Académique Documents
Professionnel Documents
Culture Documents
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 Natural Phonetics Tonetics
16 Transcriptions
21 e contents of the HPr ( of the HPh)
23 Observations on phonetic terminology
28 Guide to the gures
34 Guide to dierent types of transcriptions
36 Transcribing by hand
37 Generic symbols
39 e ocial IPA chart
41 canIPA correspondent oIPA 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 simplications and other modications), 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. 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 (), which techni-
cally can be neither voiced nor voiceless, but has more anities with the latter
type, and will schematically be represented with ), besides voicelessness ((f), ) and
voicing ((v), ).
Furthermore, we also nd lenis (or lenited] voicelessness and voicing (respective-
ly (, v), , ), and mixed phonation (or half-voicing, (), with three schematic
1.1. Phonation types and positions of the glottis.
2 + +
. lenis voicelessness . lenis voicing
(, s, h, ), A) +
(v, z, H, ")
3
. mixed phonation
(, , , ) {1 (pbX ), 2
(bpX b), 3 (bpX )}: 1 (|'a, ap'a)
. creaky voice 2 (&aa'a) 3 (a'pa, 'a|) . falsetto
(Ca) (ma)
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 choice between the three of them depends on context: after
a pause (or silence) or a voiceless C, the rst half is still voiceless; whereas the sec-
ond half, which is in contact with voiced phones, is voiced: ( ), as in German: Bett
/'bEt/ ('Et). Instead, before a pause or a voiceless C, the two halves are exchanged:
( ), as in English: Bed /'bEd/ ('b;). 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: ( ), as in Ne-
apolitan-Italian pronunciation: dato /'dato/ ('dA;&). We have added several par-
ticulars in 1.1, where there are two peculiar phonation types, too: creak ( ) 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 voice is
1.2. Dierent phonation types exemplied 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.
('pha;) ('spa;) ('ba;) ('b;) ('phl;I) ('h;&Sp)
pie spy buy bib play headship ({Amer./Brit.} Eng.)
/ /
('kT, 'T) ('wks, 'ws) ('Eka) ('wks)
actor wicks (Brit.) actor wicks (mediatic Brit.)
('fa:va) (u'gwan:to) (&ukan'tan:te) ('sud:H, -d:) ('Es:th, -t) (sa'pe:te)
fava un guanto un cantante sud est sapete (Ital.)
(&ilkap'pOt:to) (&ilkp'pOt:o) (&ugn'dan:de) (s'e;Ie) (sa'be:de)
il cappotto il cappotto (~) un cantante (~) sapete (~) sapete () (reg.)
('pe) ('p#) ('pi) ('p) ('ppa) ('pis)) ('tX) ('Atu)
pied poids puis plat peuple prisme quatre (French) alto (Lus. Port.)
('phaen) ('aen) ('li:pI) ('a:b) ('ap&Ilt) ('Ek&f)
Pein Bein lieblich aber Abbild Bergdorf (Germ.)
(5pa) (pha2U) (5t,2bu2tUU) ('1pa) (a63qa) (63pa3b)
pai paishu tingbudng bi bici bba (Chinese)
('kaan) ('khaan) ('gaan) ('gHaan) ('m21) (3i'i)
kaan khaan gaan ghaan mq? (Hindi) chichi (Japanese)
1. prelude 3
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.
1.3-5 will help to reconstruct the typical modalities for the production and
identication 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 dierent degrees of jaw opening, and dierent lip positions,
since lip rounding doubles the number of possible vocoids.
1.3. Orograms with the extreme points for vocoid articulations.
i u
1.4. Vocogram with the extreme points for vocoid articulations (and corresponding labio-
grams).
i u
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
i M [*] Y y % u high () M
I [] Y T U lower-high ()
e X [] [] + P o higher-mid ()
= +
x [] [] # lower-mid () Y{%}
E [] @ O higher-low ()
a A [] ~
low () y{%}u
0 1 2 3 4 5 6 7 8 9
Furthermore, 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
1.6. Vocoid orograms. unrounded
i M {{*}}
I {{}}
e X {{}}
x {{}}
a A
rounded
Y y % u
Y T U
{{}} + P o
{{}} #
{{}} @ O
{{}}
1. prelude 5
i a u
i u
i M {{*}} { W} Y {% } y {% } u
I {{}} { : w} {T } Y {T } U
e X {{}} { , } {{}} {+ P} {+ P} o
x {{}} { } {{}} { } # { }
E { ,} {{}} {@ } {@ } O
aA { a} {{}} { } { }
1.1.4. Also for consonants we follow a more scientic terminology, which de-
nes 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 identication 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 dierent 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 diculty, instead of the clearer articulatory terms that we use), some terms
of auditory origin, such as aricate for stop-strictive or fricative 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, ), with just one tapping. Also aps are possible, such as ([, ), which
produce a rapid contact in passing to a fronter position, as can be seen from the
illustration ( 1.14.3).
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 2-13, including the variants indicated. is is why we call them HPr con-
1.9.1. Nasal contoids.
/m
)/m /[ /M
/M / /"
/
/M / /n /n
/N / / /
/n /n /N /n
/
/~ / /N
/ / / /
/\ / /m
/,
1. prelude 7
toids, 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 only 200 in comparison with the 462 given in the HPh,
they are less than half (again, of course, considering them as mere articulations,
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 / / /
p/b (/{ Q/ p/b
]/7
t/d t/d 4/7
T/D B/ / T/D
t/d
+/_ / /8
/ k/g $/
1.10. Stop contoids.
/
p/ / / ./
k/ %/ / w/
1.11.1..
q/Q q/Q / /"
C/ c/G &/1 C/G
c/g v/ /
/
/ / /
/6 f/v 5/ f/v
w/W
/ Q/z
/D
/ /
/J /
x/ /)
X/
/
s/z
s/z / /
/ /
S/Z
/
/ X/5 x/ /
./ //\ S/q /=
/ / /d /
/ /, /y /m
F/B / /V /
/ / / /
/j / / /
/V
W/w / x/
+
/H h H
/b /d / /%
/F / /= /j
+ + + +
/S / / /
+ +
+ +
/> / /< /<
+ , semilateral approximants
+ +
/ semiapproximant / /
10 a handbook of pronunciation
/r /5 /5 {/D
/R /R /R /e
1.14.3. ap contoids.
+
/[ r/m {V}/e /
+ + +
+
/l /R / a/l
+ + + +
/ /] / /l
+ + + +
/$ / /L /
+ + + +
/L / / /)
In 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 ocial
ones. Near those marked with another slightly dierent 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, 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 dierentiate the contoids given there.
+
=
wu C ji
&1
szqQ td l r
n
B N
R L kgx
S /B
s, z , S, Z , , J x, , j, w l, {L} !,
, s, z S, Z , q, Q c, G
T, D l !,
12 a handbook of pronunciation
1.18. Table of the contoids dealt with in the HPr. (@ indicates an intermediate articulation manner not present here, 10 of the HPh.]
@
`
&
F >
>)
`
@
@
@ b { b
F
p ( p
bilabial
m[ mMM
b
bilabial rounded
palatalized bilabial
uvularized bilabial
labiodentalized bilabial
> fv 5
]
labiodental
7
labiodental rounded
S postalveolariz. labiodental r.
palatalized labiodental
f w
v W
uvularized labiodental
" n
>
pro-dental
>
>
> sz sz
> d
>d 7
q q
dental
t 4
QQ
d
l R
denti-alveolar
velarized dental
[5] [5] r
[R] [R] R
uvularized dental
t T
[] l
d D
D z
> n N n
.
alveolar
r
a
[
m
alveobilabial
labiodentalized alveolar
alveolar semi-rounded
B
semi-paltaliz. alveolar semi-r.
V
velarized alveolar
R
e
5
l $
]
[n] N
uvularized alveolar
>
> "
>
postalveolar
>
postalveolar rounded
velarized postalveolar
S X x . / S
L
>
> Z 5 \ 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
>
> [
> {D
prepalatal
+
~
_
bilabialized prepalatal
labiodentalized prepalatal
N
>
>
> j 3
>J
>
k%
palatal
n
L
,
% F
postpalatal
8
postpalatal rounded
prevelar
provelar
V w
=
provelar rounded
> )
velar
w
x
W
k
\m ,
g
ym
velar rounded
j
labiodentalized velar
velarbilabial
velaralveolar
uvulo-postalveolariz. velar r.
<
> x H hH
k
'
uvular
X
K
r
pharyngealized uvular
pharyngeal
h
laryngeal
laryngeal rounded
@
`
&
`
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 denitions) for the
neutral pronunciation of every single language.
However, the table in 1.18 provides these contoids (except 7 complex coartic-
ulations, which appear among the orogram synopses, though), for a useful overall
comparison.
1.19 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 dierent contoid, with an intermediate
phase, made up of an'articulation with two simultaneous occlusions.
1.19.
Lateral () nasal () tl, t, t () tn, t
() explosions. ((Tl, T, T)) ((Tn, T))
1.20.
Inaudible explosion.
pt ((pT)) kt ((kT))
1.1.5. Intonation is the most evasive aspect of languages and (perhaps because
of this) it is generally very badly and supercially 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, 1.21-3 will help us to present intonation, without useless
absurd and harmful complications. In fact, without scaring (but also without
baing or disorienting anyone), 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 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 rst two, as shown by the
rst two examples in 1.21, are respectively used to state or to ask something. is
kind of question is dened 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, ]
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, ]
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 dierent, from the
very start, from en'll you come back again? In fact, even before hearing // () 'll,
which is the only syntactical dierence, we can perceive that /'wEn/ (5wn:) is
already dierent 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 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 dier also because it may have a slightly weaker stress and
dierent pitch: (&wn) up to (wn), and to /wn/ (wn) 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-/ = (-n j-), as a com-
promise): en you come back again (5w~: j&khm'bk 'gn:3 3), (&w~5j;u, w~-
5j;u, w~5j;u), and en'll you come back again? (5wn j&khm'bk 'gn:3 3).
(As some readers might have observed, in our inter(dia)phonemic transcription
of English, we use both /l/ and //, although they are not in opposition, (l, );
whereas, an intraphonemic kind of transcription would give only /l/, with (l, ).
Instead, we also have // (, ), the latter occurring before // or /j/. Our transcrip-
tion is diaphonemic too, since it also shows dierences between American and
British English, as for instance in no /'nOU/ ('n;)a ('n;)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 eect, 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 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 dierences exist for the suspensive intoneme, which is the most var-
ied and imaginative one.
In 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
1.24, that shows pitch modications 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).
