore the dia-
phonemic undersigned dots): (‚i'fi;si∞)b /òe'fisiı/ di˙cil˚ ('e;]i)b /'eıe/ ele˚ ('E;]å)b /'E-
ıa/ ela˚ (ba'ta;tå)b /ba'tata/ batata˚ ('nO;vå)b /'nOva/ nova˚ ('no;vu)b /'novu/ novo˚
('ëu;vå)b /'Suva/ >uva˘ However, it is worthwhile reflecting upon (A): ('pAu) /'pau/
pau˚ (Au'das)b {(Au'ƒaS)l] /Au'òas/ audaz˚ ('mA∞)b {('mAı)l] /'maı/ mal˚ (A∞'tu;Rå)b
{(AÓ'tu;R‘)l] /Aı'tuRa/ altura˘
It is important to distinguish: seca ('se;kå)b /'seka/» çdry (f.)Ç and ('sE;kå)b /'sEka/”
çdries, drought, nuisanceÇ, cerco ('se˜ku)b /'seKku/» çsiegeÇ and ('sE˜ku)b /'sEKku/”
çI encloseÇ, lobo (']o;bu)b /'ıobu/» çwolfÇ and (']O;bu)b /'ıObu/” çlobeÇ, fosso ('fo;su)b
258 a handbook of pronunciation
/e/ (e), /e{˘}ö/ (™ö, ’eö) /o/ (o), /o{˘}ö/ (9ö, ’Úö)
/ei/ (eI), /eiö˘/ (™Iö, ’eIö) /o™/ (oU, o), /oö˘/ (9Uö, ’ÚUö)
/E/ (E, »™) /O/ (O, »ø)
/a/ (a, Au, Aı, ’å[s]ò) {/’a/ (å)} /a[i/u]{˘}ö/ (A[i/u]{˘}ö) {(≈[i/u]{˘}ö)}
7.1.1.3. Notice that nasalization slightly changes some timbres according to the
diaphonemic symbols used: in particular /aö/ (Aö) (for which, as already seen,
(≈ö) is possible as well), and /'eö, 'Èö, 'oö, 'ìö/ (™ö, 9ö) (when stressed, some-
times, (eö, Úö) may be heard too).
Besides, in word-final position, we have real diphthongs for what –too often and
too hurriedly– is transcribed simply as ç/e, Ú/Ç, ¤ /ein, oun/ ('™I«, ’eI«÷ '9U«, ’ÚU«).
Instead, in the same position, /anò/ opposes the diphthong /aunò/: ('O˜fA«) /'OKfan/
órfã˚ ('O˜fAu«) /'OKfaun/ órfão˘ Later on, we will see other diphthongs, too, both
oral and (phonetically) nasalized.
It has to be noted that in modern neutral Brazilian pronunciation, the nasaliza-
tion of vocoids is more and more evident (even in diphthongs). As already said,
nasalization occurs when a V is followed by a nasal consonant [N), in the same syl-
lable or not, ¤ both in checked and unchecked syllables; and, above all, both in
stressed and unstressed syllables: (&AmA'NA«)b /Ama'Nan/ amaˆã˚ ('bA;Nu)b /'baNu/
baˆo˚ ('fi;nu)b /'finu/ fino˘
‹ereas, in traditional neutral Brazilian pronunciation, nasalization occurs on-
ly in checked syllables (in N]˚ either stressed or not; but a çtroublesomeÇ occur-
rence of nasalization is possible (or not), in unchecked stressed syllables followed
by /N/ (and with /'aöé/ ('åöé) as well): (&ama'NA«)b;t /Ama'Nan/ amaˆã˚ ('bA;Nu,
'bå;Nu)b;t /'baNu/ baˆo˚ ('fi;nu)b;t /'finu/ fino˘
Lastly, in mediatic pronunciation, nasalization occurs, in addition to (stressed
or unstressed) checked syllables (in N), even in unchecked stressed syllables fol-
lowed by N÷ but, again, a çtroublesomeÇ occurrence of nasalization is possible even
in /’éNé/ (¤ an unchecked, unstressed syllable, followed by /N/): (&amA'NA«, &a-
ma-)b;m /Ama'Nan/ amaˆã˚ ('bA;Nu)b;m /'baNu/ baˆo˚ ('fi;nu)b;m /'finu/ fino˘
7. portuguese 259
Lusitanian peculiarities
7.1.2.1. Let us now consider û 7.2 to see the vocalic realizations of neutral Lu-
sitanian Portuguese. It is a good idea to make a close comparison of the two fig-
ures; as a matter of fact (neglecting the broken-line white markers, of the unneces-
sary Brazilian variants, (I, U, x÷ ≈): ('e;]i, -I÷ 'E;]å, -x÷ 'no;vu, -U÷ &AmA'NA, &≈m≈'N≈)b]˚ it
is clear that Lusitanian Portuguese has a higher number of both taxophones and
phonemes: at least two more (which, in an exclusively Lusitanian phonemic tran-
scription, would be ç/…, å/Ç –or, deviating from o‚cial symbols, they should rather
be ç/¢, å/Ç; but, more traditionally rendered as ç/È, å/Ç)– for (’¢, ’Û÷ '¢ò), ('å, ’‘)÷ in
addition to (Åö) /—Aö/ and (Aö˘) /'aö˘/, with occasional oppositions between them
(that, in a Lusitanian way, we ought to mark as ç/aö, åö/Ç).
Besides, (’™, ’a, ’ø) /È, A, ì/ are needed, above all, but not only, in some pre-stress
syllables (by diachronic fusion, which is no longer visible in present-day spelling:
/éé/ = /é/, deriving from a prior /é0é/ structure, with C dropping), in addition
to /’Èı˘, ’ìı˘, ’Aı˘; ’ÈKò, ’ìKò, ’AKò/, as we will see (and /’Au/, and /’Ai/ as well). Fur-
thermore, /’-Èe2÷ ’-ì™2÷ ’-e, ’-È/ are special cases, as in: (ab'd9;meI«)b (‘B'ƒO;m’I«,
±-™n)l /ab'òìmÈe2/ abdômenb˚ -ómenl˚ ('ipsi]ÚU«)b (-î]ÚU«, ±-øn)l /'ipsiıì™2/ ìpsilon˚
(&i˙k]u'zi;vi)b (&i~k9u-, -v[¢], ±-v™)l /inkıu'zive, -È/ incl¨ive˘
7.1.2.3. Here are some other remarkable cases of /A/, in Lusitanian pronuncia-
tion, by contraction: (a) /A/ à [èa a¶]˚ (a'ke;]¢) /A'keıe/ à¢ele˚ (ka'vÉIR‘) /kA'veiRa/ ca-
veira ((a, a'ke;]i, ka'veIRå)b]÷ after CC (simplified or not in their pronunciation –
and in spelling): (a'sAu«)l /A'saun/ a(c)ção˚ (fa'tu;R‘)l /fA'tuRa/ fa(c)tura ((a'sAu«, fa-
'tu;Rå)b]˘ Also (ka'm9i”S)l /kA'moins/ Camões.
Before looking at the two çfurtherÇ phonemes, let us consider the Lusitanian
pronunciation of the examples already given in Brazilian pronunciation: (d¢'fi;siı)l
/òi'fisiı/, ('e;]¢)l /'eıe/, ('E;]‘)l /'Eıa/, (b‘'ta;t‘)l /ba'tata/, ('nO;v‘)l /'nOva/, ('no;vu)l /'no-
vu/, ('Su;v‘)l /'Suva/÷ ('pAu)l /'pau/, (Au'ƒaS)l /Au'òas/, ('mAı)l /'maı/, (AÓ'tu;R‘)l /Aı'tu-
Ra/÷ ('se;k‘)l /'seka/», ('sE;k‘)l /'sEka/”, ('seÍku)l /'seKku/», ('sEÍku)l /'sEKku/”, (']o;Bu)l
/'ıobu/», (']O;Bu)l /'ıObu/”, ('fo;su)l /'fosu/», ('fO;su)l /'fOsu/”.
Furthermore (for both accents): (ka'da;ve˜)b (k‘'ƒa;v™R)l /ka'davÈK/ cadáver˚ ('sO;-
Ro˜)b (-øR)l /'sORìK/ sóror˚ (a'ma;ve∞)b (‘'ma;v™ı)l /a'mavÈı/ amável˚ ('A∞kjo∞)b ('AÓ-
kˆøı)l /'aıkwìı/ álcool˚ (vo∞'ta˜)b (vøÓ'taR)l /vìı'taK/ voltar˘ Of course, if only Bra-
zilian pronunciation were taken into account, even in these cases, no diaphoneme
with underwritten dots would be necessary, since there is no di‡erence in the real-
ization of /e, È÷ o, ì/ (e) and (o).
7.1.2.5. Here are some specific examples to show the di‡erences between the
two accents more clearly. In order to account for the pronunciation of central Por-
tugal, with Lisbon as its focal point, therefore as a model of neutral Lusitanian pro-
nunciation, it is important to know that /e/, in stressed syllables followed by /N, S,
Z, L, j, i/, takes on the pronunciation (É), which is close to the realization of the
Lusitanian phoneme ç/å/Ç (but we must lay great stress on the fact that they are
not alike {even if, occasionally, for some words or specific speakers, the di‡erence
is only very slight}). However, we must point out that even the –more regular and
etymological– pronunciation corresponding to the Brazilian one, is neutral (¤ ac-
ceptable neutral]˚ with /e, ei/ (e, eI), which is kept quite well outside the areas in-
7. portuguese 261
7.1.2.7. Another Lusitanian çoddityÇ consists of the change of /’i/ (i) -i- sequenc-
es (not /iö˘/ (iö)) into ç/…/Ç, ¤ /e/, in adjacent syllables (except for the last one,
whether stressed or not, and the first one, if absolute initial, with no C-]\ (&d¢v¢'ƒiR,
d¢v'-)l (&‚ivi'‚i˜)b /òeve'òiK/ dividir˚ (m[¢]'niStÍu)l (mi'nistRu)b /me'nistRu/ min¤tro˚
(&in[¢]'mi;Ÿu)l (&ini'mi;gu)b /ine'migu/ inimigo˚ (pÍ¢&vi][Û]'Zãa;ƒu)l (pRi&vi]e'òãa;du)b /pRe-
viıÈ'Zjaòu/ privilegiado (in the very last example, the sequence (-]Û'Z-) /-ıÈ'Z-/ results
from what has just been said previously).
However, we find the following exceptions, where i /’i/, corresponding to an
original /'i/, remains /’i/ in derivatives, as in the conditional mood of verbs in -ir:
(d[¢]&v¢ƒi'Ri‘m¨S)l (‚i&vi‚i'RiAmus)b /òeveòi'Riamus/ dividirìamos÷ and in superla-
tives: (&d¢fî'si;]imu)l (&‚ifi'si;]imu)b /òefi'siıimu/ dificìlimo˘
For rare /u/ sequences, the same is true, so we have to resort to the diaphoneme
/™/ to be able to account for this –possible– Lusitanian phenomenon: (f˙'tu;Ru, f¢-,
f¢-, fu-, f¨-, fu-)l (fu'tu;Ru)b /f™'tuRu/ futuro˘
Diphthongs
7.1.3.2. ˛erefore, with constant reference to û 7.1-2 (for both accents ¤ Bra-
zilian and Lusitanian), we now show the Portuguese diphthongs, even if in the vo-
cograms only /ei, o™/ appear as (eI, eI, ™I, oU, ÚU, 9U)b (ÉI, ÙI, 3I, ’I÷ oU, 9U, ÚU÷ 4i,
7. portuguese 263
4u)l. As a matter of fact, their second elements are (I, U), which are not the exact
realization of the phonemes /i, u/; for the Lusitanian accent, even (4i, 4u) are
placed in the vocogram, because their first element is closer, as occurs however for
/’aö/ (4ö) as well).
≈rst of all, we list them phonetically and diaphonemically: (eI)b (ÉI, ’ÙI)l /ei/,
(Ei, ’™i) /Ei/, (ai) /ai/÷ (ui) /ui/, (oi) /oi/, (Oi, ’øi) /Oi/÷ (iu) /iu/, (eu) /eu/, (Eu, ’™u)
/Eu/, (Au) /au/, (oU, o)b (o, oU)l /o™/ ((o) unstressed /ì™/).
Here are some actual examples: ('˜eIs)b ('ºÉIS)l /'Keis/ re¤˚ (]eI'ë9i«s)b (]ÙI'S9i”S)l
/ıei'Soins/ Leixõπ˚ ('˜Eis)b ('ºEiS)l /'KEis/ ré¤˚ (&˜™i'zi;tus)b (&º™i'zi;tuS)l /'KEi'zitus/ reiz-
itos˚ ('vai) /'vai/ vai÷ ('fui) /'fui/ fui˚ ('sois)b ('soiS)l /'sois/ so¤˚ ('sOis)b ('sOiS)l /'sOis/ só¤˚
(i“RøikA'm™~Ci)b (i“Røik‘'m™nt˙)l /i'ROika'mente/ heroicamente÷ ('viu) /'viu/ viu (and
vi-o as well, since the supposed di‡erence between the two is not at all phonic –in
neutral pronunciation– but derives exclusively from a wish to keep di‡erent mor-
phological forms separate), ('seu) /'seu/ seu˚ (&ëap™u'zi;Nu)b (&S‘p™u'zi;Nu)l /SapEu'zi-
Nu/ >apeuziˆo÷ (ëa'pEu)b (S‘-)l /Sa'pEu/ >apéu˚ ('mAu) /'mau/ mau˚ ('voU, —vo)
/'vo™/ vou÷ Lusitanian pronunciation prefers the reverse order, ('vo, 'voU), for the
last one.
Instead, the simplification of ('eI)b ('ÉI, ’ÙI)l /ei/ changing into ('e) does not be-
long to neutral pronunciation, although it is quite widespread, mainly in Brazilian
(and southern Lusitanian) pronunciation.
Of course, there are other actual diphthongs, as the following: (i«'f]uu)b (i”'f9-)l
/in'fıuu/ influo˚ ('mou) /'mou/ moo˚ (']uå)b (-‘)l /'ıua/ lua˚ ('˜iå)b ('ºi‘)l /'Kia/ ria˚
('boå)b (-‘)l /'boa/ boa÷ including combinations as in the true triphthongs resulting
from /i+ei, i+au/: ('vieIs)b ('viÙIS)l /'vieis/ vìeis, ('˜ieIs)b ('ºiÙIS)l /'Kieis/ rìeis, ('˜ieI«)b
('ºi’I«)l /'Kiein/ riem, and ('˜iAu«)b ('ºi4u«)l /'Kiaun/ riam.
ger and avoid people believing that, in unstressed syllables, in Lusitanian pronun-
ciation they can be reduced to ç/åu, åi/Ç (as some texts say).
Let us briefly observe (and without showing it in vocograms {although, of
course, this is unavoidable in a book entirely dedicated to the pronunciation of
Portuguese}), that when /E, O/ are half-lengthened in unchecked syllables (or, for
emphasis, even in checked syllables), besides as being realized as normal monoph-
thongs, (E;, O;), they can also be realized as doubling, or as very narrow diphthongs
– more frequently so in Brazilian pronunciation. In comparison with the articula-
tions shown in û 7.1-2, these may start from slightly raised points and reach the
indicated ones, (EE§, OO§), or else they may start from those and lower themselves
a little, crossing the border of the phone below (in the vocogram), ((Eπ°, OØ°)) – in-
cluding intermediate shifts (however, again of the opening type, although limit-
ed), that is ((Ep, OQ)).
Consonants
7.2.0. û 7.3 is the table of the Luso-Brazilian consonants, which we will exam-
ine systematically, according to manners of articulation. Instead, û 1.9-15 give the
orograms, equally grouped by manners, of all the contoids given in this book, even
as secondary, occasional, or regional variants, for the 12 languages dealt with.
postalveo-palatal
postalveo-palatal
velar rounded
protruded
labiodental
rounded
provelar
alveolar
bilabial
palatal
uvular
dental
velar
Nasals
7.2.1.1. Portuguese has three nasal phonemes: (m, n, N) /m, n, N/: ('kA;mås)b
('kå;m‘S)l /'kamas/ camø˚ ('kA;nås)b ('kå;n‘S)l /'kanas/ canø˚ ('kA;Nås)b ('kå;N‘S)l /'ka-
Nas/ caˆø˘
In addition, there are some taxophones for /ö0, ö˘, öò/, and it is useful to linger
over them, since too many descriptions keep on ignoring them completely or in
part. As a matter of fact, the only result of transcriptions like ç('Ki, 'ıesu, 'ıA, 'bRAku,
'sÚ, 'us/'uS)Ç is to make people think that Portuguese sounds almost like French
–which is decidedly misleading– since they persist in groundless phonological
characteristics.
266 a handbook of pronunciation
7.2.1.2. ˛erefore, transcriptions like those just mentioned are not trustworthy,
unless they presume that foreigners will inevitably add a nasal consonant (al-
though it is hard to believe), producing something similar to what is expected (ex-
cluding northern French people, who would then have other problems indeed).
On the other hand, this would be a very shoddy way of teaching phonetics.
˛erefore, taxophones are absolutely necessary for exact transcriptions of Por-
tuguese (and for satisfying phonic reproductions), including devoiced variants es-
pecially for the Lusitanian accent (in front of voiceless C]\ (m) + /p, b/: ('t™mpu)b
('t™≈pu)l /'tempu/ tempo˚ ('s9mbRå)b ('s9mbR‘)l /'sombRa/ sombra÷ (n) + /t, d/ (den-
tal, which could be transcribed with ((˙)), but the normal symbol is quite su‚-
cient) and (~) (for Brazilian Portuguese, before (C, ‚) + (i, ã), /t, d/ + /i, e, j/): ('A~-
Cis)b ('Ant?S)l /'antes/ antπ˚ ('v™ndå)b ('v™nd‘)l /'venda/ venda˚ ('9~‚i)b ('9nd¢)l /'on-
de/ onde÷ (˙) + /k, g/: ('nu˙kå)b ('nu~k‘)l /'nunka/ nunca˚ (']i˙gjå)b (']i˙gj‘)l /'ıin-
gwa/ lìn%a˘
Stops
7.2.2.1. ˛ere are three diphonic pairs, already seen in various examples, (p, b÷
t, d÷ k, g) /p, b÷ t, d÷ k, g/, with dental (t, d) which, in Brazilian pronunciation,
becomes stop-strictives ((C, ‚), seen previously), before /i, e, j/: ('p9mbå)b (-‘)l
/'pomba/ pomba˚ ('kA˙gå)b ('kA˙g‘)l /'kanga/ canga˚ ('t™ndå)b (-‘)l /'tenda/ tenda˚
('Ci;midu)b ('ti;miƒu)l /'timiòu/ tìmido˚ (sAu'da;‚i)b (sAu'ƒa;ƒ¢)l /sAu'òaòe/ saudade˚
('Cãa;tRu)b ('tªa;tÍu)l /'tjatRu/ teatro˘ Before front V or /j/, /k, g/ are realized as pre-
7. portuguese 267
velar, ((´, Ò÷ Úl)), by coarticulation, but it is not necessary to use these symbols, ex-
cept for Brazilian pronunciation before /i, e, j/, where we find (©, á): ('©intu)b ('kin-
tu)l /'kintu/ ¢into˚ ('tO;©i)b ('tO;k˙)l /'tOke/ to¢e˚ (iN'©ãE;tu)b (i~'kªE;tu)l /in'kjEtu/ in-
¢ieto˚ ('áiå)b ('gi‘)l /'gia/ %ia˘
˛e most interesting Lusitanian peculiarity (which is more complicated for for-
eigners, but also for Brazilans who might try to speak like Lusitanians) is constitut-
ed by voiced stops, /b, d, g/, which are realized as actual stops, (b, d, g), only after
pauses, after N (as can be seen in previous examples), and for /ıd/ (ıd), with an api-
cal contact (since both of them are homorganic articulations), or for emphasis or
precision: ('b9U«) /'boun/ bom˚ ('da) /'òa/ dá˚ ('ga;tu)l (-u)b /'gatu/ gato˚ ('kAıdu)l
(-∞-)b /'kaıdu/ caldo÷ (p¢'ƒi;ƒu)l (pe'‚i;du)b /pe'òiòu/ pedido÷ with emphasis: (p¢"di:-
du)l…
7.2.2.2. In all other cases, continuous realizations occur, (B, ƒ, Ÿ): two approxi-
mants and a constrictive respectively. ˛e diaphonemic transcription shows /òb, òò,
òg/, just to underline the di‡erence and to make their distribution adequately un-
derstood (avoiding forced inferences, which risk strengthening wrong concepts).
Of course, in Lusitanian Portuguese, the (word-initial) notation /b, ò, g/ is çpo-
tentialÇ, since it corresponds to (b, d, g) or (B, ƒ, Ÿ) according to actual contexts,
not in absolute terms. In addition, we must admit that in the case of /ıb, ıg/, and
of /Rb, Rò, Rg/, and of (more) scholarly consonant clusters as well, stop realizations
are not rare, (ıb, ıg÷ Rb, Rd, Rg), even without emphasis or desire for precision.
Some examples: ('a;B‘)l ('a;bå)b /'aba/ aba˚ (u'BRi~ku)l (u'bRi˙ku)b /u'bRinku/ o
brinco˚ (]iZ'Bo‘, ]ÛZ-)l (]iz'boå)b /ıia'boa/ L¤boa˚ ('AıBu«, 'Aıb-)l ('A∞bu«)b /'aıbun/
álbum˚ (‘R'Bi;tͪu, ‘R'b-)l (a˜'bi;tRãu)b /aK'bitRju/ arbìtrio˚ (&‘Bƒi'kaR, &‘bd-)l (&ab‚i-
'ka˜)b /abòi'kaK/ abdicar˚ (&um‘ƒ¢'ƒa;ƒ‘)l (&umade'da;då)b /umaòe'òaòa/ uma deda-
da˚ (p‘'ƒRAu«)l (pa'dRAu«)b /pa'òRaun/ padrão˚ ('aRƒju, 'aRd-)l ('a˜dju)b /'aKòwu/ ár-
duo˚ (dÛZ'ƒ3I«)l (dez'd™I«)b /òea'òein/ dπdém˘
More: (™'Ÿa;Ÿu)l (™'ga;gu)b /E'gagu/ é gago˚ (Aı'ŸOS, Aı'g-)l (A∞'gOs)b /Aı'gOs/ algoz˚
('veZŸu)l (-zgu)b /'veagu/ vπgo˚ (‘'ŸRa;s‘)l (a'gRa;så)b /a'gRasa/ a graça˚ (&‘RŸ‘'ma;s‘,
&‘Rg-)l (&a˜gA'ma;så)b /aKga'masa/ argamøsa˘
Constrictives
7.2.3.1. ˛ere are three diphonic pairs of constrictive phonemes, plus a couple
of diaphonemes (/s, a/, which we will see presently): (f, v÷ s, z÷ ëb, Sl, òb, Zl) /f, v÷ s,
z÷ S, Z/: ('fa;su)b (-su)l /'fasu/ faço˚ ('po;vu) /'povu/ povo˚ ('su∞)b (-ı)l /'suı/ sul˚ (te«-
'sAu«)b (-”-)l /ten'saun/ tensão˚ ('vA∞så)b (-Ós‘)l /'vaısa/ valsa˚ ('pE˜så)b ('pEÍs‘)l /'pER-
sa/ persa˚ (pa'se;ãu)b (p‘'sÉ;ãu)l /pa'seju/ pøseio˚ ('va;zu) /'vazu/ vøo˚ (a'za˜)b (‘'zaR)l
/a'zaK/ azar˚ (u'zi~‚ãus)b (u'zindã¨S)l /u'zindjus/ os ìndios˚ (ëe'Res)b (SÛ'ReS)l /Se'Res/ xe-
rez˚ ('ëE˜ëis)b ('SERS?S)l /'SEKSes/ Xerxπ˚ ('o;òi)b ('o;ZÛ)l /'oZe/ hoje˘
In neutral Brazilian, the diaphonemes /s, a/ behave as /s, z/, with distribution:
/s|, s[ò]=, z[ò]Ê, zòé/, while, in neutral Lusitanian, they correspond to /S|, S[ò]=, Z[ò]Ê,
zòé/ (as well as in the çCariocaÇ accent, of Rio de Janeiro, but of course realized as
(ë|, ë[ò]=, ò[ò]Ê, zòé)): ('tRas)b ('tÍaS)l /'tRas/ trás˚ (us'pais)b (uS'paiS)l /us'pais/ os pa¤˚ (i-
'sE∞su)b (?S'sEÓsu)l /is'sEısu/ excelso˚ (bas'ta˜)b (b‘S'taR)l /bas'taK/ bøtar˚ (pes'ka;då)b
(p?S'ka;ƒ‘)l /pes'kaòa/ pπcada˚ (az'mAi«s)b (‘Z'mAi”S)l /aa'mains/ ø mãπ˚ ('azmå)b
('aZm‘)l /'aama/ øma˚ (az']A«s)b (‘Z']A”S)l /aa'ıans/ ø lãs (further examples occur in
other parts of this chapter).
Approximants
7.2.4.1. ˛ere are two approximants, (ã, j) /j, w/ (semi-palatal, and semi-velar
rounded), corresponding to prevocalic unstressed >i˚ e≥ and >u˚ o≥: ('mão;]u) /'mjo-
ıu/ miolo˚ ('g]O;Rãå)b (-‘)l /'gıORja/ glória˚ ('pq9;nãå)b ('pªO;nã‘)l /'pjìnja/ peôniab˚ peó-l˚
('pqAu«)b ('p'Au«)l /'pjaun/ peão˚ (i'dE;ãå)b /i'òEja/ idéia (i'ƒÉ;ã‘)l /i'òeja/ ideia˚ (ko-
'˜e;ãu)b (k¨'ºÉ;ãu)l /ko'Keju/ correio˚ (kÚm'bO;ãu) /kom'bOju/ comboio˚ (sAm'pa;ãu)b
(sA≈-)l /sam'paju/ Sampaio÷ ('a;gjå)b ('a;Ÿj‘)l /'agwa/ á%a˚ ('pjE;tå)b ('pˆE;t‘)l
/'pwEta/ poeta˚ ('va;kju)b (-kˆu)l /'vakwu/ vácuo˚ (a˜'gji˜)b (‘R'ŸjiR, ‘R'g-)l /aK-
'gwiK/ ar$irb˚ -%irl˚ ('käimbRå)b ('kîimbR‘)l /'kwimbRa/ Coimbra˘
If pronunciation is slowed down, either for clarity or emphasis, (ã, j) /j, w/ may
change (through (j, w)) to (i, u) (and to (e, o) as well, according to spelling): ('g]O;-
Riå)b (-i‘)l, (pi'9;niå, pe-)b (pi'O;ni‘, pe-)l˚ (pi'Au«, pe-), (i'dE;iå)b (i'ƒÉ;i‘)l, (ko'˜e;Iu)b
(k¨'ºÉ;Iu)l, (kÚm'bO;iu), (sAm'pa;iu)b (sA≈-)l÷ ('a;guå)b ('a;Ÿu‘)l, (pu'E;tå, po-)b (-‘),
('va;kuu, -¨u), (&a˜gu'i˜)b (&‘RŸu'iR, -g-), (ku'im-, ko-).
Several examples show that, in Lusitanian pronunciation, by assimilation, we
have devoicing after voiceless C˚ and nasalization before nasalized V˚ (ª, ˆ÷ ', î÷ q,
ä); in Brazilian pronunciation, we generally find only nasalization, (q, ä).
Trills
7.2.5.1. Under this manner of articulation, in addition to the alveolar tap, (R)
/R/, we will treat the theoretical uvular trill, /K/, and the diaphoneme /K/ as well.
For the former, there is not much to say, except that it has a single alveolar con-
tact, in fact it may even become an alveolar approximant (¸); in Luso-Brazilian pro-
7. portuguese 269
nunciation, it occurs between V (in the same word), where it distinctively oppos-
es /K/ and after tautosyllabic C\ ('ka;Ru) /'kaRu/ caro (Ô ('ka;˜u)b (-ºu)l /'kaKu/ car-
ro]˚ ('bRa;su)b (-su)l /'bRasu/ braço˚ ('fRiu)b ('fÍ-)l /'fRiu/ frio˘
Besides (as a realization of /K/) (R) occurs, in Brazilian pronunciation, in word-
-final position too, when followed by a subsequent word-initial V˚ with resyllabifi-
cation and behavior as if in word-internal position (¤ /éKòé/=/é˘Ré/), while in Lu-
sitanian pronunciation it also occurs in final position, even before a pause or a C\
(']e 'RA∞gu)b (']e 'RAıŸu, -gu)l /'ıeK 'aıgu/ ler algo˚ (']e˜ 'poUku, 'po-)b (']eÍ 'poku, 'poU-)l
/'ıeK 'po™ku/ ler pouco˚ (nAu«']e˜)b (n4u«']eR)l /naun'ıeK/ não ler˘
Laterals
7.2.6.1. ˛ere are two lateral phonemes, in both accents, /ı, L/. For the former
we prefer the symbol /ı/, to a more generic /l/, since even before V it is realized as
a semi-velarized alveolar, (]) (or (ı) as well, ¤ completely velarized alveolar, even
if it can also be (l), which used to be the traditional neutral Lusitanian pronuncia-
tion, but only optional nowadays).
After V˚ before a pause or a C˚ in the Brazilian accent, velarized alveolar round-
ed (∞) occurs; it has a frequent vocalized variant, (u) –which, however, is hardly
neutral– often mistaken for (∞).
In the Lusitanian accent, we have (ı) (with the possible uvularized variant, (l)):
(']eICi)b (']ÉIt˙)l /'ıeite/ leite˚ ('k]a;Ru)b ('k9a;-)l /'kıaRu/ claro˚ (a'zu ]is'ku;Ru)b (‘'zu ]ZS-
'ku;Ru)l /a'zuı is'kuRu/ azul πcuro˚ ('sA∞tu÷ 'sAutu)b ('sAÓtu)l /'saıtu/ salto˚ (òe'RA∞÷
-Au)b (ZÛ'RAı)l /Ze'Raı/ geral˚ ('ta;Lu) /'taLu/ taºo˘
‹ile in Brazilian pronunciation, as already said, /ıò, ı0/ may become (u); in
Lusitanian pronunciation, /ıò/ may be followed by a more or less short — devoiced
(¢) in contact with voiceless C˘
Structures
7.3.0. ˛e interesting subjects, for this part, are: metaphony, words in connect-
ed speech, stress and intonation.
Taxophonics
7.3.1.2. Simplifying our exposition a little, we may say that metaphony oper-
ates in partially di‡erent ways with verbs and non-verbs (¤ nouns˚ adjectives and
some pronouns)˘
Furthermore, a distinction is to be made between e and o˘ For non-verbs with
stressed e˚ the endings -o˚ os /-u, -us/ may cause the closing of timbres: (ka'pe;]u[s])b
(k‘'pe;]u, -¨S)l /ka'peıu[s]/ capelo(s)˚ but (ka'pE;]å[s])b (k‘'pE;]‘[S])l /ka'pEıa[s]/ capela(s)˘
7. portuguese 271
However, not all feminine forms have /E/; actually, very often it is not so: ('ze;-
bRå)b (-BR‘)l /'zebRa/ zebra˚ ('se;då)b (-ƒ‘)l /'seòa/ seda÷ on the other hand, not all
masculine forms have /e/ either, so we find: ('bE;]u) /'bEıu/ belo˚ compared to ('ne;-
gRu)b (-Ÿ-)l /'negRu/ negro (plural and feminine as well).
7.3.1.3. For nouns with stressed o˚ only -o /-u/ (m. sg.) may cause closing: ('po˜-
ku)b ('poÍku)l /'poKku/ porco˚ while we have ('pO˜kus)b ('pOÍkuS)l /'pOKkus/ porcos
and ('pO˜kå[s])b ('pOÍk‘[S])l /'pOKka[s]/ porca(s) as well; besides: ('no;vu) /'novu/ no-
vo˚ but ('nO;vus)b (-¨S)l /'nOvus/ novos and ('nO;vå[s])b (-‘[S])l /'nOva[s]/ nova(s)˚ (fo˜-
'mo;zu)b (f¨R-)l /foK'mozu/ formoso˚ but (fo˜'mO;zus)b (f¨R-, -¨S)l /foK'mOzus/ formo-
sos and (fo˜'mO;zå[s])b (f¨R-, -‘[S])l /foK'mOza[s]/ formosa(s)÷ however, we find: (is'po;-
zu[s])b (?S-, -u, -¨S)l /es'pozu[s]/ πposo(s) and (is'po;zå[s])b (?S-, -‘[S])l /es'poza[s]/ πpo-
sa(s)˚ but (is'pO;zus)b (?S-, -¨S)l /es'pOzus/ πposos for the çcoupleÇ.
Nevertheless, we also find several cases with no variation, such as: (a'do;bu[s])b
(‘'ƒo;Bu, -¨S)l /a'òobu[s]/ adobo(s)˚ ('go˜du[s], -å[s])b (-Rƒu[S], -‘[S])l /'goKòu[s], -a[s]/
gordo(s)˚ -a(s)˘ Of course, there are cases with /O/ as well: ('mO;du[s])b ('mO;ƒu, -¨S)l
/'mOòu[s]/ modo(s)˚ ('fO;ku[s])b (-ku[S])l /'fOku[s]/ foco(s)÷ and even feminine forms
with /o/: ('go;tå[s])b (-‘[S])l /'gota[s]/ gota(s)˚ ('fo˜så[s])b ('foÍs‘[S])l /'foKsa[s]/ força(s)˘
7.3.1.4. For verbs˚ the endings which cause closing are -o˚ -a˚ -ø˚ -am /-u, -a, -as,
-aun/ (for second-conjugation forms with stress on the stem): ('de;vu, 'mo;vu {-å,
-ås, -Au«})b ({-‘, -‘S, -4u«})l /'òevu, 'movu {-a, -as, -aun}/ devo˚ movo…
In checked syllables in N˚ the e‡ect is neutralized: ('v™ndu, 'v™~‚i)b ('v™ndu,
'v™nd¢)l /'vendo, 'vende/ vendo˚ vende÷ however, in unchecked syllables, Lusitanian
pronunciation maintain the di‡erence (contrary to Brazilian pronunciation due
to nasalization): ('t™;mu, 't™;mi)b ('te;mu, 'tE;m¢)l /'temu, 'tÈme/ temo˚ teme˚ ('k9;mu,
'k9;mi)b ('ko;mu, 'kO;m¢)l /'komu, 'kìme/ como˚ come˘
For the same reason, nouns behave in the same way: ('s™;não˜)b ('sE;nãøR)l /'sÈ-
njìK/ sênior/sénior˚ ('k9;miku)b ('kO;mîku)l /'kìmiku/ cômico/cómico (with far from
unquestionable consequences on spelling, which is still overestimated).
Lastly, we have the pronouns ('e;]i[s], 'E;]å[s])b (-¢, -¢S, -‘[S])l /'eıe[s], 'Eıa[s]/ ele(s) e-
la(s)˚ ('esCi[s], 'Estå[s])b ('eSt˙, -?S, 'ESt‘[S])l /'este[s], 'Esta[s]/ πte(s)˚ πta(s) and their deriv-
atives. For other indications, and exceptions, good grammars answer the purpose
quite exhaustively (especially if they are less recent).
7.3.1.5. In colloquial Brazilian Portuguese (in common –not lofty– words), an-
other type of vowel adjustment is frequent; it is a synchronic phenomenon which
may cause pre-stressed e˚ o to be realized as /i, u/ (i, u), often (I, U) (Ô û 7.1), when
the next stressed vowel is /i, u/: (me'ni;nu, mi-, mI-)b (m¢'ni;nu)l /me'ninu/ menino˚
(&a]e'gRiå, &a]i-, &a]I-)b (&‘]¢'ŸRi‘)l /aıe'gRia/ alegria˚ (so'˜i;zu, su-, sU-)b (s¨'ºi;zu)l /su'Ki-
zu/ sorr¤o˚ (ve']u;du, vi-, vI-)b (v¢']u;ƒu)l /ve'ıuòu/ veludo˘
However, the same vowels may be realized as (™, ø) when the stress is on open-
er V\ (˜e']O;òãu, ˜™-)b (º¢']O;Zãu)l /Ke'ıOZju/ relógio˚ (ko'˜E;tu, kø-)b (k¨'ºE;tu)l /ko'KE-
tu/ correto/correcto˘
A systematic use of /e, o/ may give the impression of meticulous attention (or,
272 a handbook of pronunciation
perhaps, formality); on the other hand, a methodical use of (i/I÷ u/U÷ ™, ø) would
certainly produce something çexcessiveÇ, and strange or foreign, since lofty or rare
words must remain unchanged. Furthermore, we have for instance: foºiˆa /fo-
'LiNa/ (fu'Li;Nå)b (f¨'Li;N‘)l çcalendarÇ, but (fo'Li;Nå)b (f¨'Li;N‘)l çsmall leafÇ, which
is felt as a derivative, contrary to the other.
7.3.2.1. Clusters such as /éòé/ (where the first V is word-final {even in unstressed
monosyllables: me˚ te˚ se˚ ºe˚ ¢e˚ e˚ de˚ o˚ do, no} + initial V] produce some simpli-
fications within intonation groups, not only in rhythm groups.
Generally, /e, i/ and /u/ are realized as /j, w/, respectively, or are dropped: (en-
'tRe;]is)b (en'tÍe;]¢S)l /entRe'eıes/ entre elπ˚ ('da;gjå, '‚ãa;-)b ('da;Ÿj‘)l /òe'agwa/ de á-
%a˚ ('ãoUtRå 'koizå, 'ão;-)b ('ão;tÍ‘ 'koiz‘, 'ãoU-)l /i'o™tRa 'koiza/ e outra co¤a˚ (si˙'k]i;-
nå)b (sí~'k9i;n‘)l /sein'kıina/ se inclina˘
Further examples: ('to; d[j]esCi't™mpu)b ('to; ƒ[j]eSt˙'t™≈pu)l /'toòu este'tempu/
todo πte tempo˚ (sãa'vi;Ris)b (sª‘'vi;RZS)l /sia'viRes/ se a virπ˚ (&ist[j]™u«'˜oUbu, -'˜o;-)b
(?St[ˆ]™u«'ºo;Bu, -oU-)l /estuEun'Ko™bu/ ¤to é um roubo˚ (äA'mi; gjita']ãA;nu)b (j‘-
'mi; Ÿjît‘']ãå;nu)l /ua'migu ita'ıjanu/ o amigo italiano˘
More: (nAum'po; '‚[ã]i˜)b (n4u≈'po; 'ƒ[ã]iR)l /naumpoòe'iK/ não pôde ir˚ (©ã™'muin
t[j]a']E;gRi)b (kª™'muin t[ˆ]‘']E;ŸR¢)l /keE'muintu a'ıEgRe/ ¢e é muito alegre˚ ('‚i;si L[ã]a-
'e;]i)b ('di;sZ L[ã]‘'e;][¢])l /'òiseLe a'eıe/ d¤se-ºe a ele˚ ('deIë[is] 'ta˜)b ('dÉIS[?S] 'taR)l /'òeiSe
is'taK/ deixe πtar˚ ('òu; R[ã]i'd9;nãu)b ('Zu; R[ã]i'ƒO;nãu)l /'ZuRi i'òìnju/ jùri idôneo/idóneo˘
Besides: ('be;b[j]u 'to;du)b ('be;B[j]¨ 'to;ƒu)l /'bebuu 'toòu/ bebo-o todo˚ ('t[j]u;zåz
'muin 't[j]i;su)b ('t[ˆ]u;z‘Z 'muin 't[ˆ]i;su)l /tu'uzas 'muintu 'isu/ tu ¨ø muito ¤so˚
('gRA~ ‚ão'˜o˜, -n do-)b ('gRAn d[ã]ø'ºoR)l /'gRande o'KoK/ grande horror˚ (äAn't9;nãu)b
(j4n'tO;nãu, ä4n-)l /uan'tìnju/ o Antônio/António˘
ni˙'g™I«)b (-3I«)l /kounnin'gein/ com nin%ém÷ (']A A~'Ci;gå, ']A;~ 'C-)b (']A 4n'ti;Ÿ‘,
']A;n 't-)l /'ıan an'tiga/ lã antiga˚ (kÚU'A«sãå)b (kÚU'A”sª‘)l /koun'ansja/ com ânsia÷ (']A
a'zu∞, ']A; 'z-)b (']A ‘'zuı, ']A; 'z-)l /'ıan a'zuı/ lã azul˚ (kÚU'i;su)b (-su)l /koun'isu/ com ¤-
so˚ (u'9;meI 'A∞tu, 'ä9;-)b (u'O;m’I 'AÓtu, 'äO;-)l /un'ìmein 'aıtu/ um homem alto÷ (se-
'qE;]å)b (s’'qE;]‘)l /sein'Eıa/ sem ela˚ (nA'äE)b (n4'äE)l /naun'E/ não é˘
7.3.2.3. ‹en the syllables in contact have identical or similar C˚ we can find
some geminate, due to the dropping of the vowel element: ('fi;©[i] kÚ'mi;gu)b ('fik[˙]
k¨'mi;Ÿu)l /'fike ko'migu/ fi¢e comigo (da&ke][i]']a;du)b (d‘&ke][¢]']a;ƒu)l /òa'keıe 'ıa-
òu/ da¢ele lado˚ ('k9;m[i] 'muintu)b ('kO;m[¢] 'muintu)l /'kìme 'muintu/ come mui-
to˚ ('k9mpRuCi 'tu;du, -t 't-)b ('k9≈pÍut[˙] 'tu;ƒu)l /'kompRute 'tuòu/ compro-te tudo˚
('kAmp[u] peRi'go;zu)b ('kA≈p[u] p¢Ri'Ÿo;zu, -p pÍi-)l /'kampu peRi'gozu/ campo peri-
goso˘
More: (us'peIëi zis&tAuna'dAndu, -eIë ëis-)b (uS'pÉISZ zZS&t4un‘'ƒAndu, -ÉIS S?S-)l /us'pei-
Se zistaunna'òandu/ os peixπ πtão nadando˚ ('oUv[i] 'vO;zis, 'o;-)b ('o;v[¢] 'vO;zZS, 'oU-)l
/'o™ve 'vOzes/ ouve vozπ˚ ('pa;s[i] 'se;du)b ('pa;s[˙] 'se;ƒu)l /'pase 'seòu/ pøse cedo˚ ('pE;-
‚[i] 'tu;du, -∂ 't-)b ('pE;ƒ[¢] 'tu;ƒu, -‡ 't-)l /'pEòe 'tuòu/ pede tudo˚ ('s™~C[i] 'bu;Lå, -n∂
'b-)b ('s™nt[˙] 'Bu;L‘, -n∂ 'b-)l /'sente 'buLa/ sente buºa˘
Chiefly when the syllables contain /t, ò/, one of them may drop completely: (&fa-
ku∞'da;[‚i] ‚i']e;tRås, -a‚ ‚i-)b (&f‘k¨ı'da;[ƒ¢] ƒ¢']e;tÍ‘S, -aƒ ƒ¢-)l /fakuı'daòe òe'ıetRas/
faculdade de Letrø˚ ('pO;[‚iz &]dez]i'ga˜, 'pOd d-)b ('pO;[ƒÛZ &]ƒÛZ]i'ŸaR, 'pOƒ ƒ-)l /'pOòea
òeaıi'gaK/ podπ dπligar˚ ('kA∞[du] ‚i'kA;nå)b ('kAı[du] d¢'kå;n‘)l /'kaıdu òe'kana/ cal-
do de cana˘
More examples: ('ò™~[Ci] ‚i'fO;Rå)b ('Z™nt¢ ƒ¢'fO;R‘, -n d¢-)l /'Zente òe'fORa/ gente de
fora˚ ('ò™~[Ci] ‚i'ReItå)b ('Z™nt¢ ƒ¢'RÉIt‘, -n d¢-)l /'Zente òe'Reita/ gente direita÷ also:
('ba;[Ci]Ci, -CCi)b ('ba;[t˙]t˙, -tt˙)l /'batete/ bate-te˚ ('vEs[Ci]Ci, -sCCi)b ('vES[t˙]t˙, -tt˙)l
/'vEstete/ vπte-te˘
di‡erence as to ('kai) /'kai/ cai]˘ In Lusitanian pronunciation, again, the rare in-
stances of /’iò/ oscillate between /i/ (which is more modern and closer to spelling)
and (the more traditional) /e/: ('taksi, -s˙)l ('taksi)b /'taksi/ táxi÷ forms like /'ZuRi/ jù-
ri and /'ZuRe/ jure (which are alike in Brazilian pronunciation: ('òu;Ri)b], in the Lu-
sitanian one, may either be alike: ('Zu;R¢)l, or di‡erent: ('Zu;Ri)l (-R¢)l (respectively).
7.3.2.5. ‹en, at word boundaries, /KòK, ıòı/ come together, they normally sim-
plify: ('ma[˜] '˜ãA∞)b ('ma[R] 'ºãAı)l /'maK 'Kjaı/ mar real˚ ('ma[˜] ˜umo'Ro;zu)b ('ma[R]
ºumu'Ro;zu)l /'maK Kumo'Rozu/ mar rumoroso˚ (kA'nA[∞] ']impu)b (k‘'nA[ı] ']i≈pu)l
/ka'naı 'ıimpu/ canal limpo˚ (kA'nA[∞] ]imi'ta;du)b (k‘'nA[ı] ]¢mî'ta;ƒu)l /ka'naı ıemi-
'taòu/ canal limitado˘
˛e same occurs to /sòs, aòz, sòS, aòZ/, when they are alike: ('dE[s] 'sE;ku]us)b ('dES
'sE;k¨]¨S)l /'òEs 'sEkuıus/ dez séculos˚ ('dEs 'ëa;]is)b ('dE[S] 'Sa;]¢S)l /'òEs 'Saıis/ dez xalπ˚
(a[z]'z9;nås)b (‘Z'zo;n‘S)l /aa'zonas/ as zonø˚ (&azòA'nE;]ås)b (&‘[Z]Z‘'nE;]‘S)l /aaZa'nEıas/
ø janelø÷ however, we find (Ss)l (s)b /ss/ in cases such as: (&?Ss¢']™nt˙)l (&ise']™~Ci)b
/isse'ıente/ excelente˚ (kÍZS'seR)l (kRe'se˜)b /kRes'seK/ crπcer˘
7.3.2.6. In the Lusitanian accent, the sequences /0ı, 0R/ are often realized as
(0¢], 0¢R) (even if it is better to avoid such a pronunciation): ('f9oR÷ f¢']-÷ ≠f¢']-)l
('f]o˜)b /'fıoK/ flor˚ ('k9a;Ru÷ k¢']-÷ ≠k¢']-)l ('k]a;Ru)b /'kıaRu/ claro˚ ('g]O;Rã‘÷ g¢']-÷ ≠g¢']-)l
('g]O;Rãå)b /'gıORja/ glória÷ on the other hand, the sequences /0eı, 0eR, 0oı, 0oR/,
while behaving in exactly the opposite way, produce similar results, even if –in this
case– one can use that pronunciation without hesitation (provided one's speech
rate is not slow).
Some examples: (k¢'Ri‘÷ k'Íi‘÷ 'kÍi‘)l (ke'Riå)b /ke'Ria/ ¢eria˚ (ko'Ro‘÷ k'Ro‘÷ 'kÍo‘)l
(ko'Roå)b /ko'Roa/ coroa˚ (f¢'ROS, f'ÍOS, 'fÍOS)l (fe'ROs)b /fe'ROs/ feroz˚ (&m¢R¢'seR, m¢Í'seR)l
(&meRe'se˜)b /meRe'seK/ merecer÷ in current speech it is normal to reduce the preposi-
tions combined with the definite article pelo(s)˚ pela(s) to /pıé, pé/, also in Brazil-
ian Portuguese [èp'lo˚ po¶]\ (p]A'mAi«, &pe]A-)b (p9‘'mAi«, &p¢]‘-)l /p[e]ıa'main/ pela
mãe˚ ('p]ja˜, p]u'a˜, &pe]u-)b ('p9jaR, p9u'aR, &p¢]u-)l /p[e]ıu'aK/ /p[e]ıu'aK/ pelo ar˘
Compare also: pára ('pa;Rå)b (-‘)l /'paRa/ and para (&paRa, pRaò0, paò0, paRòé,
pRòé)b (&p‘R‘, pÍ‘ò0, p‘ò0, p‘Ròé, pÍòé)l /&paRa, pRaò0, paò0, paRòé, pRòé/: (&paRa'ka,
pRa'ka÷ &paRa'©i, pRa'©i)b (&p‘R‘'ka, pÍ‘'ka÷ &p‘R‘'ki, pÍ‘'ki)l /para'ka, pRa'ka÷ para'ki,
pRa'ki/ para cá˚ para aqui; besides, cada is generally (&kada)b (&k僑)l /&kaòa/. ˛ese
are restressed forms, when they occur in isolation (as it happens in metalinguistic
usage), in comparison with normal forms, which have no primary stress, but sec-
ondary or weak ones. ˛e same holds true for a(s)˚ da(s), na(s)… (a{s}, da{s},
na{s})b (‘{S}, d‘{S}, ‘{S})l /a{s}, òa{s}, na{s}/, and even for mas (mas)b (m‘S)l /mas/
(with a possible secondary stress for rhythmic reasons, (&måS)l).
In Lusitanian pronunciation again, also /0es, 0es/, mainly in the ending -π˚ may
be quite reduced, up to becoming intense –çsyllabicÇ– (however the dropping of
the V is an uneducated feature): ('f9o;RZS, -R°÷ -ÍS)l ('f]o;Ris)b /'fıoRes/ florπ; combining
what we have just seen above, in a fairly uneducated type of Lusitanian pronun-
ciation, we may also find (f¢']o;ÍS)l), (ºu'mA”s?S, -”s°)l (˜Ú'mA«sis)b /Ko'manses/ ro-
mancπ˚ (&º¢StAu'RAnt?S, &º°t-, -nt°)l (&˜estAu'RA~Cis)b /Kestau'Rantes/ rπtaurantπ˘
7. portuguese 275
In the Lusitanian accent, the sequences /peK, pRe/ often receive an intense C
and confuse with one another (as has happened to per%ntar˚ in comparison to
Spanish pre%ntar]˚ therefore (p”'fÉItu)l may stand for (pe˜'feItu, pRe'feItu)b /peK-
'feitu, pRe'feitu/ perfeito˚ prefeito (with additional variants (p¢Í'fÉItu, pÍ¢'fÉItu)l].
7.3.2.7. In both accents, the preposition com /koun/, followed by the articles,
is reduced to /kon-/ (kÚ), and to (ku, kj, kä) as well: (kÚu'pai, käu-, kju-)b (kÚu-
'pai, kîu-, kˆu-)l /kon-u'pai/ com o pai˚ (kÚ&umAmu'LE˜, käu-)b (kÚ&um‘mu'LER,
kîu-)l /kon-umamu'LEK/ com uma muºer˚ (kÚaz'mAu«s, käaz-, kjaz-)b (kÚ‘Z-
'mAu”S, kî‘Z-, kˆ‘Z-)l /kon-az'mauns/ com ø mãos˚ (kÚuz']o;bus, kuz-)b (-Z']o;B¨S,
k¨-)l /kon-uz'ıobus/ com os lobos˘
Even em is considerably reduced, up to ç/in/Ç: (eIm&pu˜tu'ges, im-)b (’I≈&p¨Ít¨-
'ŸeS, i≈-)l /ÈimpuKtu'ges/ em portu%ês˚ (eI˙'ka;zå, i˙-)b (’I~'ka;z‘, i~-)l /Èin'kaza/ em
casa˚ (&eqAu'si;]ãu, iAu-, q-)b (’qAu'si;]ãu, iAu-, q-)l /Èin-Au'siıju/ em auxìlio˘ ˛e dash put
before a V˚ in the diaphonemic transcription, could even be dispensed with, with-
out creating problems, since those forms are recognizable, somehow, thanks to the
dot under the symbols too, /ko™n, Èin/. Otherwise, we could use /«/, as a diapho-
neme (but in a systematic way, then, ¤ for all cases of /éöò/); however, this would
make the transcription heavier and the phonemic analysis more complicated.
In rhythm groups, sequenze /ééòé/ = (éò0é): (u&mejA'mi;gu)b (u&mej‘'mi;Ÿu)l
/umeua'migu, umewa-/ o meu amigo.
7.3.2.8. In the neutral Lusitanian accent, as the various examples have shown,
we systematically find (in addition to (≈, n, ~, ”, Ó, 9, Í, ª, ˆ), except in slow and
very precise pronunciation) even (˙, ?÷ u), when between voiceless C, or between
them and pause, and vice versa. It is quite common to hear things like: (uS&pͨf˙-
'so;RZS &p¨Ít¨'Ÿe;zZS)l (us&pRofe'so;Ris &po˜tu'ge;zis)b /uspRofe'soRes poKtu'gezes/ os profπ-
sorπ portu%ππ˘
In Lusitanian pronunciation, /eò, uò/ are fairly regularly devoiced, even com-
pletely, up to their dropping (after voiced C as well) which occurs, for /’e/, even
within words or rhythm groups: (∂S'kuÓp, d?S'kuÓp˙)l (‚is'ku∞pi)b /òes'kuıpe/ dπ-
culpe˚ (&t–vi'zAu«, &t¢ıv-, &t¢]¢-)l (&te]evi'zAu«)b /teıevi'zaun/ telev¤ão˚ (‘p'tE;s, &‘p˙'tE;s˙)l
(&ape'tE;si)b /ape'tEse/ apetece˚ (∂°'pOÍt°, ∂?S'pOÍtuS)l (‚is'pO˜tus)b /òes'pOKtus/ dπ-
portos˘
More: (S'pÍaR, &?Sp¢'RaR)l (&espe'Ra˜)b /ispe'RaK/ πperar˚ (St‘'tAı, &?S-)l (&ista'tA∞)b /ista-
'taı/ πtatal˚ (Sk‘'seR, &°k-, &?S-)l (&iska'se˜)b /iska'seK/ πcøser˚ (&'Bå;Nu, º¢)l (˜e'bA;Nu)b
/Ke'baNu/ rebaˆo˚ (‘&supÍãuRi'ƒa[ƒ] ƒu'sOı, ‘&s¨p¢&Rãu-)l (a&supeRãoRi'da[‚i] du'sO∞, -'dad
du-)b /asupeRjoRi'òaòe òu'sOı/ a superioridade do sol˚ (u'fin ∂s'≈å;n‘, ∂s¢-, d¢s¢-)l (u-
'fi~ ‚ise'mA;nå)b /u'fin dese'mana/ o fim-de-semana˚ (p‘'RE;s ≈um‘'Bo‘ i'ƒÉ;ã‘, p‘'RE;s
≈u-, p‘'RE;s¢ m-, p‘'RE;s¢ mãu-)l (reading: (pa'RE;semi uma'boa i'dE;ãå, -se mqu-)b} /pa-
'REseme uma'boa i'òEja/ parece-me uma boa ideial/idéiab˘
7.3.2.9. We will now consider some examples of /0uò/ = (±ò) (the symbols for
/0eò/ are more normal, as in ('k™nt)l ('k™~Ci)b /'kente/ ¢ente]\ ('a;tî[[°])l ('a;Ci-
mu[s])b /'atimu[s]/ átimo(s)˚ ('p9å;“[°])l ('p]A;nu[s])b /'pıanu[s]/ plano(s)˚ ('bå;”[°])l ('bA;-
276 a handbook of pronunciation
Stress
7.3.3.2. ˛en there are (actual) lexical compounds, which maintain a fairly strong
stress on their first element, fluctuating between a primary and a strengthened sec-
ondary one (according to the scale ('), (“), (&)); so we have a first /'/ followed by an-
other which is realized as (“) (or (&), if adjacent to the primary stress): (“gja˜da-
'˜oUpå, -o;-)b (-Rƒ‘'ºo;p‘, -oU-)l /'gwaKòa'Ko™pa/ %arda-roupa˚ (“agja'fO˜Ci)b (“aŸj‘-
'fOÍt˙)l /'agwa'fOKte/ á%a-forte˚ (“A∞Ci'baiëu)b (“AÓtî'BaiSu)l /'aıti'baiSu/ altibaixo˚ (&mA∞-
'kRãa;du)b (&mAÓ'kÍãa;ƒu)l /'maı'kRjaòu/ malcreado˚ (“pAu~‚i']O)b (“pAund¢']O)l /'paun-
de'ıO/ pão-de-ló˚ (“]uzobRazi']eIRu)b (-øBR‘zi']ÉI-)l /'ıuzìbRazi'ıeiRu/ l¨o-brøileiro˘
˛ere are even some particular compounds, still considered as such (rather than
crystallizations), which maintain the two elements fairly independent; first of all,
we find adverbs in /-'mente/ -mente: (“òustA'm™~Ci)b (“ZuSt‘'m™nt˙)l /'Zusta'mente/ j¨-
7. portuguese 277
7.3.3.3. In addition, nouns with the diminutive infix /-z-/ -z- follow the same
pattern: (mu&L™˜'zi;Nå)b (mu&L™R'zi;N‘)l /mu'LEK'ziNa/ muºerziˆa˚ (ku&L™˜'zi;Nå)b
(k¨&L™R'zi;N‘)l /ku'LEK'ziNa/ coºerziˆa˚ (“ø˜fA«'zi;Nå)b (“øÍf4«'zi;N‘)l /'OKfan'ziNa/ or-
fãziˆa˚ (“ëavena'zi;Nå)b (“Sav¢n‘'zi;N‘)l /'Savena'ziNa/ >avenaziˆa˚ (a&vø'zi;Nå)b (‘-
&vø'zi;N‘)l /a'vO'ziNa/ avoziˆa˚ (a&vo'zi;Nu)b (‘&vo'zi;Nu)l /a'vo'ziNu/ avoziˆo˚ (&pai'zi;-
Nu)b (-'zi;Nu)l /'pai'ziNu/ paiziˆo˚ (“9meI«'zi;Nu)b (“øm’I«'zi;Nu)l /'ìmein'ziNu/ ho-
menziˆo˚ (“9meI«za'˜Au«)b (“øm’I«z‘'ºAu«)l /'ìmeinza'Kaun/ homenzarrão˚ (˜a-
&pa'ze;Lu)b (º‘&pa'zÉ;Lu)l /Ka'pa'zeLu/ rapazeºo˚ (]e«&sø∞'zi;tu)b (]e”&søı'zi;tu)l /ıen'sOı-
'zitu/ lençolzito˘ (Once, a graphic grave accent was used in such compounds.)
As to the (phonic) stressing of verbs, it is important to examin the following
examples well; they are very significant and certainly not free from serious doubts
when consideredmerely from an orthografic point of view; they are given in sim-
ple diaphonemica transcription: /li'mitu, li'mita{s}, li'mitaun/ limito, limita(s)˚
limitam; /'bailu, 'baila{s}, 'bailaun/ bailo, baila(s), bailam; /'kauzu, 'kauza{s}, 'kau-
zaun/ causo, causa(s), causam.
Besides: /aK'kwaK/ arcuar, /aK'kuu/ arcuo, /aK'kua{s}/ arcua(s); /in'flwiK/ influir,
/in'fluu/ influo, /in'flui{s}/ influi(s), /in'flwi/ influì, /in'flwiu/ influiu, /in'flwia{s}/
influìa(s); /pa'sjaK/ passear, /pa'seju/ passeio, /pa'seja{s}/ passeia(s), /pa'sejaun/ passe-
iam; /'KiK/ rir, /'Kiu/ rio, /'Kiu/ riu, /'Kia{s}/ ria(s), /'Kiein/ riem, /'Kiaun/ riam; /òi-
'zia/ dizia, /òi'Ria/ diria, /òi'Rei/ direi.
Now, let us observe very carefully the following: /sa'iK/ sair, /sa'iòu/ saìdo, /sa'in-
du/ saindo, /'sai{s}/ sai(s), /sa'i{s}/ saì(s), /'saju/ saio, /'saja{s}/ saia(s), /sa'iu/ saiu,
/sa'ia{s}/ saìa(s), /sa'iaun/ saìam, /'saein/ saem, /sa'iRaun/ saìram, /sai'Riaun/ sairiam,
/sai'Raun/ sairão, /sai'Ria{s}/ sairia(s), /sai'Rei{s}/ sairei(s), /sa'iReis/ saìreis, /sai'Rieis/
sairìeis.
Lastly, we find çinfixedÇ futures and conditionals (which are quite çoddÇ {for
foreigners or Brazilians, too}): (&se']jÉI)l /'se'ıwei/ sê-lo-ei˚ (tÍ‘&taÍ'sªa)l /tRa'taK'sja/
tratar-se-á˚ (&diR'LjaS)l /'òiK'Lwas/ dir-ºo-ás˚ (kÚn&ta']jAu«)l /kon'ta'ıwaun/ contá-lo-ão˚
(&faR'L[ã]i‘)l /'faK'Ljia/ far-ºe-ia˚ (pu&ƒeÍ's[ª]i‘)l /po'òeK'sjia/ poder-se-ia˚ (&di']ji‘S)l /'òi-
'ıwias/ di-lo-iø˚ (kÚn&ta']äi4«)l /kon'ta'ıwian/ contá-lo-iam (reading: (&se']jeI, tRa&ta˜-
'sãa, &‚i˜'Ljas, kÚn&ta']jAu«, &fa˜'L[ã]iå, po&de˜'s[ã]iå, &‚i']jiås, kÚn&ta']äiA«)b).
Intonation
7.3.3.4. û 7.4-5 show the preintonemes and intonemes of the two neutral Por-
tuguese accents. It is important to make comparisons both between them and with
those of other languages:
/./: (çt™;Nu äAçmi;gu çmuintu simçpa;Ciku3 3)b ('tÉ;Nu ä‘'mi;Ÿu Çmuintu si≈Çpa;ti-
ku3 3)l /'teNu un-a'migu 'muintu sim'patiku./ Teˆo um amigo muito simpático.
278 a handbook of pronunciation
/ / (2 2 ç 2 2 ç 2 2 ç 2) /./ (2 ç 3 3)
/¿ / (¿ 2 2 ç 2 2 ç 2 2 ç 2) /?/ (2 ¶ 1 2)
/¡ / (¡ 2 2 ç 2 2 ç 2 2 ç 2) /÷/ (2 ç 2 2)
/˚ / (˚ 2 2 ç 2 2 ç 2 2 ç 2) /,/ (2 ' 2)
£International∞ accent
7.4.3. As shown in the figure, the vowels are (i÷ e÷ 'E, »™÷ a÷ 'O, »ø÷ o÷ u) /i, e, E, a,
O, o, u/ (we still use the diaphonemic transcription as in the preceding chapters):
('vi, 'e;li, &™la'dE;vi, 'ga;ta, 'pO;su, &pøsu'daR, 'o;ku, 'u;va) /'vi, 'eıe, Eıa'òEve, 'gata, 'pOsu,
pOsu'òaK, 'oku, 'uva/ vi˚ ele˚ ela deve˚ gata˚ posso˚ posso dar˚ oco˚ uva; including their
nasalized taxophones (i, e, A, Ú, u): ('si˙, pen'denti, 'kAntu, 'kÚnta, 'mundu) /'sin,
pen'dente, 'kantu, 'konta, 'mundu/ sim˚ pendente˚ canto˚ conta˚ mundo. Of course,
the phonemes /e, E÷ O, o/ must be keep distinct, although they are ètroublesome¶
for foreiners, including the other thorny vocalic problem –metaphony– otherwise,
we would really have a èforeign¶ pronunciation. Written e˚ o, in unstressed sylla-
bles, always correspond to /e, o/ (even for esC-); while, only for final -e(s)˚ -o(s), we
have /e, u/: ('lentis, 'mu;Rus) /'ıentes, 'muRus/ lentes˚ muros˘
7.4.4. ˛e diphthongs are simpler, as well. In fact, they all have (i, u) as second
elements: (ei, Ei, ai, Oi, oi, ui÷ iu, eu, Eu, au, ou), /ei, Ei, ai, Oi, oi, ui÷ iu, eu, Eu, au,
ou, o™/: ('rei, 'vou) /'Kei, 'vo™/ rei˚ vou; with their nasalized taxophones (ei, Ai, Úi,
ui÷ Au): ('bei˙, 'mAi˙, 'pÚi˙, 'muintu, 'nAu˙) /'bein, 'main, 'poin, 'muintu, 'naun/
bem˚ mãe˚ põe˚ muito˚ não. Certainly, (Å) would not be a feasible proposition for
the nasalized diphthongs, or for nasalized /a/. ˛ey are too far away from native-
-speaker's reality (although this would certainly not prevent understanding).
On the other hand, a kind of pronunciation which is still international, but a
little less ègenuine¶, might easily renounce the nasalization caused by /ö/ in checked
280 a handbook of pronunciation
syllables – all the more so that, in many types of Lusitanian pronunciations, this
nasalization is really very reduced, so that it corresponds to the merely phonetic one
which is present in /éö/ sequences, in most languages without distinctive/phone-
mic nasalization. Normally, this slight degree of nasalization is not indicated.
7.4.5. Beside the èo‚cial¶ phonemic diphthongs, we find a new series of diph-
thongs, derived from the vocalized realizations of //él0, élò// sequences (dia-
phonemically indicated as /éı0, éıò/). By explicitly adding only the one which
does not already coincide with the èo‚cial¶ diphthongs, we have (Ou): as in
(bRa'ziu, 'feutRu, 'mEu, 'sau, 'sOu, 'soutu, 'suu) /bRa'ziı, 'feıtRu, 'mEı, 'saı, 'sOı, 'soıtu,
'suı/ Brasil˚ feltro˚ mel˚ sal˚ sol˚ solto˚ sul.
Actually, these diphthongal realizations are more convenient, for their articula-
tory simplicity and for better understanding spoken Portuguese, because they pre-
pare foreigners for the interpretation èdilemma¶ of /éı/ as (éu) (coinciding with
many phonological /éu/ diphthongs).
7.4.8. Also the approximants, more naturally, are such, (j, w) /j, w/ (not semi-
-approximants, (ã, j)): ('glO;Rja, 'li˙gwa) /'gıORja, 'ıingwa/ glória˚ lìngua˘ As to /R/,
of course, we have (R) (even in word-final position before a pause or C, where in
neutral Bazilian we find (˜) /K/). ‹ile, for /K/, a more ènatural¶ (r) is better (af-
ter C\ /n, l, s/, diaphonemically /n, ı, s/): ('Únra, &u˙ra'pas, 'biuru, ra'laR) /'onKa, un-
Ka'pas, 'biıKu, Ka'ıaK/ honra˚ um rapaz˚ bilro˚ ralar, and (Rr), after V: ('kaRru, &uRra-
'pas) /'kaKu, uKa'pas/ carro˚ o rapaz˘ ˛en, it occurs even in word-initial position;
7. portuguese 281
Text
7.5.0. ˛e transcribed passage, †e North Wind and the Sun, given in six di‡er-
ent çnormalizedÇ versions. We start with the (Brazilian and Lusitanian) neutral
pronunciations of (neutral British) English – this is the first step of the phonetic
method (the written text is given in § 2.5.2.0). ˛e Portuguese translation follows,
in its neutral Brazilian, Lusitanian, and international Portuguese versions. ˛e
variants given in brackets might need slight adjustments, on which the reader is
invited to reflect, according to the occurring segments.
In the text, for the Lusitanian accent, we indicate (å|) /’aò/, before a pause (even
though in the isolated examples given we simply put (‘ò), in order to better show
the di‡erence with the Brazialian accent).
At the end, as usual, the version showing the English pronunciation of Portu-
guese is given; it represents the habits of neutral British speakers, who are fluent
in Portuguese (after prolonged contact with native speakers, but with no help
from the phonetic method), who have adequately learned the relative prominenc-
es, but who substantially use segmental and suprasegmental elements which are
typical of neutral British English (although, of course, a neutral accent is not so
common). Obviously, the same principle is valid for the foreign pronunciations
of English, given first.
Speakers of American English could prepare their own version both of the Por-
tuguese pronunciation of English and of their pronunciation of Portuguese, as an
excellent exercise, by listening to native speakers, best of all after recording them.
Of course, speakers of other languages could do the same thing. ˛e author would
be happy to receive their transcriptions and recordings, both in case of help
–should they need it– and to make their contribution known to others (possibly
in our website on canIPA Natural Phonetics – Ô § 0.12).
Portuguese text
7.5.3. O vento norte e o sol porfiavam sobre ¢al dos do¤ era o ma¤ forte, ¢an-
do sucedeu pøsar um viajante envolto numa capa. Ao vê-lo, põem-se de acordo em co-
mo a¢ele ¢e primeiro conse%¤se obrigar o viajante a tirar a capa seria considera-
do o ma¤ forte.
O vento norte começou a soprar com muita fùria, mø ¢anto ma¤ soprava, ma¤
o viajante se embruºava na sua capa, até ¢e o vento norte dπ¤tiu. O sol briºou
então com todo o πplendor, e imediatamente o viajante tirou a capa. E øsim o ven-
to norte teve de recoˆecer a superioridade do sol.
Você gostou da h¤toriˆa? Queremos repetì-la?
Brazilian pronunciation
Lusitanian pronunciation
International pronunciation
8.0.2. At the beginning, we think it useful to show the Cyrillic alphabet, both
roman and italic. ˛e latter especially, since the shape of some letters can be mis-
leading at first, by comparison with the Roman (or Latin) ones typical of certain
kinds of transliteration, which are given in brackets. It is not recommendable to
use the ambiguous ch for /x/, since kh is decidedly more appropriate, while keep-
ing ch for /c/ (in case the most satisfying choice of using £ /c/ and x /x/ is not
made).
For Cyrillic writing, it is important to pay particular attention to the roman >b,
e, n, p, c, y, x≥, and the italic >e, $, π, p, c, t, y, x≥.
Vowels
8.1.2. In stressed syllables, the six vowel phonemes are realized as the black
markers shown in û 8.1. Obviously, those with a white center also correspond to
certain unstressed contexts that we will see later on: (t'çi;) /t'ri/ tp$, ('F™;Rx) /'era/
<pa, ('kFak) /'kak/ kak, ('tjøt) /'tot/ tot, ('jum) /'um/ ym, ('bF…t) /'b…t/ ∫vt; be-
sides, we also have: ('qF™∆) /'qex/ zex, ('SFak) /'Sak/ wag, ('qjuk) /'quk/ µyk˘
As some examples have already shown, there are some interesting di‡erences in
our phonetic transcriptions in comparison with phonemic ones, apart from sym-
bol variations such as (™) /e/, (ø) /o/. ‹ile (i, I) /i/ remain unchanged (except in
the sequences (ãi, ãI) /’ji/, which represent >$≥ after V˚ and >e≥ in unstressed sylla-
bles), the most important element is instead the prevelar semi-approximant (F), oc-
curring before stressed /e, a, …/, and the velar rounded semi-approximant (j), be-
fore stressed /o, u/ (all of them preceded or not by a di‡erent C from /j, c, SS, ZZ/).
˛erefore, the V which are di‡erent from /i/ are realized as sequences of (F) + (™, a,
…), or (j) + (ø, u); in particular this happens in stressed syllables, even in com-
pounds and in sentences where distinct timbres occur; that is with /’e, ’o/ too.
Obviously, (F, j) are less çevidentÇ than fully prevelar, or velar rounded, approx-
imants, (j, w); (w) occurs in English ('w¤T, 'khw¤T, 'wø:{<}) /'wIt, 'kwIt, 'wO:≤/ wit,
quit˚ war. It is to be noted that (F, j) are quite di‡erent from a phonic zero, (`)
(although naïve Russian speakers would swear they cannot hear anything but (`)),
as in English ('¤T, 'kh¤T, 'ø:{<}) /'It, 'kIt, 'O:{≤}/ it˚ kit˚ oar). However, in unstressed syl-
lables, the approximants (F, j) do not occur, as will be clear from some examples
given below.
It is important to explain from the very beginning the palatalizing function of
the phonemes /i, j/ on the C which precede them. ˛us we have to pre-empt, here,
the triple division of the Russian C. We find çnormalÇ C, which are palatalizable\
/m, n÷ r÷ l/ (m, m÷ n, ~÷ R, ç÷ ı, ¬), /p, b÷ t, d÷ k, g/ (p, p÷ b, b÷ t, ⁄÷ d, Á÷ k, ©÷ g, á),
/f, v÷ s, z÷ x/ (f, f÷ v, v÷ s, À÷ z, =÷ ∆, â). ˛en there are the palatal ones, in the gener-
al sense that they have a real çpalatalÇ component, in spite of their actual realiza-
tions: /j, c, S[S], ZZ/ (j, ã÷ C[ã], ‚÷ ë[ë], ëë[ã]÷ òò). ≈nally, we have non-palatal C\ /q÷ S,
q/ (q÷ S, q), which are never palatalized (although spelling sequences like >z$, we,
µì≥ exist: (S'ju;) /S'ju/ wìù). We must always keep in mind, then, that */qi÷ Si, qi/
do not occur (any longer) but are substituted by /q…÷ S…, q…/, in spite of historical
writings with >z, w, µ≥ followed by >$, e, r, ù, ì≥˘
8.1.3. After all, /i, j/ (and /c÷ S, Z/) would be su‚cient to adequately explain
how the Russian phonemic system works, which is so typical because of palatali-
288 a handbook of pronunciation
8.1.4. ˛e four vowel phonemes /e, a, o, u/ may occur between Ç˚ and are real-
ized closer — fronter, as indicated by the grey markers, (e, Å, P, ¯) (again in û 8.1):
('je1) /'jelj/ elì, ('pÅ⁄) /'pjatj/ πrtì, ('⁄P;⁄x) /'tjotja/ tetr, (⁄¯'fak) /tju'fjak/ tù-
frk˘
˛e advancement and raising of /[0]iJ/ is decidedly more limited, (iï) (and, in
the vocogram, there is no special marker di‡erent from the one for normal (i), so
as not to render the figure too heavy): ('bi⁄) /'bitj/ ∫$tì˘
On the contrary, /…/ does not occur in this context; and this phenomenon has
cultivated the myth of the complementarity of (i, …). So they are grouped in the
same phoneme, /i/, under the delusion that a çbetterÇ phonological system has
been found – while it is exactly the contrary. çDeviantÇ schooling is hard to re-
8. russian 289
move, especially in regard to o‚cial writing (with its known and less known prob-
lems) and çrespellingÇ, which is even used in schools in addition to in çscientificÇ
works, with phonemic ambitions. ˛ey would like to find –almost alchemist's–
solutions, by using complex formulae and esoteric iconograms (which we will be
careful not to reproduce).
In û 8.1, three çpotentialÇ phonetic diphthongs, (™Ä, aå, ø∏), are also given,
which can substitute (™;, a;, ø;) (with an çinterpalatalÇ variant, (ÅÄ), for (Å;)). ˛ey
occur mainly in intonemes, in unchecked (final or internal) syllables: (å'À™;, å'À™Ä÷
-ø;, -ø∏) /f'sje÷ -jo/ bce, ('F™;∆x, 'F™Ä∆x) /'exa/ <xo, ('dFa;, 'dFaå) /'da/ da, ('Fa;tx≈, 'Faå-
tx≈) /'atam/ atom, ('ÁÅ;Áx, 'ÁÅÄÁx) /'djadja/ drdr, (d'njø;, d'njø∏) /d'no/ dno,
('gjø;ıxs, 'gjø∏ıxs) /'golas/ goloc˘ ˛ey also occur for emphasis, in which case they
are longer; thus we have in unchecked syllables (™;Ä÷ a;å, Å;Ä÷ ø;∏): (å"À™;Ä÷ -ø;∏)
/f'sje÷ -jo/ bce, ("F™;Ä∆x) /'exa/ <xo, ("dFa;å) /'da/ da, ("Fa;åtx≈) /'atam/ atom, ("ÁÅ;Ä-
Áx) /'djadja/ drdr, (d"njø;∏) /d'no/ dno, ("gjø;∏ıxs) /'golas/ goloc; and in checked
syllables (™Ä÷ aå, ÅÄ÷ ø∏): ("~™Ät) /"njet/ net!, ("qFaå1) /"qalj/ µalì!, ("pÅÄ⁄) /"pjatj/
πrtì!, ("djø∏≈) /"dom/ dom!
8.1.5. Russian also has some ço‚cialÇ diphthongs, all with a front second ele-
ment /éi/ (éi). Here we will consider those occurring in stressed syllables: ('©ii)
/'kii/ k$=, ('pei⁄I) /'pjeiti/ πe=te, ('mFai) /'mai/ ma=, ('CãÅi) /'cai/ ua=, ('bjøi)
/'boi/ ∫o=, (zå'çPi) /za'rjoi/ ‘ape=, ('djui⁄I) /'duiti/ dy=te, ('Á¯i≈) /'djuim/ dù=m,
('vF…idu) /'v…idu/ bv=dy. As can be seen, the orthography uses >=≥ for diphthongs
çconsecratedÇ by grammar. However, mainly in loanwords, others occur, as >ay≥:
('fFaust) /'faust/ Fayct˘
In unstressed syllables, we have (again with (éi)): ('ju;¬ii) /'uljii/ yle=, (Cii'kjøå-
s©ii) /cii'kofskii/ Ua=kobck$=, (tåi'gFa;) /tai'ga/ ta=ga, ('Á™;ıxi⁄I) /'djelaiti/ de-
la=te, (ui'dju;) /ui'du/ y=dy, (Á¯i'mjøåkx) /djui'mofka/ dù=mobka, ('b™;ı¢i)
/'bjel…i/ ∫elv=˘
Instead, purely graphemic çdiphthongsÇ like >$$˚ $r˚ e$˚ ee˚ ae˚ ar˚ aù˚ oe˚ ve≥
do not correspond at all to phono-diphthongs, except when the semi-approximant
(which is shown in round brackets, (é[ã]é)) falls completely. In this case, howev-
er, if the last vocalic element is /i/, the phone (I) is maintained (instead of having
(i), as in phonemic diphthongs, unless it is followed –in sentences– by (J) or /i/):
(vI'tFa;¬i[ã]I, v¢-) /vi'taliji, v…-/ b *tal$$, (vRåÀ'Ài[;ã]I) /vras'siji/ b Pocc$$, ('¬i;¬i[ã]x) /'lil-
ija/ l$l$r, (mu'=e[;ã]I) /mu'zjeji/ my‘e$/-ee (Ô (mu'=ei) /mu'zjei/ my‘e=], (å'Àe~~i[ã]I)
/a'sjenniji/ ocennee, (6'ma[;ã]I) /v'maji/ b mae, (&~IRu'SF…;mxãx) /niru'S…maja/ nepyw$-
mar, (z'nFa;ãu) /z'naju/ ‘naù, ('njø;vx[ã]I) /'novaji/ noboe˚ ('juqn¢[ã]I) /'juqn…ji/ ùµ-
nve˘
˛e following examples are bisyllabic: (på'juk) /pa'uk/ πayk, (må'ãi;) /ma'ji/
mo$, (på'F™t) /pa'et/ πo<t, (du'F™1) /du'elj/ dy<lì˘
out a pause), (å', ò’å, ’x) /a/, (u) /u/, (¢', ’¢) /…/: (CI'sF…;) /ci's…/ uacv, (pi'⁄i;) /pi'ti/ πr-
t$, (&ãIz¢'kFa;) /jiz…'ka/ r‘vka, (å'bF…;Cii) /a'b…cii/ o∫vua=, (&jIRå'ıFaS) /jira'laS/ epa-
law, (&piçimi'¬i;) /pirimi'li/ πepemel$, (&I¬Ik'tçi;CIstvx) /ilik'tricistva/ <lektp$uect-
bo, ('pjø;¬I) /'poli/ πole÷ (&gxıå'vFa;) /gala'va/ goloba, (&åvå˙'gFaÍt, -n'g-) /avan'gart/
abangapd, (k'RFasnxãx) /k'rasnaja/ kpacnar÷ (u'Rjøk) /u'rok/ ypok, ('djø;mu) /'do-
mu/ domy, (Àu'dFa;) /sju'da/ cùda, (À¯'Àu;kx⁄) /sju'sjukatj/ cùcùkatì÷ (q¢&vi¬I'zFa;-
q¢ãx) /q…vili'zats…ja/ z$b$l$‘az$r, (q¢'nFa;) /q…'na/ µena, ('dF…;R¢) /'d…r…/ dvpv˘
For >r≥ in unstressed syllables (and for >Ça≥ too), it is important to explain that
neutral pronunciation has only /i/, whereas a transversal use of (‘, È) (or even (x,
å) /’a/) is very widespread, although decidedly non-neutral. ˛is occurs both in tra-
ditional and alternative pronunciation (and indeed even for speakers who are neu-
tral, one way or another!). ˛is can occur both within lexemes, and in inflected
forms with >r+C≥ consonantal endings (influenced by the forms ending in -r, with
/’ja/, (Jx, ãx)).
˛us we have: (pIt'njø;, ≠p‘t-, ≠påt-) /pit'no/ πrtno, (jI'zF…k, ≠j‘-, ≠jx-) /ji'z…k/ r‘-
vk, (&jIz¢'©i;, ≠&j‘-, ≠&jx-) /jiz…'ki/ r‘vk$, (&pxg¬i'Á™Ó, ≠-¬È-, ≠-¬å-) /pagli'djel/ πoglrdel,
('nFa;CIt, ≠-C‘t, ≠-Cxt) /'nacit/ nauat, (å'bF…;Cii, ≠-C‘i, ≠-Cxi) /a'b…cii/ o∫vua=,
('kjøRmIt, ≠-m‘t, ≠-mxt) /'kormit/ kopmrt, ('Áe;⁄I≈, ≠-⁄‘≈, ≠-⁄x≈) /'djetim/ de-
trm, ('zFa;~IÓ, ≠-‘Ó, ≠-xÓ) /'zanil/ ‘anrl, (&bIspi'⁄i 'pÅ⁄, ≠&bÙspå'⁄i 'pÄ⁄, ≠-pÈ'⁄i) /bis-
pi'ti 'pjatj/ ∫e‘ πrt$ πrtì. (Please note the variants of -$k, as well: (s'pju+~Ik,
≠-~‘k, ≠-~xk) /s'putnik/ cπytn$k.)
çNormalÇ traditional pronunciation, for (pretonic) >e', r', Ça'≥ (¤ occurring in
syllables immediately before the stressed or tonic one in a intoneme) has ç/’e/Ç (É',
Ù'+J), which correspond to modern pronunciation /’i/ (I, iJ): (&bIspÙ'⁄i 'pÄ⁄)t, as
also happens in: (bi'çø;zx)m (bÙ-)t /bi'rjoza/ ∫epe‘a.
In vowel sequences which contain /’a/, we generally find an extension of the use
of (å): (&vååb-Rå'qFa⁄) /vaabra'qatj/ boo∫paµatì, (nåu'gFat) /nau'gat/ naygad, (&pååd-
nå'mju;) /paadna'mu/ πo odnomy, (&nååstRå'vFa∆) /naastra'vax/ na octpobax, (u&åd-
nå'vjø;) /uadna'vo/ y odnogo, (såu'Cãas[t]vxvxÓ) /sau'cas[t]vaval/ coyuactbobal,
(sååt'v™s[t]vxvx⁄) /saat'vjes[t]vavatj/ cootbetctbobatì.
Consonants
8.2. ˛e phonemic structuring we prefer for Russian only has twenty-two conso-
nant phonemes (including /ZZ/), which combine with the six vowels, as we have al-
ready seen in the previous sections. Here we will systematically consider Russian
C's according to their manners of articulation.
Let us observe that no voiced diphonic C occurs before a pause or a voiceless C:
('bjøp) /'bop/ ∫o∫, ('mjøsk) /'mosk/ mo‘g, (t'Rjup©I) /t'rupki/ tpy∫k$, (åså'dju;)
/fsa'du/ b cady. In addition, no voiceless C occurs before a diphonic voiced C (/v/
is a bit particular, Ô § 8.3.1.7): (z'bF…⁄) /z'b…tj/ c∫vtì, (zgå'RF…;) /zga'r…/ c gopv.
≈nally, in general, son(or)ants are partially devoiced before either voiceless C
or pauses; usually, between a voiceless C and a pause, they become voiceless, un-
less they become intense, in which case they are voiced or half-voiced (according
8. russian 291
to their phonic context): ('qF…Í) /'q…r/ µ$p, (Í'tFa;) /r'ta/ pta, ('kFaÍtx) /'karta/ kap-
ta, ('bjøbÍ, -;b“) /'bobr/ ∫o∫p, (_'~™p5, -;p”) /d'njepr/ Dneπp.
˛e table in û 8.2 shows the consonantal articulations of Russian, which are
necessary for an appropriate pronunciation of this language.
Instead, û 1.9-15 show the orograms of all the contoids treated in the chapters
of this book, by grouping them according to manners of articulation, including
secondary, occasional, or regional variants, for the 12 languages dealt with. ˛is
way of displaying them makes the necessary comparisons between di‡erent lan-
guages more straightforward.
protruded (non-pr.*)
velariz. postalveo-pal.
velarized alveolar
postalveo-palatal
velar rounded
labiodental
labio-palatal
labiodental
palatalized
prepalatal
laryngeal
prevelar
alveolar
bilabial
palatal
dental
velar
ö m (m) (M) (“) (n) n (~) (2)* (~) (N) (˙)
F p b (p b) t d (+ _) (© á) k g
Ô q (Q) C (‚) (⁄ Á)
ƒ (å 6) f v (f v)
_ s z ë {ò} S q (À =) (,|J) (y)
ß (â) j|(ã) (F) (∆) (j) {H}
ó R (ç)
‹ ı (¬) (L)
/c÷ S, Z÷ x÷ r÷ l/ (C÷ ë, ò÷ ∆÷ R÷ ı)÷ /m, n÷ r÷ l/+(), £÷ 5÷ ñ, Ñ) + ((º, ª, ¸, √, n))
Nasals
8.2.1. Russian has only two nasal phonemes, /m, n/, with various taxophones,
especially for the latter. ˛us: /m/ (m, m, M, “) and /n/ (m, m, M, “, n, ~, 2, N,
˙) (more specific symbols like ((˙, n, ˝)) are not necessary, but –as we will shortly
see– (º, ¸, n÷ ª, √) can be used, for a less-assimilated kind of coarticulation,
which maintains their front lingual contact –alveolar/prepalatal– while adding a
secondary articulation with no contact: bilabial, labiodental, velar).
For /m/: ('mFa;mx) /'mama/ mama, ('ıFa≈px) /'lampa/ lamπa, ('njøR≈) /'norm/
nopm, ('çit), 'çi;tM) /'ritm/ p$tm, ('çiå), 'çi;åM) /'rifm/ p$fm, (m'njøi) /m'noi/
mno=, (≈'∆Fa;) /m'xa/ mxa, (mg'ıFa;) /mg'la/ mgla, ('mik) /'mik/ m$g, ('mÅ⁄)
/'mjatj/ mrtì, ('Àe)) /'sjemj/ cemì, (Im'biÌ) /im'birj/ $m∫$pì, (åM'vjøn) /am'von/
ambon, ('FaõfxRx) /'amfara/ amfopa. /m/ can occur befor heterorganic C: (m'gıFa;)
/m'gla/ mgla, (m'njøi) /m'noi/ mno=, (≈'∆Fa;) /m'xa/ mxa; however, for /mj/, and
/m/ + /i/, we regularly have (m): ('miÍ) /'mir/ m$p, ('Àe)) /'sjemj/ cemì.
For /n/: (&jømbi'çøt, &jøª-) /ombi'rjot, on-/ on ∫epet˚ (kåM'vFa;, -¸'v-) /kan'va/
kanba, (åõ'fFas, å∞-) /an'fas/ anfac, ('njøs) /'nos/ noc, (n'RFaf) /n'raf/ npab, (kå-
'mFa~ÁI) /ka'mandi/ komande, ('~i⁄) /'nitj/ n$tì, ('kjø8) /'konj/ konì˚ (bi~'=in)
292 a handbook of pronunciation
Stops
8.2.2. ˛ere are three diphonic pairs which do not oppose before a pause or a
voiceless C: /p, b÷ t, d÷ k, g/ (p, p, b, b÷ t, ⁄, +, d, Á, _÷ k, ©, g, á). ˛e actual articu-
lation of çpalatalizedÇ /t, d/ is prepalatal which are stopstrictives, but stops before ho-
morganic (or similar) C. Sometimes, we find (+) before pauses as well): ('pjøt) /'pot/
πot, ('ıjøp) /'lop/ lo∫, ('peC) /'pjec/ πeuì, ('tjøp) /'topj/ toπì, ('gjø;ıup) /'golupj/
goly∫ì, ('bF…t) /'b…t/ ∫vt, (å'bF™;tx≈) /a'betam/ o∫ <tom, ('bi⁄) /'bitj/ ∫$tì÷ (t'Rjut)
/t'rut/ tpyt, (åq'qFa;) /at'qa/ otza, (t'kju;) /t'ku/ tky, ('vit) /'vit/ b$d, (⁄'mFa;)
/tj'ma/ tìma, (⁄'fju;) /tj'fu/ tìfy!, (pi+'¬a;, pIt-, 'pe+-¬x) /pit'lja, 'pjetlja/ πetlr, (Rå-
'bjø+~Ik) /ra'botnik/ pa∫otn$k, (åC'Cãøt) /at'cot/ otuet, ('djup) /'dup/ dy∫,
('tjødq¢) /'todq…/ tot µe, ('Áe8) /'djenj/ denì, (_'~a;) /d'nja/ dnr, ('pjø_¬I) /'podli/
πodle, (d'veÌ, Á'-) /d'vjerj/ dbepì, ('kFak) /'kak/ kak, ('pFakt) /'pakt/ πakt, ('¬øk)
/'ljok/ leg, ('kF™;txmu) /'ketamu/ k <tomy, (©Iås'©jøÍ) /kias'kjor/ k$ockep, ('áip©ii)
/'gipkii/ g$∫k$=, ('gjøt) /'got/ god, (g'nju⁄) /g'nutj/ gnytì, (nå'ái;) /na'gi/ nog$˘
Stopstrictives
8.2.3. Russian has two voiceless stopstrictive phonemes, /q, c/, but four reali-
zations, by voicing assimilation (in addition to the palatalized variants of /t, d/,
(⁄, Á), which we have already seen in § 8.2.2). Besides, before a stressed V –di‡er-
ent from /i/– /c/ is (Cã): ('qF…kñ, -%) /'q…kl/ z$kl, (å'⁄™q) /a'tjeq/ otez, (b'RFaqqx)
/b'raqqa/ ∫patza, (pıåQ'dFaR≈) /plaq'darm/ πlazdapm÷ (u'Ci;⁄i1) /u'citilj/ yu$-
telì, ('Cãas) /'cas/ uac, (p¬i'Cãø;) /pli'co/ πleuo, ('ıjuC) /'luc/ lyu, (C'¬™n) /c'ljen/
ulen, (nå‚'Áif) /nac'dif/ naud$b˘
8. russian 293
Constrictives
8.2.4. For the time being, let us introduce the three constrictive diphonic pairs,
(f, f, å÷ v, v, 6) /f, v/, (s, À÷ z, =) /s, z/, (S, q) /S, q/; apart from the palatalized taxo-
phones (of the first two pairs, but not of the last one), the labiodentals have a bila-
bial realization when they do not precede either /j, r, l/ or V (including the voiced
approximants –(F, j)– which are added, from a phonetic point of view, in front of
stressed vocoids): ('fjøn) /'fon/ fon, ('Rjøf) /'rof/ pob, (&ååtå'mFat) /afta'mat/ abto-
mat, (f'fi¬mI) /f'filjmi/ b f$lìme, ('v™Íf) /'vjerfj/ bepfì, ('vjøÓk) /'volk/ bolk,
(S'kFa6q¢) /S'kavq…/ wkaf µe, ('sjøn) /'son/ con, (Rå'sFa;) /ra'sa/ poca.
More examples: (À'~™k) /s'njek/ cneg, ('jøÀ) /'osj/ ocì, ('zjup) /'zup/ ‘y∫, (z'ıjøi)
/z'loi/ ‘lo=, (z'b™;ı¢≈) /z'bjel…m/ c ∫elvm, (I'=u≈) /i'zjum/ $‘ùm, (&bi=Ái'⁄ei) /biz-
di'tjei/ ∫e‘ dete=÷ ('SF™À⁄) /'Sestj/ wectì, ('ıjøS) /'loS/ loµì, ('qF™st) /'qest/ µect,
(q'dFa⁄) /q'datj/ µdatì, (∆å'qju;) /xa'qu/ xoµy˘
Lastly, there is another –quite particular– voiceless constrictive, (ë[ë]) /S[S]/, which
is lengthened before /é/, and has a traditional variant constituted by the corre-
sponding constrictive followed by the (sometimes very weak) stopstrictive, (ëC).
˛is is more complicated and not at all necessary in modern pronunciation (in-
deed, as we have already said, it more aptly belongs to traditional pronunciation).
Before a stressed V (when di‡erent from /i/), a (ã) is inserted: (ë'ëit)m (ë'Cit)t /S'Sit/
∑$t, (påë'ëãa;dx)m (påëC'ã-)t /paS'Sada/ πo∑ada, ('bjøÍë)m (-ëC)t /'borS/ ∫op∑.
˛e possible sequence (ëC) /Sc/ is di‡erent –but quite normal indeed– as in:
(ë'ëã™≈÷ ë'Cã™≈) /S'Sem/ c uem (Ô § 8.3.2.5).
Approximants
8.2.5.1. From a phonemic point of view, Russian has two approximants. Palatal
/j/, which is realized as fully palatal either before a stressed V, ('jé), or in initial posi-
tion in a rhythm group before a (stressed or unstressed) V, (òjé). Instead, /j/ is real-
ized as a palatal semi-approximant after a (stressed or unstressed) V before anoth-
er unstressed V\ (é’ãé) (and also in (é'ãi), for /é'ji/ >V$≥\ (må'ãi;) /ma'ji/ mo$]÷ after
C˚ it changes the sequence into a çpalatalizedÇ C, /0j/ (J) (Ô § 8.1.2-3): ('jÅ⁄) /'jatj/
rtì, (ji'vi⁄) /ji'vitj/ rb$tì, (&mxãI'kjøås©ii) /maji'kofskii/ Markobck$=, (bå1'SFa;-
ãx) /balj'Saja/ ∫olìwar˘
For emphasis, or precision, we can have (j) = (,) (semi-constrictive) and, respec-
tively, (ã) = (j). However, it is more important to note that, in non-slow speech,
the /ijé, éji/ sequences are realized as (ié, éi): ('FaRmiãx, -mix) /'armija/ apm$r,
(må'ãi;, må'i;) /ma'ji/ mo$˘ Sometimes, even in /éjé/ (éãé) sequences (without /i/),
it is not easy to adequately perceive (ã), especially in fast speech: (bå1'SFa;ãx, -a[;ã]x)
(above all in preintonemes).
tionaries and teaching texts): ('qF™∆) /'qex/ zex, ('SjøÓk) /'Solk/ welk, (&pxRå'Sjut)
/para'Sut/ πapawùt, ('qF…Ík) /'q…rk/ z$pk, ('qF…⁄) /'q…tj/ µ$tì, ('ıjøS) /'loS/ loµì˘
However, they can be followed by /j/ (j) inserted before the V (and in writing
>ì≥ appears too): (S'jøt) /S'jot/ wìet, (S'ju;) /S'ju/ wìù, ('ıjøqju) /'loqju/ loµìù˘
Besides, the palatal approximant can occur after (J), too, then it is phonemical-
ly geminated /jj/, and may have a distinctive value: (s¢ç'jø;) /s…rj'jo/ cvpìe, (p'ju;)
/pj'ju/ πìù, (Á'ja;vxÓ) /dj'javal/ dìrbol, (v'jut) /vj'jut/ bìùt, (IL'jiC) /ilj'jic/ *lì$u˘
Even /sjj, zjj/ can occur, but indicated with >œ≥ (although not always, as in cœe-
µ$tìcr, given at the end of this section): (À'j™;∆x⁄) /sj'jexatj/ cœexatì, (I='jÅ⁄) /izj-
'jatj/ $‘œrtì˘
˛e same occurs for /Jj/: (C'ja;) /c'ja/ uìr (Ô ('Cãat) /'cat/ uad, where the di‡er-
ence is clear also because of the di‡erent syllabic structure). ˛e /ljj, njj/ sequences
are realized as (Lj, Nj) (or even (¬j, ~j), which are already su‚ciently di‡erent):
(L'jøt) /lj'jot/ çs/he poursÇ lìet (Ô ('¬øt) /'ljot/ çiceÇ led or çflightÇ let]˚ (L'ju;)
/lj'ju/ lìù, (vRåN'jø;) /vranj'jo/ bpanìe, (sviN'ja;) /svinj'ja/ cb$nìr˘
˛ere are even cases where a (0) remains separated and uninfluenced by a fol-
lowing /j/; thus, if no stress sign is already present, it is necessary and su‚cient to
introduce a hyphen in the phonemic transcription (as the orthography uses >ϳ):
(åb'j™;∆x⁄) /ab'jexatj/ o∫œexatì, (s'jø;q¢qx) /s'joq…qa/ cœeµ$tìcr, (d&vu∆'ja;Rusn¢i)
/d“vux'jarusn…i/ dbyxœrpycnv=, (t“RansjI6Rå'peis©ii) /t“rans-jivra'pjeiskii/ tpancœe-
bpoπe=ck$=, (&m™q'ja;Rusn¢i) /“mjeq'jarusn…i/ meµœrpycnv=˘
£Trills∞
8.2.6. In the Russian phonemic system we find only one çtrillÇ phoneme, which
is realized either as an alveolar, (R), or a prepalatal, (ç), tap – according to contexts.
For emphasis, or speaking in a slow and clear way, as in teaching, it is possible to
hear the corresponding trills, (r, D), which may be used in these situations. But,
for a good pronunciation, these are not necessary; instead, it is more important to
respect the normal degrees of devoicing (Ô § 8.3.1.7): ('RFat) /'rat/ pad, ('pFaÍk)
/'park/ πapk, ('pøt5, -;t”) /'pjotr/ ∏etp, ('çis) /'ris/ p$c, ('çat) /'rjat/ prd, (få'nFaÌ) /fa-
'narj/ fonapì, (Ài~'⁄abÌ, -;bh) /sin'tjabrj/ centr∫pì, ('vi∆[, -;∆·) /'vixrj/ b$xpì, ('jøb-
Rxs) /'obras/ o∫pa‘, ('Fad-çIs) /'adris/ adpec˘
8. russian 295
Laterals
8.2.7. ˛ere is only one lateral phoneme, /l/, with three taxophones, (ı, ¬, L)
(independently of devoicing – Ô § 8.3.1.7); (ı) is a velarized alveolar, but it be-
comes (velarized) dental (which is possible to transcribe with ((Ï))), by assimilation
before /t, d÷ q÷ s, z/: ('ıF…;q¢) /'l…q…/ lvµ$, (s'ıjø;vx) /s'lova/ clobo, ('djuÓ) /'dul/ dyl,
('igÓ, 'i;g®) /'igl/ $gl, (s'mF…sñ, -;s∆) /s'm…sl/ cmvcl, ('¬i;qx) /'liqa/ l$za, ('qFa1) /'qalj/
µalì, ('tjø1kx) /'toljka/ tolìko, ('pFa¬mx) /'paljma/ πalìma, ('Rjub1, -;bˆ) /'rublj/
py∫lì, ('mF…ÀÑ, -;À∆) /'m…slj/ mvclì, (&pjøı'¬it-Rx) /“pol'litra/ πol-l$tpa (here, lexical
compounding shows how it prevails against phonology, since /l+l/ remain separat-
ed, (ı¬), instead of fusing into (¬¬)), (L'jut) /lj'jut/ lìùt˘ Some speakers can have a
velarized dental (or denti-alveolar) (ı) in any case, instead of velarized alveolar, not
only before dental articulations.
Structures
8.3. In these sections we will deal above all with the di‡erences between the
phonic structure and writing, which inevitably presents çsurprisesÇ, since pronun-
ciation changes, whereas orthography remains unchanging, except for o‚cial re-
forms (which however are always insu‚cient and partial, as those of Peter the
Great {1708-1710}, the Academy of Sciences {1735, 1738, 1758}, and Lenin {1917}
– ('¬e;~In) /'ljenin/ Len$n). We will treat some di‡erent distributions of certain
phonemes, in the formation of Russian words.
Taxophonics
8.3.1.1. Let us start by contrasting some (similar, but not identical) structures,
which foreigners are likely to confuse, thus compromising what they say.
Russian grammar considers the division into grapho-syllables to be quite flexi-
ble, for instance: /sis'tra/ (ÀIs'tRFa;) ce-ctpa˚ cec-tpa˚ cect-pa˘ ˛us, it is quite sur-
prising that for the division into phono-syllables, instead, the prevailing approach
perseveres in proposing (or imposing) */si'stra/, Ô § 8.0.3).
In the case of graphic geminates, for instance, we have: An-na /'anna/, but pho-
netically ('Fannx).