1.24. Tonograms of parentheses and quotations.
1.1.8. 1.25 helps us understand the relationship tone languages have with in-
tonation. As a matter of fact, the dierent 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 1.26.
erefore, intonation is added to the existing tones, by modifying them in pre-
intonemes, according to what is shown in 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, 13.9 of the HPh)
modications, which are indicated by the corresponding tonograms, for Chinese
and Japanese ( 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 dierences 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
dierent, as can be seen in the individual tonograms given.
/ / ( ) (( )) / / ( ) (( ))
/ / ( ) (( )) / /( ) (( ))
1 /5/ (5) >1 2 // ('1) >Q 3 // (2) >5 4 /6/ (63) >Z
Transcriptions
1.2.2. However, it is immediately clear that dealing with descriptions and teach-
ing indications such as /aj, aw/ are rarely faithful and highly misleading, for such
dierent realities as English (a, a) (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 /oj/ 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,
/) 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 dier-
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 3 5 will surely explain the notational
dierence between English and German. To complete the series given above, we
have English /Ii, EI, OE, OU, Uu/, and German /OY/.
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 atten reality, by only showing what is functional, but compare
dierent languages better, by considering similarities and dierences as well.
In Japanese, it is certainly better to use /M/ rather than /u/, even if there is no
possibility of confusing them, since this language has no /u/. However, the use of
/M/ highlights the dierences 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 dierent 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 dierent would be to represent them with dierent fonts:
o, o, o, o
In conclusion, many dierent 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 xed 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 allophones {a more ambiguous and less advisable term, since
it implies modications not necessarily due to combination, but simply any sort
of dierence, 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 innite, 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 classication 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 dierent languages and dialects is realized by phones
and their symbols, which make it possible to compare dierent 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 dierent phonetic val-
ues.
us plain phonemic symbols (generally chosen among the most common, as
the ocial IPA ones, International Phonetic Alphabet, or oIPA), do not repre-
sent pronunciation exactly, but rather the relationships between the phonemes of
a given language.
ey are useful for the specic 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 rules, 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 diculties 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), (), (6), (), .
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 (p'i) /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 (pi)
(since another slight dierence exists, 4.3.5).
However, the low pitch on (i) depends on its occurring in the stressed sylla-
ble of a conclusive intoneme; thus, in actual fact, we ought to transcribe it as (p-
i3 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 1.26:
(5ma) /5ma/ ma mother, ('1ma) /ma/ m hemp, (2maa) /ma/ m horse, (63ma)
/6ma/ m scold, curse (In 11, we will see that Mandarin Chinese also has a ze-
ro 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 ocial 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 ecacy of the pho-
netic method and the associated transcriptions thus the reader will come upon
the spelling only after having xed 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-
inglearning. 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 rst).
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 diers 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 best language is
spoken when the social and regional origin of the speaker are least obvious. Neu-
tral pronunciation brings this miracle to pass, sometimes even to the point of
making the identication of social and regional origin completely impossible.
e supporters of regional pronunciations, who encourage individuals to
22 a handbook of pronunciation
maintain their genuine pronunciation (even when they carry out public 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 eectively 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 ocial language nor a local dialect. In Italy and
the German-speaking countries, dialects dier 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 dierences, while accent predominant-
ly involves dierences 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 gures,
and showing commitment. e last quality is clearly indispensable: it is not enough
simply to want to know a sustained eort to learn according to an eective meth-
od is also required.
tation of the process employed by the human ear, which manages to compensate
in an extremely eective 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 eld of terminology, as well, scientic rigor is a great help, while
the vagueness and lack of clarity of certain obsolete traditions with little scientic
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 ( 5.2-3 of the HPh).
e concept of a phonetic syllable also remains overly subject to the inuence
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-
et, since otherwise there is the risk of spreading and reinforcing unmitigated er-
rors, which compromise and discourage eective 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) denitive table shown on the ocial website. Ob-
viously, this would only bring us to the level of the ocial 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 umbrella on top, in the place of the
like /c/
normal /c/, or the possible /tXS/).
As we have said in 8.5 of the HPh, we prefer not to avoid the perfectly gram-
matical use of suxal 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 suciently perceiving the sound, so
that it can be identied 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 specic
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 denition suggests nding a way to symbolize the specic 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 nd 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 ocial IPA (oIPA = an o alphabet! 7.4 of the HPh) are not
at all sucient, 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 ( 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 dierent
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 averaging the results of multiple speakers and many utterances in var-
ious contexts, while keeping in mind phonological considerations and excluding
inappropriate samples.
e discoveries of sociolinguistics also require normalization; otherwise there
is a signicant risk of irremediably creating confusion, even when scientic da-
ta are used. is fact has been responsibly demonstrated in several recent works.
We will not cite these sources (as examples to follow), simply because this ought
to be the normal way of doing things, not the alarmist or scoopist manner of
far too many publications.
At rst 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 dierent, particu-
larly since in Brazilian Portuguese (and also in Lusitanian, but with additional
dierences), even nasalized taxophones are expected ((i, e/, A, 9/, u), when fol-
lowed by (), 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, , , , , ^/ and //, , the traditional //), including the four na-
salized vowels (/, , , ^/), 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 dierences can be seen
by comparing the vocograms (or vowel quadrilaterals) of these languages, in 2-
13. e same is true for the other elements which correspond.
1.3.9. If we now consider German and English, even without going into too
much detail (the details are available, of course, in 2 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 specic chap-
ters, or to see the dierences more rapidly, consult the transcriptions at the end of
those chapters).
In German and in English, the /i/ (short i) 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/ ('bT, '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 /i:/, even for English, ).
at emerges quite clearly from this discussion is the sorry state of those gram-
mar texts and language courses which describe the German /I/, or worse still, the
English one, as a short i, as in French vite, or in Spanish listo, or in Italian tto.
en one considers that many Italian regional pronunciations have a vowel in t-
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 specically directed exer-
cises, followed by careful checks. ese exercises consist of listening to good record-
ings and recording oneself to perform merciless examinations of how much pro-
gress has been made without cheating, otherwise all eort 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 formulae refer
to two works by the present author (Manuale di pronuncia italiana {A Handbook
of Italian Pronunciation} and Dizionario di pronuncia italiana {A Dictionary of
Italian Pronunciation}, 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 biological method!
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 dicult
to understand why some phonetics 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 ( 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 eort should lead to a real understanding of the vocoid articulations and
of the dierent 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 2-13, or in the phonosyntheses of 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
spirit of a language, manifested especially through vocoids, then through pitch,
and nally through contoids. Even a single millimeter makes a notable dierence
1. prelude 29
1.4.3. Markers can also be lled 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 () for example in hut hull /'ht, 'h/ ('hT, 'h:), or feet feel ('fIiT, 'fi;I)
(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: (, A, , , , , ). 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 dierence 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
(white) markers.
For examples of these rules in practice, the reader is invited to consult the vo-
cogram of the international French accent ( 4.4.1), southern French (
4.4.3), or German (with its various accents: 5), or Brazilian Portuguese, Russian,
Arabic ( 7-10).
1.28. Markers for variants.
unstressed // () unstressed /u/ (%)
stressed or unstressed /I, / (, ) stressed or unstressed /U/ (P)
stressed /E/ (e, E)
1.4.5. Diphthongs are considered wide, when their gure 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 rst
element, but at a slightly dierent 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) ( 1.30).
en the rst 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 gure with the edge and line both dashed).
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:, a)
(short or long) monophthong and monotimbric diphthong (a, a:, aa)
For example, in British English, beers /'bIz/ ('b;) and bear(s) /'bE{z}/
('b;{}) are in the same group as beer /'bI/ ('b;), even in accents which present
(, a) for //. 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 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 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 vowel triangles) show all of the symbols
perfectly aligned along the edges, threaded along the lines (rather like pearls,
since they are all circular as well), so that they extend outside of the margins. We
nd such gures decidedly odd, given that the objective reality of vocograms is
quite dierent.
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-
nies lateral contraction lateralization a trait accompanying and characterizing
some of the approximants ( 1.13.3). Semi-approximants have a dotted horizon-
tal line ( 1.13.2).
Trills, taps, and aps ( 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 ap orograms ( 1.33). In the same gure, we can see sequential articu-
latory compositions as well, which show two rapid tappings for (r), followed by
two open positions, with a white background.
For the tap, (R), there is only one contact, for a rapid tapping; whereas, the ap,
([), consists of three quite dierent (and quick) phases: rstly, the tip is brought
behind the alveolar ridge; then, while moving forwards, it touches the ridge (and
this is the pertinent articulation); nally, the third phase corresponds to the de-
tachment at every tapping, such as the two white ones in (r).
r = + + +
1.33.
R = +
Trills, (r),
taps, (R),
and aps, ([).
[ = + +
Laterals are identiable with an arrow placed on the most fundamental part of
the articulation. If the arrow is black, the contoid is bilateral ( 1.15.1); if white,
the contoid is unilateral ( 1.15.2). If the contoid is instead a lateral tap, a white
dot is shown ( 1.15.3). (Much more is to be found in the HPh, on other types of
laterals, with dierent additions.)
1.34.
(Bi)lateral, (l), and + +
unilateral, (), articulations. l
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 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 gure. Grooved
stopstrictives have a curved line as well, representing the groove ( 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 ( 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 signies 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 dierence
becomes clear immediately. It is useful to observe carefully (in the last part of every
set of diagrams, in 1.17) the palatograms and dorsograms which show the
dierence for the voiced alveolar articulation between stop, (D), and (bi)lateral,
(l), in addition to unilateral, (), and constrictive, (), as well (although these do
not occur in the 12 languages of the HPr).
1.4.11. Dorsograms (as in 1.17, rst row) give a new perspective, which is not
longitudinal but instead transversal. ese gures are particularly helpful to illus-
trate the dierence 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 prole (as in 1.5 1.16), they
indicate the direction of the characteristic movements, as produced by particular
facial muscles. Frontal labiograms require no particular explanations ( 1.8 and
those just mentioned) the vertical gap, which is steadily larger as the jaw opens,
is fairly intuitive. By carefully observing 1.16, let us consider attentively the labi-
al dierences for (, 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 ( 1.16.1).
Another useful type of diagram is the laryngogram (as in 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 ( 1.1.5-8 1.21-3 {and, in the HPh 6.4.5.1-4 13.8-34}), as in
tones as well ( 1.26 {and, in the HPh again 6.4.4 12.17-18}), lines (or
dashes) placed at dierent 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 ('). Unstressed 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 superuous for those who desire to do phonetics
thoroughly (and not merely easily, and inevitably supercially). A small num-
ber of symbols ineluctably leads to mediocrity, whereas a large number of symbols
opens the way towards the true understanding and savoring 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, verications, and connections, as well as new perspec-
tives, can be found in the general index (contents) and in the analytical index (in-
dex), by ipping through the chapters and the sections, and by paying attention
to the tables, the lists of symbols, and the groups of gures.