/'kjaxta/ Krxta, (t'©jøt) /t'kjot/ tket, (ås'kjø;çI) /fs'kori/ bckope, (å⁄I'nu⁄) /fti-
'nutj/ btrnytì, (&påÍt'fe¬) /part'fjelj/ πoptfelì, (v'vøÓ) /v'vjol/ bbel, (v'jøt) /vj'jot/
bìet, ('©ju;gu, 'kju-) /'kjugu/ k ùgy, (k'~i;gx) /k'niga/ kn$ga, (6z'g¬at) /vz'gljat/
b‘glrd˘ In traditional (and traditional-like) pronunciation, the palatalization of
/k, g/ also spreads to cases such as the last two: (k'~i;gx)m (©'~i-)t /k'niga/ kn$ga,
(6z'g¬at)m (-á¬at)t /vz'gljat/ b‘glrd˘
More examples: (s'RFa;zu) /s'razu/ cpa‘y, (s'pRFa;vx) /s'prava/ cπpaba, ('zFaåtRx)
/'zaftra/ ‘abtpa, (zRå'Cãøk) /zra'cok/ ‘pauok, (å'ÀÅ;©ii) /f'sjakii/ bcrk$=, (À'j™Ó) /sj'-
jel/ cœel, (6'=Å⁄) /v'zjatj/ b‘rtì, (='¬iqqx) /z'liqqa/ ‘l$tìcr, (='Á™;ıx⁄) /z'djelatj/
cdelatì, (dRu='ja;) /druzj'ja/ dpy‘ìr, (S'ju;) /S'ju/ wìù, (&åtSå'gFa⁄) /atSa'gatj/ otwa-
gatì, (q'dFa⁄) /q'datj/ µdatì, (q'Áøt) /q'djot/ µdet, (q'qFa⁄) /q'qatj/ cµatì,
(C'Cã™tnx) /c'cetna/ t∑etno, (∆'¬PëëI) /x'ljoSSi/ xlectue, (IS'SjøÓkx) /iS'Solka/ $‘
welka, (À⁄i'âi;) /sti'xi/ ct$x$, (∆'vjøst) /x'vost/ xboct, (Rås'qv™t) /ras'qvjet/ pacz-
bet, (C'¬™n) /c'ljen/ ulen, (kåp'Cãø;n¢i) /kap'con…i/ koπuenv=, (CçIz'm™Rnx) /criz-
'mjerna/ upe‘mepno, (&åCëI'vjø;) /atci'vo/ otuego, (⁄'mFa;) /tj'ma/ tìma˘
Some further examples: (m'gıFa;) /m'gla/ mgla, (m'njøi) /m'noi/ mno=, (m'ıFat-
S¢i) /m'latS…i/ mladw$=, (m'~e;~i[ã]I) /m'njeniji/ mnen$e, (≈'ëãe;~i[ã]I) /m'Seniji/
m∑en$e, (m'RFa;mxÍ) /m'ramar/ mpamop, (≈'Cãaqqx) /m'caqqa/ muatìcr, (çIå-
'¬iz≈, -;zõ) /ria'lizm/ peal$‘m, (ÁIåf'RFag≈, -;gõ) /diaf'ragm/ d$afpagm, ('veÁ≈,
-;Áõ) /'vjedjm/ bedìm, (d'RFa∆), -;∆M) /d'raxm/ dpaxm, ('me;ÀIq) /'mjesiq/ mecrz,
(å'pçÅm) /f'prjamj/ bπprmì, ('kFa=8, -;=«) /'kaznj/ ka‘nì, ('peÀ}, -;À¯) /'pjesnj/ πec-
nì, (ı'bFa;) /l'ba/ l∫a, (ı'gFa⁄) /l'gatj/ lgatì, ('igÓ, 'i;g®) /'igl/ $gl, ('RF…∆ñ, -;∆%) /'r…xl/
pvxl, (b'¬™dn¢i) /b'ljedn…i/ ∫lednv=, (S'¬a;px) /S'ljapa/ wlrπa, (Í'tFa;) /r'ta/ pta,
(R'qF…;) /r'q…/ pµ$, (R've;~i[ã]I) /r'vjeniji/ pben$e, ('~™gÍ, -;g“) /'njegr/ negp, (åpi'çøt)
/fpi'rjot/ bπeped, (ç'ja;nxÀ⁄) /rj'janastj/ pìrnoctì, (6'njut[, -;t·) /v'nutrj/ bnytpì˘
≈nally, let us also observe: (åspıåk'nju⁄) /fsplak'nutj/ bcπlaknytì˘
8.3.1.3. Generally, in Russian (but not necessarily, as for instance in quick pro-
nunciation) stops and stopstrictives, in heterorganic sequences, are audibly re-
leased ((0$0)), whereas those in homorganic sequences are inaudibly released
((0æ0); only here we will use the special diacritics): (kå'Rjøp$kx) /ka'ropka/ kopo∫-
ka, (p$'⁄i;qx) /p'tiqa/ πt$za, (kup$'qju;) /kup'qu/ kyπzy, ('jøt$pusk) /'otpusk/
otπyck, (t$'kFaC) /t'kac/ tkau, (ë'ëãøt$kx) /S'Sotka/ ∑etka, (åd$'gFat$kx) /ad'gatka/
otgadka, (s'vFaÁ$bx) /s'vadjba/ cbadì∫a, (åÀig$'dFa;) /fsig'da/ bcegda, (g$'Á™;) /g'd-
je/ gde, (påd$'møt$kx) /pad'mjotka/ πodmetka, (zåt$'me;~i[ã]I) /zat'mjeniji/ ‘atme-
n$e, ('áib$nu⁄) /'gibnutj/ g$∫nytì, (k$'~i;gx) /k'niga/ kn$ga, (ÀiÁ$'mjøi) /sidj-
'moi/ cedìmo=, (d$'veÌ, Á$'veÌ) /d'vjerj/ dbepì˚ (kåg$'dFa;) /kag'da/ kogda, ('⁄øg$b¢)
/'tjogb…/ tek ∫v˘
More: (pItæ'njø;) /pit'no/ πrtno, ('b™dæn¢i) /'bjedn…i/ ∫ednv=, (&vxRå+æ'~ik) /varat-
'nik/ bopotn$k, (pi+'¬a;, pItæ-, 'pe+æ¬x) /pit'lja, 'pjetlja/ πetlr, (fu+æ'¬aÍ) /fut'ljar/ fyt-
lrp. In addition: (åp'pFaC$kx⁄) /ap'packatj/ o∫πaukatì, (åt'tju;dx) /at'tuda/ ot-
tyda, (åq'qFa;) /at'qa/ otza, (&å⁄⁄I'njuÓ) /atti'nul/ ottrnyl, (å⁄'⁄u;dx) /at'sjuda/
otcùda (also (åt'Àu-), and even (åt'su-) /at'su-/), (&åCCI'vjø;) /atci'vo/ otuego (pho-
nemic geminates –and similar homorganic sequences– are realized as phonetic
geminates with a limited duration of the second element: ((0æ0))).
8. russian 297
8.3.1.4. In addition to /q, c/ (q, Q÷ C[ã], ‚), we also find sequences like /ts, dz÷
tS, dq/ (ts, dz÷ tS, dq) (but generally /ts, dz/ (ts, dz) become /q/ {(q) and (Q)}, ex-
cept in compounding, especially with prefixes): (&åts¢'ıFa;ãu) /ats…'laju/ otcvlaù,
(påts'ıju;SxÓ) /pats'luSal/ πodclywal, (nåts'mjøt5, -;t”) /nats'motr/ nadcmotp,
('jødz¢f) /'odz…f/ ot‘vb, (&pxdzxgå'ıjø;vxk) /padzaga'lovak/ πod‘agolobok, (nåd-
'zjøÍ) /nad'zor/ nad‘op, (åt'SF™¬~Ik) /at'Seljnik/ otwelìn$k, (&åbvIt'SFa;ı¢i) /abvit-
'Sal…i/ o∫betwalv=, (&pxdq¢'gFa⁄) /padq…'gatj/ πodµ$gatì, ('tjødq¢) /'todq…/ tot
µe, (d'qju˙g¬I) /d'qungli/ dµyngl$÷ (&gxRåq'kjøi, -ts'k-) /garats'koi/ gopodcko=, (så-
'v™q©ii, -ts©-) /sa'vjetskii/ cobetck$=, (ku'pFaqqx, -atsx) /ku'paqqa, -atsa/ kyπatì-
cr, ('Á™qtvx, -tst-) /'djetstva/ detctbo (only in careful speech the variants given af-
ter a comma are possible, and are due above all to the influence of spelling). Let us
notice the di‡erences between /vj, vvj, v-j/ (v, vvj, vj), in: ('vøÓ) /'vjol/ bel, (v'vøÓ)
/v'vjol/ bbel, (v'jøÓku, 'vø-) /v'jolku/ b elky; (vj) easily becomes (v). However, they
can remain distinct in careful speech – above all in (the most) traditional one.
8.3.1.7. As we have seen from various examples, the Russian voiced diphonic
phonemes are substituted by the corresponding voiceless ones, before a pause, or
before voiceless C: (d'Rjuk) /d'ruk/ dpyg, ('vjøtkx) /'votka/ bodka, (&bIskx¬I'ba;~ii)
/biskali'banii/ ∫e‘ kole∫an$=˘ On the contrary, the voiceless phonemes are substi-
tuted by the corresponding voiced ones (or by voiced phones, in the case of /q, c/
(Q, ‚)), before a voiced diphonic C: (&å~Ig'djøt) /anig'dot/ anekdot, (&å6gx~Is-
'tFan) /avganis'tan/ Afgan$ctan, (g'djø;mu) /g'domu/ k domy, (å'⁄™Qb¢Ó) /a'tjeq-
b…l/ otez ∫vl˘
However, >b≥ /v/ alone is not su‚cient to change a preceding voiceless C into a
298 a handbook of pronunciation
8.3.1.8. Written geminate C (and >∑≥) are phonically short before a pause or
before a C: (g'RFa≈) /g'ram/ gpamm, ('vFan) /'van/ bann, (d'qF…n) /d'q…n/ dµ$nn,
(k'ıFas) /k'las/ klacc, (p'ıFaë) /p'laS/ πla∑, (pRåg'RFamn¢i) /prag'ramn…i/ πpog-
pammnv=, ('Rjus©ii) /'ruskii/ pycck$=, (g'Rjupkx) /g'rupka/ gpyππka, ('mjøën¢i)
/'moSn…i/ mo∑nv=˘
©thin lexemes, written geminate C can be realized as phonetically slightly
geminated C˚ (00): (&åsså'nFans) /assa'nans/ acconanc, ('vFannx) /'vanna/ banna,
('kFassx) /'kassa/ kacca, ('tjønnx) /'tonna/ tonna˘
However, more often CC are pronounced short: (&å⁄Is'tFat) /atis'tat/ attec-
tat, (bå'ıjøn) /ba'lon/ ∫allon, (bå'Àein) /ba'sjein/ ∫acce=n, (gRå'mFa;⁄Ikx) /gra-
'matika/ gpammat$ka, (d&çiÀI'Rjøåkx) /drisi'rofka/ dpecc$pobka, (I'¬¯;=i[ã]x) /i'ljuz-
ija/ $llù‘$r, (&kx¬Ik'⁄if) /kalik'tif/ kollekt$b, (&mi¬i'm™t5, -;t·) /mili'mjetr/ m$l-
l$metp, (&åku'pFa;q¢ãx) /aku'paq…ja/ okkyπaz$r, (&pxRå'¬e1) /para'ljelj/ πapallelì,
(pI'Rjøn) /pi'ron/ πeppon, (&çIq¢'ÀøÍ) /riq…'sjor/ peµ$ccep, ('tF™;~Is) /'tenis/ tenn$c,
(&⁄içI'tjø;çi[ã]x) /tiri'torija/ tepp$top$r, (tRå'¬eibus) /tra'ljeibus/ tpolle=∫yc, (I-
'À™nq¢ãx) /i'sjenq…ja/ <ccenz$r, (I'f™kt) /i'fjekt/ <ffekt˘
8.3.1.9. In mid-fast speech, unstressed syllables are reduced (in particular the
non-initial ones occurring immediately before a stress, or the final ones after a
stress), especially near /r, l/: (px&mxıå'Á™Ó, &pxm®å-, &pxmıå-) /pamala'djel/ πomolo-
8. russian 299
8.3.1.11. Some personal pronouns have reduced forms: (⁄i'ba;, ⁄i'ãa;, ⁄i&a, ⁄å,
⁄x) /ti'bja/ te∫r, (⁄i'b™;, ⁄i'ã™;, ⁄i&™, ⁄I) /ti'bje/ te∫e, ('vFas, &vas, vås, vxs) /'vas/ bac,
('vFa≈, &va≈, vå≈, vx≈) /'vam/ bam, ('jÅ ⁄i&b™'dFa≈, 'jÅ ⁄I'dFa≈) /'ja ti'bje 'dam/ r
te∫e dam, ('ja &vam'dFa≈, 'ja vxm'dFa≈) /'ja 'vam 'dam/ r bam dam.
Certain numbers have reduced forms too (although we only provide the mod-
els, which are used for similar forms as well): (&ÁiÀi'⁄i;, ÁiÀ'⁄i;) /disi'ti/ decrt$, (å-
'Ái;nxqqx⁄, -nxqx⁄, -nqx⁄) /a'dinaqqatj/ od$nnadzatì, (å'Ái;nxqqxt¢i, -nxqxt¢i,
-nqxt¢i, -nxqt¢i) /a'dinaqqat…i/ od$nnadzatv=, (d'vFaqqx⁄, d'vFa;qx⁄, d'vFaq$⁄)
/d'vaqqatj/ dbadzatì, (d&vFaqq¢'⁄i;, d&vFaq¢'⁄i;, dvaq'⁄i;, dvåq'⁄i;) /dvaqqa'ti/
dbadzat$, (&piÁÁi'Àat, &piÁi'-, pi[ã]i'-) /piddi'sjat/ πrtìdecrt, (pi'⁄i;ÁiÀi⁄I, -'⁄i;ÁiÀ-
⁄I, -'⁄i[;ã]iÀ⁄I) /pi'tidisiti/ πrt$decrt$, (&S¢=Ái'Àat, &S¢=ãi'-, S¢i'-) /S…zdi'sjat/ wectìde-
crt, (S¢À'⁄i;ÁiÀi⁄I, -'⁄i;ÁiÀ⁄I, -'⁄i[;ã]iÀ⁄I) /S…s'tidisiti/ wect$decrt$, ('ÀemÁiÀxt,
'Àe)Àxt) /'sjemdisjat/ cemìdecrt, (Ài'mi;ÁiÀi⁄I, -'mi;ÁiÀ⁄I, -'mi[;ã]iÀ⁄I) /si'midisiti/
cem$decrt$, ('vjø;ÀimÁiÀxt, 'vjø;Ài)Àxt, 'vjø[;ã]i)Àxt) /'vosimjdisjat/ bocemìde-
crt, (våÀ'mi;ÁiÀi⁄I, -'mi;ÁiÀ⁄I, -'mi[;ã]iÀ⁄I) /vasj'midisiti/ bocìm$decrt$.
8.3.1.12. ≈rst and patronymic names are usually reduced by frequent use, more
than by particular phonemic rules: ('jø;ÀIpxviC, -ÀipiC, -ÀIp¢C, -øÀpviC, -øÀp¢C,
-øÀpC) /'osipavic/ Oc$πob$u, (bå'çi;sxviC, -'çiÀviC, -'çi;ÀiC, -'çi;s¢C, -'çiëC, -'çië) /ba'risav-
ic/ Êop$cob$u, ('pFavıxviC, 'pFa;v®viC, -av¬iC, -avı¢C, -a;¬iC, -a;ı¢C, -a1C, -aÓC) /'pav-
lavic/ ∏ablob$u, (&å¬Ik'sFandRxviC, -and“viC, -andviC, -andçiC, -andR¢C, -a;~iC, -a;-
n¢C, &å¬Ik'sFa8C, ¬Ik'sFa8C, k'sFa8C) /alik'sandravic/ Alekcandpob$u, (vıå'Ái;mIRx6-
nx, -mIR6nx, -mIRxnx, -mIRnx) /vla'dimiravna/ Blad$m$pobna, ('fø;dxRx6nx, -dxRx-
nx, -dR*nx, -dxRnx, -d“nx) /'fjodaravna/ Fedopobna, (bå'çi;sx6nx, -'çisnx) /ba'risav-
na/ Êop$cobna, (kåR'~e[;ã]i6nx, -'~™6nx) /kar'njejivna/ Kopneebna, (få'Áe[;ã]i6nx,
-'Á™i6nx, -'Á™6nx) /fa'djejivna/ Fadeebna, (påÍ'fiç-jI6nx, -'fi;çI6nx, -'fi;çInx, -'fiçnx)
/par'firijivna/ ∏opf$pì$ebna, (så've;LãI6nx, -'ve;¬ãI6nx, -'ve;¬I6nx, -'ve;¬Inx, -'ve¬nx)
/sa'vjeljjivna/ Cabelìebna˘
In addition: ('Fannx mI'∆Faiıx6nx, -ı*nx, -ıxnx, -ınx) /'anna mi'xailavna/ Anna
M$xa=lobna, (&å¬Ik'sFand Rå¬Ik'sFandRxviC, &å¬Ik'sFand Rå¬Ik'sFandR¢C, ¬Ik'sFand Rå¬Ik-
'sFandR¢C, ¬Ik'sFand Rå¬Ik'sFa8C, ¬Ik'sFan ¬Ik'sFa8C, ¬Ik'sFan 'sFa8C, 'sFan 'sFa8C) /alik-
'sandr alik'sandravic/ Alekcandp Alekcandpob$u, (&kxnstå8'⁄i; n¢så'Fa;kxviC, -'⁄i;
nI-, &kxns'⁄i; ~I'sFakC) /kanstan'tin …sa'akavic/ Konctant$n *caakob$u, ('pFa;vI
ı¢'vFa;nxviC, -vi lI-, -vFa8C, 'pFaı 'vFa8C) /'pavil i'vanavic/ ∏abel *banob$u˘
8.3.1.13. Currently, /ji/ >e˚ r≥ becomes /i/: (jI'mju;, I-) /ji'mu/ emy, (jI'qF…;, I-) /ji-
'qi/ eµ$, (jIv'¬aqqx, Iv'¬aqqx, Iv'¬aq[q]) /jiv'ljaqqa/ rblrtìcr, (&~i[ã]iÀ'⁄™stvInnx, ~ii-
'⁄™svInx÷ -'⁄eÀ⁄v-) /nijis'tjestvinna/ neectectbenno.
In general, /ijé/ = ç/ié/Ç (Ô § 8.2.5.1): (vi'⁄i;ãx, -'⁄i;x, -'⁄ix) /vi'tija/ b$t$r, ('FaR-
miãx, -mix, -mIx) /'armija/ apm$r, (&pçiã¯'⁄iqqx, &pçi¯-) /priju'tiqqa/ πp$ùt$tìcr,
(pçi'jÅ+~iãI, pçi'Å+~iI) /pri'jatniji/ πp$rtnee, (&q¢⁄i'j™;, -⁄i'™;) /q…ti'je/ µ$t$e, (&b¢⁄i-
'ju;, &b¢⁄i'u;) /'b…tiju/ ∫vt$ù˘
Besides, usually /0j’jé/ = /’0jé/ and /ii’0é/ = /i’0é/: (&piç-jI'vjøi, &piçI'vjøi) /pirj-
ji'voi/ πepìebo=, (b¢'CãÅC-jI∆, -Å;CI∆) /b…'cacjix/ ∫vuauì$x, ('kFa=N-ju, 'kFa=~-ju,
'kFa=~u) /'kaznjju/ ka‘nìù, (¬u'bjøv-ju) /lju'bovjju/ lù∫obìù, (Svii'qFaÍ, Svi'-, SvI'-)
8. russian 301
widespread (and also relatively less far away from spelling). We will now indicate
the di‡erences which mainly fall within çtraditionalÇ pronunciation, according to
the çold Moscow pronunciationÇ which was acquired orally. It used to include
some çstrangeÇ exceptions, which in some respect contributed to di‡erentiate it
from (St) Petersburg pronunciation, generally closer to orthography. Since then,
this city has been called Petrograd, then Leningrad and now Saint Petersburg: so
to say çrename it and you'll get over it!Ç.
û 8.3. Vowel elements of traditional (and alternative) Russian.
/'i, ’0iJ, i'/ (iï)t, /’i/ (I)t /JuJ/ (%)t, /’JuJ/ (T)t
/'…, …'/ (…)t, /’…/ (¢)t, /…'/ (…)t /'u, u'/ (u)t, /’u/ (U)t
/’…...'/ (‘, È)t:a˚ /'...’…, ’…ò/ (‘)t:a {/’u/ (û, È, ‘)a}
/'JeJ/ (eï)t /'JoJ/ (+)t
/'e/ (™) {/'e˘, "e/ (™[Ä])}t /'o/ {/’o/} (ø) {/'o˘, "o/ (ø[∏])}t
/'JaJ/ (Ä, "Ä[Ä])t /a'/ (a°)t, /’a/ (x)t
/'a/ (a), /'a˘, "a/ (a[å])t
>e≥ /’iò/ (‘)t, >e, r, Ça≥ /i'/ (É', JÙ'J)t˚ >e≥ /’i[ò]/ (É)a, /’0iJ/ (Ù)a, ><≥ /ò’i/ (É)a
8.3.2.2. Modern pronunciation has (’¢) for /’…/, whereas alternative (and often
traditional) pronunciation presents opener variants, (È, ‘). Most frequently, we
find (È)a:t before a stress; (‘)a:t after it; and (È)a (‘)t in absolute final position: (&bÈ-
ta'vjøi)t (&bÈtå-)a (&b¢tå-)m /b…ta'voi/ ∫vtobo=, (&vÈ∆a'Ái⁄)t (&vÈ∆å-)a (&v¢∆å-)m /v…xa-
'ditj/ bvxod$tì, ('vF…;m‘%)ta (-m¢%)m /'v…m…l/ bvmvl, ('jø;p‘t‘)t (-p‘tÈ)a (-p¢t¢)m
/'op…t…/ oπvtv˚ (q…'vjøtnxã‘)t (-nxãÈ, -ãÉ)a (q¢'vjøtnx[ã]I)m /q…'votnaji/ µ$botnoe.
As we have already said, the timbres of /’…/ (È, ‘) are fairly similar to those of /’a/
(x), thus it is easy to think of them as the same sound and to declare their neutral-
ization (as quite a few authors have done). Actually, in general, there is such a
di‡erence, so that they are rarely confused, even in spelling, except for evident cas-
es of poor education.
For >’eò≥, in traditional pronunciation we often find (‘ò)t even for /’…ò/ (in addi-
tion to /’iò/, most surprisingly, as we have seen at the end of § 8.3.2.1): ('ıjuCS¢)m
(-‘)t /'lucS…/ lyuwe˚ ('∆ju;q¢)m (-‘)t /'xuq…/ xyµe˘ Here we may actually speak of pos-
sible neutralization between /’iò/ and /’…ò/, which was more typical in former times.
For /’u/, alternative pronunciation may have (0û, 0È, 0‘÷ JÉ, JI, JÙJ, JiJ) (in
û 8.3, (û) is indicated by a thin broken line and glossed in brackets): (&kûv‘Í'kFaq-
qx, &kÈ-, &k‘-)a (&kuv¢Í-)m /kuv…r'kaqqa/ kybvpkatìcr, (&sûmå'tjø;∆x, &sÈ-, &s‘-)a
(&su-)m /suma'toxa/ cymatoxa, ('zFa;mûq‘≈, -mÈ-, -m‘-)a (-uq¢≈)m /'zamuq…m/ ‘a-
myµem, (kx&mÙ~i'©j™;, -mi-)a (-m¯~i-)m /kamjuni'kje/ kommùn$ke, (I'me;ãÙë&ëiiÀx,
-[ã]i&-)a (-ã¯&-)m /i'mjejuSSiisja/ $meù∑$=cr, (ÀÉ'dFa;, ÀI-)a (À¯-)m /sju'da/ cùda˘
In traditional pronunciation, we can even find: (mid've;ÁÈmI)t (mId've;ÁimI)m
/mid'vjedimi/ medbedrm$˚ (miÁ've;Á‘≈)t (mId've;Ái≈)m /mid'vjedim/ medbe-
dem, ('vF…;~‘s)t (-Is)m /'v…nis/ bvnec÷ and, on the other hand, also: (Ài'vjø_~I)t (ÀI-
'vjød~x)m /si'vodnja/ cegodnr.
8.3.2.5. Traditional pronunciation has (ëC)t (or, as we have already said, (ëC)t or
(ëCë)t) for (òëë, éëëé, ëò)m /S[S]/, but in Moscow the latter –more agile– pronunciation
has always prevailed: (ëCi'kFa;)t (ëëI-)m /SSi'ka/ ∑eka, (ta'vFa;çiëC)t (tå'vFa;çië[ë])m /ta'va-
riS[S]/ tobap$∑˘ Contrary to the most frequent transliteration type (∞£), the graph-
eme >∑≥ might seem to suggest a closer articulatory relation with >w≥ (S) /S/; and
indeed a constrictive pronunciation is decidedly more recommendable: (ë'ëãÅÀ⁄jI)
/S'Sastjji/ cuactìe, (&piçi'piëëIk) /piri'piSSik/ πepeπ$cu$k (the -u$k su‚x is not inter-
preted as a separation element).
However, in modern pronunciation also (ëC) /Sc/ is normal, when a clear mor-
phemic boundary –which is shown in writing too– is present, with >cu˚ ‘u˚ wu˚
µu˚ ctu˚ ‘du≥, never >∑≥: (ë'Cã™≈) /S'cem/ c uem, (&IëCI'vjø;) /iSci'vo/ $‘ uego˘ /SS/
or /Sc/ occurs in: (vIs'njuëëIt¢i, -ëC-) /vis'nuSSit…i, -Sc-/ becnywuatv=, (bå'RjøëëIt¢i,
-ëC-) /ba'roSSit…i, -Sc-/ ∫opo‘duatv=.
But, between a lexeme and a su‚x, we generally find (ëë) /SS/: (Iz'vjøëëIk) /iz'voS-
Sik/ $‘bo‘u$k, (&piçi'beëëIk) /piri'beSSIk/ πepe∫eµu$k, ('qjøëëI) /'qoSSi/ µectue˘ Be-
tween a prefix and a lexeme, (ëC) /Sc/ is most frequent: (bië&CIıå'veCn¢i) /biScila-
'vjecn…i/ ∫ecuelobeunv=, (&IëCIÍ'⁄i⁄) /iScir'titj/ $cuept$tì, (&RxëCI'∆aqqx) /raSci-
'xaqqa/ pacu$xatìcr. In most commonly used words and when a prefix is no lon-
ger thought of as something separated, we usually find (ëë) /SS/: (Rxë'ëãøskx) /raS'Sos-
ka/ pacuecka, (ë'ëÅÀ⁄jI) /S'Sastjji/ cuactìe, (ë'ëãøt) /S'Sot/ cuet, (Ië'ëã™s) /iS'Ses/ $cue‘˘
≈nally, the phoneme /ZZ/ (òò) (which occurs in few words and only within lex-
emes) is more typical of traditional than modern pronunciation: (ò'òãøt)t (q'qjøt)m
/q'qot/ µµet, ('jeòòU)t ('j™qqu)m /'jeqqu/ e‘µy, (&U[ã]iò'òãÄ⁄)t (&u[ã]Iq'qFa⁄)m /ujiq'qatj/
ye‘µatì, ('pjøòòI)t ('pjøqq¢)m /'poqq…/ πo‘µe÷ for derivative forms of ('djøëC)t
('djøë, 'djøS⁄)m /'doS, 'doStj/ doµdì˚ we have: ('djøòòIk)t ('doZZik/ ('djøqÁIk)m
8. russian 305
8.3.2.8. Please, note the di‡erences between modern and traditional pronuncia-
tion, as far as /rJ/ is concerned: ('⁄™RmIn)m ('⁄eçmIn)t /'tjermin/ tepm$n, (⁄IÍ-
'pit)m (⁄iÌ-)t /tir'pit/ tepπ$t, ('À™Rbi[ã]x)m ('Àeçb-)t /'sjerbija/ Cep∫$r, ('v™Íf)m ('veÌf)t
/'vjerfj/ bepfì, ('Cã™RvI)m (-çvI)t /'cervi/ uepb$, ('⁄™R~i[ã]I)m ('⁄eç~-)t /'tjerniji/ tep-
n$e, ('qF™Í⁄)m (-Ì⁄)t /'qertj/ µepdì, (u'À™RÁi[ã]I)m (-eçÁi[ã]‘, -[ã]È)t /u'sjerdiji/ ycep-
d$e, (À'⁄™R¬i⁄)m (-eç¬-)t /s'tjerlitj/ cteplrdì, ('v™ÍÀi[ã]x)m ('veÌÀ-)t /'vjersija/ bepc$r,
(À'v™R=IÓÀx)m (À'veç=IÓsx)t /s'vjerzilsa/ cbep‘$lcr, (kå'm™ÍCIs©ii)m (ka'meÌCiÀ©ii)t /ka-
'mjerciskii/ kommepueck$=˘
Even in traditional pronunciation, as in the modern one, we find (RJ) after ('0é)
(with non-front V]\ ('pFaÍ⁄i[ã]x) /'partija/ πapt$r, ('jøRÁIÍ) /'ordir/ opdep, ('djuR-
~I) (-‘)t /'durni/ dypne, ('mFaR¬x) /'marlja/ maplr, ('kjuÍÀI) (-‘)t /'kursi/ kypce˘ ˛e
same occurs after unstressed V (even if they are front ones): (vIÍ'⁄e⁄) /vir'tjetj/ bep-
tetì, (ÀIR'vis) /sir'vis/ cepb$‘, (vIR'=i;ıx) /vir'zila/ bep‘$la, (=IR'~Ist¢i) /zir'nist…i/ ‘ep-
n$ctv=˘
C + (ç) are not palatalized: (v'çe;mx) /v'rjemja/ bpemr˚ (f'ç™skx) /f'rjeska/ fpecka˚
(påÍ'tç™t) /par'trjet/ πoptpet˚ (∆ån'dçi⁄) /xan'dritj/ xandp$tì˚ (pçIz'çit) /priz'rit/
πpe‘p$t˘
For reflexive forms (in -cr˚ -cì]˚ traditional pronunciation has /s/ (not /sj/), ex-
cept for final-stressed gerunds: (ba'jÄÀ)t (bå'jÅÀ)m /ba'jasj/ ∫orcì (against (&sxbi'RFa;-
ãxs)t (-bI-)m /sabi'rajas/ co∫$parcì]˘ On the contrary, in modern pronunciation re-
flexive verbs have /sj/ (À), except in the third person (in -tcr) and infinitives (in
-tìcr): (~i'Àøqqx) /ni'sjotsa, -qqa/ necetcr, (kıå'djuqqx) /kla'dutsa, -qqa/ kla-
dytcr, (u'Ciqqx) /u'citsa, -qqa/ yu$tìcr˘
8. russian 307
8.3.2.9. ˛e pronouns $x˚ $m˚ $m$ are more regularly /ix, im, 'imi/ ('i∆, i∆, I∆÷
'i≈, i≈, I≈÷ 'i;mI, &imI, &ImI), whereas traditionally they had an exceptional pronun-
ciation (for >ò$≥) with /ji/, by analogy with ego /ji'vo/, emy /ji'mu/ (but today this is
lofty, or non-neutral).
A stronghold of traditional pronunciation had >Cò$≥ as /0ò…/: (v…'tFa;¬i[ã]U)t /v…-
'taliju/ b *tal$ù, ('kF…;çI)t /'k…ri/ k *pe, by distinguishing them from (vi'tFa;¬i[ã]U)t
/vi'taliju/ B$tal$ù, ('©i;çI)t /'kiri/ K$pe˘ However, native speakers spontaneously
make things natural, by eliminating artifices, thus today a çpalatalizedÇ pronuncia-
tion prevails in both pairs of examples, without doubt because of orthography too.
˛is is a further demonstration that Russian actually has six vowel phonemes, indi-
cated in spelling (with the only problem of unstressed-syllable reductions and his-
torical changes). Purists (of course!) still consider this spontaneous pronunciation
non-neutral, but they will have to change their minds sooner or later. As a matter
of fact, a well-balanced position gives: /'kiri, ±'k…ri/ (where ± indicates a kind of çin-
tentionalÇ pronunciation – used because çit should be usedÇ).
Some other particular cases: ('sjønq¢) /'sonq…/ colnze, ('Cãustvx) /'custva/ uybc-
tbo, ('tF…;ÀiCx, 'tF…ëëx) /'t…sica/ tvcrua, (Ài'Cãas, ë'ëãas, ëas) /si'cas/ ce=uac (in the
sense of çnow, at onceÇ), (på'qFa;ıustx, -a;ı¢stx, -aÓstx, b'q-) /pa'qalusta, -l[…]sta/ πoµa-
ly=cta, (z'dRFastvui⁄I, -astui-, -asÀ⁄I, -aÀ⁄, z&dRaÀ⁄, zaÀ) /z'drastvuiti/ ‘dpabctby=-
te.
As a useful reminder, and as a drill (thus in the examples we only mark stress,
through accents), we can say that >t≥ = (`) in the graphic sequences >ctd˚ ctìd˚
ctl˚ ctn˚ ctc˚ ctck≥ (the last one is /sk/): wectìdec®tì˚ neb™ctka˚ cuactl¢-
bv=˚ wectnådzatì˚ $‘b™ctnv=˚ wectìcøt /S…s'sot/, typ¢ctck$=; >d≥ = (`) in
>‘dk˚ ‘dn≥\ πo™‘dka˚ πpå‘dn$k˚ πø‘dno˘
Besides, the sequences >tc˚ dc≥ correspond to /q/, in the pronominal endings
-t(ì)cr: (smi'jøqqx) /smi'joqqa/ cmeetcr, (smi'jaqqx) /smi'jaqqa/ cmertìcr, and
when >t˚ d≥ are followed by the endings -ck$=˚ -ctbo˚ -ctb$e\ ('Á™q©ii) /'djeqkii/
detck$=, (åt'sjuqtvi[ã]I) /at'suqtviji/ otcytctb$e, (&pRå[ã]Iz'vjøqtvx) /prajiz'voq-
tva/ πpo$‘bodctbo, (À'¬™qtvi[ã]I, s'¬-) /s'ljeqtviji/ cledctb$e˘
Lastly, in the genitive endings -ogo, -ego, >g≥ corresponds to /v/: ('b™;ıxvx) /'bjela-
va/ ∫elogo, (jI'vjø;) /ji'vo/ ego˘ ˛us the oddity of (ÀI'vjø_~x) /si'vodnja/ cegodnr
çtodayÇ (= çof this dayÇ) is only seeming.
8.3.2.11. Here we will present the di‡erences between modern and traditional
pronunciation, in reference to poststress-syllable V˚ with /’…/ (‘) for /i/ (>e≥˚ general-
ly described as ç/È/Ç, exactly like /’a/, of unstressed >a˚ o≥, greatly falsifying pho-
netic reality). ˛e –transcribed or retranscribed– examples are taken from Shapiro
(1968).
Nominative and accusative singular of neuter nouns, -e: ('pjø;¬I)m (-¬‘)t /'poli/
πole, ('mjø;çI)m (-ç‘)t /'mori/ mope, ('ve;CI)m (-C‘)t /'vjeci/ beue, (q¢'¬iëëI)m (-ëC‘)t
/q…'liSSi/ µ$l$∑e, (z'dFa;~i[ã]I)m (-[ã]‘)t /z'daniji/ ‘dan$e.
Genitive, dative, and locative singular of nouns and adjectives, -e-: ('Ài;~Ivx)m
(-~‘-)t /'siniva/ c$nego, ('Ài;~Imu)m (-~‘-)t /'sinimu/ c$nemy, (å'Ài;~I≈)m (-~‘≈)t /f'si-
nim/ b c$nem.
Nominative and accusative, singular and plural, of neuter adjectives and pro-
nouns, -e: (z'ljø[;ã]I)m (-[ã]‘)t /z'loji/ ‘loe, (z'ıF…[;ã]I)m (-[ã]‘)t /z'l…ji/ ‘lve, (tå'kjø[;ã]I)m
(ta'kjø[;ã]‘)t /ta'koji/ takoe, (tå'©i[;ã]I)m (ta'©i[;ã]‘)t /ta'kiji/ tak$e, (¬i'ba;qjI)m (-j‘)t
/li'bjaqji/ le∫rµìe, ('vjøÓCjI)m (-j‘)t /'volcji/ boluìe.
Collective numerals, -e\ (d'vjø[;ã]I)m (-[ã]‘)t /d'voji/ dboe, (t'Rjø[;ã]I)m (-[ã]‘)t /t'ro-
ji/ tpoe˘
Comparatives, -ee: (smi'¬e[;ã]I)m (Àmi'¬e;ã‘)t /smi'ljeji/ cmelee, (vIR'~e[;ã]I)m (viR'~e[;-
ã]‘)t /vir'njeji/ bepnee.
Instrumental singular non-feminine and genitive plural, -em, -eb: (u'Ci;⁄i¬I≈)m
(-‘≈)t /u'citilim/ yu$telem, ('pıFa;CI≈)m (-‘≈)t /'placim/ πlauem, (q¢'¬iëëI≈)m (q…-
'¬ië C‘≈)t /q…'liSSim/ µ$l$∑em, (b'RFa;⁄jIf)m (-‘f)t /b'ratjif/ ∫patìeb, ('SF…L-jIf)m (-‘f)t
/'S…ljjif/ w$lìeb.
Instrumental of plural nouns, -Çam$: ('kFap-¬imI)m (-¬‘mI)t /'kaplimi/ kaπlrm$,
('tju;CimI)m (-‘mI)t /'tucimi/ tyuam$, ('RjøëëimI)m (-ëë‘mI)t /'roSSimi/ po∑am$.
8. russian 309
Stress
8.3.3.1. As far as the actual stess force is concerned on the various syllables of
words in Russian sentences, we have to make the situation clear since –on this as-
pect too– confused and misleading ideas freely circulate. As a matter of fact, the
prominence of a certain syllable does not necessarily coincide with its stress(ing)˘
As is well-known, prominence (which makes a given syllable stand out from near-
by ones) is not constituted by stress alone. Rather it is a complex play of stress
force, length, pitch, and timbres of the various segments of the syllables.
Now, in Russian, the vocoids in a pre-tonic syllable (¤ the one that immediate-
ly precedes the stressed –or tonic– syllable in a rhythm group, that is the çfirst pre-
tonicÇ in Russian texts) are articulated in a fairly clear and distinct way, although
slightly less so than in stressed syllables. ˛is fact is most evident for /a/, as a mat-
ter of fact we have for instance: (&gxıå'vFa;) /gala'va/ goloba˘
Nevertheless, we must not confuse a vowel timbre with the degree of stress,
since for rhythmic reasons a secondary stress falls on the first syllable of a word,
not on the second one. Each syllable has a peculiar prominence: (&gx) because of
rhythmic stress (in spite of a decidedly more attenuated timbre – less peripheral
in the vocogram); (ıå) because of its semi-attenuated timbre and a significantly
di‡erent pitch from that of the tonic (syllable) which follows it, although not
marked by a particular symbol (Ô û 8.4). Lastly, (çvFa;) is prominent because of
primary stress, a full timbre, and a half-lengthening too (in unchecked syllable, in
addition to the insertion of (F)).
All too often, even in specific handbooks, we come across statements which
claim that the pretonic syllable, (ıå), would have the second degree of stress,
whereas all the others are weaker, above all those after the stress. However, in (s'tFa;-
Rxãx) /s'taraja/ ctapar˚ for the last two syllables, we have a timbre attenuation (as
for (&gx)) and a stress weakening (as for (ıå)), which might make them seem one
degree lower in comparison with the other two – but it is not so. ˛eir weakening
–if any– is due to the fact that, generally, words given as examples are considered
to occur in an intoneme (even if unintentionally), with a consequent reduction of
articulatory and intonational intensity on posttonic syllables.
8.3.3.3. In compounds, every element (before the last one, which firmly main-
tains its primary stress, /'/ (')) tends to keep a certain degree of stress, which we will
mark with /“/ (“) (or (&), if in contact with ('), corresponding to a possible rhythmic
stress). Even the vocalic timbre is firmly maintained, with no neutralizations,
which are typical of unstressed syllables): (&sxm哬øtxstRå'je;~i[ã]I) /sama“ljotastra'je-
niji/ camoletoctpoen$e, (kåÍ“tjøfi¬ikå'pFaÓkx) /kar“tofilika'palka/ kaptofeleko-
πalka, (“bjøÍtpRxvå_'~ik) /“bortpravad'nik/ ∫optπpobodn$k, (s“pjøÍtkRu'qjøk)
/s“portkru'qok/ cπoptkpyµok, (å“F™Rx“fjøtåÀ'~i;mxk) /a“era“fotas'nimak/ a<pofo-
tocn$mok, (I“¬™ktRx“pFaRxpx&dxgçI'vFa;⁄i1) /i“ljektra“parapadagri'vatil/ <lektpoπa-
poπodogpebatelì, (“gjøs¢z'dFat, -sI-) /“gos-…z'dat/ Goc$‘dat, (&mjøz'bFa~k, &mjøs-,
-∫k) /“moz'bank, “mos-/ Moc∫ank, (&ik'RFatkx[ã]I) /“ik'ratkaji/ $ kpatkoe, (“pjøÓCI-
8. russian 311
8.3.3.4. Obviously, in actual sentences there are unstressed words too. Mostly
they are functional monosyllables (not lexical ones) such as prepositions, conjunc-
tions, and some particles: (&åtstå'ıFa;) /atsta'la/ ot ctola, (&pxdåk'njø≈) /padak-
'nom/ πod oknom, (kåm'~™;) /kam'nje/ ko mne, (b'RFa⁄-jx i'ÀøstR¢) /b'ratjja i'sjostr…/
∫patìr $ cectpv, (~Iz'nFa;ãu) /niz'naju/ ne ‘naù, (skå'qF…;kx) /ska'q…ka/ ckaµ$-ka,
('jø2q¢) /'onq…/ on µe, (pçi'Áø+-¬I) /pri'djotli/ πp$det l$.
Also pronouns and monosyllabic adjectives can be destressed: (mjøib'RFat)
/moib'rat/ mo= ∫pat, (='Á™Àjøn, -xn) /z'djesj-on, -an/ ‘decì on, (ku'dFa v¢bI'qF…;⁄I)
/ku'da v…bi'q…ti/ kyda bv ∫eµ$te, (&viÀi'j¯8, &ve-) /visji'junj/ becì $ùnì÷ also <to:
(k'tjø F™tå='Á™;ıxÓ, Itå-) /k'to etaz'djelal, i-/ kto <to cdelal, ('vjø;njøn, -xn) /'vo-
non, -an/ bon on, ('vjø;tjøn, -xn) /'voton, -an/ bot on, (vjøt'tjut, vå-) /vot'tut,
va-/ bot tyt, (vjøn'tFa≈, vå-) /von'tam, va-/ bon tam, (jønz'nFaÓ) /onz'nal/ on
‘nal, (S&tjøjøn'tFa≈, S&tåjø-) /Stoon'tam, Stao-/ uto on tam, (&njøjøm'bF…Óta≈,
-º'b-, &nåjø-) /noom'b…ltam, -n'b-, nao-/ no on ∫vl tam, (tåk'vjøt) /tak'vot/ tak
bot (conjunction, whereas the adverb tak does not reduce), ('jøn då'ja;) /'on da-
'ja/ on da r (a conjunction again, whereas da, the adverb çyesÇ is generally stressed
('dFa;); finally: (I'mju ∆xÁb¢S'tjø;) /i'mu xadjb…S'to/ emy xotì ∫v uto.
Here is a list of the most frequent (functional) monosyllables with reduced
forms (realized with unstressed vowel taxophones, required by the context): ∫v,
∫vl, bv, bam, becì, da, kak, l$, mne, mv, ne, n$, o∫, on, ot, πo, πod, cemì, to,
tv. ˛e pronoun r çIÇ can be reduced immediately after a stressed V: (må'gu jx-
pxvI'dFa⁄) /ma'gu japavi'datj/ mogy r πob$datì, (g'Áe jxmå'gu pxıu'Ci⁄) /g'dje ja-
ma'gu palu'citj/ gde r mogy πolyu$tì, (kåg'dFa jxnåCI'nFaÓ) /kag'da janaci'nal/
kogda r nau$nal˘
However, monosyllables can be more independent about vowel timbre than un-
stressed syllables in polysyllables. Consider, for instance: ('vjø;Rxn) /'voran/ bopon
and ('vjøR-jøn) /'voron/ bop on, (&⁄i¬I'sFa;) /tili'sa/ teleca and (&⁄e¬I'sFa;) /tjeli'sa/
te leca, (jIs'~ei) /jis'njei/ rcne= and (jås'~ei) /jas'njei/ r c ne=.
Intonation
8.3.4. û 8.4 shows the preintonemes and intonemes of neutral Russian – then
we just give some illustrative examples, to use for comparisons:
/./: (çF™;tx mjøi¬¯çbi;m¢i &kx≈påç=i;txÍ) /'eta moi-lju'bim…i kampa'zitar./ >to
mo= lù∫$mv= komπo‘$top.
/?/: (¿çvF…; jI&vjø•vi;Ái¬I2 2) /¿'v… jivo'vidili?/ Bv ego b$del$?, (¿&F™tx¬I•kkjø;2 2, -∆•k-)
/¿etalik'ko?, -x'ko?/ >to legko?
/÷/: (çjeÀ¬I &jø~~Iåtçve;⁄It2 2 &janxpI'Sju; jI&vjøåqçqju;3 3) /'jesli on-niat'vjetit÷ ja-na-
pi'Su jivoaq'qu/ Ecl$ on ne otbet$t, r naπ$wy ego otzy.
çElliptic questionsÇ (¤ those without a verb) have the pattern /¿ ÷/: (¿aIgçzFa;mI-
n¢2 2) /¿aig'zamin…÷/ A <k‘amenv?
û 8.4. Russian preintonemes and intonemes.
/ / (2 2 ç 2 2 ç 2 2 ç 2) /./ (2 ç 3 3)
/¿ / (¿ 2 2 ç 2 2 ç 2 2 ç 2) /?/ (2 • 2 2)
/¡ / (¡ 2 2 ç 2 2 ç 2 2 ç 2) /÷/ (2 ç 2 2)
/˚ / (˚ 2 2 ç 2 2 ç 2 2 ç 2) /,/ (2 ' 2)
Text
8.4.0. ˛e transcribed passage, †e North Wind and the Sun, gives the neutral
accent; as always, according to the phonetic method, first comes the English text
(Ô § 2.5.2.0) with a broad Russian accent, then the Russian translation, with neu-
tral pronunciation; then follows the çtraditionalÇ accent (not recommendable any
8. russian 313
longer today).
At the end, as usual, the version showing the English pronunciation of Russian
is given; it represents the habits of neutral British speakers, fluent in Russian (af-
ter prolonged contact with native speakers, but with no help from the phonetic
method), who have adequately learned the relative prominences, but who substan-
tially use segmental and intonation elements which are typical of neutral Russian
(although, of course, a neutral accent is not so common). Obviously, the same
principle is valid for the foreign pronunciations of English, given first.
Speakers of American English could prepare their own version both of the Rus-
sian pronunciation of English and of their pronunciation of Russian, as an excel-
lent exercise, by listening to native speakers, best of all after recording them. Of
course, speakers of other languages could do the same thing. ˛e author would be
happy to receive their transcriptions and recordings, both in case of help –should
they need it– and to make their contribution known to others (possibly in our
website on canIPA Natural Phonetics – Ô § 0.12).
Russian text
9.0.1. In this chapter, we will deal with the çneutral modernÇ pronunciation of
Arabic, corresponding to the çsupraregionalÇ language used in o‚cial radio and
television broadcastings, free from local characteristics (which are typical of every-
-day life in each of the Arabic countries). In fact, as for the other languages treat-
ed in this handbook, this is the most suitable form for general teaching and learn-
ing purposes. It constitutes the basis for local varieties too (which are more or less
important from a numerical and socio-cultural point of view), with some simpli-
fications, modifications and additions.
On the other hand, of course, if substantial contact in particular areas is forseen,
it is advisable to take on some local lexical, semantic, and morphosyntactic charac-
teristics, as well as the phonic ones (which mainly regard some phonemes).
Our transliteration diverges from the more traditional ones used by Arabists
(which are far from being homogeneous, however), especially as far as long vow-
els and diphthongs are concerned: ii˚ aa˚ uu (not as i˚ a˚ u)˚ ai˚ au (but ay˚ aw on-
ly when final, or followed by y˚ w÷ therefore, they are not always ay˚ aw). Further-
more, we use ∞˚ ≠˚ É˚ Æ˚ ‚ (for /S÷ †, ∑÷ X, º/, not digraphs: sh˚ th˚ dh˚ kh˚ gh {not even
í˚ î˚ for the last two}) and ∆˚ ` (instead of –˚ j˚ for /h, Z/), while keeping =˚ ƒ˚ «˚ „
(for /t, d, †, Ã/)˚ and q˚ '˚ ‘ as well (for /›, ö, H/).
Vowels
9.1.1. Neutral Arabic has three short and three long vowels, with some remark-
able taxophones, due to the influence of certain consonants and of syllable struc-
ture. ˛ere are more variations for /a[:]/ and less for /u[:]/, while /i[:]/ is in an inter-
mediate position. ˛ere are two çdiphthongsÇ as well, /ai, au/, which are best con-
sidered as sequences of /a/ + /i, u/, since their neutral realization is obtained pre-
cisely by juxtaposing the two elements, which are subject to the influence of conso-
nants and syllable structure. ˛e influence of the local ways of speaking is very
strong, even if unintentional, in teaching recordings as well, especially for /ai, au/,
but also for the basic V˘
However, the information we will give in this chapter represents the neutral pro-
nunciation, which does not necessarily correspond to everything one may hear
even from good native speakers. Nevertheless, the reader who follows exactly what
is indicated certainly achieves the çneutralÇ pronunciation (not a regional one),
even if –for V– this usage is quite close to that of Levantine Arabic (Ô § 19.2 of
NPT/HPh], in particular for /ai, au/, seen that elsewhere they are generally real-
ized as monophthongs ((e:, o:) or, at most, as narrow diphthongs, (™I, øU)); con-
sider also Gulf Arabic and Egyptian Arabic (§ 19.1 “ § 18.25 in NPT/HPh]˚ while
9. arabic 317
9.1.2. Always bearing in mind that the unmarked value of /i{:}, a{:}, u{:}/ is (i{:},
Ä{:}, u{:}) (shown by the black markers in the vocogram of û 9.1), we should note
well –and constantly recognize– the list of the Arabic realizations, which have the
following distributions (besides, /é:/ in unstressed syllables = (é)):
9.1.3. According to the distributions just seen, û 9.1 shows the realizations of
the Arabic –short and long– vowels, (i[:], I[:], ¢[:]) /i[:]/, (Ä[:], a[:], A[:], å[:]) {and (Ø[:]),
çKoranicÇ variant, Ô û 9.2) /a[:]/, (u[:], U[:]) /u[:]/. Here are some examples: ('›¢f) /'›if/
qif˚ (›¢'›A:n) /›i:'›a:n/ qiiqaan˚ ('†I:ni) /'†i:ni:/ «iinii˚ (†A'dI:›) /†a'di:›/ «adiiq˚ ('bInt)
/'bint/ bint˚ ('fi:l) /'fi:l/ fiil÷ (†å'›IH÷ †Ø-) /†a'›iH/ «aqi‘˚ ('†Af:) /'†aff/ «a‡˚ ('tA:h) /'ta:h/
=aa∆a˚ ('baHda) /'baHda/ ba‘da˚ ('5a:HIn) /'ra:hin/ raahin˚ ('HÄ:∑a) /'ha:∑a:/ haaÉaa˚
('wÄlÄd) /'walad/ walad˚ ('bÄ:b) /'ba:b/ baab÷ ('†U:›) /'†u:›/ «uuq˚ ('Huna) /'huna:/
hunaa˚ ('†u:f) /'†u:f/ «uuf˚ (ºU'†u:m) /Xu'†u:m/ Æu«uum˚ ('fUndU›) /'fundu›/ funduq˘
û 9.1 does not show the di‡erent realizations of /ai, au/, which result from the
combination (Ä-, a-, A-) + (-i, -I, -¢) or + (-u, -U), according to context: ('bÄIt) /'bait/
bait˚ ('öÄina) /'öaina/ aina˚ ('Haini) /'Haini:/ ‘ainii˚ ('›AIl) /'›ail/ qail˚ ('fÄUz) /'fauz/
fauz˚ ('lÄUn) /'laun/ laun˚ ('HaUdÄ) /'Hauda[h]/ ‘auda-h˚ (mU'›AuwÄm, -AwwÄm)
/mu'›awwam/ muqawwam˚ (öÄu'lÄ:d) /öau'la:d/ aulaad˘
û 9.2. Some Arabic vowel variants.
9.1.4. û 9.2 shows some frequent realizations of /ai, au/: (e:, ™I÷ o:, øU), which
are very widespread outside neutral and Levantine Arabic. However, one's pronun-
ciation may still be considered neutral, although çcolloquialÇ, even if it uses such
variants, provided all other articulations are appropriate; in fact, this pronuncia-
tion is quite common indeed: ('be:t, 'b™It) /'bait/ bait˚ ('öe:na, 'ö™I-) /'öaina/ aina˚
('He:ni, 'H™I-) /'Haini:/ ‘ainii˚ ('›e:l, '›™Il) /'›ail/ qail˚ ('fo:z, 'føUz) /'fauz/ fauz˚ ('lo:n,
'løUn) /'laun/ laun˘
û 9.3. Colloquial variants and neutralizations.
/’i/ (Ù), /’a/ (‘), /’U/ (P) {/’i, ’a, ’u/ (È) ç/È/Ç possible
frequent çcolloquialÇ variants çcolloquialÇ neutralization)
û 9.3 shows further vocalic articulations, all of which are in the intermediate
realizational area (typically unused in neutral pronunciation, as can be seen in û
9.1), as happens for the variants of /ai, au/ as well. ˛e white markers indicate un-
stressed realizations of /i, a, u/, (Ù, ‘, P), which are considerably centralized. ˛e
9. arabic 319
9.1.5. Actually, not every single realization given in û 9.2-3 is necessary for a
good neutral pronunciation of Arabic. Nevertheless, if they are rationed and used
in a natural way (speaking fluently), they help to give greater çspontaneityÇ, simi-
lar to natives' speech again within a kind of colloquial neutral pronunciation. ˛is
will depend on words themselves – for instance, (Ø[:]) is more likely to occur in
çloftyÇ words.
Furthermore, still within the neutral accent (but slightly more çinternationalÇ),
there is another possibility taking a di‡erent approach, ¤ towards a reduction in
the number of vowel taxophones, as regards the realizations of /a[:]/. In fact, they
may reduce to two articulations, in stressed or unstressed syllables: (Ä[:]), for 2.3-
7, and (A[:]), for 2.1-3; thus with a slight overlapping for 2.3, variably resolved by
single speakers.
˛e vocalic realizations belonging to this kind of çinternationalÇ pronunciation
are shown in û 9.4. ˛is çclearerÇ pronunciation generally uses: (i[:], ¤[:]÷ Ä[:], A[:]÷
U[:]) (it may also have (™:, ø:) for /ai, au/), but not necessarily (Ù, ‘, P) (and least of
all (È)), seen in § 9.1.4.
˛e systematic and complete shortening of unstressed long vowels belongs to
çmodernÇ and çinternationalÇ pronunciation (and to dialects as well, generally).
However, in çKoranicÇ pronunciation, /’é:/ (appearing in phonemic transcrip-
tions) are realized as half-long, (’é;). It is not di‚cult to achieve them, if we consid-
er both types of transcription used in this chapter: (›¢'›A:n, ›¢;-) /›i:'›a:n/ qiiqaan˚
('†I:ni, -ni;) /'†i:ni:/ «iinii˚ ('HÄ:∑a, -∑a;) /'ha:∑a:/ haaÉaa˚ ('Huna, -na;) /'huna:/ hu-
naa…
û 9.4. International vowels.
/i/ (i, ¤), /i:/ (i:, ¤:) /u/ (U), /u:/ (U:)
Geographic variants
9.1.6. Among the main variations and deviations from the neutral form (Ô §
9.1.3), in certain areas, we find that /a:/ never has the (Ä:) timbre, in any context
(as often happens in Iraq and northern Lebanon). Furthermore, in an almost gen-
eral way, in several colloquial variants, the diphthongs /ai, au/ reduce to (e:, ™I÷ o:,
øU) (Ô û 9.2), from Maghreb to the Persian Gulf, except in the Levant. However,
the diphthongs are kept, generally, when they are in absolute final position or fol-
lowed by /j, w/: ('bÄIt, 'b™It, 'be:t) /'bait/ bait˚ ('lÄUn, 'løUn, 'lo:n) /'laun/ laun˚ ('HaIn,
'H™In, 'He:n) /'Hain/ ‘ain÷ ('sÄijId, 'sÄjjId) /'sajjid/ sayyid˚ (mu'SÄuw¢›, -Äww-) /mu'Saw-
wi›/ mu∞awwiq˚ ('nÄi, 'nÄj:, -J:) /'nai/ nayy˚ ('lÄu, 'lÄw, 'lÄ)) /'lau/ law˘
Especially in Maghreb, besides (unstressed) /’i, ’u/, even /’a/ may be dropped (with
possible consequent stress shifts): ('öÄnta, ó'tÄ, n'tÄ) /'öanta/ anta˚ ('›AlÄm, '›Alm)
/'›alam/ qalam˚ (li'sÄ:n, Í'sÄ:n) /li'sa:n/ lisaan˚ (sÄ'lÄ:m, s'lÄ:m) /sa'la:m/ salaam˘
Consonants
9.2.0. û 9.5 gives the articulations of modern neutral Arabic. ˛e main geo-
graphical variants of some phonemes are given in round brackets.
û 1.9-15, instead, show –gathered in groups according to manners of articula-
tion– the orograms of all contoids treated in the chapters of this book, even as sec-
ondary, occasional, or regional, variants for the 12 languages dealt with.
û 9.5. Table of Arabic consonants.
uvularized alveolar
postalveo-palalatal
uvularized dental
postalveopalatal
velar rounded
protruded
pharyngeal
labiodental
laryngeal
alveolar
bilabial
palatal
uvular
dental
velar
Nasals
9.2.1. ˛ere are two nasal phonemes, /m, n/ (m, M, n, ~, ˙, ,): (mu'mƆÄl)
/mu'ma††al/ muma≠≠al˚ (tÄm'Zi:d, -'Gi:d) /tam'Zi:d/ tam`iid˚ ('SÄms) /'Sams/ ∞ams˚
9. arabic 321
Stops
9.2.2.1. Arabic has no ç/p/Ç, but only /b/, which however is realized as (p) when
followed by voiceless C\ ('bÄ:b) /'ba:b/ baab˚ ('lÄbÄn) /'laban/ laban˚ ('haps) /'habs/
∆abs˘ On the other hand, neutral Arabic has no ç/g/Ç, either; while it has two voice-
less stops, velar and uvular, /k, ›/ (k, ›) (in phonemic opposition): ('k[h]Uskusu,
kUs'k[h]usu) /'kus“kusu/ kuskusu˚ (mIk'†a:5) /mik'†a:r/ mik≠aar˚ ('mÄlIk) /'malik/
malik˚ (›A'di:m) /›a'di:m/ qadiim˚ (öA›'dÄ:m) /a›'da:m/ aqdaam˚ ('sU:›) /'su:›/ su-
uq˚ (&öAı›UR'öÄ:n) /al›ur'öa:n/ al-Qur'aan˚ (sA›'›å:tA÷ -'›Ø:-) /sa›'›a:ta[h]/ saqqaa=a-h˘
(›) enjoys great prestige, even among speakers who do not use it, although very
frequently it is substituted by other articulations (as will be seen in § 9.2.7). Note:
('kÄlb) /'kalb/ kalb çdogÇ and ('›Alb) /'›alb/ qalb çheartÇ.
Furthermore, we have the diphonic pairs /t, d/ (t, d) and /t, d/ (t, d) (uvular-
ized), which is quite peculiar. ˛ey are dental; and /t/ may be a little çaspiratedÇ,
when at the beginning of a stressed syllable (but we will mark it only here). Often,
/t, d/ are denti-alveolar if final before a pause (but it is not necessary to use ((4, 7))):
(tÄ'dÄ:wUl) /ta'da:wul/ tadaawul˚ (Si't[h]Ä:ö) /Si'ta:ö/ ∞itaa'˚ ('had:) /'hadd/ ∆add˚ (bA-
'tå:t¢s÷ -'tØ:-) /ba'ta:tis/ ba=aa=is˚ ('dA5t, 'då5t÷ 'dØ5t) /'dart/ ƒar=˚ ('dAZIR, -GIR) /'da-
Zir/ ƒa`ir˘
Although neutral Arabic has no (g), this sound is frequent in di‡erent dialects,
as a variant of other phonemes, mainly of /Z/ and /›/. ˛erefore, it is natural that
the isolated phonemes may currently be brought to normalization, by changing
their articulations, in order to form a more homogeneous and coherent system. E-
ven the shift of /Z/ to (Z) (instead of the more çKoranicÇ –and ancient– (G)), or to
(g), is a part of this trend.
9.2.2.2. ˛e last neutral Arabic stop is /ö/ (ö), which may occur in every posi-
tion, as the other C do, and may be geminated as well. In word-initial position, it
automatically occurs when no other C is present: (öÄ'mi:n) /öa'mi:n/ amiin˚ ('sÄ:-
öIh) /'sa:öih/ saa'i∆˚ ('mÄ:ö) /'ma:ö/ maa'˚ ('zÄnnÄö) /'zannaö/ zanna'˚ (&RuöÄ'sÄ:ö) /ru-
öa'sa:ö/ ru'asaa'˚ (öÄl'öÄ:n) /öal'öa:n/ al-aan˚ ('5aös) /'raös/ ra's˚ ('bÄdö) /'badö/ bad'˚
(sÄö'öÄ:l) /saö'öa:l/ sa''aal˘
˛erefore, in Arabic, any word traditionally beginning with a V (¤ vowel), pho-
nically, begins with /ö/ (ö), because in this language all phono-syllables begin with
a C\ ('öÄb) /'öab/ ab˚ (öi'dÄ:nÄ) /öi'da:na[h]/ idaana-h˚ ('öUm:) /'öumm/ umm˘ On the
other hand, Arabic does not tolerate syllables beginning with more than one C (ex-
cept rare cases of loanwords not adapted to Arabic phonology). Consequently, in
actual pronunciation, a short V is prefixed; generally, it is /i/ (but, sometimes, /u,
322 a handbook of pronunciation
a/, in accordance with the following V]˚ and /a/ in neutral pronunciation, for the
article – while, in the dialects, we more often find (™, I)); of course, in isolated
forms, /ö/ is prefixed as well.
However, when words with CC- (¤ with an initial consonant cluster) occur –in
connected speech– after a word ending in a V˚ it is not necessary to add the V (nor
/ö/), therefore the two words are linked. If, instead, the preceding word ends in a
C˚ then the V is added, but /ö/ is not. ˛e reader is referred to grammars, where this
phenomenon (called wa«la) is generally treated quite widely. It is to be found in
connection with the article, certain verbal forms, the imperative and a dozen
nouns. Among these, the most important are: ('öIbn, -bó, -b¢n) /'öibn/ ibn˚ ('öImRUö)
/'öimruö/ imru'˚ ('öIsm, -sõ, -s¢m) /'öism/ ism˚ (öI†'nÄ:ni) /öi†'na:ni/ i≠naani\ (Ra'öÄi-
tUb 'ni:) /ra'öaitu b'ni:/ ra'aitu bnii˚ ('bÄ:bUl 'bÄIt) /'ba:bu l'bait[i]/ baabu al-baiti˘
Constrictives
9.2.3. Among the phonemes belonging to this articulation manner, we find /f/
(f) (but not the corresponding voiced phoneme): (fa'5i:d) /fa'ri:d/ fariid˚ (öIf'sÄ:d)
/öif'sa:d/ ifsaad˘ In addition, there are two diphonic pairs, which pose no problems,
/†, ∑÷ s, z/ (†, ∑÷ s, z): (†Ä'lÄ:†) /†a'la:†/ ≠alaa≠˚ ('mÄ∑IR) /'ma∑ir/ maÉir˚ ('∑Äm:)
/'∑amm/ Éamm˚ ('da5s) /'dars/ dars˚ ('za:5) /'za:r/ zaar˚ ('kÄnz) /'kanz/ kanz˘
However, there are two more diphonic pairs, with variations that may pose
some phonemic dilemmas; they are /†, Ã/ (†÷ Ã/D) and /S, Z/ (S÷ Z/G). We prefer (Ã,
Z) for their voiced members, as they are more modern and more integrated in the
phonologic system than their more çKoranicÇ variants (D, G), which are consid-
ered more prestigious (even by those who do not use them), but have di‡erent
place or manner of articulation. ˛erefore, they complicate the phonemic system.
However, they can be used – especially in a kind of pronunciation which aims at
the traditional more than at the international accent: (mah'ÃU:Ã, -DU:D) /mah'Ãu:Ã/
ma∆„uu„˚ ('ÃA:mIö, 'D-) /'Ãa:miö/ „aami'÷ ('†UR†UR, †UR'†u:R) /'†ur†ur, †ur'†u:r/ «ur-
«u(u)r˚ (Ra'ºI:†) /ra'Xi:†/ raÆii«÷ (öIS'hÄ:d) /öiS'ha:d/ i∞haad˚ (öÄS'jÄ:ö) /öaS'ja:ö/ a∞yaa'˚
(mu'SÄuwÄS, -ÄwwÄS) /mu'SawwaS/ mu∞awwa∞˚ (öÄ'SÄd:) /öa'Sadd/ a∞add˚ (RaS'SÄ:S)
/raS'Sa:S/ ra∞∞aa∞˚ (öÄS'SaRu) /öaS'Sarw/ a∞∞arw˚ (ZÄ'mi:l, GÄ-) /Za'mi:l/ `amiil˚ ('öÄZmaH,
'öÄG-) /'öaZmaH/ a`ma‘˚ ('tÄ:Z, -G) /'ta:Z/ taa`˘
In the pharyngeal place of articulation, we find the voiceless constrictive /h/ (h)
(currently, the çcorrespondingÇ voiced sound is considered to be constrictive as well,
but in neutral pronunciation, it is clearly an approximant, /H/ (H), as we will see be-
low, § 9.2.4): ('hUbbi) /'hubbi:/ ∆ubbii˚ (mah'tu:m) /mah'tu:m/ ma∆tuum˚ (mU'hAd-
d¢R) /mu'haddir/ mu∆aƒƒir˚ ('faRIh) /'farih/ fari∆˚ (fah'ha:S) /fah'ha:S/ fa∆∆aa∞˘
Arabic also has a diphonic pair of uvular constrictive trills, (º, ˜) (as will be
seen), which phonemically may be represented with the o‚cial symbols /X, º/.
Approximants
9.2.4.1. Let us first consider the least peculiar ones (although there are rather
free occurrences), /j, w/ (j, w), even realized as (i, u), for /0jò, 0wò/ (and, possibly,
9. arabic 323
for /òj0, òw0/, in colloquial variants, as no doubt in the dialects) and (ij/jj,
uw/ww), for /éjjé, éwwé/): ('jÄ:wIR) /'ja:wir/ yaawir˚ (wA'†I:jÄ) /wa'†i:ja[h]/ wa«ii-
ya-h˚ (wU'†u:l) /wu'†u:l/ wu«uul˚ ('sÄijId, 'sÄjjId) /'sajjid/ sayyid˚ (nÄu'wÄ:m, nÄw'w-)
/naw'wa:m/ nawwaam˚ ('nÄi) /'najj/ nayy˚ (mÄn'Hi:) /man'hi:/ manhiyy˚ ('öÄbu)
/'öabw/ abw (Ô ('öÄbu) /'öabu/ abu).
˛e çKoranicÇ pronunciation prefers (-jj-, -ww-), as in the cases seen above. Let
us also consider these further examples, which show us the di‡erences between
modern pronunciation and traditional çKoranicÇ pronunciation. In principle,
they coincide even with the most important cases where, even in colloquial accents
and dialects, /ai, au/ do not change into monophthongs ((™:, ø:)).
˛is happens when they are in absolute final position, and when in front of /j,
w/ or after /i:, u:/, respectively, or when final, after C˚ as the following examples
will show: ('kÄi, 'kÄj, -J) /'kai/ kay (Ô ('kÄij, 'kÄj:, -J:) /'kaij/ kayy), ('lÄijÄn, 'lÄjjÄn)
/'lajjan/ layyan˚ (jÄu'mi:jÄ, -JÄ) /jau'mi:ja[h]/ yaumiiya-h˚ ('Humi, 'Humj, -mJ) /'Humj/
‘umy˚ ('lÄu, 'lÄw, 'lÄ)) /'lau/ law˚ ('dÄuwaR, 'dÄww-) /'dawwar/ dawwar˚ (Ha'du:wÄ,
-u:)Ä) /Ha'du:wa[h]/ ‘aduuwa-h˚ ('sÄHu, 'sÄHw, -H)) /'sahw/ sahw˚ ('Hafu, 'Hafw, -f‹,
-fä) /'Hafw/ ‘afw˘
9.2.4.2. As we have said above (§ 9.2.3), the Arabic phoneme /H/ (H) is an ap-
proximant (and generally it is laryngealized as well, ((ó)), which is fairly easy to de-
tect by its lower intrinsic tonality and creaky voice): ('HaIn) /'Hain/ ‘ain˚ ('maHa)
/'maHa/ ma‘a˚ ('baHda) /'baHda/ ba‘da˚ ('naHnaH, naH'na:H) /'naHnaH, naH'na:H/ na‘na(a)‘˚
(&faHHa'li:jÄ) /faHHa:'li:ja[h]/ fa‘‘aaliiya-h˚ ('sÄlH) /'salH/ sal‘˘ It is occasionally possible
to hear some peculiar regional –non-neutral– variants, as that with a slight simul-
taneous laryngeal stop, (?), or else a pharyngealized laryngeal stop, (ö), ™:
('na?na?, na?'na:?, 'nAönAö, nAö'nA:ö).
˛e last Arabic approximant is a true laryngeal phone, /h/ (H, h), and has a very
free occurrence; it is typically (lenis) voiced (H), but it becomes (lenis) voiceless,
when near a pause, or a voiceless C˚ or when geminate: (&öItti'ZÄ:h, -G-) /öitti'Za:h/
itti`aah˚ ('mUhtaR) /'muhtar/ muhtar˚ ('hÄ:∑iHi) /'ha:∑ihi:/ haaÉihii˚ (mÄH'bu:l)
/mah'bu:l/ mahbuul˚ (&öInHi'zÄ:m) /öinhi'za:m/ inhizaam˚ (hÄf'nÄ:f) /haf'na:f/ haf-
naaf˚ ('hija) /'hija/ hiya˚ ('ka5h) /'karh/ karh˚ ('bIh) /'bih/ bih˚ ('›AH-wÄ) /'›ahwa[h]/
qahwa-h˚ (wÄh'hÄ:Z, -G) /wah'ha:Z/ wahhaa`˘
In our transliteration, taa' marbuu=a is indicated by a hyphen, -a-h (while anoth-
er –only seemingly more recommendable and less complicated– solution could be
-a–); it is pronounced (Ä, Äh) /a[h]/ (but (a, A), if preceded by C which are capable
of modifying timbres). It is to be noted that a realization with (h) represents a very
careful, çKoranicÇ pronunciation, while the normal realization of -aò is (a): ('hÄz-
zÄ) /'hazza[h]/ hazza-h (çmovementÇ) but: ('hÄzza) /'hazza/ hazza (ç(he) shookÇ).
Trills
9.2.5. Typically the voiced alveolar /r/ is realized as a velarized trill, (5), general-
ly, in stressed syllable, and as a tap, (R), in unstressed syllables. ˛ey influence the
324 a handbook of pronunciation
timbre of /a/ (a): ('5Ip›) /'rib›/ ribq˚ (maR'bu:H) /mar'bu:H/ marbuu‘˚ ('maRIh) /'mar-
ih/ mari∆˚ (mIR'5i:h) /mir'ri:h/ mirrii∆˚ (mi'5a:h) /mi'ra:h/ miraa∆˚ ('fU5fUR, fUR'fu:5)
/'furfur, fur'fu:r/ furfu(u)r˘ It is possible to hear (¸), mainly for final /r/, but this pro-
nunciation is not recommendable.
As already said above (§ 9.2.4), Arabic has a diphonic pair of uvular constrictive
trills, /X, º/ (º, ˜): (baº'Si:S) /baX'Si:S/ baÆ∞ii∞˚ ('ºaUº) /'XauX/ ÆauÆ˚ (faº'ºa:Ri) /faX-
'Xa:ri:/ faÆÆaarii˚ ('˜adÄn) /'ºadan/ ‚adan˚ (†A'˜i:R) /†a'ºi:r/ «a‚iir˚ ('bÄ:lI˜) /'ba:liº/
baali‚˚ (mÄS'˜u:l) /maS'ºu:l/ ma∞‚uul˚ (tÄ'wa˜˜Ul) /ta'waººul/ tawa‚‚ul˘
Laterals
9.2.6. ˛ere is one lateral phoneme, /l/ (l, ¬, ı) (the third taxophone occurs in
contact with /t, d, †, Ã, ›/; it occurs as a phonostyleme (¤ a stylistic phoneme),
too, in Allaah and derivatives (provided that it is not preceded by /i/): (&mutÄ'lÄö-
lIö) /muta'laöliö/ mutala'li'˚ (tÄl'bi:s) /tal'bi:s/ talbiis˚ ('lÄIl) /'lail/ lail˚ (mĬ'jÄ:n)
/mal'ja:n/ malyaan˚ (zÄl'lA:›A) /zal'la:›a[h]/ zallaaqa-h˚ (tAı'›I:h) /tal'›i:h/ talqii∆˚
(&bIsmIl'lÄ:h) /bismil'la:h/ bismillaah˚ (öAı'ıA:h) /öal'la:h/ Allaah˘ For the typical
complete assimilation of /l/ in the article al˚ see below (§ 9.3.1.1).
Geographic variants
9.2.7. Very often, /†, ∑/ are pronounced like /t, d/, mainly in big cities in North
Africa, or like /s, z/ as well, particularly in the Middle East. In each one of these
cases a phonemic distinction is lost. In Iraq, /Ã/ is (D), as in traditional and çKo-
ranicÇ pronunciations; elsewhere it is often realized as (Ã), as in Egypt and Syria,
but it may even be confused with /t, d/, above all in Maghreb.
˛e grapheme `iim˚ /Z/ (Z), has very many geographical, social, and religious,
variants. ˛e normal (Z) prevails in the Middle East and in North Africa, while
(G) (typical of Koranic reading) is used in Jordan, Saudi Arabia (typical of Bedou-
ins) and Iraq. But in some areas, as in Egypt (Cairo), Sudan and Oman, we find
(g); elsewhere, even (›), as in Luxor (southern Egypt), and (J).
For instance, /'Za:b/ `aab is ('ZÄ:b) in the Levant (Lebanon, Palestine, and Syria)
and in Africa (except Egypt and Sudan, as we have just said); it is ('GÄ:b) in Arabia,
the Persian Gulf, Jordan, Iraq and among the rural and the nomadic peoples in
Morocco.
˛e grapheme qaaf˚ /›/ (›), very often becomes (ö), above all in Cairo and oth-
er big cities; but it becomes (H, k) as well, or even (g), particularly in central-south-
ern Egypt, in rural areas of Morocco, and among Saudi Arabia Bedouins (who of-
ten, typically, change /k/ into (c)). For instance, /'›alb/ qalb may be: ('›Alb÷ 'öÄlb,
'Halb, 'kÄlb, 'gÄlb).
In the Levant, /òö/ is generally quite weak, therefore we could transcribe it as (,)
(instead of (ö)). In Maghreb, /ö/ may not even be present at all.
9. arabic 325
Structures
9.3.0. Here we will deal with assimilation, quantity, and some typical reduc-
tions of colloquial speech, still within neutral pronunciation (although with geo-
graphical variants).
Taxophonics
9.3.1.2. In colloquial speech, unstressed /i, u/ tend to drop whenever this does
not form three-C clusters. On the other hand, when in connected speech two
word-final C are followed by a word-initial C˚ then a short V˚ generally /i/, is insert-
ed to divide the cluster. In certain cases, the added V may even bear stress; and
sometimes a V inserted in a certain place may cause –or allow– the dropping of
another short V˘
Furthermore, long V tend to shorten a little in checked syllables (but less than
in unstressed syllables); and short final V in polysyllables tend to drop (above all
grammemes, and /énò/ as well, except for accusative /anò/, which may become
/a:ò/, in any position, even non-pausal). Verbs and pronouns are more likely to
maintain the V˚ since endings have distinctive and pragmatic values.
326 a handbook of pronunciation
Let us see some examples: ('hÄ:∑Ä ki'tÄ:b, 'hÄ:∑Äk 'tÄ:b) /'ha:∑a: ki'ta:b[un]/ haa-
Éaa kitaabun˚ (öÄl'hIb-RU 'tAijIb, 'tAjjIb) /öal'hibru 'tajjib[un]/ al-∆ibru =ayyibun˚ (kÄ-
'tÄptu&biHi, -tUb&Hi) /ka'tabtu 'bihi/ katabtu bihi˚ ('kÄ:nÄ fIl'bÄIt) /'ka:na fi:albait[i]/
kaana fii al-baiti˚ (Ra'öÄitUl 'bÄIt) /ra'öaitu al'bait[a]/ ra'aitu al-baita˚ (kÄ'tÄpt) /ka-
'tabt[u]/ katabtu˚ (&sÄkÄ'ki:n) /saka:'ki:n/ sakaakiin˘
9.3.1.3. As far as çpausal formsÇ are concerned, let us take from Mitchell (1990)
some examples (which we will transcribe again appropriately), keeping in mind
that this is an çorthologicalÇ phenomenon people use to pronounce words in into-
nemes, including preceding words, semantically and syntactically linked (to the
exclusion of grammemes).
˛is category includes final short V (with or without infinitive -n) – ('kÄtÄb)
/'katab[a]/ kataba˚ ('jÄktUb) /'jaktub[u]/ yaktubu˚ ('liöÄ~ 'jÄktUb) /'liöan 'jaktub[a]/
li'an yaktuba˚ (&fimÄ'da:RIs) /fi:ma'da:ris[a]/ fii madaarisa˚ (fIl'bÄIt) /fi:l'bait[i]/ fii al-
-baiti÷ ('bÄIt) /'bait[un]/ baitun˚ (fi'bÄIt) /fi:'bait[in]/ fii baitin˚ ('mUn∑u 'zÄmÄnI˙
wÄ'Zi:z, -'Gi:z) /'mun∑u 'zamanin wa'Zi:z[in]/ munÉu zamanin wa`iizin˘ However,
(accusative) -an /an/ may be substituted by /a:/: (öaH'tA:HU HIS'5i:nÄ di'na:Ra) /öaH'ta:-
hu HiS'ri:na di:'na:ra:/ a‘=aahu ‘i∞riina dinaaran˚ except for adverbs like: ('ha:lÄn)
/'ha:lan/ ∆aalan˚ ('öÄ¢dAn) /'öaidan/ aiƒan˘
Lastly, taa' marbuu=a [¤ -atin˚ -atan˚ -atun] is substituted by /a[h]/, which is
generally pronounced (Ä) (or (Äh) in a {more} çKoranicÇ style): (ZÄ'mi:lÄ)
/Za'mi:la[h]/ `amiila-h ((ZÄ'mi:lÄtUn, &ZÄmi'lÄtUn, %G-) /Za'mi:“latun/ `amiilatun]˚
('fÄtahat mÄ'da:RisÄ lÄi'li:jÄ, &mÄda'5isÄ) /'fÄtahat ma'da:“risa lai'li:ja[h]/ fata∆at
madaarisa lailiiya-h ((lÄi'li:jÄtÄn, &lÄili'jÄtÄn) /lai'li:“jatan/ lailiiyatan]˚ (&fIlZu'nÄinÄ,
-Gu-) /fi:lZu'naina[h]/ fii al-`unaina-h ((Zu'nÄinÄ&ti, &ZunÄi'nÄti, %G-) /Zu'nai“nati/
`unainati]˘
9.3.1.4. Colloquially, very often /ö/ is not maintained when it occurs within
words or at the end of words – it is dropped or changed into /j, w/, or else it
lengthens a possible preceding V\ ('miöÄ, 'miÄ, 'mijÄ, 'mi:Ä) /'miöa[h]/ mi'a-h˚ ('jÄö-
ºu∑u, 'ja:ºu∑u, ja'ºu∑u) /'jaöXu∑u, jaö'Xu∑u/ ya'ÆuÉu˚ ('5aös, '5a:s) /'raös/ ra's˚ (sÄ-
'mÄ:ö, sÄ'mÄ:) /sa'ma:ö/ samaa'˘ ‹en two hamzas occur in contiguous syllables,
the first one is maintained: ('ZÄ:öÄ &[ö]ÄZÄ'luHUm, '[ö]ÄZÄ&luHUm, [ö]Ä'ZÄlHUm, 'GÄ:öÄ,
-ÄGÄ-) /'Za:öa öaZa'luhum, 'öaZaluhum/ `aa'a a`aluhum˘
In /0j, 0w, 0r, 0l/ sequences, Arabic syllabification is heterosyllabic, /0˘j, 0˘w,
0˘r, 0˘l/: (mIt'5a:s) /mit'ra:s/ mitraas˚ ('öÄtlÄfa, öÄt'lÄfa) /'öat“lafa/ atlafa˚ ('mÄdRasÄ,
mÄd'Ra-) /'mad“rasa[h]/ madrasa-h˚ ('öÄd-la) /'öadla/ adla˚ ('ma5-jÄm) /'marjam/ Mar-
yam˚ ('mUsRIH) /'musriH/ musri‘˚ (mIs'wÄ:k) /mis'wa:k/ miswaak˚ ('mÄS-wi) /'maSwi/
ma∞wi˚ ('öÄfRaza, öÄf'Raza) /'öaf“raza/ afraza˚ ('mÄk-wa) /'makwa/ makwa˚ ('öĬ-jÄn)
/'öaljan/ alyan˚ ('öÄH-wa) /'öahwa/ ahwa˘
9.3.1.5. In final position, after C˚ the sonants (/m, n, r, l/) may be realized in
di‡erent ways, according to how accurately one speaks. From a phonemic point
of view, they are just C˚ but –phonetically– they may be plain (or devoiced, main-
ly in front of a voiceless C]˚ or intense (çsyllabicÇ), or even with a short epenthet-
9. arabic 327
ic V (like (I, ¢)): ('›¢s≈, -sõ, -s¢m) /'›ism/ qism˚ ('lÄdn, -dó, -d¢n) /'ladn/ ladn˚ ('fÄtn,
-tó, -t¢n) /'fatn/ fatn˚ ('dUHn, -hn, -Hó, -H¢n) /'duhn/ duhn˚ ('bÄdR, -d{, -d¢R) /'badr/
badr˚ ('m¢†R, -†á, -†{) /'mi†r/ Mi«r˚ ('fÄtú, -tÍ, -t¢l) /'fatl/ fatl˚ ('5AtÓ, -t–, -t¢l) /'ratl/ ra=l˘
9.3.1.6. Here are some examples of typical Arabic phonotactics: (taö'bi:n) /taö-
'bi:n/ ta'biin˚ (mÄö'†u:R) /maö'†u:r/ ma'≠uur˚ ('mÄtHUb) /'matHu:b/ mat‘uub˚ ('öĆ-
›Al) /'öa†›al/ a≠qal˚ ('mÄdºal) /'madXal/ madÆal˚ (mÄ∑'Hu:R) /ma∑'Hu:r/ maÉ‘uur˚
('tÄzHu) /'tazhu/ tazhu˚ ('mÄsZId, -GId) /'masZid/ mas`id˚ ('mÄSta) /'maSta/ ma∞ta˚
(mÄS'˜u:l) /maS'ºu:l/ ma∞‚uul˚ ('mA†5aH) /'ma†raH/ ma«ra‘˚ ('öAdZaHa, öAd'ZaHa) /'öad-
ZaHa/ aƒ`a‘a˚ ('öAdHafa, öAd'Hafa) /'öadHafa/ aƒ‘afa˚ ('öAÃlÄma, öAÃ'lÄma, -D-) /'öaÃla-
ma/ a„lama˚ (maH'Zu:n, -Gu:n) /maH'Zu:n/ ma‘`uun˚ (maH'5u:f) /maH'ru:f/ ma‘ruuf˚
(öaH'mÄ:l) /öaH'ma:l/ a‘maal˚ (ma˜'Su:S) /maº'Su:S/ ma‚∞uu∞˚ ('mA›ha) /'ma›ha/ maq-
ha˚ (mÄl'öÄ:n) /mal'öa:n/ mal'aan˚ ('jÄiöÄsu, jÄi'öÄsu) /'jai“öasu/ yai'asu˚ ('öÄuHÄma,
öÄu'HÄma) /'öau“hama/ auhama˘
Stress
9.3.2.1. For words of two or more syllables, we have the patterns given below.
It should be kept in mind that ç/à/Ç stands for çlightÇ syllables (/0é/); ç/9/Ç for
çheavyÇ ones, ¤ consisting of çheavyÇ sequences like /é00, é:0, éé0/, which are
located counting from their syllabic nucleous onwards, up to the next (belonging
to another sequence). ˛erefore, in this kind of calculation, we do not consider
them to be true phono-syllables. In addition, ç/à/Ç indicates a light or heavy sylla-
ble –indi‡erently– with no direct influence on stress assignment. ˛e symbol ç/“/Ç
indicates alternative stressing, which is substitutive (certainly not simultaneous –
as a matter of fact, one is free to choose either stress pattern from sequences show-
ing both /'/ and /“/):
2 syll. /'àà, à'9, '9à/,
3 syll. /'ààà, àà'9, à'9à, '9“àà/,
4 syll. /“àà'àà, ààà'9, àà'9à, à'9“àà, “9'ààà/,
5 syll. /à“à'ààà, àààà'9, ààà'9à, àà'9“àà, à“9à'àà, 9“àà'àà/,
6 syll. /àà“àà'àà, ààààà'9, àààà'9à, ààà'9“àà, àà“9à'àà, à9“àà'àà, 9àà'ààà/.
˛ey function as a useful point of reference, for analyses and to find correspon-
dences; in fact, it would not be easy to try to memorize them.
9.3.2.2. ˛e solution to the formulae with variants, to read along columns, pro-
vides:
/'9àà, 9'àà/÷ /àà'ààà, à'àààà/, /àààà'àà, àà'àààà/,
/àà'àà, 'àààà/, /àà'9àà, àà9'àà/, /ààà'9àà, ààà9'àà/,
/à'9àà, à9'àà/, /à9à'àà, à'9ààà/, /àà9à'àà, àà'9ààà/,
/9'ààà, '9ààà/÷ /9àà'àà, 9'àààà/÷ /à9àà'àà, à9'àààà/.
328 a handbook of pronunciation
In patterns with two possible stressings, the variants may be of two types: çEgyp-
tianÇ (Cairo), as: /'9“àà, à'9“àà, àà'9“àà, ààà'9“àà/; or çliteraryÇ (— dialectal, like
those of southern Egypt), with the following possibilities: /“àà'àà, “9à'àà/, /à“à'ààà,
à“9à'àà, 9“àà'àà/, /àà“àà'àà, àà“9à'àà, à9“àà'àà/.
For /à9“àà'àà/, we may find the patterns /à'9àààà/, /à9à'ààà/ as well (which
constitute çloftierÇ stressings, as can be seen in the story in § 9.4.2.1-2).
All the stress patterns we give belong to modern neutral pronunciation. ˛ere-
fore, one is free to choose, provided certain structural homogeneousness is main-
tained.
A dialectal peculiarity found in Lebanon (which should not be followed) puts
final stress on words ending in /é:ò, ééò/ (which, in neutral pronunciation, are not
çheavyÇ enough to bear the stress): ('mInHuma, mIn'Huma) (and (&mInHu'ma:))
/'min“huma:/ minhumaa˘
9.3.2.3. Prefixes do not influence the application of the stress rule, in di‡erent
actual words; nor does the article even when it assimilates completely. But the rule
is modified by the dropping of some (vocalic or consonantal) phones, because this
changes the syllabic structure, both within words and at their end, before a pause.
As far as secondary stress in polysyllabic words is concerned, it tends to occur
on alternate syllables, but preferring the heaviest ones, whenever possible.
Lastly, here are some examples: (Ra'su:l) /ra'su:l/ rasuul˚ (sÄ'fÄ~Z, -~G) /sa'fanZ/ sa-
fan`˚ (mu'5a:sIl) /mu'ra:sil/ muraasil˚ (ta'5assUl) /ta'rassul/ tarassul˚ ('HA:†¢mÄ, HA'†¢-)
/'Ha:“†ima[h]/ ‘aa«ima-h˚ (&mutÄ'fÄuw¢›, -Äww¢›) /muta'fawwi›/ mutafawwiq˚ (mu-
'fa:RA›A, &mufa'5A›A) /mu'fa:“ra›a[h]/ mufaaraqa-h˚ ('mÄd-Ra&sÄ, mÄd'5asÄ) /'mad-
“rasa[h]/ madrasa-h˚ ('daRa&ki) /'daraki:/ darakii˚ (&kÄtÄ'bÄHu, 'kÄtÄ&bÄHu) /“kata-
'bahu/ katabahu˚ (mUº'tÄli&fÄ, 'mUºtÄ&lifÄ) /“muX'talifa[h]/ muÆtalifa-h˚ (&öÄZÄ'lu-
HUm, 'öÄZÄ&luHUm, öÄ'ZÄlHUm, -GÄ-) /“öaZa'luhum, öa'Zalhum/ a`aluhum˘
Intonation
9.3.3. û 9.6 shows the preintonemes and intonemes of modern neutral Arabic.
˛e continuative intoneme is not very di‡erent from the suspensive one; there-
fore, generally, the pause which follows the suspensive type has greater duration.
˛e interrogative intoneme in the Levant is of the rising-falling type, /?/ (2 ' 1 2), in-
stead of that given here. It will be interesting to compare the tonograms (but also
the V and C] referring to the four geographic Arabic varieties (Ô § 9.1.1 – which
di‡er in grammar as well, but most of all in their vocabulary). We will merely give
some illustrative examples, for neutral intonation, to be used for comparisons:
/./: ('lÄ: 'öaHRIf mÄ&dinÄ'tÄkUm23) {(mÄ'di:nÄ&tÄkUm23)} /'la: 'öaHrif[u] ma“di:na'ta-
kum./ Laa a‘rifu madiinatakum.
/?/: (¿'hÄl öÄlki'tÄ:b ZÄ'di:d21) {(GÄ-)} /¿'hal öalki'ta:b[u] Za'di:d[un]?/ Hal al-kitaabu
`adiidun?
/÷/: ('jUmkIn &öÄnnA'›u:m bi'ZÄulÄ2 2) {(bi'GÄu-)} /'jumkin[u] öanna'›u:m[a] bi'Zau-
la[h]÷/ Yumkinu an naquuma bi-`aulatin…
û 9.6. Arabic preintonemes and intonemes.
Text
9.4.0. ˛e story †e North Wind and the Sun follows. According to the Phonetic
method we start with the version in Arabic pronunciation of (neutral British) En-
glish (the written text is given in § 2.5.2.0). ˛e Arabic translation follows, in its
neutral version.
At the end, as always, there is the version which gives the English pronuncia-
tion of Arabic, by neutral British speakers, fluent in Arabic (after prolonged con-
tact with native speakers, but with no help from the phonetic method), who have
adequately learned the relative prominences, but who substantially use segmental
and suprasegmental elements which are typical of neutral British English (for refer-
ence purposes, although, of course, a neutral accent is not so common). Obviously,
the same principle is valid for the foreign pronunciations of English, given first.
Speakers of American English could prepare their own version both of the Ar-
abic pronunciation of English and of their pronunciation of Arabic, as an excel-
330 a handbook of pronunciation
lent exercise, by listening to native speakers, best of all after recording them. Of
course, speakers of other languages could do the same thing. ˛e author would be
happy to receive their transcriptions and recordings, both in case of help –should
they need it– and to make their contribution known to others (possibly in our
website on canIPA Natural Phonetics – Ô § 0.12).
Arabic text
9.4.2. ˛is passage highlights the fact that the çmodern classicalÇ Arabic lan-
guage is a rather artificial concept. In fact, the currently unwritten short V have
very fluctuating realizations, due to both their presence or absence and to their
timbres (themselves) (/i, a, u/). As the number of recordings (of di‡erent speakers)
increases, the number of variations also increases (even for stressing and ortholo-
gy, ¤ the use of intonemes and pauses).
Ha'bÄ:öÄ[tIn] sÄ'mi:kÄ23œ|| 'fÄt tÄfA'›Ata2 2|’ 'HalaH ti'ba:RIs 'sÄ:b¢›[¢]2[ &fiöIZ'ba:RIl mu'sÄ:-
fIR[i]2| 'Hala 'ºalHI Ha&bÄöÄ'tiHIlÌ 'öA›-wa23||
'HA†AfÄt2 '5i:hUS SÄ'mÄ:l[i]|| bi'öA›†A 'mÄ: öIstA'tA:öÄt m¢,'›u:wa23|| wÄ'lÄ:kIn2 ì'kUllÄ-
mÄz• 'dÄ:dÄl 'HA†f[u]2 2œ|| öIz'dÄ:dÄl mu'sÄ:fIR[u] tÄ'dƆu&Ran2] &biHa'bÄ:öÄ&tIh2 2| ì'öilÄ
öÄn'öUs›¢&tA2Ô fi'jÄdIR '5i:h23œ|| &fÄta'ºallÄt 'Ham mU&hawÄlÄ'tiHa23||»Ò baH'dÄöi&∑In2|»» &sA-
tA'öÄtIS»” 'SÄmsu bi'dIföi&Ha2| &fÄmÄ'kÄ:nÄ mIn&öÄlmu'sÄ:fIR[i]2| 'öIllÄ öa,'ºala&Ha»’ Ha-
&bÄöÄ'tÄHu23»[ ì'HalÄt 'tÄu23œ|| 'wÄ 'HÄkÄ&∑Ä»Ì ö¢d'tA5Rat2 '5I:hUS SÄ'mÄ:l[i]2 2| 'öilÄl &öIHti'5a:-
f[i]2 bi'öÄnnÄS 'SÄms[Ä]2| 'kÄ:nÄt2 'hijÄl 'öA›-wa23||
¿'hÄl 'kÄ:nÄt öAı'›¢††Atu2 ¿ZÄ'mi:lÄ21|| ¿'hÄl tu'5i:d[u]2 'öÄn &nuRad'didÄ&Ha21|||)»•
10.0.2. Nasalized vowels are marked with the çtildeÇ sign (i˚ q˚ ãã); however, in
accordance with the system of marking long vowels with a macron, awkwardly,
the tilde should be put over the macron (≤˚ >˚ <]˚ as some transcribe. Nevertheless,
others use –less satisfactory– digraphic devices (aµ˚ aµ÷ añ˚ añ).
Postalveolar consonantal articulations are indicated by a dot underneath (=˚ ƒ˚
ó); postalveopalatal ones with an inverted circumflex (or ha£ek\ £˚ À˚ ∞; instead of
traditional, but ambiguous, c˚ j˚ ≈]÷ for the uvulars of Arabic and Persian loanwords
we use particular signs (q˚ Æ˚ ‚; instead of digraphs or other more complex combi-
nations, or less evident, from a visual point of view, as well). ˛ese have been men-
tioned in order to avoid both signs and also digraphs which could be too generic.
˛e latter are reserved for çaspiratedÇ consonants (ph˚ bh˚ óh…). Furthermore, we
decidedly prefer w (to v]˚ for (6, j) /w/.
are, basically, of two kinds: those that indicate /a/ only when it is really pro-
nounced, and those that show every a (or very many), as for instance in ('måt-låb)
/'matlab/ matlab (or matalaba]. However, the actual phonetic situation may be
something in between, as can be seen in § 10.3.1.1 as well.
Vowels
10.1.1. û 10.1 shows Hindi vowels, which are ten: three short, (I, å, U) /i, a, u/,
and seven çlongÇ. ˛ese are actually diphthongs: five monotimbric, with narrow
movements, but nevertheless perceptible, (ii, ee, aa, oo, uu) /ii, ee, aa, oo, uu/, and
two more evident ones (even if, generally, they are simply described as long V˚
themselves), (Ä™, ∏ø) /EE, OO/ (because native speakers think they are monoph-
thongs, including the variants given, and use them as such when learning foreign
languages).
Let us see, for now, some examples of the three short V\ ('dIn) /'din/ din˚ ('påR)
/'par/ par˚ ('kUl) /'kul/ kul÷ the three corresponding diphthongs are: ('diin) /'diin/
diin˚ ('paaR) /'paar/ paar˚ ('kuul) /'kuul/ kuul÷ the other four diphthongs are: ('beeR)
/'beer/ beer˚ ('bÄ™R) /'bEEr/ baer˚ ('booR) /'boor/ boor˚ ('b∏øR) /'bOOr/ baor˘ We will
deal again with the geographic variants of /EE, OO/ (Ô û 10.2.å) again, only here,
('bÄ™R) /'bEEr/ baer˚ ('b∏øR) /'bOOr/ baor\ ('bå™R, 'b√øR) (east: Bihar, West Bengal),
('bE™R, 'bOøR) (west: Rajasthan), ('bπER, 'bØOR) (northwest: Panjab)˘
û 10.1. Hindi vowels.
/ii/ ('ii, &i, ’i) /uu/ ('uu, &u, ’u)
/i/ (I) /u/ (U)
/ee/ ('ee, &e, ’e) /oo/ ('oo, &o, ’o)
/EE/ ('Ä™, &Ä, ’™) /OO/ ('∏ø, &∏, ’ø)
/a/ ('å, &å, ’å0, ’‘˘)
/aa/ ('aa, &a, ’a)
10.1.2. Each Hindi vowel may be (distinctively) nasalized: (t'hI) /t'hi/ thi˚ ('Hì™,
&Hì, H™) /'híí/ hãq˚ (b'HÅÅ®) /b'hÅÅ#/ bhããó˚ (k'h_9) /k'hÓÓ/ khã£÷ let us consider, for
instance ('Råg) /'rag/ rag çveinÇ ('RAg) /'rÅg/ rãg çcolorÇ. Currently, words such as
('Hå∫s) /'hans/ hans tend to be confused with ('HAs) /'hÅs/ hãs÷ but, in neutral pro-
nunciation, they must be accurately distinguished, even if (∫) has no full contact
with the alveolar ridge (Ô § 10.2.1.1-2).
‹en ae˚ ao are followed by /j, w/, they correspond to (åI, åU) /ai, au/: ('måIãa)
/'maijaa/ maeyaa˚ (tåI'jaaR) /tai'jaar/ taeyaar˚ ('HåUja) /'hauwaa/ haowaa˘ ˛e same
should hold for Sanscrit words with ae˚ ao (èai˚ au¶); but, currently, these become
/EE, OO/ (Ô § 10.2.1.2).
Intra-lexemic sequences /eeh0, ooh0/ may be realized as short vowels: ('seeH-Ra,
'seH-) /'seehraa/ seehraa˚ ('mooH-låt, 'moH-) /'moohlat/ moohlat÷ when there is gram-
memic derivation, no shortening occurs: ('leeH-jÉ) /'leehja/ leehya˚ ('mooH-na)
334 a handbook of pronunciation
å. Regional variants of ae˚ ao /'EE, 'OO/: ('E™, ∫. Attenuated taxophones of /’a˘/: (‘),
'Oø) west (Rajasthan); ('πE, 'ØO) northwest ((’0‘)), (’jÉ, ’ãÉ÷ ’j∑), ((’±ê)): ± (m, p[h],
(Panjab); ('å™, '√ø) east (Bihar, West Bengal). b[H], å, 6), ((’ôx)): (k[h], g[H], ›, ˜, r).
Consonants
laryngeal
palatal
provelar
veolar
alveolar
bilabial
palatal
uvular
postal-
dental
velar
velar
Nasals
10.2.1.1. ˛ere are two fundamental nasal phonemes, /m/ (m), and /n/ (m, n,
N, ~, ˙, ,), since the NC sequences are homorganic. Besides, (M) ((˘)} occurs be-
fore /w, F/; (∫) ((ˆ)) before /s/; (n) ((q)) before /j/; and («) {(Ö)} before /h/ – with
a rare /˙/, which would give /˙h/ (and this would show the advantage of consider-
ing sequences of Ch biphonemic, even for stops) or /ngh/ (˙gH, «H). (It is worth
observing that (M, ∫, µ, n, «) are nasals with no full contact (¤ semi-nasals): semi-
-labial, semi-alveolar, semi-postalveo-palatal, semi-palatal, and semi-provelar, re-
spectively, Ô § 9.9 of NPT/HPh˚ and (π) {(¯)} as well, given in § 10.2.1.2.)
336 a handbook of pronunciation
On the other hand, it is not necessary to use the specific symbols ((˙, n)):
('mooH) /'mooh/ mooh˚ ('såmbål) /'sambal/ sambal˚ ('tiim ma'taÅe) /'tiin maa'taaee/
tiin maataaqq˚ (såM'jaad) /sam'waad/ samwaad˚ ('nåbH) /'nabh/ nabh˚ ('kåN˛h)
/'kan˛h/ ka¿=h˚ ('kå~‚) /'kanG/ kanÀ˚ ('6åµë) /'wanS/ wan∞˚ ('ån-jÉ) /'anja/ anya˚
('på˙k) /'pank/ pa;k˚ (&I,›x'laab) /in›a'laab/ inqalaab˚ ('sI«H[x]) /'si˙h[a]/ sinh(a)˘
In some words, we find /m, n/ before heterosyllabic C\ (bê'RaÅmda) /ba'raam-
daa/ baramdaa˚ (åm'Ruud) /am'ruud/ amruud˚ ('gUm˛i) /'gum˛ii/ gum=ii˚ ('CåmCa)
/'camcaa/ £am£aa˚ ('‚Umka) /'Gumkaa/ Àhumkaa˚ ('meenka) /'meenkaa/ meenkaa
(indicated in writing by the segments m˚ n˚ instead of by anuswaar˚ a diacritic
sign).
10.2.1.2. Furthermore, in lofty Sanscrit words, also /N, ˙/ (N, Í÷ ˙) occur, and
have specific graphemes, ¿˚ ;, but generally change into /n/. ˛e more frequent
genuine realization of /N/ is (Í), a nasalized postalveolar flap, or even a nasalized
postalveolar approximant, (¯), so it is better to use the symbol of the diaphone,
(π), which incorporates these values, although it generally becomes /n/. Also (~)
has a particular grapheme, n, even if it does not represent a phoneme (as was the
case instead in Sanscrit): ('baÅm) /'baam/ baam˚ ('baÅn) /baan/ baan˚ ('baÅπ, 'baÅN,
'baÅn) /'baaN/ baa¿˚ (gx'πeeë, gx'Neeë, gx'neeë) /ga'NeeS/ ga¿ee∞˚ (‘'gåN-jÉ, ‘'gåπ-jÉ)
/a'gaNja/ aga¿ya˚ ('6aÅ˙måI, -˙mêãÉ, -«m-, -nmÄ™, -mmÄ, -mm™) /'waa˙maj/ wa;-
may˚ ('baÅ˙[gx]) /'baang/ baa;g.