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 dierent parentheses 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 simplied, changing n into u, or m into ; or l into or a. Dots
should not be omitted, and so should not be written in place of i, nor for j.
Stylistic additions and modications should be avoided. us, d must remain
dierent from and from , just as is dierent from the grapheme g, and h from
. e same goes for cases like z, which should be kept distinct from both Z and .
Moreover, (L, r, R, ) are, by the same token, quite distinct from (/H, , [, y), .
Naturally, small capitals should be avoided as well, since the symbols (, , , I, G,
, ,, , K) are dierent 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 typographic eye 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, , ), (K, , , ), (A, a, ), (,
E, ), (X, x v, , O), (r, <, ), (h, H, ), (f, j, 3 J, , , )), to the presence or ab-
sence of a sort of serif, or the type of serif present: (i, I, u, , U o, , ).
In working towards this objective, the observations in 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 accept 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 .gif
can be pronounced either /'gIf/ or /'GIf/), .
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.
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
aps), and laterals (including unilaterals and tapped laterals).
In various languages, for any manner of articulation except trills (and taps and
aps), phonetically semi- articulations are possible ( less tense with no full
contact, also for nasals, stops, stopstrictives and laterals).
38 a handbook of pronunciation
vocoid/vowel semilateral C
reduced V (in duration; = ) lateral trill/tap C
shortened V intense (syllabic) lateral trill/tap C
nasalized V @ voiceless lateral trill/tap C
devoiced V nasal C
V voiced lenis V M intense (syllabic) nasal C
voiceless lenis V voiceless nasal C
half-nasalized V sonant (or sonorant) C
rounded V % intense (syllabic) sonant C
unrounded V # voiceless sonant C
5 advanced V trill (or trill tap) C
retracted V voiceless trill (or trill tap) C
lowered V constrictive trill C
raised V tap C
normal V or under other conventions lateralized tap C
creaky V (or laryngealized) ap C
0 contoid/consonant lateralized ap C
0 reduced C (in duration; = ) median approximant C
shortened C approximant C
glottalized voiceless C, with simultaneous semi-approximant C
() lateralized approximant C
intense (syllabic) C J lateralized semiapproximant C
devoiced C peripheral approximant C
voiceless lenis C or under other conven- obstruent C (F, !, ), in diphonic pairs
tions, especially diaphonemic laryngeal approximant C
C voiced lenis C laryngeal constrictive C
= voiceless C F stop C
voiced C K semi-stop C
rounded C (slit) constrictive C
unrounded C (slit) semi-constrictive C
J palatalized C _ grooved constrictive C
velarized/uvularized C ` grooved semi-constrictive C
advanced C (generic) constrictive C
retracted C (generic) semi-constrictive C
tenser/closer C ` (slit) stop-strictive C
less tense/close C @ (slit) stop-semi-strictive C
voiced creaky/laryngealized C (slit) semi-stop-strictive C
lateral C grooved stop-strictive C
constrictive lateral C & grooved stop-semi-strictive C
voiceless lateral C I grooved semi-stop-strictive C
lateral tap C ! (generic) stop-strictive C
unilateral C (generic) stop-semi-constrictive C
1. prelude 39
1.6. For informative purposes we also show the ocial IPA chart ( 1.36). It
highlights mostly its defects and global deciencies, however.
e rst section gives the ocial pulmonic consonants, which are 58 (with a
minor addition of ten other symbols). e second small section shows ve dejec-
40 a handbook of pronunciation
~~~ (~~-~)
$ Voiceless alveolo-palatal fric.
Clicks Voiced implosives Ejectives ' Voiceless labial-velar fric. B Voiced alveolo-palatal fric.
Bilabial Bilabial as in: w Voiced labial-velar app. Alveolar lateral ap
| Dental Dental/alveol. p Bilabial Voiced labial-palatal app. Simultaneous S and x
! (Post)alveolar Palatal t Dental/alveol. Voiceless epiglottal fric.
Aricates and double articulat.
Palatoalveolar Velar k Velar H Voiced epiglottal fric. can be represented by two sym-
~ Alveol. lateral Uvular s Alveol. fricat. Epiglottal plosive bols joined by a tie bar if necess.
~ ~
Front Central Back ~
Close i y % M u 1
or >1 Extra-high or Rising
I Y U 1 High 8 1 Falling
Close-mid e + o 0 1 Mid 3 21 High rising
` 1 Low "1 Low falling
Open-mid E O <1 Extra-low 4[1 Rising-falling
Open a A Downstep (relative) Global rise
ere symbols appear in pairs, the one
Upstep (relative) Global fall
to the right (and U) is rounded.
tives (or clicks), ve injectives (or implosives, including the theoretical uvular
one), and four examples of ejectives.
en the ocial 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 (, , ) 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 nd 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 ocial diacritics (even two or three for
one symbol), which can be of some use when trying to hint at some important ar-
ticulatory dierences.
Unfortunately, this partial solution also betrays one of the fundamental princi-
ples, which used to dierentiate 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 ocial 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 oIPA transcriptions (given within
), in order to show their composition (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 oIPA followers usually omit them
in their approximate and vague transcriptions), and to highlight that it is unthink-
able to do diacritical transcriptions, as all other phonetic alphabets 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 dierences, without diacritics,
with ().
Let us recall that it is paramount to carefully observe the orograms, to constant-
ly compare them, and to nd similarities also between symbols, by starting from
the ocial ones, from which the others have been derived (although with useful
modications and some necessary substitutions). As we have already said, our rm
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).
Vocoids () () ()
() (f ) or (F )
Spread/unrounded () () ()
(i) (i) (=) () () ()
(I) (I) (=) (y) (y7 ) or (y. )
(e) (e) (=) (Y) (Y7 ) or (Y& )
() (ef ) or (EF ) () (7 ) or (& )
(E) (E) (=) (#) (Z ) or (7F ) or (Z ) or (7f )
() (a) () () (7 ) or (& )
() (7 ) or (& )
() (i&) or (i]) or ()
() (I&) or (I7) (%) (%) (=)
() (e& ) or ( ) or (e7 ) (T) (%f ) or (+F )
() (EZ ) or (eZ ) (+) (+) (=)
() (aZ ) or (E7 ) or (E& ) () () (=)
() (a7 ) or (a& ) (@) (f ) ()
() (7& ) ()
() () (=)
() (f) or (HF ) () (u7 ) or (u )
() () or (F ) () (U7 ) or (U )
() () (=) (P) (o7 ) or (o )
() () (=) () (OZ ) or (OF ) or (oZ ) or (of )
(a) (a7& ) or (f ) () (O7 ) or (O )
() (7 ) or ( )
(M) (M 7 ) or (M )
() (M Z ) or (M f ) or (M
7f ) or (7F ) (u) (u) (=)
(X) () or (7) (U) (U) (=)
(x) (Z ) or (7F ) or (Z) or (7f ) (o) (o) (=)
() () or (7 ) () (OF ) or (of )
(A) (A ) or (A7 ) (O) (O) (=)
() () (=)
(*) (M) ()
() (Mf ) or (f) Semi-rounded
() () ()
() (f) or (F ) () (ig ) or (yG)
() () () () (g ) or (%G )
() (A) () (W) (M g ) or (uG )
() (Ig ) or (YG )
Rounded (:) (I.g ) or (Y.G )
(Y) (y) () (w) (M hg ) or (UG )
() (Y) () () (eg ) or (G )
1. prelude 43
(/, =) h )
( ap
(/, j) f )
(w ({}, [) ( , R )
({V}, e)* ( , R )
Laryngeal approximant
(/, ) (# )
(h, H) (hf , Hf )
(/, ) (Hf ) Constrictive trill
(+) (h J)
(, ) (K#=H , K#H )
() (h )
() (' ) Lateralized ap
(=) (h )
() (h ) (r, m) (R: &, R:&)
( , ) ((R: &, R:&)
Lateralized approximant (+ 2 semi-)
(Bi)lateral
(/, S) (VX< &)
(/, ) ( &) ((/, l)) (lY)
(/, ) (< &) ((/, )) (lY) or (Y)
(/, ) (<< &) ((/, R)) (lY:)
(/, >) (<< &) ({a}, l) (l , l )
(/, ) (<< &) ({}, ) (l), (l ) or (), ()
(/, <) (X < &) ({}, ]) (lf), (l) or (), (f)
(/, <) (<X &< ) (/, ) (l ) or ( )
(/, ) (Xh < &) ((/, l)) (l <) or (: )
(/, $) ($) (=)
Trill ((L)) (l: J)
({}, ) (l J, l J)
(/, r) (r )
(/, L) (L) (=)
(/, 5)* (r )
(/, ) () ()
(/, 5) (r &)
({{}, D) (r+ J, r J) Unilateral
({'}, K) (K= , K) (, =)
(/, ) (l@ )
Tap (/, |) (l@ )
(/, ) (l@ )
({5}, R) (R+ , R )
(/, R)* (R ) Tap(ped) lateral
(/, R) (R < )
(/, e) (#) () (/, ) () ()
({[}, ) (R+ J, R J) (/, ) (:)
({3}, r) (KW+ , KW )
Semilateral
(/, ) (f)
(/, ) (f )
2. English
2.0. In this chapter we will deal with the American and British neutral accents
(or standard accents). We will also make a teaching proposal for an internation-
al 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 mediatic 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, aected, and Cockney accents). In a book in
progress Eli 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 ) af-
ter the diaphonemes, with examples.