Furthermore, we find the sequences /mh, nh/ (mH, nH), which (together with
/lh/ (lH) and, possibly, /˙h/, Ô § 10.2.1.1) have no o‚cial devanagari graphemes,
but combinations. In fact, they did not occur in Sanscrit, contrary to çaspiratedÇ
stops and stopstrictives, including /#h/ (®H), which comes from /Ãh/ (ÃH): (kUm-
'HaaR) /kum'haar/ kumhaar˚ ('kaÅnH) /'kaanh/ kaanh˘
Stops
quences constituted by plain stops + /h/, which is realized as (h), after voiceless C˚
or as the normal (in Hindi) voiced laryngeal approximant, (H), after voiced C (all
the more so because they are realized as heterosyllabic, instead of tautosyllabic; this
can be seen by the placement of stresses, which is more evident and logical with-
in a word or a rhythm group): ('pål) /'pal/ pal˚ (p'hål) /p'hal/ phal÷ ('baala) /'baalaa/
baalaa˚ (b'Haala) /b'haalaa/ bhaalaa÷ ('tål) /'tal/ tal˚ (t'hål) /t'hal/ thal÷ ('daa6a) /'daa-
waa/ daawaa˚ (d'Haa6a) /d'haawaa/ dhaawaa÷ ('˛åppa) /'˛appaa/ =appaa˚ (˛'håppa)
/˛'happaa/ =happaa÷ ('Ãiil) /'Ãiil/ ƒiil˚ (Ã'Hiil) /Ã'hiil/ ƒhiil÷ ('kaal) /'kaal/ kaal˚
(k'haal) /k'haal/ khaal÷ (gI'Raa) /gi'raa/ giraa˚ (gHI'Raa) /ghi'raa/ ghiraa˘
10.2.2.3. It is not unusual for /b, bh/ to be realized as (6, 6H). Besides, attenua-
tions are also possible for /k/ (∆, ◊, â), /kh/ (xh, x), /g/ (Ÿ, y, µ), /gh/ (ŸH). Before
front vowels (and before /j/), /k[h], g[h]/ are realized as prevelar, but no special sym-
bol (([”]´, [”]Ò)) is necessary. In final position, the stops may have inaudible reali-
zations: ('naak, 'naakæ) /'naak/ naak˚ ('åb, 'åbæ) /'ab/ ab÷ this is contrary to what hap-
pens to çaspiratedÇ sequences, which do not simplify, even if the laryngeal element
may be less evident, in this final position (for other C too), but it may not be miss-
ing, because it is distinctive\ ('siikh, -kh) /'siikh/ siikh˚ ('nåbH, -bH) /'nabh/ nabh
(and: ('boo‚H, -‚H) /'booGh/ booÀh˚ ('baa®H, -®H) /'baa#h/ baaóh]˘ Only within se-
quences like /0h0/, may /h/ be dropped (Ô § 10.3.1.5). Also word-final /h/ may be
attenuated (and even disappear completely): (b'jaaH, -aaH, -aa) /b'jaah/ byaah˘
Stopstrictives
10.2.3. We only find the postalveopalatal diphonic pair, /c, G/ (C, ‚) (with the
corresponding çaspiratedÇ sequences, /ch, Gh/ (Ch, ‚H), and the possibility of
(äH), as for the stops): (C‘'tUR) /ca'tur/ £atur˚ ('Raa‚) /'raaG/ raaÀ˚ ('Caal) /'caal/
£aal˚ (C'haal) /c'haal/ £haal˚ ('‚aal) /'Gaal/ Àaal˚ (‚'Haal) /G'haal/ Àhaal˘ In various
contexts, there may be frequent attenuated realizations (¤ constrictives or approx-
imants), even after a pause (or, instead, sharper ones, ¤ stops), for /c/ (ë, À÷ T), /ch/
(ëh, Àh÷ Th), /G/ (ò, ö÷ D), /Gh/ (òH, öH÷ DH).
Constrictives
10.2.4. In actual fact, we find two voiceless constrictives: /s, S/ (s, ë) (sometimes,
the latter is realized as velarized postalveolar, (∑), ¤ (ß) with velarization) and the
voiced /w/ (6, j) (whose rounded semi-velar approximant variant, (j), mainly oc-
curs after C˚ after /uu, u/, and sometimes after a pause – but, in actual fact, the
two types alternate quite freely): ('sÄ™R) /'sEEr/ saer˚ ('kooëIë) /'kooSiS/ koo∞i∞˚ (&R‘6I-
'6aaR) /rawi'waar/ rawiwaar˚ ('nii6, 'nii˘, 'niiM) (with possible nasalization, near
nasalized V]\ /'niiw/ niiw˚ (6'Råt) /w'rat/ wrat˚ ('HåUja) /'hauwaa/ haowaa˚ (s'jåRg)
/s'warg/ swarg˚ (6ê'HÅÅ, |j∑-) /wa'hÅÅ/ wahãã˘
In Sanscrit loans, we also find /ß/ (ß) (which currently becomes /S/ (ë): ('bIß, 'bIë)
338 a handbook of pronunciation
/'biß/ bi«; on the other hand, /s/ becomes (ß), before /˛/: ('kå߲) /'kas˛/ ka«= (shown
in writing as well). In Persian, Arabic, and English loans, we find /F, z/ (å, z) too
(which currently become /ph, G/ (ph, ‚)): (åê'›iiR, phê-, -'kiiR) /Fa'›iir, -'kiir/ fa-
qiir˚ ('åUt, p'hUt) /'Fut/ fut˚ (ba'zaaR, ba'‚aaR) /baa'zaar/ baazaar˘
Approximants
10.2.5. In this articulatory manner, there are two fundamental phonemes, /j,
h/ (j, ã÷ H, h); (the semi-palatal) (ã) occurs between V˚ in unstressed syllables: ('jee)
/'jee/ yee˚ (dH'jaÅn) /dh'jaan/ dhyaan˚ ('lIe, 'lie, 'lIãe) /'lijee/ liyee˘ ‹en in unstressed
final position (with an inherent a]˚ its sound is (-ãÉ, -™); when preceded by i˚ it
drops, often closing /i/ {(I) = (i)}, except in careful pronunciation: (s‘'måãÉ, s‘'må™)
/sa'maj/ samay˚ (k'ßåãÉ, -å™, k'ë-) /k'ßaj/ k«ay˚ ('CaaãÉ, 'Caa™) /'caaj/ £aay˚ (k'ßåt-RIãÉ,
kßåt'RIãÉ, -IÉ, -iÉ, kë-) /k'ßatrij/ k«atriy˘
Generally, the /h/ phoneme is (H)÷ while (h) occurs with voiceless çaspiratedÇ
consonants. Near nasalized V˚ /h/ becomes nasalized: ('Haathi) /'haathii/ haathii˚
('gåRH) /'garh/ garh˚ (&pÄHÉ'laa) /pah'laa/ pahlaa˚ ('teeRÉH, te'RÄH) /'teerah/ teerah˚
(HAs'naa) /hÅs'naa/ hãsnaa˚ ('bÅÅH) /'bÅÅh/ bããh˚ (kx'HÅÅ) /ka'hÅÅ/ kahãã˘
˛ere are two further approximants, for Persian and Arabic loans, for which the
o‚cial uvular constrictives symbols are often used, /X, º/ (or, even less precisely,
the velar ones, /x, Ÿ/); they are uvular approximants (˜, ˜) (but, for the voiced one,
more often, the uvular tap, (r), occurs). However, currently they become /k, g/ (k,
g): (d‘'˜ål, d‘'kål) /da'Xal/ daÆal˚ ('˜aana, 'kaana) /'Xaanaa/ Æaanaa (Ô (k'haana)
/k'haanaa/ khaanaa]˚ ('baar, 'baa˜, 'baag) /'baaº/ baa‚ (Ô ('baag) /'baag/ baag]˘
£Trills∞
10.2.6. In this category, we have an alveolar tap, /r/ (R), and a postalveolar flap,
/#/ (®), which oppose distinctively (and sometimes the former may be stronger, a
true trill (r), or –on the contrary– weaker: an approximant (¸); often, the latter is
weaker than normal: an approximant (Í), similar to British English (>), but with
no lip rounding at all): (H‘'Rå®) /ha'ra#/ haraó˚ (lå®'kaa) /la#'kaa/ laókaa˘ ˛ere is al-
so the sequence /#h/ (®H, ÍH): (bå®'Haa, -Í'Haa) /ba#'haa/ baóhaa (Ô (bê'®aa, -'Íaa) /ba-
'#aa/ baóaa]˘
Laterals
10.2.7. ˛ere is only one lateral phoneme, /l/ (l, $, ¬); besides, there is the se-
quence /lh/ (lH): ('laat) /'laat/ laat˚ (mê'laal) /ma'laal/ malaal˚ ('kå¬ C‘'loo) /'kal ca-
'loo/ kal £aloo˚ ('kå¬ jÉ'HÅÅ 'aao) /'kal ja'hÅÅ 'aaoo/ kal yahãã aaoo˚ ('Ãaa$Ãa) /'ÃaalÃaa/
ƒaalƒaa˚ ('kuula) /'kuulaa/ kuulaa˚ ('kuulHa) /'kuulhaa/ kuulhaa˚ ('deelHi) /'deelhii/
Deelhii˘
10. hindi 339
Structures
10.3.0. We will consider here the relationships between segments, syllabic struc-
ture, stress, and intonation. Most examples –re-transliterated and transcribed– are
based on recordings expressly made (taken from Shukla, 2000).
A typical Hindi pronunciation shows a particular kind of paraphonic setting
with breathy voice §ÿ@, or, less positively, with tense voice ≠§÷@.
Taxophonics
10.3.1.1. As for V˚ which oppose as short and çlongÇ, or rather (very) narrow
diphthongs (generally represented by geminating symbols, Ô § 10.1.1), gemina-
tion is distinctive for C too: (pê'taa) /pa'taa/ pataa˚ ('påtta, påt'taa) /'pattaa/ pattaa˚
(bê'Caa) /ba'Caa/ ba£aa˚ ('båCCa, båC'Caa) /'baccaa/ ba££aa˚ (U'see) /u'see/ usee˚
('Usse, Us'see) /'ussee/ ussee˘ ˛is fact changes the syllabic structure, consequently,
often even stress changes (although without phonemic relevance).
We have already mentioned and demonstrated in our examples, that –in
Hindi– sequences of two or three C syllabify moving the last one to the beginning
of the next syllable: ('såt-jÉ) /'satja/ satya˚ ('gåd-Ha) /'gadhaa/ gadhaa˚ (åbH'Råk) /abh-
'rak/ abhrak˚ ('ëUk-l‘) /'Sukla/ ∞ukla˚ (åkh'®aa) /akh'#aa/ akhóaa. Obviously, in case
of isolated initial sequences, the syllabicity scale joins the elements in one syllable,
although some slight di‡erence is maintained; however, if they are internal, they
divide into two syllables, including the preceding V\ (‚'Hiil) /G'hiil/ Àhiil. ˛e
geminates are (0˘0) between V˚ but (0:˘) + C˚ even for /j, w, h, r, #, l/ (in this last
context, especially short stops and stopstrictives are often realized as (0:˘) + C˚ thus
neutralizing the di‡erence with geminates): ('bUÃ:Ha) /'buÃÃhaa/ buƒƒhaa˚ (6Id[:]-
'ãaal‘ãÉ, -lÄ™, -lÉ) /wid'jaalaj/ widyaalay˘ If the number of C in the sequences ex-
ceeds three, the last two move to the beginning of the second syllable: (såt'jaaRth-
pR‘&kaë) /sat'jaarthprakaaS/ Satyaarthprakaa∞.
Another interesting phenomenon, that complicates the description and the ac-
quisition of this language, concerns the epenthesis of a vowel; therefore, an attenu-
ated vocoid is inserted (Ô § 10.1.3, although here we indicate it simply as an audi-
ble o‡set, ($)), not only in complex consonant clusters, but also –in not quick pro-
nunciation– at the end of words, even after a single consonant: (‘'gåR$) /a'gar/ agar˚
('uupåR$) /'uupar/ uupar˚ (Cål$'naa) /cal'naa/ £alnaa˚ ('phuul$) /'phuul/ phuul˚ ('ëåR$-
båt) /'Sarbat/ ∞arbat˚ (båd$'maaë) /bad'maaS/ badmaa∞˚ (lå®$'kaa) /la#'kaa/ laókaa˚
(khI®$'kii) /khi#'kii/ khiókii˚ (åbH'jaas$) /abh'jaas/ abhyaas˚ (s‘'måã$) /sa'maj/ samay˚
(Rakh$'naa) /rakh'naa/ rakhnaa˚ (Ug$'naa) /ug'naa/ ugnaa˚ ('uugH$na;) /'uughnaa/
££ghnaa˚ ('på®$) /'pa#/ paó˚ ('på®H$) /'pa#h/ paóh…
/d'juut/ dyuut˚ (dH'jåni, -∑'nii) /dh'wani/ dwani˚ (dH'RU6) /dh'ruw/ dhruw˚ (Ã'joo®-
Hi, Ãjo®'Hii) /'Ãjoo#hii/ ƒyooóhii˚ (C'jUt) /c'jut/ £yut˚ (‚'jooti) /G'jooti/ Àyooti˚ (k'jaa)
/k'jaa/ kyaa˚ (k'jaath) /k'waath/ kwaath˚ (kh'jaab) /kh'waab/ khwaab˚ (g'RaÅm)
/g'raam/ graam˚ (g'laÅni) /g'laani/ glaani˚ (gH'RaÅπ, -N, -n) /gh'raaN/ ghraa¿.
More: (˜'jaal, kh'j-) /X'jaal/ Æyaal˚ (H'Raas) /h'raas/ hraas˚ (ëmê'ëaÅn) /Sma'Saan/
∞ma∞aan˚ (ë'RaÅnt) /S'raant/ ∞raant˚ (ë'låth) /S'lath/ ∞lath˚ (s'kåndH) /s'kandh/ skandh˚
(skhx'lån) /skha'lan/ skhalan˚ (s'tåR) /s'tar/ star˚ (s'thål) /s'thal/ sthal˚ (s'mIt) /s'mit/
smit˚ (s'naÅn) /s'naan/ snaan˚ (s'jaaH) /s'jaah/ syaah˚ (s'jåRg) /s'warg/ swarg˚ (s'Rå߲a)
/s'ras˛aa/ sra«=aa˚ (z'jaada) /z'jaadaa/ zyaadaa˚ (6jåt'haa) /wjat'haa/ wyathaa˚ (6'Råt)
/w'rat/ wrat˘ In words of Sanscrit origin, we find /òkß/ as well (but people current-
ly change it into (kë)): (k'ßåπ, k'ßåN, k'ëån) /k'ßaN/ k«a¿˚ (k'ßiiR, k'ë-) /k'ßiir/ k«iir˘
Furthermore, for initial sequences with /òs0[0]/, current pronunciation prefixes
an epenthetic vowel, generally (I, I) – but (å, å) is possible as well (and also (‘, ‘),
with the other variants seen, Ô § 10.1.3), before non-front vowels: (s'tRii, Is-, Is-)
/s'trii/ strii˚ (sp'huuRti, Is-, Is-, ås-, ås-, ‘s-, ‘s-) /sp'huurti/ sphuurti˚ (s'naÅn, Is-, Is-, ås-,
ås-, ‘s-, ‘s-) /s'naan/ snaan (also with /S/: (ë'look, Ië-, Ië-, åë-, åë-, ‘ë-, ‘ë-) /S'look/ ∞look]˘
10.3.1.4. Short (I, å, U) /i, a, u/, in final position, have the peculiarity of fading
(Ô § 10.1.3), to a point in which they drop as well, as happens to /a/, (å÷ å, ‘, ê, x,
∑÷ `), or strengthen, ¤ (i, u), simply as closer realizations of /i, u/, or actually be-
coming /ii, uu/. In this case, stress shifts are also possible, according to the weight
of the syllables constituting given words: ('Rå6i, R‘'6ii) /'rawi/ rawi˚ ('tIthi, tIt'hii)
/'tithi/ tithi˚ ('ëIëu, ëI'ëuu) /'SiSu/ ∞i∞u˚ ('6åsu, 6ê'suu) /'wasu/ wasu˘ ˛e change Vò =
VVò is not considered completely neutral, in spite of being very widespread indeed
(with many supporters too).
≈nally, as many examples have already shown, sequences of /'ééö/ + /é, ò, 0/
are realized as ('é–ö): ('HaÅni) /'haani/ haani˚ ('kaÅm) /'kaam/ kaam˚ (ak'RaÅnt)
/aak'raant/ aakraant˘
10. hindi 341
10.3.1.5. Dealing with consonants, we have already seen the assimilatory char-
acteristics within words. ˛e same holds in sentences, for words linked by mean-
ing, especially if joined in rhythm groups, as in: ('tiim ma'taÅe) /'tiin maa'taaee/ ti-
in maataaqq˚ ('tiim pê'piite) /'tiin pa'piitee/ tiin papiitee˚ ('tiiN 'Ãaal) /'tiin 'Ãaal/ ti-
in ƒaal˚ ('tii˙ kx'mål) /'tiin ka'mal/ tiin kamal˚ ('tii˙ 'gaaÉ) /'tiin 'gaaj/ tiin gaay˚
('tii~ 'CaaÉ) /'tiin 'caaj/ tiin £aaj˚ ('tii~ '‚aal) /'tiin 'Gaal/ tiin Àaal˚ ('tiin ë‘'Raab) /'ti-
in Sa'raab/ tiin ∞araab˚ ('tiin 'jaaR) /'tiin 'jaar/ tiin yaar˚ ('tiiM 'jåRß, -Rë) /'tiin 'warß/
tiin war«˚ ('tii, ›x'saai, -˙ kx-) /'tiin ›a'saaii/ tiin qasaaii˚ ('tii« 'Haath) /'tiin 'haath/
tiin haath˘
Besides, both within words and sentences, in not slow –nor emphatic– speech,
for diphonic C˚ voicing assimilation is regular, to the second element of a sequence
(while a possible /h/ is lost): ('ågbåR, åg'båR) /'akhbar/ Akhbar˚ (tåz'biiH) /tas'biih/
tasbiih˚ ('CUb 'bÄ™˛ho, 'CUb b™˛'hoo) /'cup 'bEE˛hoo/ £up bae=hoo˚ ('Chåb gx'jaa)
/'cap ga'jaa/ £hap gayaa˚ ('saad 'bÄ™˛ho, 'saad b™˛'hoo) /'saath 'bEE˛hoo/ saath bae-
=hoo˚ ('åp 'tåk) /'ab 'tak/ ab tak˚ ('Haad 'doo) /'haath 'doo/ haath doo˚ ('aakpåR) /'aag-
par/ aak par˚ ('eeg 'dIn) /'eek 'din/ eek din˚ ('‚åpki) /'Gabki/ Àab ki˚ ('baa˜ 'kaa, -k
'k-) /'baaº 'kaa/ baa‚ kaa˘
Stress
10.3.2.1. In Hindi, stress position is not distinctive; in fact, the same speaker,
in di‡erent occasions, may stress di‡erent syllables of the same word. Furthermore,
these fluctuations also depend on the placing of words in sentences, on nearby
words, on orthological highlighting and emphasis. Still more important is the fact
that stress is distributed among rhythm groups, usually moving back from the end,
according to syllabic çweightsÇ. ˛is holds for isolated words as well, but always
with a certain flexibility. For instance, we normally have ('HIndi) /'hindii/ hindii˚
but (HIn'dii&ke, -&ka) /hin'diikee, -kaa/ hindii kee˚ hindii kaa˘
On the other hand, given its non-distinctiveness, native speakers use stress as
something fluctuating (without fully realizing it), frequently alternating, for rhyth-
mic reasons, within rhythm groups. After all, it is the same thing for segmental
duration and for syllabic pitch, in languages where they are not distinctive: they
may change quite freely, without real problems. In Hindi intonemes (both for the
three marked and the unmarked one, or continuative), the terminal posttonic syl-
lable, generally, bears a secondary stress, which complicates the (already unsettled)
di‡erentiation and identification of stress strength on the di‡erent non-light sylla-
bles (but sometimes on the light ones too).
zations of /a/ (‘, É, x, ∑, ê) do not count (Ô § 10.1.3), being usually more elusive,
(‘, É, x, ∑, ê), and are just a mere physical support, necessary to make a word pro-
nounceable, but they may drop completely.
Furthermore, we have mid syllables˚ constituted by VC (¤ a short V and a C]˚
or by VV\ ('bål) /'bal/ bal˚ (k'jaa) /k'jaa/ kyaa˚ ('HIndi) /'hindii/ hindii˚ ('ëåjja) /'Saj-
jaa/ ∞ayyaa˚ including çdisyllabicÇ syllables in counting morae, as (b'Haai) /b'haaii/
bhaaii (ç/aa/ + /ii/Ç = four morae).
Besides, there are heavy syllables˚ formed by VCC˚ or VVC (more rarely VVCC
as well): ('ånt) /'ant/ ant˚ ('aÅm) /'aam/ aam˚ ('aaRt) /'aart/ aart˚ ('6aaëp) /'waaSp/
waa∞p˘
10.3.2.3. Stress assignment, in isolated words, is done on the basis of the identi-
fication of the çheaviest syllableÇ, as in: (U'paadHi) /u'paadhi/ upaadhi˚ (U'paÅnt)
/u'paant/ upaant˚ (&ad-jo'paÅnt) /aadjoo'paant/ aadyoopaant˚ (kx'lii) /ka'lii/ kalii˚
(kx'maÅn) /ka'maan/ kamaan˚ ('kaÅmna) /'kaamnaa/ kaamnaa˚ (ko'πaaRk, ko-
'NaaRk, ko'naaRk) /koo'Naark/ koo¿aark˚ ('˜aÅndani) /'Xaandaanii/ Æaandaanii˚
(‚'jaÅmIti) /G'jaamiti/ Àyaamiti˚ (&tabe'daaR) /taabee'daar/ taabeedaar˚ (&tIgU'naa)
/tigu'naa/ tigunaa˚ (p&RIthji'Raa‚-Ra&so) /prithwii'raaGraasoo/ prithwiiraaÀraasoo.
More: (bån'duugba&zi) /ban'duukbaazii/ banduukbaazii˚ (måR'daÅ˙gi) /mar-
'daangii/ mardaangii˚ ('maÅndHa&ta) /'maandhaataa/ maandhaataa˚ (&såt-jåpR‘'kaaë)
/satjapra'kaaS/ Satyaprakaa∞˚ (såt'jaaRth-pR‘&kaë) /sat'jaarthprakaaS/ Satyaarthprakaa∞˚
(&sa˙go'paÅ˙g) /saangoo'paang/ saangoopaang˚ (sa'maÅn) /saa'maan/ saamaan˚ ('HaÅ-
ni) /'haani/ haani˘
If a word has more than one non-light syllables of the same weight, there are two
possibilities: the stress preferably hits either the last syllable but one or the last but
two (or even the very last, especially if constituted by VV˚ as happens more often
within a sentence, with particles and postpositions): (ë‘'kUnt‘&la, ë‘&kUnt‘'laa) /Sa-
'kuntalaa/ ∞akuntalaa˚ (gx'®eeRI&ãa, gx&®e;RI'jaa) /ga'#eerijaa/ gaóeeriyaa˚ (bê'HaadU&Ri,
bê&HadU'Rii) /ba'haadurii/ bahaadurii˚ (&mêHab'HaaRåt) /mahaab'haarat/ mahaa-
bhaarat˚ (jUd'HI߲hIR, &jUdHI߲'hIR, -ë˛-, -ë'˛-) /jud'hi߲ir/ Yudhi«=hir˚ (s‘'HuulI&ãåt, s‘&Hu-
lI'jåt) /sa'huulijat/ sahuuliyat.
More examples: (a'ëiiR-jad, &aëiR'jaad) /aa'Siirwaad/ aa∞iirwaad˚ ('å›xl&månd, &‘-
›xl'månd) /'a›lmand/ aqlmand˚ (&ana'kaÅni, a&naka'nii) /aanaa'kaanii/ aanaakaanii˚
(dH'jåni, -∑'nii) /dh'wani/ dhwani˚ (påR'jåRti, &påRjåR'tii) /par'wartii/ parwartii˚
('båRtån, båR'tån) /'bartan/ bartan˚ ('bås-ta, bås'taa) /'bastaa/ bastaa˚ (b'Haala, bHa-
'laa) /b'haalaa/ bhaalaa˚ (s‘'mIti) /sa'miti/ samiti˚ (sa'laÅna, &sala'naa) /saa'laanaa/
saalaanaa˚ (så~'Caalån, &så~Ca'lån) /san'caalan/ san£aalan˚ (&HeRap'heeRi, He&Raphe-
'Rii) /heeraap'heerii/ heeraapheerii˘
10.3.2.4. ˛e same stress pattern occurs in inflected and derived words as well:
(&båd-HI'kÚÚ) /badhi'kÚÚ/ badhikõõ˚ (&låg-HU'tåR) /laghu'tar/ laghutar˚ (&ëUCI'tåm) /Su-
ci'tam/ ∞u£itam˚ (mås'leega, &måsle'gaa) /mas'leegaa/ masleegaa˚ ('laapê&ta;, &lapê'taa)
/'laapataa/ laapataa˚ (&gxlI'jaaRa, gx&lIãa'Raa) /gali'jaaraa/ galiyaaraa˚ (pa'gålpån, &pa;-
gål'pån) /paa'galpan/ paagalpan˚ (sUn'dåRta, &sUndåR'taa) /sun'dartaa/ sundartaa˚
(b™H'naapa, &b™Hna'paa, b™&H™na'paa) /bah'naapaa/ bahnaapaa˘
10. hindi 343
10.3.2.5. Lexemic compounds, normally, have the following stress pattern (' “)
/' “/ (since the second stress, which is generally more attenuated, may sometimes
reach a degree of prominence which is relatively slightly more perceptible than a
secondary stress): (R‘'sooig“Hår) /Ra'sooiig“har/ rasooiighar˚ ('deeënI“ka[a]la) /'deeSni-
“kaalaa/ dee∞nikaalaa˚ ('6Iëj∑“ko[o]ë) /'wiSwa“kooS/ wi∞wakoo∞˚ ('duuRdåR“ëi[i]) /'duur-
dar“Sii/ duurdar∞ii˚ ('dI¬-‚‘“la[a]) /'dilGa“laa/ dilÀalaa˚ ('nImn‘lI“khIt) /'nimnli“khit/
nimnlikhit˚ ('CåndR‘“ka[Å]nta) /'candra“kaantaa/ £andrakaantaa˚ ('CåndR‘“ka[Å]nta-
“sånt‘ti) /'candra“kaantaa“santati/ £andrakaantaasantati.
More: (mê'Haasåb“Ha[a]) /ma'haasab“haa/ mahaasabhaa˚ ('HIndumê“Ha[a]såb-
“Ha[a]) /'HInduma“haasab“haa/ mahaasabhaa˚ ('tuu 'kåRm“Hi[i]M j'jåkti&HÄ) /'tuu
'karm“hiin w'jaktihEE/ tuu karm-hiin wyakti hae˚ ('tuu 'kåRm“Hi[i]«&HÄ) /'tuu 'karm-
“hiinhEE/ tuu karm-hiin hae˚ ('aÅm“ja[Å]m) /'aam“waam/ aam-waam˚ ('kaÅnÚ-
“ka[Å]n) /'kaanÚÚ“kaan/ kaanõõ-kaan˚ ('kåÅm“ka[a]‚) /'kaam“kaaG/ kaam-kaaÀ˚ ('Ä™-
sa“6Ä[™]sa) /'EEsaa“wEEsaa/ aesaa-waesaa˘
Complete reduplications always keep both stresses: ('laal 'laal) /'laal'laal/ laal-
-laal˚ (d'HiiRe d'HiiRe) /d'hiireed'hiiree/ dhiiree-dhiiree˘
Intonation
10.3.4. û 10.4 shows the preintonemes and intonemes of the neutral Hindi lan-
guage. We will merely give some illustrative examples, to use for comparisons. Let
us observe that the posttonic syllables of the interrogative ((2 1)) and suspensive
((2 2)) intonemes, which are in the higher parts of the mid and high bands (as will
be clear from û 10.4), are generally accompanied by falsetto (for all kind of voic-
es), which is marked with (Ú) after the intonemes, in phonotonetic transcriptions.
In addition, paraphonically, there is a peculiar type of çIndian voiceÇ, with murur
344 a handbook of pronunciation
voice §ÿ@ or, in the broadest accents (≠), with tense voice §÷@:
/./: (&mìçHIndi bolçtaa–Hu3 3) /míí'hindii bool'taahuu./ Mãq hindii booltaa h££.
/?/: (¿k'jaa6™h pUs'tåkpå®H R‘'Hii&HÉ21Ú) /¿k'jaawah pus'takpa#h ra'hiihEE?/ Kyaa
wah pustak paóh rahii hae?
/÷/: (‚åb'mì™ ‚ål'dii ‚ål'dii bol'taa&Hu2 2Ú ap'meeRi 'baat n‘'Hii &s‘måCçtee3 3)
/Gab'míí Gal'dii Gal'dii bool'taahuu÷ aap'meerii 'baat na'hii samaGh'tee./ êab
mãq Àaldii-Àaldii booltaa h££, aap meerii baat nahii samaÀhtee.
û 10.4. Hindi preintonemes and intonemes.
/ / (2 2 ç 2 2 ç 2 2 ç 2) /./ (2 ç 3 3)
/¿ / (¿ 2 2 ç 2 2 ç 2 2 ç 2) /?/ (2 ' 2 1)
/¡ / (¡ 2 2 ç 2 2 ç 2 2 ç 2) /÷/ (2 ç 2 2)
/˚ / (˚ 2 2 ç 2 2 ç 2 2 ç 2) /,/ (2 ' 2)
Text
10.4.0. ˛e story †e North Wind and the Sun follows. According to the Pho-
netic method we start with the versions in Hindi pronunciation of (neutral Brit-
ish) English (the written text is given in § 2.5.2.0). ˛e Hindi translation follows,
in its neutral version.
At the end, as always, there is the version which gives the English pronuncia-
tion of Hindi, by neutral British speakers, fluent in Hindi (after prolonged con-
tact with native speakers, but with no help from the phonetic method), who have
adequately learned the relative prominences, but who substantially use segmental
and suprasegmental elements which are typical of neutral British English (for refer-
ence purposes, although, of course, a neutral accent is not so common). Obviously,
the same principle is valid for the foreign pronunciations of English, given first.
Speakers of American English could prepare their own version both of the Hin-
di pronunciation of English and of their pronunciation of Hindi, as an excellent
exercise, by listening to native speakers, best of all after recording them. Of course,
speakers of other languages could do the same thing. ˛e author would be happy
to receive their transcriptions and recordings, both in case of help –should they
need it– and to make their contribution known to others (possibly in our website
on canIPA Natural Phonetics – Ô § 0.12).
10. hindi 345
Hindi text
10.4.2. Uttarii hawaa aor suuraÀ is baat par Àhagaó rahee thee ki ham doonõõ
mqq zyaadaa balwaan kaon hae. Itnee mqq garam £oogaa pahnee eek musaafir udhar
aa niklaa. Hawaa aor suuraÀ doonõõ is baat par razii hoo gayee ki doonõõ mqq see
Àoo pahlee musaafir kaa £oogaa utarwaa deegaa wahii zyaadaa balwaan samÀhaa
Àaayeegaa.
Is par uttarii hawaa apnaa puuraa zoor lagaakar £alnee lagii. Leekin wah Àaesee
Àaesee apnaa zoor baóhaatii gayii waesee waesee musaafir apnee badan par £oogee koo
aor bhii zyaadaa kas kar lapee=taa gayaa. Ant mqq hawaa nee apnii koo∞i∞ band kar
dii. Phir suuraÀ teezii kee saath niklaa aor musaafir nee turant apnaa £oogaa utaar
diyaa. Is liyee hawaa koo maannaa paóaa ki un doonõõ mqq suuraÀ hii zyaadaa bal-
waan hae.
Kyaa tumhqq yah kahaanii a££hii lagii? Kyaa phir see sunaa££?
10.4.3. (&Utt‘çRii H‘'6aa2 ç∏øR çsuu–Rå‚2 2Ú| Is'baat2 &påR‚H‘'gå® R‘&Heçthee2 2Ú| ki'Håm
'doÚnÚ&me2| zçjaada bål'6aÅn2| çk∏9n–HÄ3 3|| It'nee&me2| gxçRåm Cooçgaa &pÄHÉçnee2 2Ú|
'eek2 mUçsaa&åIR2 2Ú| Ud'HåR2\ çaa nIkçlaa3 3|| H‘ç6aa2 2Ú| ç∏øR 'suu&Rå‚2| çdoÚnÚ2 2Ú Is'baat2
&påRRaçzii Hogxçjee2 2Ú|| kiçdoÚ&nÚ çmee&se2 2Ú &‚opåHçlee mU'saa&åIR2| &kaCoçgaa3 3| ìUçtåR-
6a deçgaa2 2Úœ|| 6ê'Hii2| zçjaada bål'6aÅn2 såm‚'Haa ‚aãeçgaa3 3||
çIs&påR2 2Ú| ì&Utt‘çRii H‘'6aa2œ åpçnaa çpuuRa 'zooR2 l‘çgaakåR Cålçnee l‘çgii3 3|| 'lee&kIn2|
'6ÄH2 ìç‚Ä™se '‚Ä™&se2 &åpnaçzooR2 2Ú\ bê®çHaati gxçjii2 2Úœ| ç6Ä™se '6Ä™&se2\ mU'saa&åIR2|
ìåpçnee bêçdåm 'påR2œ| çCooge 'koo2| ç∏øR bçHii zçjaada 'kås&kåR2| l‘çpee˛ta gxçjaa3 3||
çånt 'mee2| H‘ç6aÅne çåpni çkoo&ëIë2 2Ú| çbånd kåRçdii3 3|| p'hIR2| 'suu&Rå‚2 teçzii keçsaath
nIkçlaa3 3|| ç∏øR mUçsaaåIR 'nee2| çtURånt çåpna Coçgaa3 3 ìUçtaaR dIçjaa3 3œ| IsçlI&ãe2 2Ú| H‘-
ç6aa koçmaÅnna pê'®aa2| &kiUnçdoÚnÚ çmee2 2Ú| çsuuRå‚ 'Hii2 zçjaada bålç6aÅn–HÄ3 3||
¿çkjaa tUm'Hee2| ¿çjÄh kx'HaÅ&ni2 ¿åC:çhii l‘'gii21Ú| ¿çkjaa phIRçsee sU'naa&u21Ú|||)
346 a handbook of pronunciation
11.0.2. We will see that Chinese syllables always begin with a contoid, even
when they lack a real initial. In the case of close vowels, //òi, òy, òu// (whereas /M/
only occurs after the initials /q[h], s/}, we actually have /ji, ¥y, wu/ (recognized by
pinyin too: yi˚ yu˚ wu).
˛e other vowels, /e, X, o, a/, are preceded by (ö) (or, but less advisably, by the
following voiced phones (y, H, r, «) – respectively, velar semi-approximant, {lenis}
laryngeal approximant, uvular tap, and semi-provelar nasal).
11.0.3. Let us observe that the Chinese phonetic çtraditionÇ still follows the
phonic çfantasiesÇ of the Swedish sinologist Karlgren, who thought that Mandarin
had two çretroflexÇ (or çapicalÇ) vowels, rendered with çq, QÇ. Actually, the former
is simply a back-central vocoid, (M) (where neither çretroflexionÇ nor the lamina
are active); the latter, on the other hand, is precisely a postalveolar approximant
contoid found in the initial (⁄) />/ r˚ changed into the corresponding intense
(çsyllabicÇ) contoid, (Ö). ˛e same çphonetic traditionÇ also lists another rare
çretroflex (mid) vowelÇ, rendered with çæ&Ç which in actual fact is just the sequence
(öX⁄) /X>/ er˘ It seems surprising, to say the least, that the Journal of the Internation-
al Phonetic Association has accepted and published (in 2003), as an çillustration of
the IPAÇ, a paper which still proposes çq, QÇ –although ipa-like masqueraded as ç(≥)Ç
in both cases!– by describing them as ç(apico-)laminal denti-alveolarÇ (where no
groove is rightly mentioned) and çapical post-alveolarÇ approximants respectively.
˛e latter is correctly described (¤ can(Ö)), while the former is still not (¤ can(¡));
there is a great (articulatory, auditory, and acoustic) di‡erence between (¡) and
(M).
348 a handbook of pronunciation
11.0.4. Continuing, we have e (X, É, ‘, x, ∑) /X/, with its çvariantÇ (™) /e/, in con-
tact with /j, ¥, i/, and (É) in /ae/. But it is decidedly safer to consider /X/ and /e/ as
two separate phonemes, given the great phonetic di‡erence, because of interpho-
nemic reasons (even if, from an intraphonemic point of view, ¤ just within the
Chinese language, they could –or should– be considered as taxophones of the
same phoneme). In addition, o (ø, Ö) /o/, with its çvariantÇ (U) /u/ – which, on the
contrary, is a di‡erent phoneme. Instead, o –in contact with labials [mo˚ bo˚ po˚
fo˚ ou˚ wo/Cuo]– could be considered as a variant of /X/, as many Chinese phonolo-
gists do, by using ç(X, È)Ç. However, the occurrence of the syllables o˚ yo˚ lo (al-
though rare) requires that a phoneme /o/ is introduced, which will obviously oc-
cur near labials as well. ˛en we have a (a, Å, A) /a/, with its variants (jEn, ¥Än),
for ian˚ üan˘
≈nally, this phonetics çà la chinoiseÇ even has a çzero (graphic) vowelÇ, when
pinyin çsmartlyÇ economizes on vowels, by writing un for /wXn/ (w‘n), and iu for
/jou/ (jPU), ui for /wei/ (wÙI), where a vowel is definitely present, phonically! No
doubt, uen˚ iou˚ uei would have been much better. ˛anks to the use of diapho-
nemes, then, more realistically we will also have the diphthongs /ae, ao/ (ÅÉ, AÖ),
instead of a more abstract representation as /ai, au/: ai˚ ao (pinyin stays on the
fence: -i˚ -o)!
11.0.5. We will now present the o‚cial romanization. Starting from the (conso-
nantal) initials, which are 23 (with a double occurrence of y˚ with two di‡erent
values), we will not present them alphabetically, but by logical phonic groups, ac-
companied by their phonemic and çsimpleÇ phonetic transcriptions. ˛e latter is
çsimpleÇ, because it shows only one variant (whereas, below, we will provide all
necessary details):
b /p/ (p), p /ph/ (ph), m /m/ (m), f /f/ (f)÷
d /t/ (t), t /th/ (th), n /n/ (n), z /q/ (q), c /qh/ (qh), s /s/ (s), l /l/ (l)÷
˜ /fi/ (fi), > /fih/ (fih), ´ /ß/ (ß), r />/ (⁄)÷
j /q/ (Â), q /qh/ (Âh), x /s/ (¿)÷
g /k/ (k), k /kh/ (kh), h /h/ (˜)÷
w /w/ (w), y /j/ (j) “ /¥/ (¥).
11. chinese 349
11.0.6. ˛e finals are 39 (with a triple occurrence of i˚ with three di‡erent val-
ues). ˛ey are presented according to their graphic syllabic nucleus. ˛is time,
they are given in alphabetical order, even in the seven cases where the romaniza-
tion has not been a happy one (as we will show in >ç Ç≥). Also the exclamations e
/e/ (™), yo /jo/ (jø) occur (with various tonemes, including çzeroÇ). ˛e three finals
which do not occur without an initial are marked with an asterisk (*):
å
a /aò/ (a), ia˚ òya /jaò/ (ja), ua˚ òwa /waò/ (wa)÷
ai /ae/ (ÅÉ), uai˚ òwai /wae/ (waÉ)÷ yai /jae/ (sole word, yái)
an /an/ (Ån), ian˚ òyan /jan/ (jEn), uan˚ òwan /wan/ (wÅn), üan˚ òyuan (j/q/x +
uan) /¥an/ (¥Än)÷
a« /a˙/ (a,), ia«˚ òya« /ja˙/ (ja,), ua«˚ òwa« /wa˙/ (wA,)÷
ao /ao/ (AÖ), iao˚ òyao /jao/ (jaÖ).
™
e /Xò/ (X, Xx|, ’x|), ie˚ òye /jeò/ (j™), üe˚ òyue (j/q/x + ue) /¥eò/ (¥™)÷
ei* /ei/ (™I) {(ÉI)}, en /Xn/ (Én), er /X>/ (X⁄)÷
e«* /X˙/ (x,), ue«˚ òwe« /wX˙/ (w∑,) {(wø,)}.
¤
i /iò, òjiò/ (i, ji), (s-/z-/c-) /Mò/ (M) > ï (with ü]≥, (´-/˜-/>-) /ˇò/ (Ö) > ï ≥÷
in˚ òyin /in/ (In, òjIn)÷ i«˚ òyi« /i˙/ (¤˙, òj¤˙)÷
ui˚ òwei /wei/ (wÙI) {(w¤i)} > uei ≥.
ø
o /oò/ (ø, øx|), ou /ou/ (øU) {(∑U)}, uo˚ òwo /woò/ (wø, wøx|) {(w∑, w∑x)}÷
o«* /u˙/ (U,) > u« ≥, io«˚ òyo« /ju˙/ (jU,) {(¥U,)} > iu«˚ yu« ≥.
¨
u˚ òwu /uò, òwuò/ (u, wu)÷ ü˚ òyu (j/q/x + u) /yò, ò¥yò/ (y, ò¥y)÷
un˚ òwen /wXn/ (w‘n) > uen ≥; ün˚ òyun (j/q/x + un) /yn, ò¥yn/ ([ò¥]Yn) {([ò]¥In,
[òj]Yn)}÷
iu˚ òyou /jou/ (jPU) {(j¨u)} > iou ≥.
Vowels
11.1.2. We first find the close front phoneme, /i/: /[òj]i, [òj]in, [òj]i˙/ ([òj]i, [òj]In,
[òj]¤,) (C/y)i˚ (C/y)in˚ (C/y)i«\ (5ji) /5ji/ yi˚ ([~ii) /¶ni/ n`˚ (4jIn) /•jin/ yìn˚ (5ÂIn)
/5qin/ jin˚ (7j¤¥) /6ji˙/ yí«˚ (5¤,) /5qi˙/ ji«˘
˛e close front-central rounded, /y/, has a fairly limited distribution, ju˚ qu˚ xu˚
yu˚ nü˚ lü (with or without -n), /[¥]y, [¥]yn/ (¥y, ¥Yn) [variants with -n˚ (¥In, jYn)]:
(5Ây) /5qy/ ju˚ ([·yy) /¶qhy/ qà˚ (7¿y) /6sy/ xú˚ ([¥yy) /¶¥y/ yà˚ (5ÂYn, 5Â¥In)
/5qyn/ jun˚ (4·Yn, 4·¥In) /•qhyn/ qùn˚ (7¿Yc, 7¿¥Ic) /6syn/ xún˚ (4¥Yn, 4¥In,
4jYn) /•¥yn/ yùn˘
More conveniently, pinyin could have used çzü˚ cü˚ sü˚ yüÇ, by phonemically in-
terpreting these syllables as we do: (Ây, Âhy, ¿y) /qy, qhy, sy/. It could also have
avoided introducing an allograph without dieresis, by using yü for o‚cial yu (¥y)
/¥y/ as well. Indeed, /¥y/ could even be /jy/; but since the phoneme /¥/ is present,
it is more adequate to use it. In this way, as it seems convenient, it could also be
used to show –even phonemically– the possible oscillations between (¥Yn, jYn,
¥In): /¥yn/, and ç/jyn, ¥in/Ç). ˛erefore, it could have been possible to write
çz/c/siu«Ç, instead of the o‚cial j/q/xio« for (ÂjU,, ÂhjU,, ¿jU,) /qju˙, qhju˙,
sju˙/ (thus also è-u«¶˚ instead of the o‚cial -o« for (0U,) /0u˙/).
11.1.3. ˛en we find the close back-central (without lip rounding!), /M/, which
only occurs in /qM, qhM, sM/ (qM, qhM, sM) zi˚ ci˚ si\ (7qœ) /6qM/ zí˚ (4q·M)
/•qhM/ cì˚ (5sM) /5sM/ si (but we could extend its use to a sequence like ç/M>/Ç, in
place of /ˇ/, for /fiˇ, fihˇ, ߡ, >ˇ/ (fiÖ, fihÖ, ßÖ, ⁄Ö) ˜i˚ >i˚ ´i˚ ri\ ([fiÎÎ) /¶fiˇ/ ˜`˚
(5fi·Ö) /5fihˇ/ >i˚ (4ßÖ) /•ßˇ/ ´ì˚ (7⁄Î) /6>ˇ/ rí]˘
˛e last close phoneme is back rounded, /u/ (u, U), and we are obliged to use
two di‡erent graphemes, because of the nature of pinyinization: u˚ o« (instead of
a more logical u«]\ ([wuu) /¶wu/ wà˚ (4lu) /•lu/ lù˚ (5kU,) /5ku˙/ go«˚ (7fiU¥) /6fiu˙/
˜ò«˘
/aò/ (a) (ò)a÷ /jaò/ (ja) ia˚ òya÷ /waò/ (wa) ua˚ òwa
w‘n, x,, w∑,) e˚ er˚ en˚ un/wen˚ e«˚ ue«/we«. ˛e last one has the variant (wø,)
(which, if necessary, could be indicated as /wo˙/): (7ö$[*|]) /6X/ è˚ (5öÉn) /5Xn/ en˚
(4öX⁄) /•X>/ ér˚ (7fËc) /6fXn/ fèn˚ ([w‘‘c) /¶wXn/ w‘n˚ (7f*¥) /6fX˙/ fè«˚ (5w∑,÷ 5wø,)
/5wX˙/ we«˚ (7w≤¥÷ 7wø¥) /6wX˙/ wè«˘
≈nally, we have /o/ (öø[x|]) o˚ which can occur in the syllables /o, jo/ (ø[x|], jø[x|]),
in addition to normal /mo, po, pho, [0]wo/ ([ø[x|], (ø[x|], (hø[x|], [0]wø[x|]), mo˚
bo˚ po˚ wo/Cuo\ (5[ø[x|]) /5mo/ mo˚ (4(ø[x|]) /•po/ bó˚ (7(·ø[*|]) /6pho/ pò˚ ([wøø,
[wø*|) /¶wo/ wœ˚ (5fiwø) /5fiwo/ ˜uo˚ (Ò˜A,5jø) /•ha˙5jo/ há«yo˚ (7öø[*|]) /6o/ ò! ‹en
final, before a pause, /X, o/ normally have the accompaniment (x), as can be seen from
various examples. Instead of /mo, po, pho/, we could even pose /mwo, pwo, phwo/,
for ([ø[x|], (ø[x|], (hø[x|]), while maintaining /mou, phou/ (møU, phøU) (/pou/
does not occur), by dissimilation.
11.1.5. Lastly, we find the open vowel, /a/, with its distributions and various
taxophones, /[0]aò, jan, ¥an, [w/0]an, [j]a˙, wa˙/ (öa, jEn, ¥Än, [w/0]Ån, [j]a,, wA,):
(5öa&ëa) /5aqa/ aza˚ (7t·a) /6tha/ tà˚ (4jEn) /•jan/ yán˚ ([tjEEc) /¶tjan/ di∑n˚ (7¥Éc)
/6¥an/ yuàn˚ (5·¥Än) /5qh¥an/ quan˚ ([öÅÅc) /¶an/ ∑n˚ (4p·Ån) /•phan/ pán˚ (4p·a,)
/•pha˙/ pá«˚ ([wÅÅc) /¶wan/ w∑n˚ (5ja,) /5ja˙/ ya«˚ ([wAA¥) /¶wa˙/ w∑«˘
Diphthongs
11.1.6. Moving to diphthongs (Ô û 11.3), we have /[w]ei, [w]ae, [j]ao, [j]ou/ (™I,
wÙI, ÅÉ, waÉ, jaÖ, AÖ, øU, jPU) ei˚ ui˚ (u)ai˚ (i)ao˚ ou˚ iu÷ for /[w]ei, [j]ou/ there are
also the possible variants (shown in û 11.4, which are not necessary for a good
pronunciation, and will be given only here), (ÉI, w¤i, ∑U, j¨u): (5f™I÷ 5fÉI) /5fei/ fei˚
(5twÙI÷ 5tw¤i) /5twei/ dui˚ ([öÅÅË) /¶ae/ ∑i˚ (7waË) /6wae/ wài˚ ([öAAÖ) /¶ao/ ∑o˚ ([˜AAÖ)
352 a handbook of pronunciation
/¶hao/ h∑o˚ (4jaÖ) /•jao/ yáo˚ (5öøU÷ 5ö∑U) /5ou/ Ou˚ ([møøU÷ [m≤≤U) /¶mou/ mœu˚
(5tjPU÷ 5tj¨u) /5tjou/ diu˘
û 11.3. Mandarin diphthongs.
/5hX>/ heir˚ (5t·jå⁄) /5thja>/ tianr˚ (4mjå⁄) /•mja>/ miáor˚ (7q·w$Y) /6qhwX>/ cuòr˚
(4t·wX⁄) /•thwX>/ tóur˚ (^ßÅm5p·wX⁄) /5ßan5phwX>/ ´anpor˘
û 11.5. Vowels with />/ (⁄) -r.
Vowel reductions
can be seen, (¤) applies to /i[n/˙], y[n]/: (7tiœ∂i), (-3d¤) /6titi/ dídi˚ (5kwÅ~&ñi), (-3B¤)
/5kwansi/ guanxi˚ (7¿jaœ¸y), (-3¸i, -3©¤) /6sjaqy/ xiàqu÷ and (û), to /M[>], u[˙]/: (7ji-
œΩœ), (-3zŒ) /6jisM/ yísi˚ (7⁄Ë∫œËÎ), (-3ΩŒ) /6>Xnߡ/ rèn´i˚ (5ßu&Ñu), (-vû) /5ßufu/ ´ufu˘
û 11.6. Reductions in syllables with çzeroÇ tones (in quick speech, not in an intoneme; where-
as, in an intoneme, the distinct timbers of û 11.1-3 are used).
11.1.10. In addition, we have (È), for /[j/¥]e, [w]Xn/, and (Ȥ) for /[w]ei/: ([Âj™™-
^©ã™), (-1©ã™, -1©ãÈ) /¶qjeqje/ ji‘jie˚ (7öÅËœ⁄Ëc), (-3⁄Ëc, -3⁄Èc) /6ae>Xn/ àiren˚ (7m™I-
œm™I), (-3mȤ) /6meimei/ mèimei÷ then (X), for /X, [w]X˙, [w]o/, and ([j]X¨), for
/[j]ou/: (5kX&âX), (-2gX) /5kXkX/ gege˚ (çt¤¥^2X), (-1"X) /¶ti˙fiX/ d`«˜e˚ (5¿jEN&Ëx,),
(-2ΩX,) /5sjanßX˙/ xian´e«˚ (7Âj™œm$), (-3m$) /6qjemo/ jièmo˚ (çnwÅ©^∆jø), (-1hjX)
/¶nwanhwo/ nu∑nhuo˚ (7¥U¥œ∂øU), (-3d$¨) /6¥u˙tou/ yò«tou˚ (4p·x,&jøU), (-2ãX¨)
/•phX˙jou/ pé«you˘
Lastly, we have: (å) in /a[n], [w]a, [w]ae/: (7paœÊa), (-3bå) /6papa/ bàba˚ (5ma&ma),
(-2må) /5mama/ mama˚ (ç¿i^∆wan), (-1hjån) /¶sihwan/ x`huan˚ (7t·ÅËœthaË), (-3tåË)
/6thaethae/ tàitai÷ and (√) in /a˙, [j]ao/: ([wÅÅ∫^Ëa,), (-1Ω√,) /¶wanßa˙/ w∑n´a«˚
(5fiÖ&∂AÖ), (-2d√Ö) /5fiˇtao/ ˜idao˘
û 11.7. Further reductions in syllables with çzeroÇ tones (in even quicker speech, not in an
intoneme; then, in an intoneme, the realizations of û 11.6 are also possible).
Besides, (É) in /ae, ja/ (É, ãÉ): (7t·ÅËœthÅË), (-3tåË, -3tË) /6thaethae/ tàitai˚ (4⁄É~-
&¸ãa), (-2©ãå, -2©ãÉ) /•>Xnqja/ rénjia÷ and (‘) for /X/ an for any other /a/ a: (5kX&âX),
(-2g‘) /5kXkX/ gege˚ (çt¤¥^ËX), (-1Ω‘) /¶ti˙ßX/ d`«˜e˚ (7öÅËœ⁄*c), (-3⁄‘c) /6ae>Xn/ àiren˚
(5¿jEN&Ëx,), (-2Ω‘,) /5sjanßX˙/ xian´e«˚ (7paœÊa), (-3bå, -3b‘) /6papa/ bàba˚ (5maœma),
(-3må, -3m‘) /5mama/ mama˚ (ç¿i^∆wan), (-1hjån, -1hj‘n, -1hÏn) /¶sihwan/ x`huan˚
(çwÅ∫^Ëa,), (-1Ωå,, -1Ω‘,) /¶wanßa˙/ w∑n´a«˘
11.1.12. û 11.8 shows the vowel realizations that are su‚cient for an çinterna-
tionalÇ pronunciation of Chinese, without taxophones (thus, without either atten-
uations or neutralizations such as those given in û 11.6-7; equally, we do not even
mark creaky voice, which is çsuperfluousÇ in an international kind of pronuncia-
tion). So we have only eight monophthongs (and four diphthongs, obtained by
combining five of the eight basic ones). ˛ey are a satisfactory result, once (M, x)
have been learned.
˛erefore, we have: (5ji) /5ji/ yi˚ ([~ii) /¶ni/ n`˚ (•jin) /•jin/ yìn˚ (5Âin) /5qin/ jin˚
(7ji˙) /6ji˙/ yí«˚ (5Âi˙) /5qi˙/ ji«˘
˛en: (5Ây) /5qy/ ju˚ ([Âhyy) /¶qhy/ qà˚ (7¿y) /6sy/ xú˚ ([¥yy) /¶¥y/ yà˚ (5Âyn)
/5qyn/ jun˚ (•Âhyn) /•qhyn/ qùn˚ (7¿yn) /6syn/ xún˚ (•¥yn) /•¥yn/ yùn˘
Besides: ([wuu) /¶wu/ wà˚ (•lu) /•lu/ lù˚ (5ku˙) /5ku˙/ go«˚ (7fiu˙) /6fiu˙/ ˜ò«
(•j™) /•je/ yé˚ (5tj™) /5tje/ die˚ (7¥™) /6¥e/ yuè˚ ([¿¥™™) /¶s¥e/ xu‘÷ and (7öx) /6X/ è˚ (5öxn)
/5Xn/ en˚ (•öX⁄) /•X>/ ér˚ (7fxn) /6fXn/ fèn˚ ([wxxn) /¶wXn/ w‘n˚ (7fx˙) /6fX˙/ fè«˚ (5wx˙)
/5wX˙/ we«˚ (7wx˙) /6wX˙/ wè«˘
More: (5mø) /5mo/ mo˚ (•pø) /•po/ bó˚ (7phø) /6pho/ pò˚ ([wøø) /¶wo/ wœ˚ (5fiwø)
/5fiwo/ ˜uo˚ (Òha˙5jø) /•ha˙5jo/ há«yo˚ (7öø) /6o/ ò! and (5öa&qa) /5aqa/ aza˚ (7tha)
/6tha/ tà˚ (•jan) /•jan/ yán˚ ([tjaan) /¶tjan/ di∑n˚ (7¥an) /6¥an/ yuàn˚ (5Âh¥an)
/5qh¥an/ quan˚ ([öaan) /¶an/ ∑n˚ (•phan) /•phan/ pán˚ (•pha˙) /•pha˙/ pá«˚ ([waan)
/¶wan/ w∑n˚ (5ja˙) /5ja˙/ ya«˚ ([waa˙) /¶wa˙/ w∑«˘
11.1.13. And with />/: ([fiÖÖ) /¶fiˇ/ ˜`˚ (5fihÖ) /5fihˇ/ >i˚ (•ßÖ) /•ßÖ/ ´ì˚ (7⁄Ö) /6>ˇ/
rí÷ (5kx⁄) /5kX>/ ger˚ ([pxx⁄) /¶pX>/ b‘nr˚ (•h≈⁄) /•hX˙>/ hé«r˚ ([naa⁄) /¶na>/ n∑r˚
(7pa⁄) /6pa>/ bànr˚ (5qÅ⁄) /5qa˙>/ za«r÷ (5hwa⁄) /5hwa>/ huar˚ (7mja⁄) /6mja>/
miànr÷ and (•pha⁄) /•pha>/ páir˚ (7Âh¥x⁄) /6qh¥X>/ qúr˚ (7ßwx⁄) /6ßwX>/ ´úr˚
(7qhx⁄) /6qhX>/ cír˚ ([qxx⁄) /¶qX>/ z`r˚ (5kwa⁄) /5kwa>/ guanr˚ (5kw≈⁄) /5kwX˙>/
go«r˚ (5swxn&n¥x⁄) /5swXnn¥X>/ sunnür˚ (•¿i&fwx⁄) /•sifwX>/ xìfur˚ (^fiÖ•ljx⁄) /5fiˇ-
•ljX>/ ˜iliùr˚ (•Âjx⁄) /•qjX>/ jiér˚ (^ha2pa[thwxx⁄) /5hapa¶thwX>/ habatu`r˚ (7tu˙-
œÂj≈⁄) /6tu˙qjX˙>/ dò«ji«r˚ (7jx⁄) /6jX>/ yínr˚ ([sxx⁄) /¶sX>/ s`r˚ (5hx⁄) /5hX>/ heir˚
(5thja⁄) /5thja>/ tianr˚ (•mja⁄) /•mja>/ miáor˚ (7qhwx⁄) /6qhwX>/ cuòr˚ (•thwx⁄)
/•thwX>/ tóur˚ (^ßam5phwx⁄, -n5p-) /5ßan5phwX>/ ´anpor˘
For diphthongs, we have: (5f™i) /5fei/ fei˚ (5tw™i) /5twei/ dui˚ ([öaa™) /¶ae/ ∑i˚
(7wa™) /6wae/ wài˚ ([öaaø) /¶ao/ ∑o˚ ([haaø) /¶hao/ h∑o˚ (•jaø) /•jao/ yáo˚ (5öøu) /5ou/
Ou˚ ([møøu) /¶mou/ mœu˚ (5tjøu) /5tjou/ diu˘
356 a handbook of pronunciation
Consonants
distribution. Phonology, in order to be useful, must be explicit and not at all eso-
teric (although this would not be wrong diachronically).
In the table of û 11.9, which gives the consonantal articulations of Mandarin
Chinese, to save space, we show as /05/ (whereas here we use ç/0[h]/Ç = /0, 0h/)
the consonants which are distinct phonemically. In fact, simple consonants oppose
çaspirationÇ sequences (Ô Stops, § 11.2.2, and Stopstrictives, § 11.2.3).
Instead, û 1.9-15 show the orograms, grouped by manners of articulation, of
all the contoids given in the chapters of this handbook for the 12 languages dealt
with, including secondary, occasional, or regional variants.
û 11.9. Table of Chinese consonants.
palatal (* prepalat.)
postpalatal rounded
labiodental round.
bilabial rounded
bilabialized pre-
velar rounded
postalveolar
labiodental
laryngeal
alveolar
bilabial
palatal
uvular
dental
velar
ö m ([) (M) (n) n (N) (~)* (˙) ,
F p5 (b) ((5 {) t5 (d) k5 (g) (ö)
Ô q5 (Q) fi5 (") (Â5 ©)
ƒ f (v) (5 ç) (·)
_ s (z) ß (Ω) (¿ B)
ß ⁄ j|(ã) ¥|(%) (∆) w|(j) (˜) h
‹ l
/˙/ (,), />/ (⁄), /h/ (·, h, ˜, ∆)
Nasals
11.2.1.2. Here are some examples: (5mAÖ) /5mao/ mao˚ (5[ø) /5mo/ mo˚
([nAAÖ) /¶nao/ n∑o˚ (4p·Én) /•phXn/ pén˚ (4p·x,) /•phX˙/ pé«÷ (œ˜Ëc7ö$[*|]) /¶hXn6X/
h‘n è˚ (Ò·¥Än[t·ii) /•qh¥an¶thi/ quánt`˚ (œÂIc[qAAÖ) /¶qin¶qao/ j`nz∑o˚ (0twÅn-
7ljEc) /6twan6ljan/ duànliàn˚ (Ò⁄Ém4mIn) /•>Xn•min/ rénmìn˚ (Òlw‘m5pÅn) /•lwXn-
5pan/ lùnban˚ (0k·aü[faa) /6khan¶fa/ kànf∑˚ (^pjE~4Âi) /5pjan•qi/ bianjì˚ (0ÂI∫[fiÎÎ)
/6qin¶fiˇ/ jín˜`˚ (0p·ÅÊ4Â¥™) /6phan•q¥e/ pànjué˚ (œpjEÊ4¥y) /¶pjan•¥y/ bi∑nyù˚
(ÒkÅ∫[⁄ÅÅc) /¶kan¶>an/ g∑nr∑n˚ (ÒpI˙[kwÅÅc) /5pin¶kwan/ bingu∑n˚ (0ÂjE˙7wA¥)
/6qjan6wa˙/ jiànwà«˚ (ÒljE,5˜wÅn) /•ljan5hwan/ liánhuan˘
©th /˙/: (Òji7lu Òp·¤,5öÅn) /•ji6lu •phi˙5an/ yìlú pì«'an˚ (^fx,7mjEc) /5fX˙6mjan/
fe«miàn˚ (œk·U¥7pu) /¶khu˙6pu/ kœ«bú˚ (0la,7f™I) /6la˙6fei/ là«fèi˚ (^fx,[tii) /5fX˙-
¶ti/ fe«d`˚ (œk·U¥7¿i) /¶khu˙6si/ kœ«xí˚ (Òp·¤,4¥y) /•phi˙•¥y/ pì«yù˚ (^ßx,7⁄Î) /5ßX˙-
6>ˇ/ ´e«rí˚ (^fx,7ßÅc) /5fX˙6ßan/ fe«´àn˚ (Òp·¤,[kwøø, -[kwø*|) /•phi˙¶kwo/ pì«-
guœ˘
In an çinternational-likeÇ pronunciation, it is su‚cient to use (˙), instead of (,),
and also (n) before /i, y÷ j, ¥/ (and, possibly, before other consonants too]˘
Stops
11.2.2.1. We have three voiceless phonemes, /p, t, k/, which occur as either
plain, /0/, or in çaspirationÇ sequences, /0h/, with distinctive values. Before see-
ing some examples, we have to explain the nature and consistency of both voice-
lessness and çaspirationÇ.
In fact, except in slow (or very accurate) speech, /pé, té, ké/, are realized as (p,
t, k) only in fully stressed syllables (or after pauses), whereas in half-stressed sylla-
bles they are (Ê, ∂, â) or, in unstressed syllables, even (b, d, g). On the other hand
–in a parallel way– /phé, thé, khé/, in fully stressed syllables (or after pauses), are
sequences of voiceless stops and a laryngeal constrictive (decidedly çstrongÇ, (p·,
t·, k·) (however, it is possible to find a semi-constrictive as well, ((0Ö)), which is
weaker). In half-stressed syllables, we have sequences with a laryngeal approximant,
(ph, th, kh); whereas, in unstressed syllables, we can even find plain voiceless stops,
(p, t, k). We choose to adopt this kind of normalized transcription (although os-
cillations are frequent indeed, including sequences with a voiceless laryngeal semi-
-approximant, ((0h))).
A degree of distinctivity between çaspirationÇ and çnon-aspirationÇ, however, is
guaranteed. In fact, we have: /0h/ ('=·, &=h, ’=), /0/ ('=, &Ò, ’Ê), with strong and
weak extremes meeting, as (=). Although they coincide phonetically, they remain
distinct functionally, from a phonemic point of view. Before /o/ (not /ou/, by dis-
similation) we find ((), which sounds rather di‡erent from the common (p) of
other languages.
11.2.2.2. Let us now see the following examples: (5pIn) /5pin/ bin˚ (4(ø) /•po/
bo˚ (7(·ø) /6pho/ pò˚ (4tj™) /•tje/ dié˚ (4k·wÙI) /•khwei/ kuì˚ (^p·¤,5p·a,) /5phi˙-
5pha˙/ pi«pa«˚ (0t·jAÖ[wuu) /6thjao¶wu/ tiàowà˚ (5k·x,) /5khX˙/ ke«÷ (7paœÊa|,
7pa3bå) /6papa/ bàba˚ (7tiœ∂i|, 7ti3dI) /6titi/ dídi˚ (5kX&âX|, 5kX2gÈ) /5kXkX/ gege˚ (7ßÎ-
11. chinese 359
Stop-strictives
11.2.3. As far as this manner (of articulation) is concerned, in our analysis (with
the same normalized characteristics of voicing and çaspirationÇ as for Chinese
stops) we have /q, qh/ ('q, &ë, ’Q÷ 'q·, &qh, ’q) z˚ c˚ in addition to ('Â, &¸, ’©÷ '·,
&Âh, ’Â) j˚ q˚ before /i, y÷ j, ¥/ (Ô § 11.2.0), and /fi, fih/ ('fi, &2, ’"÷ 'fi·, &fih, ’fi) ˜˚
>˘
Here are some examples (expressly chosen with final syllables without tonemes):
(5qÅÉ) /5qae/ zai˚ (7q·wÅc) /6qhwan/ cuàn˚ (5Ây) /5qy/ ju˚ (4·¥U,÷ 4·jU,)
/•q¥u˙/ qió«˚ (7fi™I) /6fiei/ ˜èi˚ (4fi·Ö) /•fiˇ/ >ì÷ (4pi&ëM|, 4pi2Qû) /•piqM/ bìzi˚
(4¥Yn&qhÅÉ|, 4¥Yn2qåÉ÷ 4jYn-) /•¥ynqhae/ yùncai˚ (7tjEÊœ¸i|, 7tjEÊ2©¤) /6tjanqi/ diàn-
ji˚ (5t·jE~&Âhi|, 5t·jE~¤) /5thjanqhi/ tianqi˚ (4q·ÅÉ&2Ö|, 4q·ÅÉ2"Ö) /•qhaefiˇ/ cái-
˜i˚ (4p·™I&fihÉn|, 4p·™I2fiÈn) /•pheifihXn/ péi>en˘
Constrictives
11.2.4. ˛ere are three constrictive phonemes. Again they are all phonemically
voiceless, but with the usual voicings which depend on syllable-strength attenua-
tion. ˛us we have: /f/ ('f, &Ñ, ’v) f, /s/ ('s, &Ω, ’z) s and ('¿, &ñ, ’B) x, /ß/ ('ß, &Ë, ’Ω) ´\
(4fu) /•fu/ fù˚ (4ma&ÑÅn|, 4mavån) /•mafan/ máfan˚ (7swÙI÷ 7sw¤i) /6swei/ suí˚ (5fx,&Ωu|,
5fx,2zû) /5fX˙su/ fe«su˚ (4¿¥™&ñi|, 4¿¥™2B¤) /•s¥esi/ xuéxi˚ (5ßu) /5ßu/ ´u˚ (4lja,&ËÖ|,
4lja,2ΩÖ) /•lja˙ߡ/ liá«´i˘ Before /o/ (not /ou/) we find (5) (as happens for other la-
bial phonemes): (45ø{x}) /•fo/ fó˘ ˛erefore, we could even pose /fwo/ instead of /fo/
(but /fou/ (føU), by dissimilation).
Approximants
mi-velar rounded (j): (5ja,) /5ja˙/ ya«˚ (4tj™) /•tje/ dié˚ (5Âj™) /5qje/ jie˚ (çli^ÊjEn|,
çli1bãEn, -1bãÉn) /¶lipjan/ l`bian÷ ([¥yy) /¶¥y/ yà˚ (7·¥™) /6qh¥e/ què˚ (7tiœñ¥U¥|, 7ti-
3B%U¥) /6tis¥u˙/ díxio«÷ (4wA,) /•wa˙/ wá«˚ (5˜wÙI÷ 5˜w¤i) /5hwei/ hui˚ (çt·AÖ-
^lw‘n|, çt·AÖ1ljÈn) /¶thaolwXn/ t∑olun˘
≈nally, we find /h/ ('˜é, &∆é, ’hé) (in addition to ('0·, &0h, ’0`), already seen for
stops and stop-constrictives): ([˜AAÖ) /¶hao/ h∑o˚ (5öÉ˙&∆AÖ|, 5öÉn2h√Ö) /5Xnhao/ en-
hao˘ In theory, we could have two di‡erent phonemes in the inventory of conso-
nants: ç/x, h/Ç, with ç/x/Ç = ('˜é, &∆é, ’hé); but it is better to have only one, with
the taxophones already seen (which are di‡erent although auditorily quite simi-
lar, but with a phone of contact, (h)). Traditionally, the Chinese prefer to consid-
er initial consonants in a unitary way, with ç/05/Ç di‡erent from /0/, intraphone-
mically; but this choice does not present any real advantage from an interphone-
mic point of view, which is interested in several languages of the world – quite the
contrary!
Laterals
11.2.6. For the last manner (of articulation) necessary for Mandarin, we find
just one phoneme, /l/ (l) (which remains unchanged even before /i, y÷ j, ¥/): (4lÅÉ)
/•lae/ lái˚ (4lU,) /•lu˙/ ló«˚ ([ljEEc) /¶ljan/ li∑n˚ (7ly) /6ly/ lǢ
Structures
11.3.0. We will deal, in particular, with the typical tones (tonemes as distinctive
elements), including a çzeroÇ ton(em)e and their combinatory variants (taxotones),
in addition to degrees of stress, and finally intonation, which overlaps tone charac-
teristics, by changing them further (as we will see).
A typical Mandarin Chinese pronunciation shows a particular kind of voice
with a paraphonic setting with raised larynx §Æ@, not only for wemen. Other Chi-
nese languages, instead, use a normal (or modal) laryngeal setting, or else one with
11. chinese 361
Taxophonics
11.3.1.2. As we have said, Chinese words are formed by initial and final ele-
ments. However, their combinations have some limitations; in fact, among all the-
oretical possibilities, only 55% are actually used, for slightly over 400 typical sylla-
bles (which are regularly indicated in teaching textbooks, with some oscillations
for the less frequent ones). Here we will report only the most remarkable character-
istics.
(öX⁄) /X>/ er occurs with no initial; /f/ has a fairly limited distribution: it does
not occur before /i, j÷ y, ¥÷ w/ and some other phonemes; whereas, only in loan-
words /v/ is found (which is thus a xenophoneme added to the (v) taxophone of
/f/). Besides, (Â[h], ¿) (with other variants already seen) are taxophones of /q[h],
s/, which only occur before /i, j÷ y, ¥/, and are spelled j˚ q˚ x˘ Equally, /M/ only oc-
curs after /q[h], s/; and /ˇ/ after /fi[h], ß, >/÷ /o/, only after /m, p[h], f, w/ (and in
362 a handbook of pronunciation
some particular forms after /j, l/ or with a çzeroÇ segment); and /y/, only after /j,
¥÷ l/, /n/ (~), or after /q[h], s/ (Â[h], ¿).
≈nally, we would like to add that /m, p[h], f/ do not occur before /X, M, w÷ y,
¥÷ >÷ [j]u˙/ or some other segments; neither do /fi[h], ß/ occur before /i, j÷ y, ¥/ (or
some others), nor /k[h], h/ before /i, j÷ y, ¥/ or /o, wX˙/. ˛ere are some other more
specific limitations; but this is the spirit of Mandarin phonotactics – /n, ˙/ are the
only, very frequent, syllable-final consonants (besides, />/, which is extremely rare).
˛ere are some remarkable reductions of words such as the following (which are
quite normal, except in intonemes or in formal speech): (4ßXm, 4ßXõ, 4ßX2mÈ, 4ßX-
2mX) /•ßXnmX/ ´énme˚ ([q$$C, çq$1õ, çq$1mÈ, çq$1mX) /¶qXnmX/ z‘nme˚ ([wøøC,
çwøøõ, çwø1mÈn, -1mÉn) /¶womXn/ wœmen˚ (5t·am, 5t·aõ, 5t·amÈn, -mÉn) /5tha-
mXn/ tamen˘ Contrary to some other Chinese languages, in Mandarin there is a
somewhat exceptional occurrence of /m/ in final position, as indeed that of (õ),
which we find in : (4õ) /•Xm/ µ çreally?/what??Ç, (7≠) /6Xm/ û çyes/I understandÇ.
Stress
11.3.2.2. In sentences, however (as will be seen from the text in § 11.4.2.2 as
well), there are frequent oscillations with regard to actual stresses on di‡erent sylla-
bles, both for rhythmic reasons and prominence interplay, and even for communi-
cative purposes, even when they are not çdistinctiveÇ. ˛us, for instance, /5ßa˙¶kan-
¶fiu6ji¶fiX/ ´a«g∑n˜ày협 can have two frequent realizations, (5ßa,œâÅ∫çfiu0ji-
[fi$$, 5ßa,œâÅ∫œ2u7jiœ2$); equally, even forms such as the following often oscillate:
(0ji[tjEEc, 7jiœ∂jEc) /6ji¶tjan/ yídi∑n˚ (0(ø4f™I, 7(ø&Ñ™I) /6po•fei/ pòféi˚ (^qM7wÙI, 5qM-
0wÙI) /5qM6wei/ ziwèi˚ (0¿Iü[fuu, 7¿IüœÑu) /6sin•fu/ xínfù˚ (^ÂIn5t·jEn, 5ÂIn^thjEn)
/5qin5thjan/ jìntian˚ (0öÅË[lii, 7öÅËœli) /6ae¶li/ àil`˚ (^kÅm5p™I, 5kÅm^Ê™I) /5kan5pei/
ganbei!
All this can happen both in intonemes and preintonemes. And, what is more,
even opposite cases are frequent, that is syllables with çzeroÇ tone which bear a
strong stress, verging on mid pitch (again independently of any intoneme): (4p·x,-
&jPU, &p·x,'jPU) /•phX˙jou/ pé«you˚ (7tÅ∫œËÎ, 0∂Å∫'ßÖ) /6tanߡ/ dàn´i˚ (4ßÖ&∆øU, &ßÖ-
'˜øU) /•ßÖhou/ ´ìhou˘ In addition, even forms bearing a tone (even written) are
weakened, as happens to the grammemes and locatives in the following examples:
(5fiwø0Ëa¥, -&Ëa,) /5fiwo 6ßa˙/ ˜uo ´à«˚ (6ti0ñja, -&ñja) /6ti 6sja/ dí xià˚ (5wuœli, 5wu-
&li) /5wu ¶li/ wu l`˚ (4t·¤,œli, -&li) /•thi˙ ¶li/ tì« l`˘
Tones
11.3.3.1. ˛e most remarkable aspect of the various Chinese languages (and one
of the most serious problems for foreigners) resides in the use of çtonesÇ. In fact, in
languages such as Mandarin, a di‡erent pitch on a syllable is able to make its mean-
ing change completely. ˛erefore, we have to speak of tonemes for distinctive ele-
ments of a phonemic system, and of tones for their actual realizations, including
taxotones, ¤ combinatory variants, as can be seen from the examples given below.
In order to draw attention to their importance, and not to confuse concepts and
words, let us choose the example of ([mÅÅË) /¶mae/ m∑i çto buyÇ and (7mÅË) /6mae/
mài çto sellÇ. We now add some classical examples (giving some of the most fre-
quent meanings, among the possible ones, since transliterations may unify di‡er-
ent words {as phonic transcriptions do too}, represented by di‡erent characters,
or çideogramsÇ): (5ma) /5ma/ ma çmotherÇ, (4ma) /•ma/ má çhempÇ, ([maa) /¶ma/
m∑ çhorseÇ, (7ma) /6ma/ mà çscold, curseǢ
As some transcriptions have already shown, tonemes 3 and 4 are realized with
creaky (or laryngealized) voice, when their çlowÇ taxophones occur: ([), (ç) and (7)
(with secondary stress, (0)), but not when the çhighÇ ones occur: (•) and (6). To-
nemes 1 and 2, which are çhighÇ, (5), (4), always have normal voice quality.
A prudential warning is necessary about some diagrams which appeared in cer-
tain publications, not only in çtourist-likeÇ books, since they are too often com-
pletely wrong – so they are of no help at all to unfortunate readers…
Mandarin is relatively simple, in comparison with other Chinese languages, al-
so as far as taxotones are concerned. In fact, Mandarin has only four marked to-
nemes, traditionally indicated and listed as 1 (5) /5/ >1≥ o\ (5ji) /5ji/ yi÷ 2 (4) /•/ >Q≥ ó\
364 a handbook of pronunciation
(4wu) /•wu/ wù÷ 3 ([) /¶/ >5≥ œ\ ([¿¥™™) /¶s¥e/ xu‘÷ 4 (7) /6/ >Z≥ ò\ (7·ja¥) /6qja˙/ qià«˘
û 11.10 shows –for the four tonemes– the four basic tones, in addition to three
taxotones, which are as important phonetically (or rather çtoneticallyÇ): in addi-
tion to ç3=2 (¤ 3+3 = 3+2)Ç (•) /¶/ >5≥ œ (¤ a toneme 3 –occurring before another
3– which becomes almost a tone 2 {although it is generally –but falsely– identified
with an actual tone 2}), we have a çhalf 3Ç, (ç) /¶/ >5≥ œ (¤ a 3 before one of the oth-
er three), and a çhalf 4Ç, (6) /6/ >Z≥ ò (¤ a 4 before another 4). As can be seen, from
both tonograms and phonetic transcriptions which will follow, the last two are ac-
tually realized as the first half of their full forms.
In a kind of çinternationalÇ pronunciation, toneme 2 can be realized as (•) (in-
stead of (4) ((-1))), as the o‚cial taxotone of /¶/ followed by another /¶/. ˛e pitch
characteristics of the other ton(em)es ought to be very similar to the o‚cial ones,
even for the çzeroÇ toneme, as we have done in our transcriptions (although for
the latter, somehow, a realization of mid pitch {and weak stress} could be su‚cient,
(2), again in this type of çinternationalÇ pronunciation).
û 11.10. Mandarin tonemes and tones.
1 /5/ (5) >1≥ 2 /•/ (4) >Q≥ 3 /¶/ ([) >5≥ 3‘ /¶/ (ç) >5≥ 3“ /¶/ (•) >5≥ 4 /6/ (7) >Z≥ 4‘ /6/ (6) >Z≥
11.3.3.2. Obviously, although at first all this might seem to be a useless compli-
cation, in actual fact, it is a real facilitation for contextual realizations. As a mat-
ter of fact, it is a simplification, from a tonetic point of view, which is visible on-
ly thanks to accurate phonetic (or rather phonotonetic) transcriptions, whereas
phonemic (¤ phonotonemic) transcriptions, as well as graphemic renderings, do
not change at all, since their entity, their essence, remains the same.
As can be seen from the following examples, and –above all– from û 11.11, the
allotones (¤ taxophones) of tonemes 3 and 4 present a simplification, by assimila-
tion.
û 11.11. Fundamental taxotones.
/¶ 5/ = (– 5)
/¶ •/ = (– 4)
/¶ ¶/ = (Ò [)
/¶ 6/ = (– 7)
/6 6/ = (0 7)
11. chinese 365
In fact, toneme 3, ([) = (ç), /¶/ >5≥˚ shortens (even as its segmental duration does),
when it occurs before tonemes which are di‡erent from itself: (œ˜wø5fi·X) /¶hwo-
5fihX/ huœ>e˚ (œt*¥4¥y) /¶tX˙•¥y/ d‘«yù˚ (œk·AÖ7ßÎ) /¶khao6ߡ/ k∑o´í÷ on the other
hand, when it occurs before itself, /¶ ¶/ >5 5≥˚ it has the taxotone (•): (ÒßøU[pjAAÖ)
/¶ßou¶pjao/ ´œubi∑o˘
Even in this case, it is right to keep its tonemic and graphemic representation
unchanged, since, should we modify them, they would lead us to think of di‡er-
ent structures, not only from a phonic point of view, but also lexically and, then,
semantically.
Toneme 4 resorts to a (halved) taxotone before itself: (0fa,7Âja) /6fa˙6qja/ fà«-
jià˘ Tonemes 1 and 2 do not have any variants; and, although the taxotone of 3 (+
3) is tonetically (more) similar to toneme 2, it nevertheless remains a taxotone of
toneme 3 (except in the çinternationalÇ accent {which is a simplification}).
û 11.12. Taxotones of particular tonemic sequences.
/5 • 5/ = /5 5 5/ (5 — 5)
/5 • •/ = /5 5 •/ (5 — 4)
/5 • ¶/ = /5 5 ¶/ (5 — [)
/5 • 6/ = /5 5 6/ (5 — 7)
/• • 5/ = /• 5 5/ (4 — 5)
/• • •/ = /• 5 •/ (4 — 4)
/• • ¶/ = /• 5 ¶/ (4 — [)
/• • 6/ = /• 5 6/ (4 — 63)
/5¶ ¶/ = (5 Ò [) = (5 — [)
/•¶ ¶/ = (4 Ò[) = (4 — [)
/¶ ¶ ¶/ = (• Ò[) = (• — [)
366 a handbook of pronunciation
11.3.3.3. As far as sequences of three tonemes are concerned, there are some
di‡erences (and they must be indicated and applied too) in comparison with tone-
mic (and graphemic) representations. û 11.12 shows what happens to sequences
of /5 •/ followed by one of the other four tonemes: (5tU,^nÅM5fx,) /5tu˙•nan5fX˙/
do«nán-fe«˚ (5sÅ~^~jE~4Âi) /5san•njan•qi/ sannián-jì˚ (5¿jEN^⁄É~[Âjaa¥) /5sjan•>Xn-
¶qja˙/ xianrén-ji∑«˚ (5¿i^∆U,7ßÎ) /5si•hu˙6ߡ/ xihó«´í.
û 11.13 shows sequences of /• •/ followed by one of the other four tonemes:
(4m™I^laM5fa,) /•mei•lan5fa˙/ Méi Lánfa«˚ (4˜ÅÉ^m™I4lÅÉ) /•hae•mei•lae/ hái méi
lái˚ (4wÅ~^Âh¥Än[tUU¥) /•wan•qh¥an¶tu˙/ wánquán dœ«˚ (4¿Y~^ja,7ÂjEc) /•syn-
•ja˙6qjan/ xùnyá«-jiàn˘
≈nally, û 11.14 shows the behavior of the sequence /¶ ¶/ preceded by one of the
first three tonemes: (5sÅ~^jEn[Âj<<Y) /5san¶jan¶qjX˙>/ Sany∑n Ji‘«r˚ (4˜ÅN^Ëu-
[pjAAÖ) /•han¶ßu¶pjao/ hán´àbi∑o˚ (•wø^j™[jÕÕU) /¶wo¶je¶jou/ wœ y‘ yœu˘
/5 ”/ (5 &) /• ”/ (4 &) /¶ ”/ (ç —) /6 ”/ (7 –)
/5 ” ”/ (5 2 –) /• ” ”/ (4 2 –) /¶ ” ”/ (ç 1 —) /6 ” ”/ (7 3 –)
11. chinese 367
/5 ” 6/ (5 1 7) /• ” 6/ (4 1 7) /¶ ” 5/ (ç 2 5) /¶ ” •/ (ç 2 4) /¶ ” •/ (ç 2 7)
Intonation
11.3.4.2. Besides çisolatedÇ tones (typical of separate words {as examples are}
and of a suspensive intoneme, and of an emphatic preintoneme as well, which are
given in black), û 11.19 also shows the taxotones (given in grey) of the four
marked tonemes, as they are modified when they are realized in conclusive (/./ (13)),
interrogative (/?/ (31)), and continuative (/,/ (2)) intonemes. It is obvious that the to-
netic notation used is –necessarily– just an indication of the deformation under-
gone by actual tones. Of course, in non-tonal languages, the use of tonetic sym-
bols is –automatically– much more precise, since it is free from any phonemic val-
ue applied to lexical pitch.
In addition, û 11.20 shows in grey all the allotones occurring in the four prein-
tonemes. ˛e four marked in black, in the emphatic preintoneme, correspond to
the four çisolatedÇ tones. ˛ey are also useful for comparisons, which must be
done very carefully, in order to appropriately memorize and master them in actu-
al sentences (at first as a simple identification step, then as active drills as well).
Undoubtedly, they require a good deal of commitment and constancy. But any
e‡ort will certainly be rewarded. Of course, paraphonics further modifies things
depending on attititudes, feelings, Â.
11.3.4.3. Lastly, we will see three examples to start an accurate analysis of into-
nation which is always in an indissoluble relation with tones. Given the complex-
ity of this enterprise, we will use the same basic sentence, with the same tonemes.
Of course, other tonal combinations are to be experimented as well, on the basis
of the examples provided:
/./: (5t·a0ËÎ 4ji2gX Ò∆X7ßÅc2dX 5ku&~ja,13) /5thaߡ •jikX •hX6ßantX 5kunja˙./ Ta ´í yìge
hé´àn de gunia«.
/?/: (¿5t·a0ËÎ 4ji2gX Ò∆X7ßÅc2dX 5ku&~ja,2ma31) /¿5thaߡ •jikX •hX6ßantX 5kunja˙ma?/
Ta ´í yìge hé´àn de gunia« ma?
/÷/: (5t·a0ËÎ 4ji2gX Ò∆X7ßÅc2dX 5ku&~ja,^ 7tÅ∫3ΩÎ çwø1bu 7öÅËœtha13) /5thaߡ •jikX •hX6ßan-
tX 5kunja˙÷ 6tanߡ ¶wopu 6aetha./ Ta ´í yìge hé´àn de gunia«, dàn´i wœ bù ài ta.
û 11.19. ˛e four tonemes combined with the four intonemes.
//()
/¿/ (¿ )
/¡/ (¡ )
/˚/ (˚ )
Text
11.4.0. ˛e story †e North Wind and the Sun follows, given in four di‡erent
çnormalizedÇ versions. We start with the (neutral) Chinese pronunciation of (neu-
tral British) English – this is the first step of the phonetic method (the written text
is given in § 2.5.2.0). ˛e Chinese translation follows, in its neutral and çinterna-
tionalÇ versions.
Please, note the creaky vowels before pauses (¤ with creaky phonation type, or
laryngealization, (ü); whereas in Mandarin creaky voice is in relation with low
tones, also for the voiced consonants belonging to the same syllable coda). ˛ese
creaky vowels confer great çauthenticityÇ, together with a (paraphonic) postdorsal
setting, that is with a slight but permanent –or frequent– raising of the dorsum
towards the velum together with the raising of the larynx §# Æ@.
At the end, as always, there is the version which gives the English pronuncia-
tion of Chinese, by neutral British speakers, fluent in Chinese (after prolonged
contact with native speakers, but with no help from the phonetic method), who
have adequately learned the relative prominences, but who substantially use seg-
mental and intonation elements which are typical of neutral British English (for
reference purposes, although, of course, a neutral accent is not so common). Ob-
viously, the same principle is valid for the foreign pronunciation of English, giv-
en first.
Speakers of American English could prepare their own version both of the Chi-
nese pronunciation of English and of their pronunciation of Chinese, as an excel-
lent exercise, by listening to native speakers, best of all after recording them. Of
course, speakers of other languages could do the same thing. ˛e author would be
happy to receive their transcriptions and recordings, both in case of help –should
they need it– and to make their contribution known to others (possibly in our
website on canIPA Natural Phonetics – Ô § 0.12).
370 a handbook of pronunciation
Mandarin text
11.4.2. Yœu yí huì b‘ife« gen tàiyá« ˜è«zàinar ˜e«lún ´éide b‘n´i dà,
´uo˜e ´uo˜e láile yìge zœudàorde, ´en´a« >uan˜e yìjiàn hòu páozi. Tamen
li∑ jiú ´a«lia« h∑o le ´uo, è´éi né« xian jiào ˜ège zœudàorde b∑ tade páozi tuo-
le xiàlai a, jiú suàn ´éide b‘n´i dà¶.
H∑o, b‘ife« jiú ´`qi dà jín lái j`ngua j`ngua, k‘´í ta guade yuè líhai, nèige rén
b∑ páozi guœde yuè j`n; dào mòli∑or b‘ife« méile fázi, ˜`h∑o jiú suànle. Yíhu`r tài-
yá« jiú >ulái rèrerde yì ´ài, nèi zœudàorde m∑´à« jiú b∑ páozi tuole xiàlai. Suœ-
y` b‘ife« bú né« bú >é«rèn dàod` hái´i tàiyá« b` ta b‘n´i dà.
N` x`huan ˜è ´œu érge ma? Wœmen zài >à« yì biàn ma?
A short appendix
12.0.2. In Japanese, even the pitch of the di‡erent morae is distinctive. ˛ey
form syllables, words, phrases, and sentences. ˛e example just seen shows that
pitch is not marked in spelling, that is in the transliteration, as well as in normal
hiragana writing – (3âi'må2˙åœnå, 3âi'må3˙åœnå÷ 3âi'må2˙å&nå) /hiRaù˙aùna÷ -naå (which is
added to characters, çideogramsÇ, that were taken from Chinese).
Normally, this does not happen in the other type of writing, katakana (3kÄ'tå-
3kåœnå, 3kÄ'tå2kåœnå) /kataùkaùna/, which is generally used in teaching and scientific
textbooks, in order to çhintÇ at the pronunciation of onomatopeic terms and re-
cent loanwords (more recent than the Chinese ones, which adapted to Japanese),
or for stylistic reasons.
Instead, in phonotonemic transcriptions, we indicate with Å the point after
which the voice goes from a mid pitch to a low one. We call this akusento (Ô §
12.3.2.1), by using the term taken from English [accent]˚ in order to indicate this
particular pitch phenomenon. As a matter of fact, it is not properly a stress ele-
ment, since, in actual fact –as we will see– both pitch patterns and segmental con-
374 a handbook of pronunciation
Vowels
12.1.1. Japanese has only five vowels, which can be distinctively short or long
(or rather doubled, being realized almost as monotimbric diphthongs), and they can
combine into di‡erent kinds of sequences (as happens in a slow pronunciation of
Spanish in Saavedra or La Habana (ßaa'B™;ƒRa, laa'Ba;na). However, too often
Japanese pronunciation is hastily çdescribedÇ as having the vowels of Spanish and
the consonants of English. We will see that this is not the case at all.
û 12.1 shows the actual articulations of the five vowels: (i, ™, å, ø, M÷ ii, ™™, aa,
øø, MM) /i, e, a, o, M÷ ii, ee, aa, oo, MM/ i˚ e˚ a˚ o˚ u÷ ii˚ ee˚ aa˚ oo˚ uu˘ Since çlongÇ
vowels are phonemic sequences (and phonetic geminations), the marker of (short)
/a/ (å) is grey, seeing that it is articulated as a less open vocoid (even in stressed syl-
lables). ˛e most problematic Japanese vowel –of course for non-native speakers–
is (M) /M/ u˚ which lacks the typical lip rounding of (u) so widespread in many
languages. Besides, it is articulated with the tongue dorsum in a fronter position
than (u) in most languages; in fact, it is back-central, not simply back.
Mostly, the younger generations articulate /M/ in a further front position, as
high central –with partial or full rounding, (˚) or (%), respectively– but we always
notate (M), since those are youth characteristics which generally change into (M),
sooner or later. ˛is peculiarity continues in relative time, without really chang-
ing the structure. Obviously, real non-neutral pronunciations are a di‡erent thing.
Let us now see some examples for each phoneme: (3i'™) /ieå ie˚ (Çáiô2køø) /giô-
koo/ ginkoo˚ (3så'6i3¿ii) /sabisiùi/ sabi´ii÷ (Ç™i2™P) /eieô/ eien˚ ('™3˙åø) /eù˙ao/ egao˚
(3kå'må2t™) /kaRate/ karate÷ (Çåi) /aùi/ ai÷ (çQå¿2¿:i) /zassi/ zas´i˚ (3må'tø) /mato/ ma-
to÷ ('n™3kø) /neùko/ neko˚ (3ø'©i3mM) /okiùRM/ okiru˚ (3ø'tø2kø) /otokoå otoko÷ (3sM'mi)
/sMmiå sumi˚ (2åçkM3må) /aùkMma/ akuma˚ (3mM'¿i) /mMsiå mu´i.
12. japanese 375
/a/ (å)
/aa/ (aa)
12.1.3. Besides, neutral Japanese has the peculiarity of presenting vowel devoic-
ing. Indeed, there are two partially di‡erent degrees.
˛e first type is complete, and produces voiceless vocoids, that is with no vibra-
tion of the vocal folds, as also happens with consonants, such as (voiceless) (s) /s/
in comparison with (voiced) (z) /z/. More appropriately, these voiceless vowels are
lenis too (or lenited), ¤ only the arytenoids are open, as for (h). ˛is concerns (i,
¨) between voiceless consonants, or between a voiceless consonant and a pause,
when in low-pitched syllables (but occasionally even in non-low-pitch ones, how-
ever never on çaccentedÇ morae, after which pitch becomes low), and never in in-
terrogative post-intonemes (which shows a raising of the basic pitch): (3©i'¿i) /kisiå
ki´i˚ (3k¨2Âi'6i2mM) /kMtibiRM/ ku>ibiru˚ (3hå'nå2¿i) /hanasiå hana´i˚ (3âi'tø3q¨)
/hitoùtM/ hito_u.
˛e second type, or degree, of devoicing is half-voicing, which concerns the oth-
er three vowels, but acts less systematically. In fact, we find (Ä, ), especially in the
first syllable of words, when followed by syllables containing the same vowel (more
rarely /e/ (É) is devoiced too): (3kÄ'tå2nå) /katanaå katana˚ (3t'kø2mø) /tokoRo/ toko-
ro÷ (3kÉ'så2nåi) /kesanai/ kesanai˘
˛e second type again applies, instead, to all vowels before a pause, either short
or long, preceded by any consonant. Obviously, /i, M/ become (i, ¨), when they
are in the condition of total devoicing, in the context (=é|). ˛erefore, before a
pause, the last vocoid is (î, É, Ä, , ¯) (even in diphthongs). It is important to state
that this type of devoicing is syntagmatic, which means that we have one vocoid
–not two– which begins as voiced and ends as voiceless. So the first part is voiced,
whereas the second one is voiceless: ((é‚)) – and this is hinted at by the pause con-
text, even if the actual pause is short. Examples relating to this phenomenon can
be found in the passage of the Text section, § 12.4. In simpler terms, we could say
that the auditory e‡ect is almost that of a very short (h), ((éh)) or a semiapproxi-
mant ((éh)).
For emphasis, strictly speaking (or for other paraphonic implications connect-
376 a handbook of pronunciation
ed with states of mind), often an actual (éh|) sequence may be heard: (˚Çsøø3d™3s¨-
œkåh2) /˚soùodesMka,/ Soo desu ka?!
Consonants
postpalatal
prepalatal
laryngeal
provelar
alveolar
bilabial
palatal
dental
velar
Nasals
12.2.1.1. At the beginning of syllables, Japanese has three possible nasal pho-
nemes: /m/ (m) m (bilabial), /n/ (n) n (alveolar; realized, however, as prepalatal,
(~), before /i, j/, by assimilation): (3~i'mø2nø) /nimono/ nimono˚ and /˙/ (˙) g (ve-
lar): (3å'˙å3k¨) /a˙aùkM/ agaku.
But we must add at once that, within words, /˙/ (˙) g (which is sometimes tran-
sliterated as ä) can be systematic only in the most neutral type of pronunciation,
after /é, ô/. In fact, nowadays the oscillation between /g ˙ ˙/ (g ˙ Ÿ ˙ ˙) is very
widespread, but with much fluctuation among people and words. However, no
native speaker systematically has only (g) /g/. In (sentence, phrase, lexeme, or
grammeme) initial position, (g) /g/ occurs, even for speakers who possess (˙) /˙/;
12. japanese 377
for (enclitic) ga we normally find (˙å) /˙a/; for ga (conjunction) we have (gå), but
(˙å) is possible too, even after a pause.
Again, there is (g) /g/ –above all– in loanwords, in onomatopeic expressions, in
Chinese reduplicate words, and after certain proclitics (which are obviously het-
erorganic): ('ái3˙å) /giù˙a/ giga˚ (Çáiô3˙å) /giùô˙a/ ginga; (3mi'ái) /migi/ migi˚ (3må'gM-
2mø) /magMRo/ maguro˚ (3møôÇg™P) /moôgeùô/ mongen˚ (3må'gø) /magoå mago˘
12.2.1.2. An even bigger phonic problem regarding nasals arises from a fourth
Japanese nasal phoneme, ¤ çmoraic nÇ, /ô/ (sometimes transliterated as â]˚ which
always occurs in a syllable coda and has a prevailing articulation as semi-provelar
(¤ provelar semi-nasal, with no actual contact between the dorsum and the ve-
lum). It is phonetically more energetic, ¤ intense (P).
It occurs in three positions: (1) before continuous consonants (¤ those produced
with an incomplete occlusion of the oral cavity), that is /s, z÷ j, µ÷ h/ (s ˙ ¿, z ˙ B÷
j, µ÷ h ˙ F ˙ â), (2) before a vowel (which is heterosyllabic, of course), and (3) in
final position before a pause (or, again, before continuous consonants, or vowels)\
(3™P'såP) /eôsaô/ ensan˚ (Çhå«2¿jå) /haôsja/ han´a˚ (ÇMN2jM) /MôjM/ un'yu˚ (ÇkåP-
2µå) /kaôµa/ kanwa˚ (3s™™ÇsåP3âi) /seesaùôhi/ seisanhi˚ (Çs™P3âjå3k¨) /seùôhjakM/ sen-
hyaku˚ (3gø's™P3F¨) /goseùôhM/ gosenfu˚ (ÇtåP3i) /taùôi/ tan'i˘
However, by assimilation, /ô/ has other realizations too. In fact, it is articulated
as a homorganic nasal (to a following consonant and is always intense): (1) (õ, ó÷
›, ô), before the correspondent stops /p, b÷ t, d/ (p, b÷ t, d) and /k, g/ (k, ©÷ g, á):
(Çsåõ3pø) /saùôpo/ sanpo˚ (Ç¿iõ2bMP) /siôbMô/ ´inbun˚ (ÇMó2t™P) /Môteô/ unten˚
(Çt™N3©i) ((-3£i)) /teùôki/ tenki˚ (Çtåô3kå) /taùôka/ tanka÷ and (2) (ó, «) before dental
or bilabialized prepalatal stopstrictive taxophones of /t, z/ (q, Q) and (Â, ©):
(3båóÇQåi) /baôzaùi/ banzai˚ (Ç™«2ªøø) /eôtjoo/ en>oo˚ (Ç~i«2©iP) /niôziô/ nin-
jin÷ and, naturally, (ô) before the frequent stop(semi){con}strictive variant, (wå,
˜å), of /ka/ (kå) ka\ (Çtåô3wå, -3˜å) /taùôka/ tanka.
In addition, we have (3) (õ, ó, «, ô) before nasals, /m, n, ˙/ (m n g) and again
(ó) before r /R/ (¬), ™: (ÇMõ3m™™) /Mùômee/ unmei˚ (3åóÇnåi) /aônaùi/ annai˚ (Ç~i«-
2~i2k¨) /niônikM/ ninniku˚ (Çåô2˙åi, -3˙åi) /aô˙ai, aôù-/ angai˚ and (3åó'¬å2k[¨]3¿i)
/aôRakMùsi/ anraku´i˘
Stops
12.2.2.1. Japanese has three (voicing) diphonic pairs for stops: the voiceless are
/p, t, k/ (p, t, k) (as already seen, often /k/ is (w, ˜) + /a{a}/; occasionally /t/ be-
comes (th) + /a{a}, o{o}/; and often /p/ is (ph), in the context between /ô/ and /i{i},
a{a}/): (3kåõ'påi, 3wå-, 3˜å-, -'phåi) /kaôpai/ kanpai˚ ('tå3kø÷ 'thå-) /taùko/ tako˚ (3©ip-
'p:M) /kippM/ kippu˚ (3µå'tå÷ -'thå) /µataå wata˚ (Çk™õ3pøø) /keùôpoo/ kenpoo
(word-initial /p/ only occurs in loanwords and onomatopeic forms). In emphatic
speech, there is more çaspirationÇ.
˛e greatest çoddityÇ regards /t/ (t) which, by assimilation, before /i, j/ (i, j) is
realized as a bilabialized prepalatal stopstrictive (Â) (transliterated as >˚ even if /j/
378 a handbook of pronunciation
(j) remains; but it is important to note that lip rounding is reduced, by assimila-
tion to /i, j/, but the phone remains di‡erent from (⁄)): (3Âi'Âi) /titiå >i>i˚
(Ǫåó2tø) /tjaôto/ >anto˘ In addition –and even more çstrangelyÇ– /t/ (t), before
/M/ (M) is realized as a dental stopstrictive (q) (transliterated _): (3qM'Ni) ((-∞i))
/tM˙i/ _ugi˘
Let us observe, once and for all, that the çpalatalÇ realizations of /k, g, ˙/, ¤ (©,
á, N), are instead çpostpalatalÇ, or retracted palatal, and can be represented better
with çspecialÇ symbols, ((£, 8, ∞)). ˛ey may safely be used (although, more often,
(©, á, N) are used), after stating that they are realized in the rear part of their artic-
ulatory space, we define çpostpalatalÇ, which in any case remains distinct from the
çprevelarÇ articulation, (´, Ò, ”), typical of most languages before front vowels and
(j), as in English (('´hI;i, '>™ÒjÈlå)) {Am. Engl. (('<™ÒjÈl≥))} (('s¤”¤˙)) /'kIi, '<EgjÈlÈ≤, 'sI˙-
I˙/ key˚ regular˚ singing.
˛us, we have: ((3£ip'p:M, 3mi'8i, 3qM'∞i)) /kippM, migi, tM˙i/ kippu˚ migi˚ _ugi˚
or (3©ip'p:M, 3mi'ái, 3qM'Ni), provided (Ni) remains di‡erent from (~i) (therefore,
/ni/ (~i) should not be rendered as if it were ç(Ni), as we find in quite a few publica-
tions).
12.2.2.2. For /b/ (b) b˚ the variant (6) is more common (¤ a voiced bilabial con-
strictive), which occurs after vowels, especially in non-slow and non-formal pro-
nunciation; less frequently the corresponding approximant, (B), is used: (3sM'6™-
3mM) /sMbeùRM/ suberu˚ (Çbåi2kåi) /baikai/ baikai˚ (Ç©jMõ3bi) /zjMùôbi/ junbi˘
˛e phoneme /d/ (d) d poses no problems, except that genuine Japanese words
never have *di, *dyV and *du˚ substituted by ji˚ /zi/ (©i, Bi) jV˚ /zjé/ (©jé, Bjé)
and zu˚ /zM/ (QM, zM): (Çdåi3BiP÷ -3©iP) /daùiziô/ daijin˚ (3©jM'zM÷ -'QM) /zjMzMå
juzu˘ Its normal distribution, non-emphatic and non-slow, is (éB[j]÷ |©[j], «©[j]).
Also /g/ (g) g poses no problems, apart from a complementary (or alternative)
distribution with /˙/ (˙) [g˚ which sometimes is transliterated as ä˚ as already said):
('gø3gå3k¨) /goùgakM/ gogaku˘
Word-initial vowels, both at the beginning or in the middle of phrases and sen-
tences, are generally preceded by (ö), especially for emphasis or to separate vowels
of adjoining words.
˛is fact will be indicated prevailingly in connected transcriptions, such as those
in § 12.4. In addition, especially (but not only) in women's pronunciation, short
utterance-final vowels, mostly with a suspensive intoneme, can be followed by (ö[æ])
(¤ with or without {an audible} release) as an alternative pronunciation instead of
a possible more çnormalÇ partial devoicing of the last vocoid: ('[ö]ø3kÉ|, '[ö]ø3k™[ö]|)
/oùke/ oke˚ (3[ö]å'sM|, 3[ö]å'sM[ö]|) /asMå asu˘ In the passage in § 12.4.2.3, a couple of
cases are indicated.