Vowels /EI/ /eI/ dZ /'dEI/ ('D;I)
/aE/ /aI, I/ y /'flaE/ ('fla;)
/i/ /i, I/ lady /'lEIdi/ ('lIDi) /OE/ /OI/ boy /'bOE/ ('b;)
/I/ /I/ bit /'bIt/ ('bT) /aO/ /aU/ cow /'kaO/ ('kha;)
/E/ /e, E/ let /'lEt/ ('lT) /OU/ /U, oU/ go /'gOU/ ('g;)a ('g;)b
// /, a/ hat /'ht/ ('hT) /Uu/ /u:, u/ who /'hUu/ ('hU;u)a ('h;u)b
// /, / hut /'ht/ ('hT)a ('hT)b
// /O, A:, A/ hot /'ht/ ('hAT)a ('hT)b // /I, , , / wies /'wISz/ ('wS)a
/U/ /U/ book /'bUk/ ('bk) ('wS)b
/u/ /u, U/ inuenza /Influ'Enz/ (&Mfl- /;/ /, A:/ last /'l;st/ ('lsT)a ('lA;sT)b
'nz)a (-)b /A;/ /A:, / pasta /'pA;st/ ('phA;sT)a
// // ano`er /'n/ ('n)a ('n- ('phsT)b
)b contain /kn'tEIn/ (khn'Th;In) /;/ /, O:/ so /'s;/ ('sO:)a ('s;)b
/A:/ /A:, A/ spa /'spA:/ ('spA:) /O;/ /O:, / false /'fO;s/ ('fO;s, 'fAs)a
/O:/ /O:, O/ law /'lO:/ ('lO:)a ('l:)b ('f;s, 'fs)b
/;/ /r, !:, :r, r, / hurry /'h;<i/
/Ii/ /i:, i/ see /'sIi/ ('sI;i) ('h;i)a ('h>i)b
2. english 47
/:</ /:r, :r, r, !:/ furry /'f:<i/ ('f;i)a /K:/ /, O:/ repertory /'<EptK:<i/ ('<p-
('f;>i)b &T<i)a ('>pT>i, ->i)b
/:/ /:, :, :r, r, !:/ fur /'f:/ ('f:)a // /U, , T, / regular /'<Egjl/ ('<g-
('f:)b jl)a ('>gjl, -gj-)b
/A:/ /A:, A:r/ car /'kA:/ ('khA:<)a ('khA:)b
/O:/ /O:, O:r/ door /'dO:/ ('D:<)a ('D:)b Consonants
// /r, , r, r, / wonder /'wnd/ /m/ /m/ some /'sm/ ('sm:)a ('sm:)b
('wnD)a ('wnD)b /n/ /n/ sun /'sn/ ('sn:)a ('sn:)b
/i/ /i, I, ir, Ir, I, i, ir, i/ // // su /'s/ ('s:)a ('s:)b
happier /'hpi/ ('hpi)a ('hpi)b // /m, m, m/ rhy`m /'<I/ ('<-
/I/ /I, Ir, Ir, I, Ir, I/ here /'hI/ )a ('>-)b
('hI;)a ('h;)b // /n, n, n/ cotton /'kt/ ('khAT)a
/E/ /e, E:, er, Er, e, er, E/ `ere ('kh-)b
/'E/ (';)a (';)b
/U/ /U, Ur, Ur, U, Ur, U/ moor /p/ /p/ pa$ /'pk/ ('phk)
/'mU/ ('mU;)a ('m;)b (with further /b/ /b/ ba$ /'bk/ ('bk)
pronunciations) /t/ /t/ o /'tUu/ ('ThU;u)a ('Th;u)b
/u/ /u, U, ur, Ur, u, u, U, /d/ /d/ do /'dUu/ ('DU;u)a ('D;u)b
ur, u:/ rescuer /'<Eskju/ ('<skj)a /k/ /k/ came /'kEIm/ ('kh;Im)
('<skj)b /g/ /g/ game /'gEIm/ ('g;Im)
/Ii/ /i:, i:r, i:, i:, i:r, i:/ seer /c/ /tS/ >ain /'cEIn/ ('ch;In)
/'sIi/ ('sIi)a ('sIi)b /G/ /dZ/ Jane /'GEIn/ ('G;In)
/EI/ /eI, eIr, eI, eI, eIr, eI/ /f/ /f/ f[ /'fjUu/ ('fj;u)
plZer /'plEI/ ('phlI)a ('phlI)b /v/ /v/ vi[ /'vjUu/ ('vj;u)
/aE/ /aI, I, aIr, aI, aI, aIr, // // wreath /'<Ii/ ('<Ii)a ('>Ii)b
aI/ liar /'laE/ ('la)a ('la)b // // wreathe /'<Ii/ ('<I;i)a ('>I;i)b
/OE/ /OI, OIr, OI, OI, OIr, OI/ /s/ /s/ ice /'aEs/ ('as)
employer /Im'plOE/ (m'phl)a (m- /z/ /z/ \es /'aEz/ ('a;)
'phl)b /S/ /S/ dilution /d'lUuSn/ (D'lUuSn)a
/aO/ /aU, aUr, aU, aU, aUr, aU/ (D'luS)b
tower /'taO/ ('Tha)a ('Tha)b /Z/ /Z/ delusion /d'lUuZn/ (D'lUuZn)a
/OU/ /U, oUr, oU, oU, oUr/ (D'luZ)b
slower /'slOU/ ('sl)a ('sl)b
/</ /r/ rate /'<EIt/ ('<IT)a ('>IT)b
/Uu/ /u:, u:r, u:, u:, u:r/ doer
/j/ /j/ yate /'jEIt/ ('jIT)
/'dUu/ ('DUu)a ('Du)b
/w/ /w/ wait /'wEIt/ ('wIT)
/</ /r, r, r, / wonderi /'wnd- /h/ /h/ hate /'hEIt/ ('hIT)
<I/ ('wnD<, -D<)a ('wnD>, /l/ /l/ late /'lEIt/ ('lIT)
-N>)b
// /t, / ci /'sIi/ ('s[i)a ('sTi)b
/I</ /Ir, Ir, Ir/ heari /'hI<I/ ('h<-
// /, r/ car /'kA:/ ('khA:<)a ('khA:)b
)a ('h>)b
// /j, / n[ /'nUu/ ('nU;u)a ('nj;u)b
/E</ /er, er, Er, E:r/ cari /'kE<I/
// /h, / when /'wEn/ ('wn: 'hwn:
('kh<)a ('kh>)b
'Wn:)
/U</ /Ur, Ur, Ur/ curi /'kjU<I/
('khj<)a ('khj>)b (with f. pron.) // /l/ lull /'l/ ('l,:)a ('l:)b
// /, e, E/ dictionary /'dIkSn<i/ // /l/ little /'lI/ ('l[)a ('lT)b
('DkS&<i)a ('DkS>i)b
48 a handbook of pronunciation
Vowels
American monophthongs
2.1.2.1. 2.1 shows the American monophthongs (the British ones are given
in 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/, () //, (A:) /A:/, (A) //, () //, (O:) /O:/, () /U/. Although in the mediatic
American accent ( 2.4.2.2) /A:/ and // 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 dierent, even if chiey only in
terms of length. Furthermore, in this way we can keep a diaphonemic relation
with the British neutral accent. And nally, 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 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:)): ('hT)
/'hIt/ hit, ('js) /'jEs/ yes, ('m;n) /'mn/ man, ('fA;) /'fA:/ fa`er, ('lAT) /'lt/ lot,
('<n:) /'<n/ run, ('lO:n) /'lO:n/ lYn, ('phT) /'pUt/ put.
For the sake of descriptive precision, although generally variations do not re-
quire dierent symbols, it is worthwhile to notice the taxophones of /I, E, , ,
U/; please note the two that change their symbols as well, in comparison with (,
): ('bE:, 'h,:) /'bE, 'h/ bell hull.
2.1.2.2. e gr\ markers in the vocogram also show three variants of /I, U, O:/
(I, U, :) (compared to normal (, , O:)). ey occur in (stressed or unstressed) syl-
lables checked by //: (I) /I/, ([j]U) /[j]U/, (:<) /O:/. ere are no dierences in
terms of the nuclear element (or, perhaps, nucleal element) in () /E/, (A:<)
/A:/, because they coincide with the black markers for /E, A:/ (, A:). Instead, the
2. english 49
typical American realization of /:/ is through an intense contoid (:), rather than
a vocoid as in British pronunciation ((:)).
Examples: ('hI;) /'hI/ here, ('phjU;) /'pjU/ pure, ('D:<) /'dO:/ door, (';)
/'E/ `ere, ('fA:<) /'fA:/ far, ('f:) /'f:/ fur A possible pronunciation with (, )
(instead of (I, U), for /I, U/} does not sound strange, although it is not the
most frequently heard.
For /O:</ (that is, before vowels: /O:</), the realization is (O:<) (dierent 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 dierence
between /I<, E<, U</ and /I, E, U/ (taken into consideration above) should
be carefully noted. Here are a few examples: ('h<)a ('h>)b /'hI<I/ heari,
('b<)a ('b>)b /'bE<I/ beari, ('D<)a ('Dj>)b /'dU<I/ duri
It must be also noticed that, in American pronunciation, the phonemic sequence
/jU/ has the variant /j:/ (more typically mediatic): ('phjU; 'phj:) pure (the vari-
ant /[j]:</ occurs for /jU<, U</, too): ('khj<is 'khj;is) curious, (n'D<ns -'D;-
ns) endurance, ('D< 'D;) duri is is also true of other consonants with a
palatal component: ('SU; 'S:) /'SU 'S:/ sure but ('phU;, -:<) /'pU, -O:/ poor.
e dierence between ('mO;<n) /'mO:nI/ morni and ('mon, 'mo;<nI)
mourni (which, by now, belongs only to traditional American pronunciation)
is no longer neutral, nor is it modern, but simply regional. As a matter of fact, they
are now both pronounced ('m;<n) (('m;n)b) /'mO:nI/ (and the traditional
pronunciation of mourni could be shown diaphonemically as /'mOnI/ or
/'mo:nI/).
2.1.2.3. e four white markers show the unstressed realizations of the remain-
ing three phonemes: (i) /i/, () /u/, (, ) //. ey occur in the following contexts
/i/ at the end of lexemes: ('lIDi[]) /'lEIdi[z]/ lady ladies, ('ni&) /'EniI/ aJ-
`i, and before vowels: (<i'kT) /<i'kt/ react /u/ chiey before vowels: (&sc'I-
Sn) /sIcu'EISn/ situation (Th'Gi) /tu'Gi/ to Algy A few words ending in -ue
[value atue virtue iue tiue] may have a reduced variant, especially in a prein-
toneme, which might usefully be represented with the diaphoneme /u/: ('vlj[u],
'sTc[u]) /'vlju, 'stcu/ value atue
nally, // is the most frequent unstressed vowel phoneme of the English lan-
guage (in particular American, Oceanian, and South African): (f'ThAg<f) /f'tg-
<f/ #otogra#er. In words like ('sf)a ('sf)b /'sOUf/ sofa we can see that a
50 a handbook of pronunciation
nal // 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
( 2.2). Here we have nine bla$ markers (for nine either stressed or unstressed
phonemes): () /I/, () /E/, () //, (A:) /A:/, () //, () //, (:) /O:/, () /U/, (:)
/:/. We nd: ('hT) /'hIt/ hit, ('js) /'jEs/ yes, ('m;n) /'mn/ man, ('fA;) /'fA:/
fa`er, ('fA:) /'fA:/ far, ('lT) /'lt/ lot, ('>n:) /'<n/ run, ('l:n) /'lO:n/ lYn, ('D:)
/'dO:/ door, ('phT) /'pUt/ put and ('w:) /'w: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:/, we can see that it is pronounced exactly as fa`er ('fA;) (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:
diimilation ( 2.3.3.5).
Let us quickly add that even for (:) /O:/, as in ('w:[]) /'wO:[z]/ war(s) by now,
2. english 51
f
// = /I/ () /[]/ (, *, |)
/O:/ (:), /O;/ (:, )
/E[]/ (, ) /:</ (:>), /:/ (:), /;</ (:>)
f
2.1.3.2. For British English, /I, E, U/ are given in 2.3: ('h;) /'hI/ here,
('phj;) /'pjU/ pure, (';) /'E/ `ere We have already seen that in British En-
glish a simple vocoid, with no contoid, occurs in cases like ('f:) /'f:/ fur It is to
be noticed that, in this type of pronunciation, the phonemic sequence /[j]U/, by
this time, is almost exclusively substituted with its variant /[j]O:/: ('phj: 'phj;)
pure, ('ph: 'ph;) poor, ('S: 'S;) 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;>is, n'Dj;>ns, 'Dj;>, 'S:)
pure curious endurance during sure (at that time, of course, the neutral pronun-
ciation had () /I/ for modern (i) /i/: ('khj;>s), curious). Another curiosity
is that, in the rst half of the twentieth century, the same pronunciation (j:) /j:/
(which today no longer occurs, except as an old-fashioned variant) was a variant of
/I, I</: ('D; 'Dj:) dear ('h; 'hj:) here (sn's; -'sj:) sincere ('khl>ns 'khlj;-
>ns) clearance however, it still remains as a secondary variant in ('j; '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/, (, )
//. ey occur in the same contexts /i/ at the end of a lexeme: ('lIDi[]) /'lEIdi[z]/
lady ladies, ('ni&) /'EniI/ aJ`i, and before a vowel: (>i'kT) /<i'kt/ react
/u/ occurs chiey before vowels: (&sc'IS) /sIcu'EISn/ situation (Th'Gi) /tu-
'Gi/ to Algy e possibility of nding /'vlju/ value ( 2.1.2.3) is less common
in British English.
nally, // is the most frequent unstressed vowel phoneme in British English,
too, also because of the vocalization of /, :/): (f'Thg>f) /f'tg<fz/ #o-
togra#ers ((f'ThAg<f)a) ('>c) /'<Icd/ Ri>ard (('<c)a), (Th'g;) /t'gOU/
to go ((Th'g;)a). Of course, we also nd (|) /, / (before pauses): ('sf)
/'sOUf/ sofa {('sf)a}, ('A;fT) /';ft/ after {('[;]fT)a}.
2.1.3.5. It is useful to recall here (although this is also true of the preceding cas-
es of //) that, in neutral British pronunciation (as well as Oceanian and South
African), the normal realization of // is (): (ph'f;mns) /p'fO:mns/ perform-
ance (&s'Th;In) /s'tEIn/ ascertain (h'hzbn) /h'hzbnd/ her husband Of
course, this holds good unless in absolute nal position before a pause, (|), or nal
in a rhythm group, or with the grammemes /z, d/, (, , ): ('h>) /'hI<I/
heari ('h; w'j;u) /'hI wI'jUu/ here wi` you, ('b;) /'bIz/ beers, ('h;> n-
';) /'hI n{d}'E/ here and `ere
e modern neutral British pronunciation of /E/ does not change any longer
according to context, but it is always () (with a strong tendency to a long mon-
ophthong, through (), up to (:) /E:/, as has, for a few generations, already hap-
pened to the previous /O/): ('b;[]) /'bE[z]/ bear(s) For /A:, :, O:/, as we
have seen, // is lost and realized through a zero phone, (`): ('khA:[]) /'kA:[z]/
car(s), ('f:[]) /'f:[z]/ fur(s), ('D:[]) /'dO:[z]/ door(s).
In absolute nal position and before pauses, we nd: ('b;|) /'bI|/ beer,
('khj;|) (and (-j:|)) /'kjU|/ cure and (f'Thg>f|) /f'tg<f|/ #otogra#er
But, if nal in a rhythm group, we have: ('b; f'mI;i) {('bI; f'mI;i)a} /'bI f-
2.3. British // diphthongs.
/I, I, I</ (>, z+, D+, +|) /j/U, -U, -U</ (j>, jz+,
jD+, j+|) (+ /jO:, -</}
'mIi/ a beer for me, (&k'khj;> 'z 'g;, -j:> 'z-) {(&k'khjU; 'z 'g;, -j: 'z-)a}
/t'kjU 'Izt 'gUd/ `at cure isn't good, (f'Thg>f 'DDT) ((f'ThAg<f 'DD-
T)a) /f'tg<f '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 (
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/,
(a) /aE/, () /OE/, (a) /aO/, () /OU/, (Uu) /Uu/: ('ThI;i) /'tIi/ tea, ('D;I) /'dEI/ dZ,
('ha;) /'haE/ high, ('b;) /'bOE/ boy, ('na;) /'naO/ now, ('g;) /'gOU/ go, ('hU;u)
/'hUu/ who Besides, we need: (iI) /Ii/, (ju) /jUu/, (Uu, u) /Uu/, ([j]uU) /[j]Uu/:
('fi;I, 'fiI) /'fIi/ feel, ('ju) /'jUu/ you`, ('nU;u, 'n;u) /'nUu/ n[, ('ju;U, 'juU)
/'jUu/ yule, ('khu;U, 'khuU) /'kUu/ cool
e other diphthongs, even if followed by //, do not change much their com-
ponents, apart from those with front second elements; besides (except for /OU/,
which has only //, in neutral pronunciation), they freely uctuate between // and
// (with a possible realization of // as // ()): ('s;, 's) /'sEI/ sail, ('fa;,
'fa) /'faE/ le, ('b;, 'b) /'bOE/ boil, ('fa;, 'fa) /'faO/ fowl but only
('s;) /'sOU/ soul.
2.4. American diphthongs.
/Ii/ (iI) /jUu/ (ju, jUu)
/Ii/ (Ii) /[]Uu/ (Uu, u), /[j/]Uu/ ([j/`]uU)
'G;2 5w 'g;3 3) /'nOU. 'GOU, 'wOUnt 'gOU./ No, Joe won't go. e rst element of
/OU/ is central and unrounded, (), while in American pronunciation it is back
and rounded, (): ('n;23 'G;2 'w 'g;23)a. In the British accent, at the be-
ginning of the twentieth century, (o) was widespread; until the fties it was (P),
always with lip rounding, while (, ), at that time, sounded rather aected.
2.5. British diphthongs.
/Ii/ (iI) /[j/]Uu/ ([j]u)
/Ii/ (Ii) /[j/]Uu/ ([j]uU)
/OE[]/ (, )
/EI[]/ (I, )
/OU[]/ (, )
/aE[]/ (a, a) /aO[]/ (a[])
2.1.4.3. In the vocogram showing /I, E, [j]U/ ( 2.3), there are also /aE,
aO/. As a matter of fact, in a typical British pronunciation (besides remaining sta-
ble, as in American English), both can frequently reduce to /a/ (a;) (and
(a;|)): ('fa, 'fa|) and ('fa;, 'fa;|) /'faE[z]/ re(s). Otherwise, /aO/ can be-
come /A:/ (A;) (and (A;|)), up to coincide with /A:[]/: ('Tha, 'Tha|),
('Tha;, 'Tha;|) (also ('ThA;, 'ThA;|), besides ('ThA:[])) /'taO[z]/ tower(s).
In this kind of pronunciation (sometimes dened smoothing), even the rarer
/EI, OE, OU/ can always be lessened, respectively, to /E:, O:, :/. us
/E:/ (up to coincide with /E/): ('lI, 'lI|), ('l;, 'l;|), ('l;[]) (and also
('l;[], 'l:[])) /'lEI[z]/ lZer(s); /O:/ (up to coincide with the old /O/): (m-
'phl, -|), (-;[], -;|), /Im'plOE[z]/ employer(s); /:/ (up to coincide
with /:/): ('m, 'm|), ('m;, 'm;|), ('m:[]) /'mOU[z]/ mower(s) (
('fa[], 'Tha[], 'lI[], m'phl[], 'm[])a).
Vowel diaphonemes
2.1.5.3. Two other diaphonemes, /;, ;/, are more important in distinguishing
between the American and British modern neutral accents. e rst, /;/, shows
the dierence between //a (but it often behaves like a long monophthong /:/)
and /A:/b (chiey before /f, , s/ and before /0/), as in: ('g<[;]sp)a ('g>A;sp)b
/'g<;sp/ grasp ('l[;]sT)a ('lA;sT)b /'l;st/ la ('ph[;]s)a ('phA;s)b /'p;s/ pa ('h[;]f)a
('hA;f)b /'h;f/ half ('[;]fT)a ('A;fT)b /';ft/ after ('ph[;])a ('phA;)b /'p;/ pa`
56 a handbook of pronunciation
2.1.5.5. ere are another couple of diaphonemes, /A;, O;/, which are relatively
less signicant, because /A;/ is used especially in words of foreign origin, written
with an a\ ('phA;sT)a ('phsT)b /'pA;st/ paa, (&vIiT'nA:m, -'-)a (-'n;m)b /vIiEt-
'nA;m/ Vietnam
ile /O;/ occurs in particular in words written with ausC aunC alC\ (O;'sTI;,
A-)a (;'sT;, -)b /O;'stI/ auere, ('hO;nT, 'hAnT)a ('h;nT)b /'hO;nt/ haunt ('sO;T,
'sAT)a ('s;T 'sT)b /'sO;t/ 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 //
// = /I/, chiey in a more rend and more conservative way of speaking. By the
same token, in British English, there are cases of // = //, chiey in a less rened
and more innovative way of speaking. So, we nd a greater convergence at a more
up-to-date and modern level.
Here are a few examples: (s'vI;)a (s'v;)b /s'vI/ severe, ('ksGnT, 'gz-)a
(-G-)b /'EksGnt, 'Egz-/ exigent, (&khAmp<'h~Sn)a (&khmp>'h~S)b /kmp<-
'hEnSn/.
2.1.6.2. It may be a good idea to take stock of the situation about some (real or
seeming) suxes and prexes, because we still nd old-fashioned and outdated tran-
scriptions, especially in bilingual dictionaries. e modern neutral pronunciation,
British too, by now, has // (while /I/ sounds quite pompous) in: -ace ('phls)
/'pls/ palace -ate ('chAklT)a ('chklT)b /'cklt/ >ocolate -iI ('hpli) /'hp-
li/ happiI -i ('khwAn[[][i)a ('khwnTTi)b /'kwni/ quanti
en /-i/ is preceded by /s/, it can often become /-sti/. So, a notation like /-si/
includes both possibilities, while excluding that the two diaphonemes /, / may
work together, because of contextual incompatibilities. at means that, if // falls,
then // automatically becomes /t/, because it is preceded by /s/): necei univer-
si veloci /n'sEsi, jUun'v:si, v'lsi/.