Constrictives
by coarticulation) before i˚ /i/ (i) (transliterated as ´) and before yV˚ /jé/ (jé) (tran-
sliterated as ´V˚ although /j/ (j) does not disappear at all): (3¿jå'¿iP) /sjasiô/ ´a-
´in˘
˛e corresponding voiced sound, /z/ z˚ is (éz) (slow — careful: (éQ)) and (|Q,
óQ). Hence, it is realized as a dental constrictive, between vowels, either in words
or sentences, either in normal or fast speech. However, after a pause or /ô/, it is re-
alized as a stopstrictive: (3kå'z™) /kaze/ kaze (slow — careful: (3kå'Q™)), (3QM'åP)
/zMaô/ zuan˚ (Çs™ó3Qø) /seùôzo/ senzo˘
In addition, we find /z/, before i˚ /i/ (i) (transliterated as j˚ instead of z] and be-
fore yV˚ /jé/ (jé) (transliterated as jV˚ although /j/ (j) remains), which is realized as
prepalatal rounded (again, with reduced rounding) (éBi, éBjé) (slow — careful:
(é©i, é©jé)) and (|©i, |©jé÷ «©i, «©jé): ('FM3Bi) /hMùzi/ fuji˚ (3©i't™P) /ziteô/
jiten˚ (Çkå«2©i) /kaôzi/ kanji˘
Approximants
12.2.4.1. Japanese has three approximants. ˛e first, /j/ (j) y˚ is (voiced) palatal:
(3så2jø'nå3må, -2må, 3så2jøø'nå3må) /sajonaùRa, -Ra, sajoonaùRa/ sayonara (-yoo-)˚ (3jå2¿i'©i,
3jå¿:'©i) /jasikiå ya´iki˚ (3mi'å2kø) /mijako/ miyako˚ (Çjøø2©ªMM) /jookjMM/ yoo-
kyuu˚ (Ç¿jMM3¿i) /sjMùMsi/ ´uu´i˘
As can be seen from the examples, it remains unchanged in word-initial posi-
tion /òjé/ (òjé), whereas it is realized as a phonetic çzeroÇ, (`), when preceded by
i˚ /i/: /ijé/ = (ié). In non-slow speech, the same is possible for /ejé/ = (™é), or (™ãé);
thus we will mark (™ãé) (by using a palatal semi-approximant symbol): (3h™'ãå) /he-
jaå heya.
After the voiceless stop phonemes /pjé, tjé, kjé/, /j/ is devoiced, (ª): (pª, ª, ©ª);
but it remains (j) after other consonants (even if voiceless, /s, h/ (¿, â) s˚ h]˚ and
without being absorbed by /t, s, z, h/: (ª, ¿j, Bj, ©j, âj) (in spite of translitera-
tions such as >, ´, j). Sequences such as *yi, *ye do not occur.
£TrillsÇ
12.2.5. Japanese has one phoneme of the çtrillÇ type, which is similar to Spanish
r /R/ (R), as in interpretar˚ /inteRpRe'taR/ (in&teRpRe'taR). A realization like this could
be su‚cient for a fairly good pronunciation of Japanese, all the more so because
that is indeed one of the possible realizations. However, it is better to learn the
two most typical articulations given shortly (which are alveolar again), and use
them instead of (R).
˛e first taxophone of /R/ r is (m), lateral flap (or lateralized flap – Ô û 10.13 in
NPT/HPh, besides û 1.14.2-3 in this handbook), which occurs after vowels (even
within sentences): (3kå'µå2må) /kaµaRa/ kawara˚ (3™'mi) /eRiå eri (the same phone al-
so occurs in American English before (≥), ™ ('b™m≥) /'bEùÈ≤/ better]˘ A non-lateralized
flap is possible too, ([): (3kå'µå2[å) /kaµaRa/ kawara˚ (3™'[i) /eRiå eri (which is the
main American phone, in all contexts without (≥), ™ ('b™[i) /'bEùi/ Betty]˘ In Japa-
nese it is better to use (m), although ([, R) are possible too.
˛e second taxophone of /R/ r is (¬), lateral tap (or tapped lateral, û 1.15.3),
which occurs after /ô/ (even within sentences) or after pauses: (Çb™ó2¬i) /beùôRi/ ben-
ri˚ (3¬M'i2Bi, -2©i) /RMizi/ ruiji˚ (3¬™2©i'¿i, -©:'¿i) /Rekisi/ reki´i˚ (Ǭøó3¬i) /RoùôRi/ ronri˘
For the sake of simplicity, we could say that the di‡erence between the two reali-
zations consists in di‡erent degrees of lateralization. In fact, (m) is less lateralized,
as lateralization is an added, or secondary, component (let us say 1ˇ’); whereas, for
(¬) lateralization is prevailing, or primary (let us say 2ˇ’).
Occasionally, lateral realizations can be heard as well, (l) (which is still alveo-
lar), and postalveolar ones (of various manners of articulation – respectively: flap,
tap, stop, flapped lateral, lateral: (®, e, Ã, », $). Obviously, these phones need not
be actively acquired – it is su‚cient to be simply able to recognize them.
£Palatalization∞
12.2.6.1. Before /i, j/, the phonemes /n, ˙÷ t, k, g÷ s÷ z÷ h/ have peculiar but neces-
sary realizations: (~i, Ni) /ni, ˙i/ ni, gi÷ (Âi, ©i, ái) /ti, ki, gi/ >i, ki˚ gi÷ (¿i÷ Bi, |©i,
«©i÷ âi) /si÷ zi÷ hi/ ´i÷ ji÷ hi\ (Ç~i«2©jMM) /niôzjMM/ ninjuu˚ (3kå'Ni2mM) /ka˙iRM/
kagiru÷ (3Âi'™) /tieù/ >ie˚ (3©i'™2mM) /kieRM/ kieru˚ (3ái'µå2k¨) /giµåkM/ giwaku÷ (3¿i-
'må) /simåå ´ima÷ (3å'Bi) /azi/ aji˚ (3©i'mi) /zimiå jimi˚ ('må3âi) /maùhi/ mahi˘
In addition, we find: (~j, Nj) /nj, ˙j/ ny, gy÷ (ª, ©ª, áj) /tj, kj, gj/ >, ky˚ gy÷ (¿j÷
Bj, |©j, «©j÷ âj) /sj÷ zj÷ hj/ ´÷ j÷ hy\ (Ç~jMM2˙å2k¨) /njMM˙akM/ nyuugaku˚ (ÇåN-
2Njå) /aô˙ja/ angya÷ (ǪMM3ªø) /tjMùMtjo/ >uu>o˚ (3©ªMMÇ©ªMM3¿jå) /kjMM-
kjMùMsja/ kyuukyuu´a˚ (ÇájMM2~jMM) /gjMMnjMM/ gyuunyuu÷ (3¿i«Ç~jMM3s™™)
/siônjMùMsee/ ´innyuusei÷ (ÇkåN2jMM) /kaôjMM/ kan'yuu˚ (3©iÇdøø3¿jå) /zidoùos-
ja/ jidoo´a˚ (3âjå'kM) /hjakMå hyaku˘
Again, we must remember that, for /k, g, ˙/ + /i, j/, the actual articulation is
çpostpalatalÇ ((£, 8, ∞)) (rather than fully palatal, (©, á, N)), and that (j) remains.
cess). So, we regularly have: (émi, |¬i, ó¬i) /Ri/ ri÷ (mi, pi) /mi, pi/ mi, pi÷ (bi, 6i)
/bi/ bi. ˛us: (3ø26i'™2mM) /obieRM/ obieru˚ (3mi'nø2mi) /minoRi/ minori˚ (3¬i'sø2k¨) /Ri-
sokM/ risoku˚ (Ç¿ió3¬i) /siùôRi/ ´inri˚ (Ç™õ2pi2q¨) /eôpitM/ enpi_u˘
In addition, /0jé/ (0jé) CyV, ChV\ (mj, |¬j, ó¬j) /Rj/ ry÷ (mj, pª) /mj, pj/ my˚ py÷
(bj, 6j) /bj/ by: (Ç™ó3¬jø) /eùôRjo/ enryo˚ (ǬjMM) /RjMùM/ ryuu˚ (3¬øp'p:ªå2k¨) /Roppja-
kM/ roppyaku˚ (Çbjøø) /bjoùo/ byoo˚ (ÇbMõ2mjå2k¨) /bMômjakM/ bunmyaku˚
(3QåiÇmjøø) /zaiRjoùo/ zairyoo˘
£Gairaigo∞ – loanwords
12.2.7. As in any language, even in Japanese, loanwords (of which about 10,000
are of English origin) require some adaptation to the syllable structure (which is
based on morae in Japanese) and new phonemic combinations for new sounds.
For these typical adaptations, let us consider two examples: (2kMçmå36M) /kMùRa-
bM/ çclubÇ˚ (3s[¨]tøÇmåi3©[i]) /sMtoRaùiki/ ç(workers') strikeÇ˚ (3s[¨]tøÇmåi3k[¨]) /sM-
toRaùikM/ ç(baseball) strikeÇ. It is clear how the Japanese syllabic structure changes
original monosyllables, with consonant clusters, into actual polysyllables.
Among new combinations, in the traditional phonotactics (which is quite rigid
and with a fairly limited number of possibilities), the most common are: (Çpaa3ti)
/paùat-i/ çpartyÇ˚ (3di'm™3k[¨]3taa) /d-iReùkMtaa/ çdirectorÇ˚ (ÇÂ[ª]™«3©i) /tjeùôzi/
çchangeÇ˚ (Ç©[j]™t3t:ø) /djeùtto/ çjetÇ˚ ('¿[j]™3F[¨]) /sjeùhM/ çchefÇ˚ (3FM'i2mM2mM,
-3mM3mM, 'âi-, 'Fµi-, 'Fi-) /hMiRMmM, h[M]iùRMmM/ çfilmÇ˚ (Ç¿iõ3Fø3~ii) /siùôhMo-
nii/ çsymphonyÇ˚ (3kåóÇqøø3n™) /kaôtMoùone/ çcanzoneÇ.
Some examples clearly show that, besides placing some phones into new combi-
nations, certain sequences are slightly çdenipponizedÇ (as the possible dropping,
in these words but not in genuine ones, of (j) after prepalatal {rounded} articula-
tions), becoming slightly more çinternationalÇ.
Structures
Taxophonics
12.3.1.1. ˛e basic things have already been said. In fact, we have seen devoiced
vocoids (Ô § 12.1.3), and the few taxophonic characteristics regarding Japanese
consonants.
382 a handbook of pronunciation
Gemination
12.3.1.2. We must state that a çmoraÇ coincides with a light syllable, as the one
formed by a short vowel (/i, e, a, o, M/), or by /ô/, or else by the first element of a
geminate consonant (/é-0-0é/ (0-0:)).
A half-heavy syllable corresponds to a geminate vowel (/ii, ee, aa, oo, MM/) or a
diphthong, or to a short vowel + /ô/ (/éô/) or + the first element of a geminate con-
sonant (whose second element belongs to the following syllable, together with its
vowel).
Instead, a heavy syllable presents a geminate vowel (or diphthong) + /ô/ (/ééô/)
or + the first element of a geminate consonant: /éé0[-0é]/.
As we have seen from various previous examples, in Japanese, vowel length is dis-
tinctive, ¤ short and çlongÇ or rather geminate (or doubled) vowels: (sø) /so, soå
so (one mora), (Çsøø) /soo, soùo/ soo (two morae); ('tø3©i) /toùki/ toki, (Çtøø3©i) /toùo-
ki/ tooki. Even consonant length is distinctively present, as in (3©i't™) /kite/ (from ki-
ru /kiRM/ çto wearÇ) and /kiteå kite (from kuru /kMRM/ çto comeÇ), both with two
morae; (3©it't:™) /kitte[ù]/ (çstampÇ) and ('©it3t:™, 3©it't:™) /kiùtteå kitte (from kiru /ki-
ùRM/ çto cutÇ), both with three morae: /ki-t-te/; (3gå'kå) /gaka/ gaka çartistÇ, (3gåk-
'k:å) /gakka/ gakka çlessonÇ.
From a phonetic point of view, a doubled consonant consists of two morae: the
first one coincides with the first element of the gemination (even if it is actually
shorter, (t)), whereas the second one (which is decidedly longer, since actually
lengthened, (t:)) constitutes another mora together with the vowel that follows it:
(©it-t:™) (although here we have omitted the pitch features given above). In fact,
/sotto/ is (3søt't:ø) sotto çsoftlyÇ (while a similar Italian word sotto /'sotto/ çunderÇ
is ('sot:to) in an intoneme, or ('sotto) in a preintoneme). Let us observe well –and
listen even more carefully to– the di‡erence between (0:0) and (00:). Both for
Japanese /sotto/ or for Italian /'sotto/, however, we always have two phono-sylla-
bles, even if Japanese /sotto/ has three morae.
çMoraicÇ n is always postvocalic, but it can also be followed by a vowel (and it
is transliterated as Vn' V˚ in order to make people realize we are dealing with /éôé/
(éP-é), not with VnV /éné/ (é-né), and the same goes for Vn'yV /éôjé/ (éP-jé),
which is di‡erent from VnyV /énjé/ (é-~jé)): (Çt™ó3døø) /teùôdoo/ tendoo (Ô, a sim-
ilar Italian word, tendo /'tEndo/, or ('tEn:do) in an intoneme).
However, in Japanese, in addition to /éôé/ (éP-é) and /éné/ (é-né), as in (ÇåP3i)
/aùôi/ an'i˚ ('å3~i) /aùni/ ani, we can also have /éô˙é/ (éô-˙é) (which is the combina-
tion of /éô/ and /˙é/): (Ç™N2Ni) /eô˙i/ engi and /éôné/ (éó-né) too (combination
of /éô/ and /né/): (Çåó2nå) /aôna/ anna˚ (Çå«3~i) /aùôni/ anni (Ô Italian: /'anna, 'an-
ni/ ('an{:}na, 'an{:}ni) Anna, anni]˘ Let us add this –not useless– example: (ÇhøP3jå)
/hoùô-ja/ hon'ya˘
˛erefore, the structure of /ô/ (õ, ó, «, N, ô, P) (one mora) is di‡erent from the
geminate one /0-0é/ (two morae, or three including the first vowel: /é-0-0é/).
However, there is no di‡erence for the counting of morae.
12. japanese 383
12.3.2.2. Only in the following table (which relates to û 12.3), we will show a
tonemic (å) and a tonetic pattern (∫, where (3ø) indicates a low-pitched syllable,
with the vowel timbre of /o/; whereas (2ø) indicates mid pitch). We consider them
to be more useful and convenient, in order to describe and learn/teach. We add
pattern (©) that is most recommendable in transliterations which do not ignore a-
kusento, when no transcriptions are used.
According to the general principles of not explicitly writing unmarked prosod-
ic elements, in transcriptions, the notation (2) –for mid pitch– could or should be
left out. However, it is certainly more useful to show it, all the more so because in
actual examples it is much less obtrusive than in the table. In any case, in pattern
(∆) it has been left out.
For useful comparisons, we will add the most widespread patterns used in trans-
literations (∂, ™), and the one used in katakana moraic transcription (ƒ) with typic-
ally oriental graphic complexities. We also show a phonotonetic pattern turned
into a more çorthodoxÇ one (Ÿ), originating from pattern (™). To indicate any mo-
ra, here we use (ø), /o/, o˚ $ (the last one to çindicateÇ katakana, in ƒ):
12.3.2.3. In type-å tonetic transcriptions, the first instance (/oooo/) di‡ers from
the second (/ooooå), because for the latter we also indicate the succeeding lower-
ing (which –for obvious reasons– is absent in type-∫ transcriptions). Indeed, it is
not objectively present in actual reality, when no words follow (as we will see short-
ly). In type-© transliterations, the accent could even be acute [ó]˚ but the grave one
is to be preferred since it can show actual movements better – in fact, the pitch
falls from mid to low.
Type-∂ and type-™ transliterations reflect the first ones in a more abstract way:
tonemic and tonetic (å, ∫); even the katakana transliteration –or çtranscriptionÇ–
(ƒ) shows the same characteristic, but in a more abstract way in comparison with
real transcriptions (å, ∫, Ÿ).
We do not use transcriptions of the type /oŒooo/ (^øŒøøø), which some use
though (on the wake of type ∂ and type-™ transliterations), because if syllables/mo-
rae were really pronounced in a high pitch, instead of the mid one, the result
would not be at all convincing. Let us add that in certain textbooks it is possible
to find both types ∂ and ™ with katakana, and type ƒ with transcriptions.
As far as the indication or pitch variants is concerned, as we have done in §
12.0.2, for hiragana ((3âi'må2˙åœnå, 3âi'må3˙åœnå÷ 3âi'må2˙å&nå) /hiRaù˙aùna÷ -naå) and ka-
takana ((3kÄ'tå3kåœnå, 3kÄ'tå2kåœnå) /kataùkaùna/), the most important thing is to in-
dicate them (unless there are usage di‡erences). Our phonotonetic transcription
also shows their (wider or narrower) di‡usion and recommendability (which is to
be interpreted gradually – from the first onwards).
On the other hand, in a pronouncing dictionary, which would be worth publish-
ing (using a transliteration, followed by the o‚cial writing and, of course, by an
IPA transcription, certainly a phonemic one), preferences could and should be
shown, by indicating /hiRa˙aùna, hiRaù˙ana, hiRa˙anaå and /kataùkana, katakaùna/,
obviously in a shortened form: /hiRa˙aùna, -Raù-÷ -naå and /kataùkana, -kaù-/. Of
course, in an introduction, the precise phonetic and tonetic characteristics would
be fully treated, with accurate transcriptions, and with all the necessary voco-
grams, orograms, and tonograms, Â.
12.3.2.4. A short –mono-moraic– syllable may have two pitch possibilities: /ne/
ne çsound, toneÇ (absence of akusento), /neå ne çrootÇ (presence of akusento), but
tonetically they are both çnon-lowÇ (¤ said in a mid pitch: ('n™); the same goes for
/ki/ ki çspiritÇ, /kiå ki çtreeÇ, ('©i), and /ha/ ha çleaveÇ, /haå ha çtoothÇ, ('hå).
In the case of two morae, we can have (Çsøø) /soùo/ soo çmonkÇ, (Çsøø) /soo/ soo
çvilla, innÇ, which are monosyllables realized right as (Çsøø), with half-low pitch –
but slightly falling or rising, respectively, since they combine mid and low, or low
and mid, pitch (Ô û 12.4).
12. japanese 385
In unstressed bimoraic syllables, we find (3) (which is low, but raised up to the
border with the mid band, as can be seen from the figure, in comparison with the
low pitch of (ç)), as in: (3s™PÇs™™) /seôseùe/ sensei˚ (Çkøø3âii) /koohiùi/ koohii.
For two syllables formed by adding –to a monomoraic monosyllable– particles
such as (˙å, µå) /˙a, µa/ ga˚ wa˚ which are akusento-less (since their pitch depends
on what precedes, even if {when pronounced metalinguistically in isolation} they
are obviously ('˙å, 'µå) /˙a, µa/), we obtain respectively: (3n™'˙å) /ne˙a/ ne ga and
('n™3˙å) /neù˙a/ ne ga, (3©i'˙å) /ki˙a/ ki ga and ('©i3˙å) /kiù˙a/ ne ga, (3hå'˙å) /ha˙a/ ha
ga and ('hå3µå) /haùµa/ ha wa.
It is fundamental not to believe that Japanese has two çtonemesÇ – low and mid.
As a matter of fact, akusento is not at all a kind of actual pitch height (as it is not
stress either). On the contrary, it is a pitch fall. It is a sort of çcatatonic pointÇ, af-
ter which the pitch falls, passing to the low band, as the examples clearly show.
But above all, akusento is either present or absent. In English, (the position of)
stress is phonemic; whereas it is not so in Japanese. Besides, in English, pitch de-
pends only on intonation (and paraphonics); whereas, in Japanese, pitch is phone-
mic and fundamental.
˛is language has mid pitch until an akusento comes, after which the pitch be-
comes low. If no akusento occurs, the pitch remains mid. On the contrary, stress
in Japanese is not phonemic and depends on a complex interplay of various fac-
tors, such as the presence or absence of akusento, where it is placed, and the sylla-
ble structures of the rhythm group.
Naturally, the tonograms su‚ciently highlight that, in a rhythm group (or in
an isolated word), the first mora is low and contrasts with the second one, which
is mid; provided that (as we have already seen) the first mora is not followed by an
akusento, in which case it is mid and what follows is low.
Again with two morae, we also have ('hå3nå) /haùna/ hana çedgeÇ, (3hå'nå) /hanaå
hana çflowerÇ, (3hå'nå) /hana/ hana çnoseÇ (all bisyllabic), as for the monosyllabic
examples with a particle seen above.
As soon as we add a particle, the e‡ect of akusento is immediately clear: (3n™'µå)
/neµa/ ne wa çsound, toneÇ, ('n™3µå) /neùµa/ ne wa çrootÇ, ('hå3nå3˙å) /haùna˙a/
hana ga çedgeÇ, (3hå'nå3˙å) /hanaù˙a/ hana ga çflowerÇ, (3hå'nå2˙å) /hana˙a/ hana ga
çnoseÇ; and so on.
û 12.4. Movements in syllables with morae of di‡erent pitch.
£ Ç /éé, éô/ =
≥ (Çéé, ÇéP), (3éé, 3éP)
£ Ç /éùé, éùô/ =
386 a handbook of pronunciation
Stress in Japanese
12.3.2.6. True problems begin with bisyllables, though. In fact, there are di‡er-
ences between (3å'm™) /ame/ ame çcandyÇ, ('å3m™) /aùme/ ame çrainÇ, and (3n™'mM)
/neRM/ neru çto sleepÇ, ('n™3mM) /neùRM/ neru çflannelÇ. Bisyllables of two morae,
that is with two light syllables, are stressed on the second syllable, unless akusento
follows the first one, which is then stressed: (3k'kø) /koko/ koko˚ (3å'Bi) /azi/ aji˚
(3M'™) /Me/ ue˚ (3i'M) /iM/ iu˚ (3ø'i) /oi/ oi and (3ø'tø) /otoå oto˚ (3å'¿i) /asiå a´i˚
(3mM'må) /mMRaå mura˚ (3qM'Ni) /tM˙iå _ugi˚ (3¿i'ø) /sioå ´io˚ (3i'™) /ieå ie˘ How-
ever, we have: ('dø3m™) /doùRe/ dore˚ ('å3©i) /aùki/ aki˚ ('qM3må) /tMùma/ _uma˚ Â.
˛ree-mora bisyllables are stressed on the heaviest syllables (¤ with more morae
than others), although there are some oscillations that we will indicate. It is impor-
tant to accurately observe akusento di‡erences (which are pitch di‡erences), in
phonemic transcriptions, since sometimes they are the only actual di‡erences (but
û 12.4 must be carefully considered): (3ø'møi) /omoi/ omoi˚ (3øÇmøi) /omoùi/ omoi˚
(3kå'søø) /kasoo/ kasoo˚ (3kåÇsøø) /kasoùo/ kasoo˚ (3i'måi) /iRai/ irai˚ (2içmåi) /iùRai/ irai˚
(3å'øi) /aoi/ aoi˚ (3åÇøi) /aoùi/ aoi˚ (3¿i'åi) /siai/ ´iai˚ (3©i'nøø) /kinoo/ kinoo˚ (3jø't™™)
/jotee/ yotei˚ (2âiç˙åi) /hiù˙ai/ higai (but ('©i3Njøø) /kiù˙joo/ kigyoo]˚ (2kMçmøø) /kM-
ùRoo/ kuroo˚ (2båçm™™) /baùRee/ baree˘
More: (3©iÇkåi) /kikaùi/ kikai˚ (3¿iÇk™P) /sikeùô/ ´iken˚ (2içk™P) /iùkeô/ iken˚ (3i'k™P)
/ikeô/ iken˚ (2gøçz™P) /goùzeô/ gozen˚ (2QMç6øP) /zMùboô/ zubon˚ (Çkøø2Bi) /koozi/
kooji˚ (Çkøø3Bi) /koùozi/ kooji˚ (Ç™N2Ni) /eô˙i/ engi˚ (Çåi2då) /aida/ aida˚ (Çhåi3mM)
/haùiRM/ hairu˚ (Ç©ªøø3tø) /kjoùoto/ Kyooto˚ (Çkåi2Ni) /kai˙iå kaigi˚ (Çdåi3k¨) /daùi-
12. japanese 387
kM/ daiku˚ (Çb™ó3¬i) /beùôRi/ benri˚ (Çmió2nå) /miôna/ minna˚ (3miq'q:M) /miqqMå
mit_u˚ (3åk'k:å) /akka/ akka (but: ('åk3k:å) /aùkka/ akka˚ ('¬™¿3¿:jå) /Reùssja/ res´a˚
where pitch prevails on other factors).
Generally, in four-mora bisyllables, stress falls on the first syllables, unless it is a
light one (¤ with just one mora) or there is an akusento after the second one (or if
the first is only çhalf-heavyÇ, ¤ checked by (=˘=:)): (Ç©jMM26jøø) /zjMMbjoo/ juu-
byoo˚ (Ç©jMM36jøø) /zjMùMbjoo/ juubyoo˚ (Çhøø2køø) /hookoo/ hookoo˚ (Ç©jMM-
3døø) /zjMùMdoo/ juudoo˚ (Çkøø2Bjøø) /koozjoo/ koojoo˚ (3køøÇBjøø) /koozjoùo/ koo-
joo˚ (3tåiÇFMM) /taihMùM/ taifuu˘
More: (Çk™ó2tøø) /keôtoo/ kentoo˚ (3k™óÇtøø) /keôtoùo/ kentoo˚ (Çs™ó2tøø) /seô-
too/ sentoo˚ (Çs™ó3tøø) /seùôtoo/ sentoo˚ (3s™PÇs™™) /seôseùe/ sensei˚ (Ç©iP2™P) /kiôeô/
kin'en˚ (ÇjMM26iP) /jMMbiô/ yuubin˚ (3gåk'k:øø) /gakkoo/ gakkoo˚ (3tø©'©:ªMM) /tok-
kjMM/ tokkyuu (but: (ÇbøÂ3Â:ªåP) /boùttjaô/ boc>an]˚ (3™'©iiP) /ekiiô/ ekiin˚ (3¿ip-
'p:åi, 3¿i-) /sippai/ ´ippai˚ (Çtøø2©ªøø) /tookjoo/ Tookyoo˘
12.3.2.9. In five-syllable words, which are generally not simple words, stress de-
cidedly tends to fall on the last but one syllable, except for particular compound-
ing or devoicing: (3kå&tå2zM'k™3mM) /katazMkeùRM/ katazukeru˚ (3i&nå26içkå3mi) /inabiù-
kaRi/ inabikari˚ (3ø&tø2køçnø3kø) /otokoùnoko/ otokonoko˚ (3n™&Bi2måçµå3¿i) /nezi-
maùµasi/ nejimawa´i˚ (3jå&må2nøç6ø3mi) /jamanoùboRi/ yamanobori˚ (3ø&kM2mi'mø-
2nø) /okMRimono/ okurimono˚ (3ø&©ªå2kM'så2må, œø2©ªå2k¨-) /okjakMsama/ okyaku-
sama˚ (3ø&t™3qMçdåi3såP) /oteùtMdaisaô/ ote_udaisan˚ (3å&m™2mi'kå3BiP, -3©iP) /ameRi-
ka ùziô/ amerikajin˘
More: (3s™ó&tå2kM'mø2nø) /seôtakMmono/ sentakumono˚ (3i&Âi2~i'Âi2BjMM) /iti-
nitizjMM/ i>ini>ijuu˚ (œd™N2©i2s¨'tøø36M) /deôkisMtooùbM/ denki-sutoobu˚
(œÂªMM2˙ø2kMÇmjøø3mi) /tjMM˙okMRjoùoRi/ >uugokuryoori˚ (œkøø2då«Ç©jMM3tå-
3k¨) /koodaôzjMùMtakM/ koodan-juutaku˚ (œdåõ2bøø's™3qM36i, çdåõ2bøø 's-) /daô-
booseùtMbi/ danboo-se_ubi˚ (œø2må'µå2mi&såP) /omaùµaRisaô/ omawarisan˚ (3~jMM-
&˙å2kM2¿iÇk™P) /njMM˙akMsikeùô/ nyuugaku-´iken˚ (œmå2ÂiÇåi2¿i3qM, -i3q¨) /ma-
tiaùisitM/ ma>iai´i_u˘
12.3.2.10. In verbs in -ru generally stress is on the preceding syllable and akusen-
to immediately after it: (œø26ø'™3mM) /oboeùRM/ oboeru˚ (œkåô2˙å'™3mM) /kaô˙aeùRM/
kangaeru˚ (œå2qM'må3mM) /atMmaùRM/ a_umaru˚ (œ¿i2må'6™3mM) /siRabeùRM/ ´iraberu˚
12. japanese 389
12.3.2.11. Since there are no real rules for akusento and stress in compounds,
especially for less long ones, we just give some examples, to rouse reflection (also
on the paramount importance that a real pronouncing dictionary would have {by
using IPA} for akusento˚ stress, segments, devoicing, Â): (3FMjM'˙™3¿[i]3©i) /hMjM-
˙eùsiki/ huyuge´iki˚ from (3FM'jM) /hMjMå huyu˚ ('k™3¿[i]3©i) /keùsiki/ ke´iki˘
Other examples: (3jM&~jMM2kM'då3mø3nø) /jMnjMMkMdaùmono/ yunyuukuda-
mono˚ from (3jM'~jMM) /jMnjMM/ yunyuu˚ (3kM'då3mø3nø) /kMdaùmono/ kudamo-
no÷ and more: (œjMM2jå2k™'zø3må) /jMMjakezoùRa/ yuuyakezora˚ from (ÇjMM2jå2k™)
/jMMjake/ yuuyake˚ ('sø3må) /soùRa/ sora÷ (3så&tø2˙øçkø3mø) /sato˙oùkoRo/ satogokoro˚
from (3så'tø) /sato/ sato˚ (3k'kø3mø) /kokoùRo/ kokoro.
Furthermore: (3~i'˙M3mM3må) /ni˙MùRMma/ niguruma˚ from ('~i) /niå ni˚ (3kM-
'mM2må) /kMRMma/ kuruma÷ (3å'så3kå3z™) /asaùkaze/ asakase˚ from ('å3så) /aùsa/ asa˚
(3kå'z™) /kaze/ kaze÷ (3iP2s[¨]&tåó2tøÇkøø3âii) /iôsMtaôtokoùohii/ insutantokoohi˚
from (3iP2s[¨]Çtåó3tø) /iôsMtaùôto/ insutanto˚ (Çkøø3âii) /koohiùi/ koohi÷ (œgåi2kø'kM-
3BiP, -3©iP) /gaikokMùziô/ gaikokujin˚ from (Çgåi2kø2k¨) /gaikokM/ gaikoku˚
(Ç©iP) /ziùô/ jin÷ (œgåi2kø&kM2BióÇtøø3mø3k¨, -2©ió-) /gaikokMziôtoùoRokM/ gaiko-
kujin-tooroku˚ from (œgåi2kø'kM3BiP, -3©iP) /gaikokMùziô/ gaikokujin˚ (Çtøø2mø2k¨)
/tooRokM/ tooroku˘
/jasasiùi hoùô/ yasa´ii hon (without modifications), (Çii 3Biç6i3©i, 3©i-) /iùi zibiki/ ii
jibiki (from (3©i'6i2©i) /zibikiå, with lowering of -biki]˚ (3sM'zM3¿ii 3h™çãå) /sMzM¿iùi
hejaå suzu´ii heya (in (3h™'ãå) /hejaå, -ya is lowered), (3s™PÇs™™œd™3s¨, -s:, -s) /seôseùe
desM/ sensei desu (with stress reduction on ('d™3s¨, -s:, -s) /deùsM/ and lowering of
de-]˘
More: (3tø't™2mø 2ø'mø2¿iœmøi çhøP) /totemo omosiRoùi hoùô/ totemo omo´iroi
hon (from (3ø'mø2¿iœmøi), with raising of o- and lowering of hon]˚ (3âi'Bjøø2~i 2F¨-
&kM2zå'qM2nå 'møó2dåi) /hizjooni hMkMzatMna moôdai/ hijoo ni fukuza_una
mondai (with raising of fu- and mon-]˚ ('møt3t:ø 3jåçsMi çhøP) /moùtto jasMùi hoùô/
motto yasui hon (with lowering of -sui and hon]˚ (3ø'hå2jøø 2gø'zåi2må3s¨, -s:, -s) /o-
hajoo gozaimaùsM/ ohayoo gozaimasu (with raising of go-]˘
Some others: (3å'mi3˙å3tøø 3gøçzåi3s¨, -s:, -s) /aRiù˙atoo gozaimaùsM/ arigatoo gozai-
masu (with lowering of -zaima-\ everything is low, except -ri-]˚ (Çdøø3mø 3åçmi3˙å3tøø
3gøçzåi3må3s¨, -s:, -s) /doùomo aRiù˙atoo gozaimaùsM/ doomo arigatoo gozaimasu
(with lowering of -ri- and -zaima-\ everything is low, except doo-˚ which is half-low).
In addition: (çM2Âi 'kå2må '™3©i çmå3d™| 3å&mM2©i'må3s¨, -s:, -s) (in the case of fast
speech with no breaks: (çM2Âi 'kå2må '™3©i çmå3d™ 3åœmM3©içmå3s¨, -s:, -s); but, separate-
ly, word by word: (3M'Âi, 3kå'må, '™3©i, 'må3d™, 3å&mM©i'må3s¨, -s:, -s)} /Mti kaRa eùki maù-
de aRMki maùsM/ u>i kara eki made arukimasu (please, note the stress change in the
first rhythm groups, mainly due to /i/ devoicing).
Also: ('~i3Bi&kå3må\ Çså«3©i çmå3d™| 3©i'mM3¿jø3~i\ 3içmå3s¨, -s:, -s)÷ possibly with
fewer breaks, in less slow speech: ('~i3Biœkå3må çså«3©i çmå3d™| 3©i'mM3¿jø3~i 3içmå-
3s¨, -s:, -s). On the other hand, in separate rhythm groups, we have: ('~i3Bi, 3kå'må,
Çså«3©i, 'må3d™| 3©i'mM3¿jø3~i, 3i'må3s¨, -s:, -s) (with modifications on kara]˚ /niùzi
kaRa saùôzi maùde ziùmMsjoni imaùsM/ niji kara sanji made jimu´o ni imasu÷ ('jø3Bi
3måçd™3~i\ 3k'kø2~i 2©i't™ 2kM2dåÇsåi) (slower: ('jø3Bi 3måçd™3~i\ 3kø'kø2~i\ 2©i't™ 2kM2då-
Çsåi); on the other hand, in separate rhythm groups, we have: ('jø3Bi, 3måçd™3~i, 3kø-
'kø~i, 3©i't™, 3kM2dåÇsåi), with stress modifications on made ni˚ in comparison with
made]˚ /joùzi maùdeni kokoni kite kMdasaùi/ yoji made ni koko ni kite kudasai˘
-be- may give a similar prominence e‡ect. But, of course, it is inferior to that of the
syllable -ma-˚ and decidedly inferior to that of ta˚ and even less so than in the sylla-
ble -´i-˚ with devoicing up to the loss of syllabicity: (3¿i, 3¿:, 3¿).
In the case of loanwords, the interplay of pitch and stress (as well as of syllabic
weight and akusento placing) often undergoes some hierarchic reversal, by mov-
ing closer to the original (stress) pattern: (Çkøø3âii) /koohiùi/ koohii (çco‡eeÇ), (2hø-
çt™3mM) /hoùteRM/ hoteru (çhotelÇ).
Intonation
12.3.3.1. ‹en actual Japanese sentences are pronounced, the pitch of their
single rhythm groups is partially modified by intonation (too), which is added
with its characteristics, according to the four types of intoneme.
û 12.5 shows the Japanese preintonemes and intonemes, with their characteris-
tics. Every preintoneme forces the pitch of its individual tones, by canalizing them
into the indicated shapes (which, in more sophisticated notations than necessary,
could even be marked with small rings, as we will see shortly). A normal preinto-
neme is compressed (/ / ( ), ((” ))), the interrogative is raised (/¿ / (¿ ), ((» ))), where-
as the imperative one is falling (/¡ / (¡ ), ((»’ ))); finally, an emphatic preintoneme is
non-compressed, (/˚ / (˚ ), ((ˇ ))).
Again in û 12.5 it is possible to see the modifications of the intonemes: the con-
clusive is falling (/./ (13)), the interrogative is rising (/?/ (31)), the suspensive is extend-
ed (/÷/ (^)), whereas the continuative intoneme is compressed (/,/ (2)).
û 12.6 shows the modifications undergone by light and heavy syllables (on two
morae of di‡erent pitch {Ô û 12.4}), when the four Japanese intonemes superim-
û 12.5. Japanese preintonemes and intonemes.
pose on them. Mainly the last syllable of an intoneme undergoes these changes,
whether it be stressed (as in these examples) or unstressed. As can be seen, a conclu-
sive intoneme makes the last syllable lower and slightly falling; an interrogative
one makes it raise by giving it a clear rising movement. A suspensive intoneme
practically does not modify anything; whereas, a continuative one slightly com-
presses it towards the mid band.
Before moving to the illustrastive examples, let us observe that, in Japanese,
questions are made by adding the particle ka (kå) /ka/ at the end, and using an in-
terrogative intoneme with total questions, but a conclusive (or a continuative) one
with partial questions. ˛is is the most recommedable and common pattern. How-
ever, since ka is very recognizable, as an interrogative element, a simple conclusive
intoneme can be used, even with total questions, or an interrogative one even with
partial questions. Lastly, above all colloquially, ka need not necessarily be used, in
which case then an interrogative intoneme is necessary with total questions.
Lastly, here are three examples to illustrate the most recommendable use of
marked intonemes:
/./: (3µå&kå2mi2måÇs™óœd™3¿i3tå13) /µakaRimaseùô desita./ Wakarimasen desita.
/?/: (¿3µå&kå2mi'må3¿i3tå3kå31) /¿µakaRimaùsitaka?/ Wakarimasita ka?
/÷/: (3FM'jM çdåt3t:å 3k™çm™3døö^ 3i2©i'må3¿i3tå13) /hMjMù datta keRedo÷ ikimaùsita./ Fu-
yu datta keredo, ikimasita.
Typically, Japanese has a peculiar paraphonic setting, with lowered larynx §æ@,
mostly for men.
û 12.6. Pitch movements for light or heavy Japanese syllables, with the four intonemes.
Text
12.4.0. ˛e story †e North Wind and the Sun follows, given in four di‡erent
çnormalizedÇ versions. We start with the (neutral) Japanese pronunciation of (neu-
tral British) English – this is the first step of the phonetic method (the written text
is given in § 2.5.2.0). ˛e Japanese translation follows, in its neutral phonotonet-
ic and phonotonemic versions, since it is important in this language to see its aku-
sento and the actual phonotonetic rendering, to make useful reflections.
At the end, as always, there is the version which gives the English pronuncia-
tion of Japanese, by neutral British speakers, fluent in Japanese (after prolonged
12. japanese 393
contact with native speakers, but with no help from the phonetic method), who
have adequately learned the relative prominences, but who substantially use seg-
mental and intonation elements which are typical of neutral British English (for
reference purposes, although, of course, a neutral accent is not so common). Ob-
viously, the same principle is valid for the foreign pronunciation of English, giv-
en first.
Speakers of American English could prepare their own version both of the Japa-
nese pronunciation of English and of their pronunciation of Japanese, as an excel-
lent exercise, by listening to native speakers, best of all after recording them. Of
course, speakers of other languages could do the same thing. ˛e author would be
happy to receive their transcriptions and recordings, both in case of help –should
they need it– and to make their contribution known to others (possibly in our
website on canIPA Natural Phonetics – Ô § 0.12).
Japanese text
Phonotonemic transcription
Phonotonetic transcription
Vowels
13.1 Esperanto (¤ E-o] presents only the five most normal and natural vowel
phonemes (which are the most widespread in the various languages of the world):
(i, ™, a, ø, u) /i, e, a, o, u/ (Ô û 13.1). ˛e five E-o vowels are realized very much
like the five (stressed) vowels of Spanish, as in: ('iR, 'tR™s, 'maR, 'døs, 'tu) /'iR, 'tRes,
'maR, 'døs, 'tu/ ir˚ tres˚ mar˚ dos˚ tù. In the most international pronunciation of Es-
peranto, e˚ o always have the timbres (™, ø), both in stressed and unstressed sylla-
bles. Here are some examples: ('tRi˙ki, '™b-l™, 'ka;Ra, 'dø;mø, 'u;nu) /'trinki, 'eble, 'ka-
ra, 'domo, 'unu/ trinki˚ eble˚ kara˚ domo˚ unu çto drink, perhaps, dear, house, oneÇ.
˛e o‚cial Esperanto diphthongs are of the (éi, éu) /éi, éu/ type, and are repre-
sented with Vj˚ V¨, even though, of course, other vowel sequences occur, which
from a phonetic point of view are real diphthongs as well (Ô § 3.1.2-3, about Italian
diphthongs, and NPT/HPh: § 1.16 “ § 2.10-1), as, for instance, /eo, io/ in (&g™øgRa-
'fiø) /geogra'fio/ geografio çgeographyÇ. Given the nature of Esperanto orthogra-
phy, a more logical solution çVi, VuÇ was not available, because of its word-stress
assignment rule (as we will see in § 13.3.4).
/a/ (a)
13. esperanto 397
Consonants
velar rounded
protruded
labiodental
prepalatal
laryngeal
alveolar
bilabial
palatal
dental
velar
Nasals
13.2.1. Esperanto has two nasal phonemes, /m, n/ (m, n): ('mø;nø, 'f™ndi) /'mo-
no, 'fendi/ mono˚ fendi çmoney, to splitÇ. ‹ile /m/ never assimilates: (™m'fa;zø,
&m™mkøn's™Rvø) /em'fazo, memkon'servo/ emfazo˚ memkonservo çemphasis, self-
conservationÇ, there are various possible assimilations for /n/ (M, ~, ˙), which ren-
der the pronunciation more fluent and natural: (iM'f™;Ra, 'ma~Gi, si~'jø;Rø, 'SRa˙-
kø) /in'fera, 'manGi, sin'joro, 'Sranko/ infera˚ mangi˚ sinjoro˚ sranko çinfernal, to
eat, gentleman, cupboardÇ (including (n, ~) ((˙, n)): ('s™ndi, 'ma~Gi) /'sendi,
'manGi/ sendi˚ mangi çto send, to eatÇ).
However, between lexemes (including prefixes) no assimilation takes place (on
the contrary, we also find secondary stresses, even on syllables contiguous to a
stressed one): (&kun'm™;ti, &kun'v™;ni, &s™n'pa;ga) /kun'meti, kun'veni, sen'paga/ kun-
398 a handbook of pronunciation
Stops
Stopstrictives
13.2.3. In addition, there are three stopstrictives: /q÷ c, G/ (the last two are a
diphonic pair): (q) c (as in German: ('hInq) /'hInq/ Hintz, not as in English:
('h¤nts) /'hInts/ hints); then, there are (c) c and its voiced counterpart, (G) g (as in
English: ('f™c, 'Gø;Ù) /'fEc, 'GOE/ fetch˚ joy).
˛ey always maintain the values we have seen, before any vowel or consonant:
('q™nt, 'pa;qø) /'qent, 'paqo/ cent˚ paco çhundred, peaceÇ, (c™'va;lø, f™'li;ca) /ce'va-
lo, fe'lica/ cevalo˚ felica çhorse, happyÇ, (GaR'd™;nø, 'pa;Gø) /Gar'deno, 'paGo/ gar-
deno˚ pago çgarden, page (of book)Ç. ˛e spelling dz represents a (rare) sequence,
(dz), not the voiced counterpart of c (which would be (Q), as in Italian ('QE:Ro)
/'QEro/ zero çzeroÇ]\ ('™dzø) /'edzo/ edzo çhusbandÇ.
Constrictives
13.2.4. ˛ere are three diphonic pairs of constrictives and a rare voiceless velar
one (which should be withdrawn from modern and future Esperanto, Ô § 13.4.3-
13. esperanto 399
Approximants
13.2.5.2. ˛e other two approximants, /j, w/ (j, w), are represented by j and ¨˚
and correspond to English ('j™s, 'w¤n:) /'jEs, 'wIn/ yes˚ win\ ('j™s, 'vø;jø, si~'jø;Rø,
'w™stø) /'jes, 'vojo, sin'joro, 'westo/ jes˚ vojo˚ sinjoro˚ ¨esto çyes, road, gentleman,
westÇ.
However, when j and ¨ are not followed by any vowels in the same word, they
are realized as the corresponding vowels /i, u/ (i, u) (and it is to be avoided for
spelling to lead people to think that, in such positions, they have to pronounce
what are actually normal and real vowels as consonants): ('tui, 'Raitøi, 'nau, 'laudi)
/'tui, 'raitoi, 'nau, 'laudi/ tuj˚ rajtoj˚ na¨˚ la¨di çimmediately, rights, nine, praiseÇ.
Apart from this last case of greater phonetic naturalness, i and u always need to be
distinguished from j and ¨ before vowels: (mi'™;lø) /mi'elo/ mielo çhoneyÇ, ('mj™;lø)
/'mjelo/ mjelo çspinal cordÇ, (fi'askø, &mani'™;Rø) /fi'asko, mani'ero/ fiasko˚ maniero
çfiasco, mannerÇ.
Let us also consider the following cases, for which (as will be seen in § 13.2.5.2)
j is forced to have two di‡erent values owing to the stress-assignment rule: ('pl™idø)
/'pleido/ plejdo çplaidÇ, but (pl™'ja;do) /ple'jado/ Plejado çPleiadÇ.
In lexical derivation, when ¨ is followed by a vowel, we have two possibilities,
/w, u/: ('na;wa, 'naua) /'nawa, 'naua/) na¨a çninthÇ; the same goes for (bal'da;wa,
-aua) /bal'dawa, -aua/) balda¨a çpromptÇ, from ('baldau) /'baldau/) balda¨ çsoonÇ.
400 a handbook of pronunciation
However, in compounding, we only have ¨ /u/: (lau'i;gi) /lau'igi/) la¨igi çto adaptÇ.
All this must not be unduly modified by the di‡erent analogical extensions of the
various ethnic languages.
13.2.5.3. In addition, even for stressing, let us consider cases (that we present
here in advance) such as: ('j™s, 'jam) /'jes, 'jam/ jes˚ jam çyes, alreadyÇ and ('i™s, 'iam)
/'ies, 'iam/ ies˚ iam çsomeone's, onceÇ, and (sø'i;fi, 'søilø, ba'lau, 'baldau) /so'ifi, 'soi-
lo÷ ba'lau, 'baldau/ soifi˚ sojlo˚ balau!˚ balda¨ çto be thirsty, threshold, sweep!,
soonÇ. To write ¨ has always been a severe problem both for the press and typewrit-
ing (and today for computers, too, unless some special fonts have been installed).
˛e first o‚cial Esperanto publication by Zamenhof appeared in 1887. ˛e Inter-
national Phonetic Association was founded in 1886 and the first version of the In-
ternational Phonetic Alphabet [IPA] was produced in 1888. ˛ese coincidences ex-
plain why they were not aware of each other's existence. Today, however, the par-
allelism between (j, w) and j˚ w is quite evident and it spontaneously leads to the
substitution of ¨ with w˚ which would certainly be to the advantage of E-o itself.
It is neither weakness nor betrayal at all… It would just be common sense!
Trills
13.2.6. ˛ere is an alveolar tap, (R) (which, for emphasis, can oscillate with a
true trill (r), alveolar as well): ('Ra;nø, 'køRpø, 'tR™;) /'rano, 'korpo, 'tre/ rano˚ korpo˚
tre çfrog, body, very (much)Ç. Any other pronunciation of /r/ (although frequent,
because of the di‡erent national languages used by Esperanto speakers) is not neu-
tral.
Laterals
13.2.7. Lastly, we find a pure alveolar lateral, (l), with no particular nuances (so
it is di‡erent from the English {non-prevocalic} l sound as in ('w™ı:{Ω}) well(s)]\
('la;nø, mul't™;ga) /'lano, mul'tega/ lano˚ multega çwool, very veryÇ. Not even a
slight palatalization (ç(¬)Ç, perhaps before /i/) would be acceptable: ('li;gi) /'ligi/ li-
gi çto bindǢ ˛e only normal assimilation occurs before dentals, (l) ((l)), and be-
fore postalveo-palatal protruded consonants, within lexemes or in unstressed gram-
memes, (¬) ((L)): ('alta, 'fa¬ci) /'alta, 'falci/ alta˚ falci çhigh, mowÇ.
Structures
˛erefore, it must have strict rules, which are coherent and systematic, yet clear
and simple, without concessions to any language, or to groups of languages.
Most of all, any peculiarities of one's mother tongue should be avoided. In fact,
if this is not done, in the end all Esperantists will speak their own çdialectÇ of Es-
peranto. As a matter of fact, this is what happens when people use E-o without con-
sidering its phonic aspect. ˛us what they obtain is quite close to the linguistic
Babel which E-o seeks to solve.
In neutral Esperanto pronunciation stressed vocoids are realized as half-long in
(either word-internal or word-final) unchecked syllables, when they occur in prom-
inent positions, ¤ in intonemes, as usually at the end of sentences: ('sa;na, 'a;) /'sa-
na, 'a/ sana˚ A çhealthy, (the letter) AÇ. In checked stressed syllables (either in word-
-internal or word-final position), Esperanto vocoids are always short: ('paRtø) /'par-
to/ parto çpartÇ. Besides, Esperanto diphthongs are always short, as in German
(while in English they are long when not followed by voiceless consonants or by
unstressed syllables, as in ('w™;I{Ω}, 'w™;I∂÷ 'w™IT, 'w™IDå/-D≥) /'wEI{z}, 'wEId÷ 'wEIt,
'wEIdÈ≤/ way(s)˚ wade÷ wait˚ waider]\ ('™uRø) /'euro/ e¨ro çEuroÇ, ('ti™) /'tie/ tie
çthereÇ.
Consonant clusters
13.3.2. Some Esperantists strive to follow çto the letterÇ the indications that Za-
menhof (1962•, the inventor of E-o) used to give enthusiastic people in answer to
their questions and doubts. However, it is to be understood that, when he stressed
the point that çeach letter must be pronounced clearly separated from the neigh-
boring onesÇ, his sole aim was to make people avoid overly marked ethnic pronun-
ciations. Surely, he had no clear intention to lead them to reflect on the Esperanto
phonic structure, which is certainly neither based on avant-garde theories nor ex-
pressed in strict phonetic terms (least of all phonemic ones!).
Here are some examples of di‡erent current ethnic pronunciations (even by ex-
pert and fluent Esperantists): (™u'Rø;pø) E¨ropo *(°'˜∏pP, j¨È'>‘¨p‘¨)÷ ('laudi) la¨-
di *('l∏_i, 'lø;DIi, 'lAÈdi, 'laodi, 'ıOudi)÷ ('pa~-jø) panjo *('pa;No, 'phπni‘¨)÷ ('lø˙g™)
longe *('lØ˙, 'lO˙¢, 'lA˙™I)÷ ('vøRtø) vorto *('vO‰to, 'vø;T‘U, 'fø<to, 'v∏ÆtP)÷ ('sta;Ri) stari
*('Sta:ºi, 'sT™‘>i, 'stØORi)÷ (la't™mpø) la tempo *(ı√'tEm:b‘, l√'thEmp{h}P)÷ (in't™nsa)
intensa *(in'dEn:Qa, ì'tÕ:sa, in't™nså)÷ (la'pa;qø) la paco *(la'Faq:qP, la'baqqso, lÅ-
'pÅt†o)÷ (mi'Gø;jas) mi gojas *(miG'GOj:jasse, mi'Z∏:jasse)÷ (™b'l™;qø) ebleco *(eb-
'b]™q:qø, ji'b¬™;qå)÷ (la'ku;bø) la kubo *(la'guubbo, la'h¯:BP)÷ (li'b™;lo) libelo *(ıib-
'bE;™ı‘, li'B™;lo)÷ ('kRu;cø) kruco *('kru:So, k'Rju;Cx)÷ ('pa;Si) pasi *('paS:Si, 'paaßi)÷ (mi-
'p™;tas) mi petas *(mi'bE;Edas:, mi'phe:t{h}as)÷ ('j™s) yes *('á™ß, 'jÄEÀ).