Besides, we have: -le ('hpls)a ('hpls)b /'hOUpls/ hopele -ne ('gDns)
/'gUdns/ goodne For -e /s/ prevails, chiey 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<nss, -s)a (p>n'ss, 'ph>nss)b. To end with, -let ('b<Is-
lT)a ('b>-)b /'b<EIslt/ bracelet -ret ('skA;<lT)a ('skA;lT)b /'skA:lt/ scarlet
2.1.6.4. On the contrary, regularly we have /I/ in: -ic(s) (f'n[k[s])a (-Tk[s])b /f-
'nEIk[s]/ #onetic(s) -ical (s'Th<k)a (->-)b /s'tI<Ik/ sarical -i ('sTnD)
/'stndI/ andi -ip ('f<~Sp)a ('f>-)b /'f<EndSIp/ friendip -ive (n'Thns)
/In'tEnsIv/ intensive
is group includes: -age ('vl) /'vIlIG/ village -i ('glS) /'IglIS/ Engli -i
('lgwsT) /'lIgwIst/ liui e- and i- (at the beginning of words, in unchecked
syllables): ('lvn) /I'lEvn/ eleven ('Thjn) /I'tljn/ Italian im- and in- (in
checked syllables): (m'ph;<TT, -<T)a (-;T- -T)b /Im'pO:tt/ important (n-
'sT;) /In'stEd/ inead In some American pronunciations, all these words may
have either /I/ or //.
As far as // is concerned, Oceanian and South African English are more like
American English.
Usually, // is unstressed, being the weak vowel par excellence. ere are two
forms, however, that are very often heard even with stressed // (except in formal
pronunciation): ('chD<n, 'ch-, 'ch-, 'ch-)a (->-)b /'cId<n, 'c-, 'cU-/ >il-
dren ('DzT, 'DT, -) /'dzt, 'd-/ doesn't (this is given in an example in 2.6.4,
too). Let us consider also the possibility of restressing for emphasis, as in: Oh, I
didn't buy it: it wasn't the ('I;i, ') dre, it was ju a (';I, ') 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: 2.6. It is useful to
make regular reference to it, in order to thoroughly understand the English conso-
nantal system.
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 nal
positions: ('n;Im) /'nEIm/ name, ('n:) /'In/ `in, (':) /'I/ `i, ('b;) /'b/
ba, (b'khm)a (b'khm)b /b'kmI/ becomi, ('s) /'sII/ sii,
('lO: 'g;, 'lA;)a ('l; 'g;)b /'l; 'gOU/ lo ago, ('sT<[k])a ('s>-)b
/'st<EI/ re`eni
In unstressed syllables, after /t, d , s, z/, there is a typical intense (syllabic)
realization of /n/ {() //; after /, /, we nd ((2))}. It is denitely worthwhile to
use it in (dia)phonemic transcriptions, as well, although clearly its origin is /n/:
('khAT, -)a ('khT, -)b /'kt/ cotton, ('wD) /'wUd/ wooden, (';)a (';)b
((-2)) /':/ ear`en, ('m:< 'T)a ('m:)b ((2-)) /'mO: n't/ more `an `at,
('f[;]s)a ('fA;s)b /'f;s/ faen, ('chz)a ('chz)b /'cOUz/ >osen After /0t,
0d/, it is best to use /n/: ('wnsTn) /'wInstn/ Winon ('hDn)a ('h-)b
/'hOUdn/ Holden
Furthermore, we have /n/ (= (n)a ()b) after /S, Z/: (ph<&nnsi'ISn)a (--, -S)b
/p<nnsi'EISn/ pronunciation (ph'<Zn)a (->Z)b /p'<IZn/ Parisian In a prein-
toneme, (n) can easily become (), chiey after /S, Z c, G/.
After other consonants, /n/ (n, n) is more usual: ('l~cn)a ('l-)b /'lncn/
lun>eon (<'lGn)a (>-)b /<'lIGn/ religion ('<bn)a ('>bn)b /'<Ibn/ ribbon,
('~jn)a ('~jn)b /'njn/ onion, ('ThIkn) /'tEIkn/ taken
en speed is higher, especially in a preintoneme, we can go as far as (c, G
p, b k, g): ('l~cn, -c)a ('l-)b (<'lGn, -G)a (>-)b ('pn, -p)a
('-)b /'OUpn/ open ('<bn, -b)a ('>-)b ('ThIkn, -k), (';<gn, -g)a (';g-)b
/'O:gn/ organ On the other hand, when speed is lower, or for emphasis, // can
easily become (n): ('ls, -sn) /'lIs/ lien
Regularly, we have: (n[D]n) /n[d]n/: ('lnn) /'lEnn/ Lennon, ('lnDn)a ('ln-)b
/'lndn/ London but, ('wAnT, -n, -nTn, 'wO;n-)a ('wnTn -n)b /'w;ntn/
wanton
2.2.1.2. Aimilation 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;im)a ('>-)b
/'d<Iimd/ dreamed, ('D<mT)a ('>-)b /'d<Emt/ dreamt, ('sm&Tham)a ('sm-)b /'sm-
taEmz/ sometimes, ('s;Im 'kha;n) /'sEIm 'kaEnd/ same kind, (sm'Gli) /sm'GEli/
some jelI ('s;IM 'vs, -m '6s) /'sEIm 'vOEs/ same voice, ('s;IM 'fkT, -M] 'f-, -m
'kT) /'sEIm 'fkt/ same fact, ('khMfT, -M]f-, -m-)a ('khMfT, -M]f-, -m-)b
/'kmft/ 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): (m, 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: ('w;<m, -mp)a ('w;m-)b /'wO:m/ warm`, ('mTi, 'mpTi) /'Em-
ti/ emp, ('sms, 'smps) /'sIms/ Sim(p)son.
Even // resists well: ('<O:, '<A;)a ('>;)b /'<;d/ wroed, ('w&bk) /'wI-
bk/ wiba$ Only for the sux -i () /I/ (but at a non-neutral level, not to
be followed), can we have /In, n, /: ('sl -n -n) /'sElI/ selli. We can also
nd: ('l, -k) /'lE/ le` ('khsTn, -ks-, -z-, -T) /'kInstn/ Kion (-
'za[i, g-)a (-Ti)b /'zaEi/ anxie, ('Ss, 'k-) /'Ss/ 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'blk) /Im'blk/ in bla$ (M-
'vaT) /In'vaEt/ invite (M'fkT M]-) /In'fEkt/ infect ('bAks, -)a (-ks)b /In-
'bks/ in `e box ('Th -t) /'tEn/ ten` ('Thns -ts) /'tEns/ tense ('phns,
-ts)a (-)b /'pIns/ pincer (for (()) see below).
For some time, /nz/ has been simplied (but /ndz/ can always be restored, ac-
cording to current spelling): ('wnz, -ndz)a (-)b /'wInz/ Windsor ('lnzi, -dzi)
/'lInzi/ Linds\ Even in /nz/, a /d/ can be inserted: ('khln:, -n:d) /'klEnz/ cleanse
('khlnz, -ndz) /'klEnzI/ cleansi. But this is less and less recommendable,
specially with grammemes: ('Thjn -nd -) /I'tljnz, -/ Italians ('GA;n
-nd)a ('G;-)b /'Gnz/ John's
Furthermore, we have (~) (postalveopalatal, ((n))): ('~c) /'Inc/ in> (~'chan)a
(-)b /In'caEn/ in ina ('I~G) /'EInG/ ael ('Th~Sn -~cn -~+Sn)a (-~S
-~c -~+S)b attention But we nd (~) (prepalatal) before /j/ (if this is hetero-
syllabic): (khm'ph~jn) /km'pnjn/ companion ('~jn)a ('-)b /'njn/ onion
Besides: ('hn<i)a ('hN>i)b /'hEn<i/ Henry (n'Th<nsk)a (N'h>-)b /In't<InsIk/ in-
trinsic ('Th 'g:, 'g;)a ('g:)b /'tEn 'g:z/ ten girls . Also (m'wn[m])a
(-nT)b /In'wIn/ in winter should be noticed.
A better transcription for (~c, ~G, ~+S) would be ((nc, nG, ntS)), with ((n)) and
even ((t)) (for a postalveopalatal stop). It was stated above that dental (()) could
be represented with (n), as well, chiey 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'n;&Tha;m)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 dierent auditorily, so more precise symbols could even be avoided, with
no real drawbacks.
In lexical composition, as well as for the negative prex 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, (); labiodental, (); velar, (n): (&mb'lIivb, &-)a (&mb-, &-)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): (Th'D;I)
/t'dEI/ todZ, ('DsT&TUuT)a ('DsT&TjuT)b /'dEsttUut/ deitute However, before /,
s, z/, /t, d/ become dental (t, d): ('It) /'EIt/ eigh`, ('w) /'wId/ wid`,
('hts) /'hts/ ha_, ('h;d) /'hEdz/ heads
On the contrary, before /</ (>)b, in British English, /t, d/ become postalveolar,
always due to assimilation: ('Th<a;)a ('h>a;)b /'t<aE/ try, ('khnT<i)a ('khN>i)b
/'knt<i/ country, ('D<k)a ('>k)b /'d<Ik/ drink
For /t<, d</, however, several pronunciations are possible, mostly with /t, d/ real-
ized as stop-strictives (or aricates): in American English ([h]<, < C[h]<, <
c[h]<, G<)a; in British English ([h]>, > [h]>, "> C[h]>, > c[h]>, G>)b.
Furthermore, a (homorganic) constrictive realization is possible for /</ (which is
then, respectively, alveolar or postalveolar rounded, (, )): (T[h], D)a ([h], )b.
So: ('Th<a;, 'h<-, 'Ch<-, 'ch<-, 'Th-)a ('h>a;, 'h>-, 'h>-, 'Ch>-, 'ch>-, 'h-)b /'t<aE/
try, ('khnT<i, -<i, -C<i, -c<i, -Ti)a ('khN>i, ->i, ->i, -C>i, -c>i, -i)b /'knt<i/
country, ('D<k, '<-, '<-, 'G<-, 'D-)a ('>k, '>-, '">-, '>-, 'G>-, '-)b /'d<Ik/
drink
All these pronunciations are possible as neutral ones too, although opinions re-
garding their correctness may be inuenced by spelling. On the other hand, from
a structural point of view, /c<, G</ could represent a tting parallelism with /S</,
as in ('S<k)a ('S>-)b /'S<Ik/ rink
2.2.2.2. Some other transformations of /t, d/ are much more signicant; indeed,
although to foreigners (T, D) might seem more marked, 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 complicate things, but rather to make them easier.
So, /t, d/, before /p, b, m, w/, generally, become (p, b): (&p'm;n) /t'mn/
`at man (&p'b;) /t'bOE/ `at boy ('nAp 'wn:)a ('np 'wn:)b /'nt 'wn/ not
one; likewise, before /k, g/, they become (k, g): ('wAk k~j'DU;u, 'w-)a ('wk k~-
j'D;u)b /'wt knj'dUu/ what can you do? (&k'g:, -;)a (-:)b /t'g:/ `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): ('khp, -]f)a ('kh-)b /'kpfU/ cupful, ('Ab6is, 'A7v-)a
('-)b /'bvis/ obvious
Prevelar articulations, which are automatic by coarticulation, need not be ex-
pressly written down: ('k, 'gT) {((', 'T))} /'IkI, '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 aspirated (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/: (Th-
'D;I) /t'dEI/ todZ, ('phIik) /'pIik/ peak (but: ('spIik) /'spIik/ speak] ('ThIk) /'tEIk/
take (but we have: ('sTIk) /'stEIk/ ake] ('kh;)a ('kh;)b /'kE/ care (but: ('sk;)a
('sk;)b /'skE/ scare]. Nevertheless, one should notice: (ms'ph<n[[])a (ms-
'ph>nT)b /mIs'p<Ind/ misprinted (with dierent phono-syllables and dierent
morphemes).
American t // ([, m)
2.2.3.2. Let us now turn to the contexts where /t/ (T) becomes // ([, m), in nor-
mal (not slow, nor particularly careful) speaking.
Between a stressed (or unstressed) vowel and another vowel, or (, ): ('b[i)a
('bTi)b /'bEi/ Bet, ('vIi[)a ('vIiT)b /'vIiOU/ veto, (&vz'bl[i)a (-Ti)b /vIz'bIl-
2. english 63
Unexplosion
2.2.4.1. Notably, in English, stops (both voiced and voiceless) are unreleased,
chiey after /[m, n, , , ]/, before pauses or consonants. is means that their
third phase ( their o-glide) is inaudible, incomplete. It is useful to put the dia-
critic () after a proper symbol, to show unreleased contoids, especially at rst 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 /, /.
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 ($): ('bA;)a ('b;)b /'bb/ Bob ('khp)a ('khp)b
/'kp/ cup ('khmp) /'kmp/ camp ('hT) /'ht/ hat ('h;n) /'hnd/ hand
('<;)a ('>;)b /'<Ed/ red ('S;<T)a ('S;T)b /'SO:t/ ort ('fO;T, 'fAT)a ('f;T, 'fT)b
/'fO;t/ fault ('blk) /'blk/ bla$ ('bk) /'bk/ bank ('D;) /'dIg/ dig. en
said in isolation, as plain examples, before a pause, they are released\ ('bA;$)a
('b;$)b ('h;n$) ('S;<T$)a ('S;T$)b ('bk$), .
Before a consonant, they are unreleased: ('kT) /'kt/ act ('kT)a (-)b /'kt/
actor ('<;b)a ('>;b)b /'<bd/ rubbed, ('Tkns, 'kk-) /'tkIns/ Atkinson,
64 a handbook of pronunciation
('b;g) /'bEgd/ begged ('sTApT)a ('sTpT)b /'stpt/ opped, ('hAT&dA, -O;, --)a ('hT-
D, --)b /'htd;g/ hotdog ('sTAp 'DUu)a ('sTp 'Du)b /'stp 'dUuI/ op doi
More examples: ('Gk 'phi;I, -iI) /'Gk 'pIi/ Ja$ Peel, ('bA;b 'g;)a ('b;b
'g;)b /'bb 'gOUz/ Bob goes ('skATln)a ('skT-)b /'sktlnd/ Scotland (p'bats
T'b-) /It'baEts/ it bites (T'ThIks, 'Th-) /It'tEIks/ it takes (k'kh<i T'kh-)a (->-)b
/It'k<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 better). In fact, for /t, d/ (+ /p,
b k, g/, in addition to (pp, pb p, bb kk, kg k, gg)), the alveolar contact can
be maintained, by adding either a bilabial, (B, ), or a velar, (, ), coarticulation.
A bilabial or velar articulation is also possible, to which an alveolar coarticulation
can be added: respectively (Q, ) or ($, ).
2.2.4.2. e so-called nasal and lateral explosions are included in this group.
It is essential that transitions from /t, d/ to /n, l, / are direct, with no o-glide
similar to aspiration, and even with no insertion of vocoids.
erefore, we have: ('phTni) /'pItni/ Pitn\, ('wD) /'wUd/ wooden, (khm-
'phlIiTli, -li) /km'plIitli/ completeI, ('l[)a ('lT)b /'lI/ little, ('swnD) /'swInd/
~indle.
Laryngeal stop ()
2.2.5.1. It is a good thing to include the laryngeal (or glottal) stop () in the
symbols inventory of the English phonological system, even if, strictly speaking,
there are no (classical) minimal pairs, in order to declare its phonological status.
e fact is that it is important, too, to have () from a descriptive and teaching
point of view.
In the (American and British) neutral pronunciation () is used, when there is
some emphasis, before vowels, especially stressed ones: (ts'O;f)a (-';-)b /Its'[]O:f/
it's Yful! In British pronunciation, () may be used even to avoid the insertion of
a non-etymological /</ (intrusive, at the end of 2.2.9.4): ('l: n';D) /'lO: n[d]-
'O:d/ lY and order instead of the frequent ('l:> n';D) (('lO: n';<D)a).
--, -T-)a (-:)b /'fUtbO:/ football, ('khk&gT, --, -T-)a (--)b /'ktgt/ catgut ('at ';,
-)a (-;)b /'aOt 'E/ out `ere, ('nT&SE, --)a ('nT&S, --)b /'ntSE/ nutell
Let us also notice: (a'kh[;]nT {'DUuT}, -n, -n {--, -})a (-A;n- {-uT, -})b /aE-
'k;nt {'dUuIt}/ I can't [do it] (wi'wnT {'DUuT}, -n, -n {--, -})a (-n- {-uT, -})b
/wi'wOUnt {'dUuIt}/ we won't [do it] In absolute nal position, () is not used, ex-
cept for /It, t/, and this only in informal speech: (wi'wAnT, -O;nT)a (-nT)b /wi-
'w;nt/ we want (wi'wAn[[]T, -T, -)a (-nTT, -)b /wi'w;nIt/ we want it ('ThkT,
-)a (-T, -)b /'tIkt/ ti$et.
However, the change /t/ = () is less frequent before /h/: ('IT 'hts -) /'EIt 'hts/
eight ha_ It sometimes occurs in /t/: ('khAT, -)a ('khT -)b /'kt/ cotton
In /nt/, it is more frequent (but it is only possible with /, /): ('sk<nT, -,
-[[]n)a ('sk>nTn -)b /'sk<ntn/ Scranton ('khlnT, -, -Tn)a ('khlnTn -)b
/'klIntn/ Clinton ('ET, -, -[n)a ('T, -Tn -)b /'klIntn/ Elton
/t/ = () is no neutral pronunciation, before /, , /: ('b[i)a ('bTi)b /'bEi/ Bet,
('bm)a ('bT)b /'bE/ better, ('l[)a ('lT)b /'lI/ little (('bi), ('b)a, ('b)b,
('l)). Nevertheless, it can occur in /t'/: (&nAT'Iivn, &nA['-, &nA'-)a ('n-)b /nt'Ii-
vn/ not even It is accepted even for it before a pause: ('ThIkT, -) /'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): ('sp 'phaD, -)a ('s-, -D)b /'sOUp 'paOd/ soap
powder, ('bk&khIs, -&kh-) /'bUkkEIs/ bookcase, ('bk 'gA;<D, -)a ('gA;D)b /'bk
'gA:d/ ba$ garden
British glottalization
Lenitions
/ ladder ( ('lm) /'l/ latter which may have a slight dierence in duration,
() shorter than a semi-chrone, (;)), ('<aD '<am) /'<aEd/ rider ( ('<am) /'<aE/
writer] ('bDi 'b[i) /'bEdi/ beddy ( ('b[i) /'bEi/ Bet] However, such a pronun-
ciation may not be considered neutral, though very widespread, but only mediat-
ic.
Another (and neutral) way to keep a dierence, partially recovers the voiceless-
ness of //: ('lb '<ab 'bAi), or better still totally: ('lr '<ar 'bi) (as men-
tioned and illustrated above), with no lengthening of the vocoid before /d/ ([, m).
In informal Briti pronunciation, chiey in monosyllables of low semantic val-
ue, /t/ may be realized as ([): (Th&gT'ph>zT, -&g[-) /tgE'p<Ezt/ to get a
present (&gT'mc, &g[-) /g'mc/ got a mab (&khwaT&bT'h>fk,
&khwa[&b[-) /kwaEbIv't<fIk/ quite a bit of trac (T'kh;Tmi, [-) /It-
'k:tmi/ it occurred to me (&s;T*'G;, &s;[*-) /sO:v'Gb/ sort of job ('wT -
'phTi, 'w[ -) /w'pIi/ what a pi ('nT nli's, 'n[ n-) /nOUnli'Is/ not
onI `is (bT';s 'T, b['-) /b'O:sOU 't/ but also `at (bTa'D;u, b[a-)
/baE'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 aricate, in favor of a more descriptive
and tangible (even checkable) term, such as preopped conrictive, which we will
presently reduce to op-rictive, after explaining that they are unitary phones, or
sounds, 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 rst half of a stop-strictive consonantal phone is a short stop, while
its second half is a short fricative one (or, better, a constrictive one). e place
of articulation is determined by the second component, to which the rst 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 monographed symbols. Of course, the second one is the more spe-
cic, so the rst 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 suciently 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 aspirated, as /p, t, k/ are (although most
68 a handbook of pronunciation
native phoneticians do not say that, in the least): ('chmni) /'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;)a ('G;)b /'GG/
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$)): ('wAc$ 'khfi 'wO;c$)a
('wc$ 'khfi)b /'w;c 'kEfli/ wab carefulI ('wc$ 'chI;i 'Wc$ 'hw-) /'wIc
'cIiz/ whi> >eese ('lA:<$ 'ch;c)a ('lA:$ 'ch;c)b /'lA:G 'c:c/ a large >ur>.
e only possible reduction may be in changing the rst stop-strictive with the
corresponding constrictive (fricative) one, (S, Z): ('wS 'chI;i 'WS 'hw-), ('lA:<
'ch;c)a ('lA: '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;<~G 'GUus, -~Z, 'A<-)a ('>~G
'Gus, -~Z)b /';<nG 'GUus/ orae 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: ('fc, 'f, 'fc) /'fEc/ feb,
('b~c, -~, -~c) /'bEnc/ ben>, ('s;cT, -T, -cT) /'s:ct/ sear>ed, ('s;cmi, -mi,
-cmi) /'s:cmi/ sear> me, ('fcT, -T, -cT) /'fEcIt/ feb it, ('Thic, -, -c)
/'tIic/ tea>er.
nally, chiey in British English, /c/ may become (S), before a pause or a C\
('fS, 'b~S, 's;ST, 's;Smi) (examples we have already seen) and ('khc 'khS)
/'kc/ cab (5wc 'bk3 3 5wS 'bk3 3) /'wIc 'bUk/ whi> book?