Paying excessive attention to spelling can lead people to uselessly force them-
selves to always unnaturally realize the grapheme n as (n), in every possible con-
text, only because the way it is written seems to require such an articulation. On
the contrary, other –even more evident– di‡erences which elude control are un-
consciously produced, like those we have just seen.
However, the fact of writing n before all consonants except p˚ b certainly does
not mean that (n) has illogically to be maintained, against (mp, mb). It is quite
the contrary, even if it has been expressed in a rather rudimentary way. Indeed, all
languages having a homorganic /ö/ to a following consonant (¤ with (mp, Mf,
~c, Ná, ˙k) Â) show this fact by using the only other unitary grapheme which is
a phoneme as well: m /m/. Otherwise, they would even write np and nb, as is done
by serious orthographies of heterorganic dialects (like, in general, those of north-
ern Italy).
Ending with some Esperanto examples, we have: ('kømbi, køm'pR™;ni) /'kombi,
kom'preni/ kombi˚ kompreni çto comb, to understandÇ, (køM'v™;ni, køM'f™;si)
/kon'veni, kon'fesi/ konveni˚ konfesi çto be suitable, to confessÇ, (køn'du;ki, 'køn-
tRau) /kon'duki, 'kontrau/ konduki˚ kontra¨ çto drive, againstÇ, (kø~'c™Rtø) /kon-
'certo/ koncerto çconcertoÇ, (kø˙'gR™;sø, kø˙'ka;va) /kon'greso, kon'kava/ kongreso˚
konkava çcongress, concaveÇ.
But (&kunm™'t™b-la, &kun&pRøduk'ta;dø, &kun&bata'lantø) /kunme'tebla, kunpro-
duk'tado, kunbata'lanto/ kunmetebla˚ kunproduktado, kunbatalanto çcombinable,
coproduction, comrade in armsÇ; in fact, they belong to di‡erent lexemes, because
they are compounds.
13. esperanto 403
Syllabification
Word stress
Sentence stress
13.3.5. Let us now give some thought to the stressing of utterances. As it would
be absurd in ethnic languages to stress every single word appearing in a sentence,
such is the case even in E-o, whose grammemes (¤ grammatical words, lacking a
real semantic value) are completely unstressed, unless they are voluntarily empha-
sized for some particular reasons.
A sentence like Mi estas la amiko de via frato çI'm your brother's friendÇ would
certainly not be *('mi; 'ö™stas 'la; öa'mi;kø 'd™; 'via 'fRa;tø), which would –rightly–
make people hate E-o as something unbearable! (On the other hand, even for for-
eign languages, it is not rare to hear such things, but this depends on incomplete
learning, as well as on widespread ignorance of phonetics and its advantages.) A
more appropriate rendition of the example given would be: (mi&™stasl{a}a'mi;kø
&d™via'fRa;tø).
As can be seen from previous transcriptions, even in compounds, we find di‡er-
ent stress-degrees in their components, depending on their semantic relevance.
However, the main stress of compounds falls on the stem vowel of the last element.
Quite evidently, a slow and strained delivery, with too many stresses on gram-
memes as well, in addition to the inevitably high frequency of the (sole) auxiliary
verb esti˚ can before long become a real nightmare if estas˚ estis˚ estos… are always
stressed in a rather mechanical way.
˛erefore, accurate speakers carefully avoid falling into this nasty habit. In fact,
they will destress, even completely, the forms of esti˚ by introducing the stylistic
refinement of using instead the allomorphs with 'st-: (&mistas&ci'ti™, &listøsf™'li;ca)
/mistasci'tie, listosfe'lica/ Mi 'stas ci tie˚ Li 'stos felica çI'm hereÇ, çhe'll be happyÇ
(either only when pronouncing or when writing as well – by providing a consider-
able visual-mnemonic help). Even the phrase (tiø'™stas) /tio'estas/ tio estas çthat isÇ,
when it is not emphasized, surely gains by becoming (&tiøstas…) /'tiostas/ tio 'stas.
Obviously, it is quite the contrary for: (¿&cu'™sti2| ¿&au"n™; '™sti23) /¿cu'esti,| ¿au"ne 'es-
ti./ Cu esti, a¨ ne esti? çTo be, or not to be?Ç.
Unfortunately, this use is not generally followed, since the backward-looking
ideas that characterize most language teaching (starting from one's own language)
do not manage to clearly separate the graphic level (which is secondary) from the
phonic one (which is primary). All this leads people to believe that reduced forms
are a signal of corrupted and slovenly language. ˛is is the case of (&aÙSDÈÑ'ThÖ¨ı-
DÈm, -ø¨-), which is quite normal for I should have told them˚ while –on the con-
trary– something like *(5a;Ù S¨fl&hπv'ThÖ¨ıd&∑™m, -ø¨-) would be quite unusual in-
13. esperanto 405
13.3.6. Returning to our first example, we will have the following realization:
(&mistasla'mi;kø d™&via'fRa;tø), which is possible to write as: Mi 'stas l'amiko de via
frato. ˛e elision of la is o‚cial, although it is often erroneously limited to poet-
ry.
Likewise, personal pronouns, which frequently appear at the end of sentences,
do not need to be always and mechanically stressed. On the contrary, only for em-
phasis or contrast may they have a strong stress, otherwise they are destressed and
enclitic (although written as separated forms, as in English, but not in many oth-
er languages); other monosyllabic particles behave in the same way: (Si'vi;dis&lin) /Si-
'vidislin/ si vidis lin çshe saw himÇ, ('da˙kønal&vi) /'dankonalvi/ dankon al vi
çthank youÇ, ('tiø&ci) /'tioci/ tio ci çthisÇ, (¿'cu;&n™21) /¿'cune?/ cu ne? çisn't it?, aren't
you?, doesn't he?, haven't they?Ç…
It is worthwhile comparing the following cases: (Si'vi;dis 'lin) /Si'vidis 'lin/ si vidis
lin çshe saw himÇ, ('da˙køn al'vi;) /'dankon al'vi/ dankon al vi çthank you!Ç, (&tiø-
'ci;) /tio'ci/ tio ci çthis hereÇ, (¿&cu'n™;21) /¿cu'ne?/ cu ne? çisn't that so?Ç.
˛e case of ju (mal)pli…, des (mal)pli… is very interesting for stressing: ('ju; &pli-
vipa'Rø;las2 2 'd™s &malpli&mikøm'pR™;nas23) /'ju plivipa'rolas÷ 'des malplimikom'pre-
nas./ ju pli vi parolas, des malpli mi komprenas çthe more you speak, the less I un-
derstandÇ, ('ju; &pli'mult™2 2 'd™s &pli'bø;n™23) /'ju pli'multe÷ 'des pli'bone/ ju pli multe˚
des pli bone çthe more, the betterÇ.
Intonation
13.3.7. û 13.3 shows the preintonemes and intonemes of neutral E-o (free from
any influence from ethnic languages), so we will see only three fundamental ex-
amples:
/./: (&ili'øft™ køn'sultas &l{a}™sp™'RantaM &vøRt'a;Røn23) /ili'ofte kon'sultas l{a}espe-
û 13.3. Esperanto preintonemes and intonemes.
'rantan vort'aron./ Ili ofte konsultas la esperantan vortaron ç˛ey often consult the
Esperanto dictionaryÇ.
/?/: (¿&cu&ili'øft™ køn'sultas &l{a}™sp™'RantaM &vøRt'a;Røn21) /¿cuili'ofte kon'sultas
l{a}espe'rantan vort'aron./ Cu ili ofte konsultas la esperantan vortaron? çDo they of-
ten consult the Esperanto dictionary?Ç.
/÷/: (&ili'øft™ køn'sultas &l{a}™sp™'RantaM &vøRt'a;Røn2 2| &s™d'tu;t™ &s™n'fRukt™23) /ili'ofte
kon'sultas l{a}espe'rantan vort'aron÷| sed'tute sen'frukte./ Ili ofte konsultas la esper-
antan vortaron, sed tute senfrukte ç˛ey often consult the Esperanto dictionary,
but that's of no use.Ç
have already said (§ 13.0.1). Besides, before Unicode, computers needed special
fonts to be installed, or particular programs to create new characters or to modi-
fy existing ones in order to use them, and they are still necessary for the many
things which are not yet part of Unicode.
But even in this way, these six letters are still a problem for orthographic correc-
tion or alphabetical arrangement. Besides, few normal typewriters could combine
4, and only for small letters; to say nothing of T, for ¨, too often replaced by 5, or Z,
or Q, or 4: à, ú˚ ù˚ û (unless one studies hard to become… a magician).
˛e problem could be overcome by simply modifying not the language (!), nor
its phonemic system (!), but merely its spelling, although for many people this
seems to mean –instead– to çalterÇ the very essence of E-o itself (whereas this is
normal for natural languages)! ˛us some evolution and structural simplification
would be quite logical, as in natural languages. ˛e way has already been indicat-
ed by Zamenhof himself, although it should be followed with rigorous criteria.
On the other hand, we find an appealing and reassuring parallelism with the In-
ternational Phonetic Alphabet, which consists in the use (already suggested by var-
ious scholars) of w˚ x˚ instead of ¨˚ h˘ Even for c˚ g˚ j˚ s, Zamenhof himself was
aware of the anti-modernism inherent in these signs, and suggested that –if there
were typographical di‚culties (which is not merely hypothetical)– they could be
substituted with ch˚ gh˚ jh˚ sh˚ hh˚ u (for h˚ ¨].
As usual, it is hard to manage to separate the prevailing graphic level from the
underrated phonic level. But it is even more di‚cult to separate personal
emotionality from linguistic objectivity.
Esperanto £26∞
13.4.3. Certainly, the most modern solution –and most useful to E-o itself–
would be to move on to Esperanto ç26Ç, (&™sp™'Rantø &dud™k's™s), which uses only
the twenty-six letters of the Roman alphabet, by renouncing –with more gains
than losses– the two rarest and less useful sounds: (x, Z) /x, Z/ h˚ j. ˛us the phone-
mic inventory will be simplified, by removing the source of some main problems,
when we consider the di‡erent phonemic systems of the various languages of the
world: it is not always easy to distinguish (h, x÷ G, Z) (as few languages systemati-
cally do).
Obviously, it is not su‚cient to delete. It is necessary to merge the forms con-
taining these two signs (and sounds) with words having more similar sounds. ˛is
means: (x) = (k÷ h), (Z) = (G÷ j). It is already a common practice to e‡ect the for-
mer substitution: (x) = (k), as for kaoso˚ kemio˚ jakto çchaos, chemistry, yachtÇ; al-
so the change (Z) = (G) has already been recorded: jazo = gazo çjazzÇ. ˛is same ap-
plication needs only to be generalized. ‹en similar words containing (k, G) al-
ready exist, (h, j) will (have to) be used; otherwise, the shape of the new words
which would coincide will be slightly changed; or else they will have to be substi-
tuted, if necessary.
In the case of horo çchorusÇ, since we already have horo çhourÇ and koro çheartÇ,
the new form koruso has been added in order to finally replace it. For holero çchol-
eraÇ, since kolero çangerÇ is already there, the solutions can be: holero or kolro
(ç26Ç). In a series of oppositions like justa çjust, fairÇ and gusta çright, exactÇ, we
may insert juro çlawÇ, juro = guro çvowÇ, jako çjacketÇ, jako = gako çjackÇ as well.
As far as jeto çjetÇ, geto çjettyÇ, jeto çthrowÇ are concerned, if this last one becomes
('G™;tø) yeto (ç26Ç), the second example can safely become (G™'t™ø) yeteo (ç26Ç),
approaching the related form (mø'l™ø) moleo çmoleÇ.
A really concrete case is ajo çthingÇ (and the corresponding su‚x aj˚ with an
analogous meaning, which is very widely used) and ago çageÇ. Here, so as not to
risk overlapping with the exclamation aj! çow!, ouch!Ç, from which in E-o it is cer-
tainly possible to derive ajo ça cry of painÇ, and without thinking of changing a
whole series of words already formed with aj (and some more in the future), it
would be convenient –and unproblematic– to change ('a;Zø) into ('a;Gø) [ajo =
ago = ayo ç26Ç), and ('a;Gø) into ('aiGø) [ago = ajgo = ajyo ç26Ç) or probably, bet-
ter still, into ('™iGø) [ago = ejgo = ejyo ç26Ç).
any consideration about the çperversionÇ of the language, since they are not at all
natural, nor necessary).
Obviously, as seen above, ¨ would become w (w), ('na;wa, 'naua) [na¨a = nawa
ç26Ç), çninthÇ; however g would become y (G), without causing much ado, given
that even in Spanish, in many contexts and variants, for y we find a very similar
pronunciation, ('pa;Gø) [pago = payo ç26Ç) çpage (of book)Ç. In addition, c would
become q (c), and here again we find similar pronunciations in Chinese and Al-
banian, (c™'va;lø) [cevalo = qevalo ç26Ç) çhorseÇ. (On the other hand, does Espe-
ranto not want to be the second language for çallÇ peoples?) ≈nally, x (having no
further need to indicate (x) or to function as a diacritic) would be used for (S),
which has a similar sound in Portuguese, Catalan, Basque, Chinese, Maltese, and
Sicilian (besides in Old Spanish), ('fi;Sø) [fiso = fixo ç26Ç) çfishÇ.
It would be a matter of becoming accustomed to the value of the new letters;
after all only for Slavists is c for (q) actually familiar, ('q™;lø) celo çaimÇ. For in-
stance, is not h a real çnothingÇ for most Romance-language speakers: ('ha;vi) havi
çto haveÇ (and consequently a real problem)?
˛ose who defend the preservation of (Z, x), because they are useful to render
some personal and place names of certain languages (with similar but not identi-
cal sounds), should also worry about all the other sounds that E-o does not man-
age to render (which are many more)… ˛e exchange of j and y (with the value of
j /G/ and y /j/), although more in tune with a widespread English-like graphemic
tradition, would push E-o away from both IPA usage and from the orthography of
many languages.
Text
13.5.0. ˛e story †e North Wind and the Sun follows, given in a number of
çnormalizedÇ accents, mostly as useful introductory examples of the phonetic meth-
od of natural phonetics. As always, first there is the version which gives the typical
Esperanto pronunciation of English. ˛en follows the actual Esperanto text, show-
ing neutral E-o pronunciation.
Esperanto text
13.5.2. Iam, la norda vento kaj la suno disputis, car ciu el la du asertis esti pli for-
ta ol la alia. En tiu momento, ili ekvidis vojaganton, kiu anta¨eniris volvite en sia
mantelo. La du disputantoj, do, decidis, ke konsideratos pli forta tiu el la du kiu suk-
cesos igi la vojaganton forpreni sian mantelon.
La norda vento ekblovis tre forte; sed, ju pli gi blovis, des pli la vojaganto mallozigis
sian mantelon, kaj, finfine, la kompatinda vento devis rezigni. Tiam, la suno montri-
gis en la cielo, kaj, mallonge poste, la vojaganto, al kiu estis varme, forprenis la mante-
lon. Tiel, la norda vento devis agnoski, ke la suno estas pli forta el la du.
Cu vi satis la historion? Cu ni ripetu gin?
Esperanto pronunciation
13.5.3. ('iam2 la'nøRda 'v™ntø2 &kaila'su;nø2| dis'pu;tis23 'caR2 ìciu &™lla'du;2 a's™Rtis23
&™sti&pli'føRta2 &øll[a]a'lia23œ| ™n&tiumø'm™ntø2 2| &ili&™k'vi;dis23 &vøja'Gantøn23 &kiuan&taw™n-
'i;Ris23 vøl'vit™ ™n&siaman't™;lø23|| la'du &dispu'tantøi2 ì'dø;2œ d™'qi;dis23 &k™køn&sid™'Ra;tøs2
&pli'føRta2 2|| &tiu™lla'du;2| &kiusuk'q™;søs2 &igila&vøja'Gantøn2 2| &føR'pR™ni &siamman't™;-
løn23||
la'nøRda 'v™ntø2 2| &™k'bløvis &tR™'føRt™23| 's™d2 'ju; &pliGi'blø;vis2 2| 'd™s2 &plila&vøja'Gan-
tø2 &mallø'zigis &siamman't™;løn23| &kaifiM'fi;n™2| la&kømpa'tinda 'v™ntø2| &d™visR™'zig-
ni23|| 'tiam2| la'sunø møn'tRi;Gis2 &™nlaci'™;lø23| &kaimal'lø˙g™ 'pøst™2 2| la&vøja'Gantø2 ìal-
&kiu[™]stis'vaRm™2 2œ ˚&føR'pR™;nis23 ˚&laman't™;løn23|| 'ti™l2| la'nøRda 'v™ntø2 ˚&d™vis-ag'nøs-
ki23 &k™la'su;nø2 ˚&™stas&pli'føRta23 ì&™lla'du;23œ||
¿&cuvi'Sa;tis21 ¿la&histø'Riøn2| ¿&cuniRi'p™;tu&Gin21|||)
British pronunciation
American pronunciation
Italian pronunciation
13.5.7. ('i;am2 la'nOrda 'vEn:to2 &kaila'su:nø2| dis'pu:tis23 'ca:r2 ì'ciu ella'du2 a'sEr:tis23
&™stipli[f]'fOr:ta2 &øllaa'li;a23|œ en&tiumo5men:tø12| &iliek'vi:diz23 &voja'Gan:to˙23 kiu&antawe-
'ni:Riz23 vol'vi:te en&siaman'tE:lo23|| la'du[d] dispu'tantoi2 ì'dO[d]2œ deq'qi:dis23 &kekkon&si-
de'ra:tos2 pli[f]5fOr:ta12|| &tiuella'du2| &kiusuk'qE:sos2 'i;gi la&voja5Gan:ton12| foR'prE;ni
&siamman'tE:lon23||
la'nOrda 5vEn:to12| ek'blO;vis tR™'fOr:te23| 'sEd:2 'ju[p] pli[G]Gi5blO:vis12| 'dEs:2 pli[l]la&vo-
412 a handbook of pronunciation
French pronunciation
13.5.8. ('iÅm2 lÅ'n∏˜dÅ 'vìntP2 &©ÅilÅ's¯nP2| _isçp¯+is3 3 [T]'SÅ:˜2 ì[T]'Si¯ ™llÅ'd¯2 ÅçsEÆ-
+is3 3 &™s+ipúi'f∏ÆtÅ2 &ÖllÅ;çliÅ3 3|œ 3n&+i¯mÖ5mìntP1 1| &ili™©çvi_is3 3 &vÖ,ÅçZÕntÚn3 3 &©i¯ÕntÅ-
°eçni˜is3 3 vÖl'vite 3n&siÅmÕnçtElP3 3|| lÅ'd¯ &_isp¯'tÕntPi2 ‘'dP2’ d™[t]çsi_is3 3 &©ekÚn&side-
'˜ÅtPs2 &púi5f∏ÆtÅ1 1|| &+i¯™llÅ'd¯2| &©i¯s¯k'sEsPs2 &iáilÅvÖ,Å5ZÕntÚn1 1| fÖÆ'püeni &siÕmÕnçtE-
lÚn3 3||
lÅ'n∏˜dÅ 5vìntP1 1| ™k'bl∏vis tü™çf∏Æte3 3| 'sEd2 ',¯ &púi[D]Zi5bl∏vis1 1| 'dEs2 &púilÅ&vÖ,Å'ZÕn-
tP2 &mÅllO'ziáis &siÕmÕnçtElÚn3 3| &©Åifin'fine2| lÅ&kÚ≈pÅ+indÅ 'vìntP2| &devisºeçzigni3 3||
'+iÅm2| lÅ's¯nP mÚn'tüi[D]Zis2 &3nlÅ[T]çS¿elP3 3| &©ÅimÅl'lÚNáe 5p∏ste1 1| lÅ&vÖ,ÅçZÕntP2 ‘Åú-
&©i¯'Es+is 5vŘme1 1’ ˚fÖÆçpüenis3 3 ˚&lÅmÕnçtElÚn3 3|| '+i™l2| lÅ'n∏˜dÅ 'vìntP2 ˚&devi-sÅgçn∏s-
©i3 3 &©elÅ's¯nP2 ˚&™stÅs&púiçf∏ÆtÅ3 3 ì&™llÅ'd¯3 3œ||
¿&[T]S¯vi•SÅ+is1 1 ¿lÅ&istÖ'ºiÚn2| ¿&[T]S¯&ni˜i•pet¯1 1 ¿'[D]Zin2|||)
German pronunciation
Spanish pronunciation
13.5.10. ('ian2 la'nøRƒa 'B™nto2 &kaila'ßu;no2| diß'pu;tiß3 3 'CaR2 ì'Ciu e{l}la'ƒu2 a'ß™Rtiß3 3
&™ßti&pli'føRta2 &ø{l}la'lia3 3œ| en&tiumo'm™nto1 1| &ilieŸ'Bi;ƒifi3 3 &Bo,a',anton3 3 &kiuantawe'ni;-
13. esperanto 413
Brazilian pronunciation
13.5.11. ('iAn2 ]AçnO˜då 'v™ntu2 &kai]å'su;nu2| ‚isçpu;Cis3 3 'ëa˜2 ìçëiu e{∞}]å'du2 açsE˜-
Cis3 3 &™sCi&p]i'fO˜tå2 &o{∞}]åç]iå3 3œ| e~&CiumÚçm™nt¨2 2| &i]ie©içvi;‚is3 3 &voãaçòAntÚn3 3
&©iuAn&tajeçni;Ris3 3 vo∞çviCi e«&siAmAnçte;]Ú«3 3|| ]åçdu &‚ispu'tAntji2 ì'dO2œ ‚içsi;‚is3 3
&©ekÚ«&side'Ra;tus2 &p]içfO˜tå2 2|| &Ciue{∞}]å'du2| &©iusu©i'sE;sus2 'iái ]a&voãaçòAntÚ«3 3| fo˜-
çpR™;ni &siAmAnçte;]Ú«3 3||
]AçnO˜då çv™nt¨2 2| &e©içb]Ovis &tR™çfO˜Ci3 3| 'sE‚i2 'ãu &p]iòiçb]O;vis2 2| 'dEs2 &p]i]a&voãa-
çòAntu2 &mA{∞}]oçziái{s} &siAmAnçte;]Ú«3 3| &kaifi«'fi;ni2| ]å&kÚmpaçCindå 'v™nt¨2| &de-
vi{s}˜eçzigni3 3|| 'CiA«2| ]åçsunu mÚn'tRi;òis2 &en]åçëje;]u3 3| &kaimA{∞}ç]9Nái çpOsCi2 2| ]å-
&voãaçòAntu2 ìA∞&©iuçEsCiz çva˜mi2 2œ ˚fo˜çpR™;nis3 3 ˚&]AmAnçte;]Ú«3 3|| 'Ci™∞2| ]AçnO˜då
'v™ntu2 ˚&d™vizaáiçnOs©i3 3 &©e]å'su;nu2 ˚&™stås&p]içfO˜tå3 3 ì&e{∞}]åçdu3 3œ||
¿&ëuvi¶ëa;Cis12 ¿&]åisto'RiÚ«2| ¿ëu&ni˜i¶pe;tu12 ¿'òi«2|||)
Lusitanian pronunciation
13.5.12. ('i4n2 ]‘'nORƒ‘ 'v™nt[u]2 &kai]‘'su;nu2| d¢SÇpu;t?S3 3 'SaÍ2 ì'Siu e{ı}]‘'ƒu2 ‘ÇsEÍtNs3 3
&™StN&p]i'fOÍta2 &o{ı}]aÇ]i‘3 3œ| en&tiumu'm™ntu2 2| &i]i¢kÇvi;ƒ˙S3 3 &voã‘ÇZAntÚn3 3 &kiuAnt‘j¢Çni;-
RÛS3 3 voı'vit¢ e”&si4m4nÇte;]Ú«3 3|| ]‘'ƒu &ƒÛSpu'tAntji2 ì'ƒO2œ ƒ¢Çsi;ƒÛS3 3 kNkÚ”&siƒ¢'Ra;tuS2 &p]i-
'fOÍt‘2 2|| &tiue{ı}]‘'ƒu2| &kiusuk'sE;sus2 'iŸi ]‘&voã‘ÇZ4ntÚ«3 3| fuÍ'pRE;ni &si4m4nÇte;]Ú«3 3||
]‘'nORƒ‘ 'v™ntu2 2| ek'b]OvÛS &tR™ÇfOÍt˙3 3| 'sEƒ2 'ãu &p]iZÛ'b]O;vÛS2 2| 'dES2 &p]i]‘&voã‘ÇZAntu2
&mA{ı}]u'ziŸÛ{S} &si4m4nÇte;]Ú«3 3| &kaifi«'fi;n[˙]2| ]‘&kÚ≈p‘'tind‘ 'v™nt[u]2| &devÛ{Z}º¢ÇziŸ-
ni3 3|| 'ti4«| ]‘'sunu mÚn'tRi;ZÛS2 &en]‘ÇSje;]u3 3| &kaimA{ı}']9˙g¢ 'pOSt[˙]2 2| ]‘&voã‘ÇZAnt[u]2
ìAı&kiu'EStÛZ ÇvaRm¢2 2œ ˚fuÍÇpRE;nÛS3 3 ˚&]‘m4nÇte;]Ú«3 3|| 'ti™ı2| ]‘'nORƒ‘ 'v™nt[u]2 ˚&ƒ™viz‘Ÿ-
ÇnOSki3 3 &ki]‘'su;nu2 ˚&™St‘S&p]i ÇfOÍt‘3 3 ì&e{ı}]‘ǃu3 3œ||
¿&Suvi'Sa;tÛS21 ¿&]aiStu'RiÚ«2| ¿Su&niºi'pe;tu21 ¿'Zi«2|||)
414 a handbook of pronunciation
Russian pronunciation
Arabic pronunciation
Hindi pronunciation
Chinese pronunciation
Japanese pronunciation
Utilizable bibliography
Prelude (é 1)
o‚cial IPA, in transcriptions which are always prosodically deficient, whereas in two pages
{arranged in three: 558-60} it tries to give çaccurateÇ descriptions, which only reveal the
method's heaviness and complexity, with results that cannot be suggested and çlanguagesÇ
which are unrecognizable even to native speakers, when one tries to reproduce them as rep-
resented; oƒIPA).
malmberg, b. (1974) Manuel de phonétique générale˘ Paris: Picard (IPA).
o'connor, j. d. (1973) Phonetics˘ Harmondsworth: Penguin (IPA).
schubiger, m. (1977) Einführung in die Phonetik˚ Berlin: De Gruyter (IPA).
smalley, w. a. (1964”) Manual of Articulatory Phonetics˘ Terrytown (~¥): Practical
Anthropology (with 33 non-enclosed {18 cm, 19 cm/s} reels, lasting 32 hours; non-IPA).
thomas, j. m. c. “ bouquiaux, l. “ cloarec-heiss, f. (1976) Initiation à la phonétique˘ Paris:
puf (completed by the vinyl record of bouquiaux “ cloarec-heiss “ thomas; expanded
IPA).
trubeckoj, n. s. (1939) Grundzüge der Phonologie˘ Göttingen: Vandenhoeck “ Ruprecht –
(1969) Principles of Phonology˘ Berkeley: Univ. of California Press (non-IPA and sometimes
mixed, so that some symbols can have {very} di‡erent values).
westermann, d. “ ward, i. c. (1990 {1930/31»}) Practical Phonetics for Students of African
Languages˘ London/New York: Kegan Paul International in association with the Interna-
tional African Institute (IPA).
English (é 2)
American English Pronunciation Program (1999) New York: Living Language/Random House
(graphemic booklet, but with 6 audiocassettes).
Australian Learners Dictionary (1997) Sydney: National Centre for English Language Teaching
and Research (gives the distributions of Australian pronunciation; IPA).
BBC Pronouncing Dictionary of British Names (1983”, 1971») Oxford: O. Univ. Press (British
pronunciation; also IPA).
canepari, l. (in progr.) English Pronunciation*. München: Lincom Europa (with 2 neutral and
2 mediatic pronunciations {¤ American and British}, in addition to the international one
and 5 further neutral pronunciations, together with several native regional accents from all
over the world and several broad foreign accents, too; canIPA]˘
dauer, r. m. (1993) Accurate English˘ Englewoods Cli‡s: Regents/Prentice Hall (American pro-
nunciation; with 4 audiocassettes; almost IPA).
Gage Canadian Dictionary (2000) Gage Educational Publishing Company: Vancouver (gives
Canadian pronunciation; IPA).
gimson, a. c. “ cruttenden, a. (2001•) An Introduction to the Pronunciation of English˘
London: Arnold (now: çcruttenden, a. <Gimson's Pronunciation of English≤Ç (!); still the
most recommendable for British pronunciation, although it has not reached the declared
–and hoped for– updating of symbols and conceptions, and in spite of some electronic lit-
tle bungles in the latest two editions; IPA).
gimson, a. c. “ ramsaran, s.m. (1982) An English Pronunciation Companion˘ Oxford: O.
Univ. Press (British English; with an audiocassette; IPA).
jones, d. (1960Ô, 1918») An Outline of English Phonetics˘ Cambridge: He‡er (British pronun-
ciation; meanwhile, pronunciation has changed and symbols have been improved, but it is
still worthwhile reading, rather than many more recent books; IPA).
jones, d. (2003»•, 1917») English Pronouncing Dictionary˘ Cambridge: C. Univ. Press (as the
previous edition, the present one, edited by P. Roach “ J. Hartman “ J. Setter, in addition
to British pronunciation, gives the American one; but it has lost much of the original spir-
utilizable bibliography 415
it, by standardizing the transcriptions {Ô Jones “ Gimson “ Ramsaran}; there is also a ver-
sion with a ©∂, which allows one to listen to the pronunciations and to search for homo-
phones by typing in a transcription; IPA). (˛e former version of the Cambridge Advanced
Learner's Dictionary˚ for American pronunciation {assigned to Hartman and to a çcleverÇ
editorial sta‡} had monstrosities like blaster with /sÿ/, that is ç/sù/Ç!)
jones, d. “ gimson, a. c. “ ramsaran, s. (1988»[) English Pronouncing Dictionary˘ London:
Dent (British; now pronunciation has changed a bit, but it is still worthwhile consulting
evenly, to have a taste of all the nuances indicated and to be able to enter the spirit of the
language; IPA).
kenyon, j. s. (1950»Ò, 1924») American Pronunciation˘ George Wahr: Ann Arbor (American
English; meanwhile, pronunciation has changed, but it is still worthwhile seeing; there is
also an çaugmentedÇ edition by others, but with no real additions, 1994»”; IPA).
– “ knott, t. a. (1953) A Pronouncing Dictionary of American English˘ Springfield, µåßß.:
Merriam (American pronunciation; meanwhile, certain pronunciations have changed, but
it is still worthwhile consulting; IPA {while, incredibly –in the 3rd millennium– mostly in
America, non-IPA dictionaries are still being published!}).
Macmillan English Dictionary for Advanced Learners (2002) Oxford: Macmillan Education
(strangely, this dictionary çcreated in Britain and the usaÇ only gives British pronunciation
and with only tiny detail; however, it is useful for the –British– stressing of lexical colloca-
tions; IPA).
Macquarie Dictionary, †e (1997’) Sydney: ˛e Macquarie Library (encyclopedic, gives
Australian pronunciation; IPA).
Merriam-Webster's Collegiate Dictionary (2003»») Springfield, Mass.: Merriam-Webster
(American pronunciation; interesting for its frequent phonetic variants, although it shows
absurd çsecondary stressesÇ, since they are marked for almost every non-attenuated V÷ unfor-
tunately non-IPA).
Oxford Advanced Learner's Dictionary of Current English (2000•) Oxford: O. Univ. Press
(British pronunciation with only major American di‡erences; it shows cases of çmarkedÇ
stresses for certain lexical collocations; IPA).
Oxford Advanced Learner's Dictionary of Current English – Encyclopedic Edition (1992) Oxford:
O. Univ. Press (British pronunciation with only major American di‡erences; it shows cas-
es of çmarkedÇ stresses for certain lexical collocations; IPA).
quaggiato, m. e. (1997/98) Fonetica inglese neutra e del Sud. Univ. of Venice, graduation the-
sis guided by L. Canepari (canIPA]˘
ragazzini, g. (1995’) Dizionario inglese {–italiano e italiano–inglese}. Bologna: Zanichelli
(British pronunciation with only major American di‡erences; its transcriptions have been
revised by L. Canepari and A. Venturi but çreducedÇ by the editorial sta‡, however it pres-
ents the innovation of /E/ instead of /e/ and very many çreduced formsÇ, more than in pro-
nunciation dictionaries, although less than in this HPr÷ IPA).
Random House Dictionary of the English Language (†e)˚ Unabridged (1987”) New York:
Random House (American; reliable for secondary stresses; but non-IPA).
Random House Webster's Collegiate Dictionary (1997”) New York: Random House (American;
reliable for secondary stresses; but non-IPA).
roach, p. (2000’) English Phonetics and Phonology. Cambridge: C. Univ. Press (British pro-
nunciation; with 2 audiocassettes; IPA).
santipolo, m. (1997/98) A Socio-phonetic Description of Some Varieties of South-eastern British
English. Univ. of Venice, graduation thesis guided by L. Canepari (canIPA]˘
turrin, n. (1996/97) †e Accents of Northern England. Univ. of Venice, graduation thesis guid-
ed by L. Canepari (canIPA]˘
upton, c. “ kretzschmar, w. a. jr. “ konopka, r. (2001) †e Oxford Dictionary of Pronunci-
416 a handbook of pronunciation
ation for Current English˘ Oxford: O. Univ. Press (although it is the latest to be projected,
it lags behind its predecessors. In addition, it is exactly the opposite to the very convenient
diaphonemic and interphonemic transcriptions – as a matter of fact, besides uselessly repeat-
ing also çidenticalÇ forms, in addition to those with slightly di‡erent phonetic renderings,
for every variant given it repeats the whole transcription {and not continuing the line, but
starting a new paragraph}, instead of indicating –more clearly– the sole di‡erences {failing
then to give the fundamental immediacy}. Besides, it uses di‡erent criteria and symbols for
the two accents {¤ British and American}, leading the reader to think there are di‡erences
even where actually there are none, as for instance also for secondary-stress markings after
primary stresses, which are indicated in the American but not in the British pronunciation,
for forms which are instead absolutely identical such as teacake˘ ˛erefore, considering the
great amount of blank space and its many useless transcriptions, it uses twice the number
of pages actually needed: for instance, for the article a˚ it uses 10 lines {ten!} to give –in sub-
stance– ça /'EI, È/Ç; besides, it mixes up neutral pronunciations and others which are not (yet)
neutral, for ex. dune with British variant identical to June, ç/dZu:n/Ç, or latter shown, in the
American pronunciation, exactly like ladder, ç/'lπdÈr/Ç, not even as a variant; çmixedÇ IPA]˘
Webster's New Encyclopedic Dictionary (1993) New York: Black Dog “ Leventhal (American
pronunciation; it marks destressable monosyllables; non-IPA).
wells, j. c. (1982) Accents of English˘ Cambridge: CUP (3 vols.; extensive survey however
mainly based on available written sources; IPA).
wells, j. c. (2000”) Longman Pronunciation Dictionary˘ Harlow: Longman (British “ Ameri-
can pronunciations; together with çJones»•Ç it provides a reliable survey, especially for Brit-
ish English; for American English, /Ø:/ has been removed while it was present in the first e-
dition {1990}, although it could be more useful than /πr/, which is still there, whereas /o:r/
has become /oUr/, which is more çeconomicalÇ, but far less appropriate; IPA).
Italian (é 3)
bonazzi, i. (1989) Dizione e qualità espressiva della voce˘ Torino: Centro Maier (with 6 very
short audiocassettes; traditional pronunciation; hypodiacritical spelling).
camilli, a. “ fiorelli, p. (1965) Pronuncia e grafia dell'italiano. ≈renze: Sansoni (simplified
ì).
canepari, l. (1970) Ùtæljæn præn≠nsÙeÙxn˚ in çLe Maître PhonétiqueÇ, 133:6-8 (as all MPh con-
tributions, fully transcribed in IPA].
– (1986’) Italiano standard e pronunce regionali˘ Padova: ©¬™¨π (with 2 enclosed audiocassettes,
the second one is about regional pronunciation; almost canIPA]˘
– (2000 {amended “ modified reprint} – 1999») Dizionario di pronuncia italiana. Bologna:
Zanichelli (60,000 forms with transcription and pronunciation variants, which correspond
at least to 180,000 actual words; pronunciations given: modern neutral, traditional neutral,
acceptable, tolerated, slovenly, intentional and lofty; canIPA]˘
– (2004 {amended “ modified reprint} – 1999”) Manuale di pronuncia italiana˘ Bologna: Za-
nichelli (with 2 enclosed audiocassettes; it introduces modern neutral pronunciation, in ad-
dition to the traditional one, besides other types, including 22 regional koinés; canIPA]˘
costamagna, l. (1996) Pronunciare l'italiano˘ Manuale di pronuncia italiana per stranieri. Pe-
rugia: Guerra (with 4 audiocassettes and a ©∂; canIPA]˘
fiorelli, p. (1965) Córso di pronùnzia italiana˘ Padova: Radar (with 14 vinyl records; tradition-
al pronunciation; hypodiacritical spelling]˘
migliorini, b. “ tagliavini, c. “ fiorelli, p. (1981”) Dizionario d'Ortografia e di Pronunzia.
Torino: ™®¤ (still with I and J mixed together; it also contains family and place names; 1st
ed. 1969 with a vinyl record; non-IPA).
utilizable bibliography 417
muljaCiÇ, Z. (1972) Fonologia della lingua italiana˘ Bologna: Il Mulino (traditional pronuncia-
tion; IPA]˘
tagliavini, c. (1965) La corretta pronuncia italiana˘ Bologna: Capitol (with 26 vinyl records;
traditional pronunciation; hypodiacritical spelling]˘
Unfortunately, in the Journal of the International Phonetic Association (2004, 117-21), an im-
plausible çdescriptionÇ of Italian was published (showing incredibly mixed and contrasting
features for a native speaker, independently from some queer transcriptions), which simul-
taneously presents both northern, central “ southern Italian characteristics and even a num-
ber of foreign ones. ˛e text wording itself of †e North Wind and the Sun is a çnewÇ tran-
slation from English, for instance with il vento del nord çthe North WindÇ instead of la tra-
montana. An answer was given that four Italian çscholarsÇ had approved it for publication.
˛e only recording used for the transcription is partially better (as one can immediately
hear), being somehow less çpossessedÇ than the çtranscriptionÇ provided in the paper.
French (é 4)
armstrong, l. e. (1932) †e Phonetics of French˘ London: Bell (reprinted many times; in spite
of its years, although pronunciation has slightly changed, it still remains good; IPA).
boch, r. (1995’) Dizionario francese {–italiano e italiano–francese}. Bologna: Zanichelli (its
transcriptions have been revised by L. Canepari but çreducedÇ by the editorial sta‡, how-
ever it presents the innovation of çneutralÇ /Õ, Ú/ instead of the old and outdated /˙, Ó/; IPA]˘
canepari, l. (in prog.) French Pronunciation* (with 4 national neutral pronunciations, besides
a mediatic and an international one, together with several native regional accents from
Europe and Canada, and some broad foreign accents, too; canIPA˘ Perhaps also as Les pronon-
ciations du français).
capelle, g. (1968) Le rythme et l'intonation de la phrase française˘ Paris: Didier (a vinyl record
and booklet, with no notation).
carton, f. (1979”) Introduction à la phonétique du français˘ Paris: Bordas (almost IPA).
coveney, a. (2001) †e Sounds of Contemporary French˘ Exeter: Elm Bank Publications (not
too reliable for things di‡erent from French, reported from controversial written sources;
IPA).
gleissner, h. (1964) Französische Aussprache˘ Leipzig: Verlag Enzyklopädie (a vinyl record and
booklet; IPA]˘
kaneman-pougatch, m. “ pedoya-guimbretière, e. (1989) Plaisir des sons˘ Paris: Alliance
Française/Hatier/Didier (with 4 audiocassettes; with no explicit intonation; IPA]˘
lerond, a. (1980) Dictionnaire de la prononciation˘ Paris: Larousse (inconveniently separates
common and proper names into two sections; IPA with some additional symbols).
malmberg, b. (1976[) Phonétique française˘ Malmö: Läromedel (IPA]˘
tranel, b. (1987) †e Sounds of French˘ Cambridge: C. Univ. Press (with an audiocassette;
IPA]˘
walker, d. c. (1984) †e Pronunciation of Canadian French˘ Ottawa: Univ. of O. Press; with-
out intonation; almost IPA]˘
warnant, l. (1987’, 1962/66») Dictionnaire de la prononciation française˘ Paris/Gembloux:
Duculot (previous editions had an enclosed vinyl record and a less çblank-pageÇ format; in-
conveniently separates common and proper names into two sections; IPA]˘
– (1996) Orthographe et prononciation en français˘ Gembloux: Duculot (derived from the pre-
ceding title, gives the 12,000 words which are not pronounced according to their orthogra-
phy; but does not give any proper names, which are often more problematical; IPA]˘
418 a handbook of pronunciation
German (é 5)
Spanish (é 6)
Unfortunately, no Spanish pronouncing dictionary exists yet! Even Antonio Quilis' meritorious
project of preparing one, with two neutral accents: Iberian and American (also with a ©∂ version,
with the possibility of recording one's voice and comparing it with the model provided) could not
be fulfilled…
canepari, l. (in prog.) Spanish Pronunciation*. München: Lincom Europa (with 6 or 7 na-
tional neutral pronunciations and an international one, together with several native regional
accents from Europe and America and some broad foreign accents, too; canIPA˘ Perhaps al-
so as Las pronunciaciones del español).
dalbor, j. b. (1980”) Spanish Pronunciation˘ Fort Worth: Holt, Rinehart and ©nston (Amer-
ican with Iberian indications; çnumericalÇ intonation with pitch levels; non-IPA]˘
Gran Diccionario de la lengua española (1985) Madrid: ߟ™¬ (Iberian; the only dictionary with
generally reliable transcriptions for Iberian Spanish, in spite of many attempts to make them
not to appear so; almost IPA]˘
miotti, r. (1995-96) Descripción fono-tonética de las variedades regionales del español de América
y España˘ Univ. of Venice, graduation thesis guided by L. Canepari (Iberian and American,
utilizable bibliography 419
Portuguese (é 7)
canepari, l. (in prog.) Portuguese Pronunciation*. München: Lincom Europa (with {Brazilian
and Lusitanian} neutral pronunciations and an international one, together with several na-
tive regional accents from Europe, America and Africa and some broad foreign accents, too;
canIPA˘ Perhaps also as As pronùncias do português).
cristófaro silva, t. (2001”) Fonética e fonologia do português˘ São Paulo: Contexto (Brazilian;
with a ©∂; IPA]˘
cuesta, p. vázquez “ da luz, m. a. mendes (1971’) Gramática portuguesa˘ Madrid: Gredos
(Lusitanian with Brazilian indications; non-IPA]˘
Dicionário da lìngua portuguesa (1981) Porto: Porto Editora (Lusitanian; does not transcribe,
but indicates with italic letters in brackets occurrences of /'e, 'o/ [ê˚ ô]˚ cases of unstressed
ç/E, a, O/Ç [è˚ à, ò] –which is one of the uses of our /È, A, ì/– and consonant clusters˚ includ-
ing x).
Dicionário da lìngua portuguesa contemporânea (2001), Lisboa: Editorial Verbo (Lusitanian;
too much proudly presented as the publishing enterprise çof the beginning of the 21st cen-
tury and of the 3rd millenniumÇ, which began before the French Revolution and finally led
to this two-volume dictionary; it does indicate metaphony, however, giving only plurals
with di‡erent vowel timbres, and hiding them at the end of the entry, not showing them
immediately after the headword, together with the other transcriptions, which include fe-
minine forms; in addition, it puzzles a lot because –in the 3rd millennium– it displays such
things as ç/õ, g, $, R, ], l/Ç, instead of /…, g, S, K, ı, L/, and even ç/Ω, ó, ö, Z/Ç instead of /E, å,
O, Z/; semi-IPA).
gärtner, e. “ thomaz jayme, j. (1984) Portugiesische Aussprache˘ Leipzig: Verlag Enzyklopädie
(Brazilian; a vinyl record and booklet; almost IPA]˘
giangola, j. p. (2001) †e Pronunciation of Brazilian Portuguese˘ München: Lincom Europa
(without intonation; hybrid IPA with (y, w) for (jé,éi÷ wé, éu))˘
irmen, f. “ cortes kollert, a. m. (1995) Taschenwörterbuch _ Portugiesisch–Deutsch {Deutsch
–Portugiesisch}˘ Berlin: Langenscheidt (Lusitanian; indicates metaphonetic variations, but
not systematically; presents /e/ + /S, Z, N, L, j, i/, for (É), but ç(Ai)Ç, for (3I, ’’I) and for (Ai,
’4i); IPA]˘
mea, g. (2003”) Dicionário de Português–Italiano˘ Porto: Porto Editora (Lusitanian; in the pres-
ent çreprintÇ the transcriptions have been added {whereas earlier they were only present in
the reverse part}, which is phonemic even if between ( ), with (b, d, g) in every context; it
420 a handbook of pronunciation
distinguishes (léı), but uses ç(!)Ç in place of (ı); in addition, it gives ç(éj, éw)Ç for (éi, éu),
and ç(–)Ç for (–ö), instead of /éö/; it indicates unstressed V which do not undergo reduc-
tion, but not always in a reliable way; metaphony is indicated if new separated entries are
present; IPA)˘
Pequeno dicionário {italiano–português} português–italiano (1993) São Paulo: Melhoramentos
(Paulista {pron. of São Paulo}; indicates some metaphonetic variations; IPA]˘
Pocket {English–Portuguese} Portuguese–English Dictionary (2001’) Glasgow: Collins (Carioca
{pron. of Rio de Janeiro}; indicates some metaphonetic variations; IPA]˘
tagliavini, c. (1938) Grammatica elementare della lingua portoghese˘ Heidelberg: Groos (Lu-
sitanian; çcanaÇ old-style IPA).
viana, a. r. gonçalves (1903) Portugais. Phonétique et phonologie _ Morphologie _ Textes˘ Leip-
zig: Teubner (Lusitanian; still the most reliable work, in spite of several inevitable misprints;
archaic IPA]˘
– (1912) Vocabulário Ortográfico e remissivo da lìngua portuguesa˘ Paris/Lisboa: Aillaud/Alves
(still the constant reference work of various authors and lexicographers for the even limit-
ed phonetic cues; current spelling with some additional diacritics. For the Portuguese or-
thography, it bases itself on a rather rational structuring, followed by several semi-reforms,
with qualities and failings {for foreigners, the latter would not be a problem any longer if
somebody were determined to prepare a Portuguese pronouncing dictionary, by taking ac-
count of the problems of metaphony and of Lusitanian peculiarities, which are completely
ignored by present-day o‚cial spelling, whereas that of the intermediate time was definite-
ly more careful}).
Russian (é 8)
Unfortunately, no IPA Russian pronouncing dictionary exists yet, with full transcriptions of the
di‡erent words (also avoiding archaic-Soviet morphonological –or orthographical– symbols,
mainly based on the Cyrillic alphabet).
Arabic (é 9)
Unfortunately, no Arabic pronouncing dictionary exists yet, with full IPA transcriptions of the
di‡erent words.
Hindi (é 10)
Unfortunately, no [IPA] Hindi pronouncing dictionary exists yet˘
chao, y. r. (1948) Mandarin Primer˘ Cambridge (µåßß.): Harvard Univ. Press (with vinyl rec-
ords; non-IPA, but with çtone-lettersÇ {now o‚cialized but trivialized by the çreformÇ of the
IPA}).
cheng, c-c. (1973) A Synchronic Phonology of Mandarin Chinese˘ ˛e Hague: Mouton (almost
IPA]˘
dow, f. d. m. (1972) An Introduction to the Pronunciation of Chinese˘ Edinburgh: E. Univ. Press
[IPA, but with not too reliable transcriptions).
– (1972”) An Outline of Mandarin Phonetics˘ Canberra: Australian National Univ. Press [IPA,
but with not too reliable transcriptions).
kratochvìl, p. (1968) †e Chinese Language Today˘ London: Hutchinson (çChinese/Karlgren-
izedÇ IPA: with çqÇ for (M), that is with a vocoid still considered as if it were the intense con-
toid (Ç); on the other hand, with the çvowel QÇ for the true intense contoid (Ö)]˘
norman, j. (1988) Chinese˘ Cambridge: C. Univ. Press [çSino-likeÇ IPA: with ç„, ÓÇ for (M, Ö)]˘
Practical Chinese Reader (1985) Beijing: ˛e Commercial Press (with audiocassettes; çChi-
nese/KarlgrenizedÇ IPA: with çqÇ for (M), that is with a vocoid still considered as if it were
the intense contoid (Ç); on the other hand, it has the çvowel QÇ for the true intense contoid
(Ö)]˘
qin, z. (1980) On Chinese Phonetics˘ Beijing: ˛e Commercial Press [çChinese/KarlgrenizedÇ
IPA: with çqÇ for (M), that is with a vocoid still consideres as if it were the intense contoid
(Ç); on the other hand, it has the çvowel QÇ for the true intense contoid (Ö)]˘
shan pao li, j. (1975) Introduction to the Chinese Language˘ Taipei: Ke Ji Publishing Co. [mod-
erately çSino-likeÇ IPA]˘
speSnev, n. a. (1980) Fonet$ka k$ta=ckogo r‘vka. Len$ngpad: *‘datelìctbo Len$ngpadc-
kogo Yn$bepc$teta (a very hybrid IPA]˘
Unfortunately, in the çJournal of the International Phonetic AssociationÇ itself Standrad Chi-
nese (Beijing) has appeared {2003, 109-112}, which is semi-Karlgrenized IPA, with both (M,
Ö) rendered as (≥), but described as (¡) and (Ö), including many more transcriptional queer
inaccuracies.
utilizable bibliography 423
Japanese (é 12)
Esperanto (é 13)
canepari, l. (1998”) Non c'è proprio niente da dire sulla pronuncia e la grafia dell'esperanto?˚ in
La linguistica, le lingue pianificate e l'Esperanto˚ a special issue of çL'EsperantoÇ 44-55
[canIPA]˘
cresswell, j. “ hartley, j. “ sullivan, l. h. (1987) Teach Yourself Esperanto˘ Sevenoaks:
Hodder “ Stoughton (with an audiocassette of çAnglo-EsperantoÇ; spelling).
kalocsay, k. “ waringhien, g. (1980[) Plena analiza gramatiko de esperanto˘ Rotterdam: ¨™å
(spelling).
wells, j. c. (1978) Lingvistikaj aspektoj de esperanto˘ Rotterdam: ©™∂/¨™å (spelling).
– (s. a.) Esperanto Pronunciation Disc˘ London: ∫™å (a vinyl record and leaflet; spelling).
zamenhof, l. l. (1962•) Lingvaj respondoj˘ Marmande: ™ƒ™ (spelling).