2.2.8.1. Also for this manner of articulation (as for the stop-strictive one, rather
than aricate), 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/: (a'kh[;]m
b'lI;iv&T, -iV&-, -i&-)a (-A;m b-)b /aE'k;n b'lIivt/ I can't believe `at ('gvmi
'fa;, 'gVmi, 'gmi) /'gIvmi 'faEv/ give me ve! (a'S;<TDT, aV-, a-)a (-;T-)b
/aEv'SO:tdt/ I've ortened it (in spite of an information loss in comparison with
I ortened it), ('la;vz 'Th<b, -Vz, -z, -)a (-, 'Th>-)b /E'laEvz 'tE<-
b/ `eir lives are terrible (in spite of the ambiguity with `eir lies are terrible),
('f:2 'fa;2 'sks2 'sv2 'IT2 'fa;V2 'sk2 'sb2 'fa;2 'sk2 's2)a ('f:2)b /'fO: 'faEv 'sIks
'sEv 'EIt/ four, ve, six, seven, eight And, let us note: ('G 'phT, 'fa;6 'bts)
/'GEf 'pIt, 'faEv 'bIts/ Je Pit ve bits.
/, / are slit dental (whereas in American English a prodental or interdental
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 dierent, the ocial sym-
bols, (, ), may be sucient. It is important for foreigners to acquire this slit ar-
ticulation well: ('k) /'Ik/ `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: (&k)a (-)b; the same can occur for the //, (), of grammemes: `e `is
Besides, in quick informal speech, // in forms such as `e `at `is bo` is
normally assimilated: ('wts z'Tha;m, 'wAts)a ('wts)b /'wts 'taEm/ what's `e
time?, ('wts 'zT, 'wAts)a ('wts)b /'wts 't/ what's `at?, ('bs 'sa;d)a ('bs)b
/'bOU 'saEdz/ bo` sides (hiz'wIm, hiz-, hizz-)a (-T)b /hiz'wIt/ he's `e
waiter ('m;<n, -, -)a (-;n-)b /In'mO:nI/ in `e morni ('O: -
'Tha;m, 'O:-, 'O:l-)a (':)b /'O: 'taEm/ all `e time
2.2.8.2. /s, z/ are grooved dental constrictives, usually pronounced with the tip
of the tongue raised, denti-alveolar, so that they could be transcribed with (s,
z), especially for comparative purposes, in order to emphasize the dierence be-
tween (s, z), pronounced with the tip of the tongue lowered. On the other hand,
native speakers themselves may indierently have one articulation or another, even
vacillating, so the plain symbols can safely be used: ('s;Im) /'sEIm/ same, ('<z)a
('>z)b /'<OUzz/ roses, ('sz)a ('sz)b /'sIzz/ sciors
For word-initial sm- sn- sl- (as well as for the non-autochthonous sr-), /s/ (s) is
normal (contrary to some other languages): ('smO:)a ('sm:)b /'smO:/ small
('snIk) /'snEIk/ snake, ('slIip) /'slIip/ sleep ((&s<Ii'lA;k, s<-, &S<Ii-)a (-k)b /s<Ii'lA;-
k, s<I-, S<Ii-/ Sri Lanka).
For dis- followed by a voiced stop, there are several possibilities, both phonetic
and phonemic: (&Dsb'lI;i, -s-) /dIsb'lIiv/ disbelieve (Ds'D;In, -s'-, -'D-, -z'D-, -'sT-)
/dIs'dEIn, -z'd-, -s't-/ disdain (Ds'gsT, -s'-, -'g-, -z'g-, -'sk-)a (-sT)b /dIs'gst, -z'g-, -'sk-/
disgu (Ds'ga;, -s'-, -'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 (('Sp))b /'Sps/ ops (('k)) /'ks/ `anks
(('bE:D))a (('b:D))b /'bEz/ bells (('hn:dD)) /'hndz/ hands. But it is sucient to tran-
scribe: ('SAps)a ('Sps)b, ('ks, 'bE:, 'h;nd)a ('b:)b.
In American English, the sequences /s, z/ are realized as (<, <) (with alveolar
constrictives): ('f;T) /'f:st/ r ('ph;) /'p:s/ purse (h'k;T) /h'sk:t/ her skirt
(f'mO: 'phIip) /f'smO: 'pIip/ for small people ('D:<) /'dO:z/ doors ('f:) /'f:z/
furs In British English, /s<, z</ generally become (>, >): ('nju&>iI) /'nUuz<Ii/
n[sreel.
More often, /S Z/ are produced with the tip of the tongue raised, as apico-
-postalveo-palatal rounded contoids, so that more suitable symbols, contrastive-
ly, could be ((x, )). But, as several natives pronounce them with the tip of the
tongue lowered, (S, Z) will be sucient. 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]: ('Sp) /'SIp/ ip, ('phlZ)a (-)b /'plEZ/ 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 (-p)b /Is'Sp/ `is op, (S'GA:<)a (-A:)b /Is-
'GA:/ `is jar, (S'jI;)a (-;)b /Is'jI/ `is year, (Ii'SU;u)a (-;u)b /Iiz'SUuz/ `ese
oes, ('hSi, 'hSSi) /'hzSi/ has e? even /sc/ = (Sc) is possible: ('khwscn, -Sc-)
/'kwEscn/ queion
Generally, with you your there is a complete fusion between elements: (a'mS,
-)a (a'mS -)b /aE'mIsju/ I mi you, (hi'nIiD Z'hEp, -G - -U-)a (Z'hp, -G -
-;-)b /hi'nIidz jK:'hEp/ he needs your help, (Z's;I, - &-, --) /zju'sEI/ as you sZ.
Let us also consider: ('ZT -) /z'jEt/ as yet. (Making use of diaphonemes like /,
K:/ may help in reducing the space of a phonemic transcription, chiey 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) uninuential raising of the tip of the tongue towards the
postalveolar region. Instead, the British sound is decidedly postalveolar, (>), in the
specic meaning of an area after the alveolar one, approached by the tip of the
tongue (not by the lamina, as in the IPA ocial 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 ( 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 dierent.
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 dierent auditorily, while remaining articulatorily ra-
ther dierent.
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 (>, >).
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, (>, >)
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 poalveolar corresponds to the ocial one
retroex, 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 modication.
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 zero, as r is pronounced only before vowels:
('<I;)a ('>;)b /'<I/ rear ('<;)a ('>;)b /'<E/ rare ('<:<)a ('>:)b /'<O:/ roar.
In American pronunciation, //, preceded by vowels or consonants, is realized
as (). It is the same even for /</ (and, by and large, for /</); /:, :<, ;</ are real-
ized as (:): ('m;D)a ('m;D)b /'m:d/ murder ('m;D<, -D)a ('m;D>, ->)b
/'m:d<, -d</ murderer Also (<, :<)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
dierence between (, ), and the actual fact that, before (a heterosyllabic) /j/, not
only does () not occur (as, instead, it does before any other consonants) but it is
prepalatal, (), ('m-jn). In other contexts, it is alveolar, (l), as in: ('lUusaT)a ('lu-,
'lju-)b /'lUusaEt, 'l-/ 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 ()) is used: ('fAl-i)a ('fl-i)b /'fli/ folI ('fAl-)a ('fl-)b /'flOU/
follow ('Thl-m) /'tElIm/ tell him ('fIilT) /'fIilIt/ feel it, ('fO:l 'aT)a ('f:l)b /'fO:l 'aOt/
fall out ('bl 'vn) /'bIl 'Evnz/ 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, /; although (to be precise) it would be useful to
use at least four symbols, (l, , ], ) (besides () in British {and British-like} pronun-
ciation, before /<, t<, d</), in addition to three intense ones, (, , ). Furthermore,
at least in specic descriptions, the taxophones with dental contact, ((l, )), should
be indicated, and also those with lip-rounding after V or C which have a labial
component): ((t, , 3, , , )). 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;i) /'lIiv/ leave, ('l[;]sT)a ('lA;sT)b /'l;st/ la,
('lk) /'lUk/ look, ('l;n) /'lOEn/ loin. To be rigorous, lip-rounding also occurs be-
fore rounded V, by coarticulation: (('tk, 't;n)); 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 strange, exactly as for /k, g/ followed either by front V or
by /j/, or else by rounded V or by /w/: the articulations ((, )) and ((k, g)), respec-
tively, are natural and automatic: (('hT)) ('khT) /'kIt/ kit, (('T)) ('gT) /'gEt/ get,
(('j;u)) ('khj;u) /'kjUub/ cube, (('kW:<))a ('kh:<) (-:)b /'kO:/ core, (('gUus))a
(('gus))b ('g-) /'gUus/ goose, (('kWaT)) ('kwaT) /'kwaEt/ quite.
erefore, in particular for //, an objective pronunciation, obtained by juxta-
posing /p/ and //, for instance, would produce an eect 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 //: ('fi -ti) ((-ti))
/'fIi/ l`y ('fO;s -ts, 'fA-)a ('f;s -ts, 'f-)b ((-ts)) /'fO;s/ false ('wES -TS, -c)a
('wS -TS, -c)b /'wES/ Wel is occurs with /z/, as well, but more rarely: ('bE:
-:d)a ('b: -:d)b /'bEz/ bells ('fi;I -d, 'fiI-) /'fIiz/ feels. On the other hand,
the reverse simplication may be heard, too, as in: ('fi;Id -, 'fiI-) ((-)) /'fIidz/
elds ('bEts, -s)a ('bts, -s)b ((-s)) /'bEts/ 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 ()) 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 dierently 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 //: /vEI'kEISn/a /v-/b vacation /nOU'vEmb/a /n-/b novem-
ber Even the article a often has a full timbre (which could sound too formal in
British pronunciation): /z'gUd '<Iiz, -zEI-/a /-z-/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 sux
-ile is /- -aE/a /-aE/b: /'hstA/ ('hAsT, -a)a ('hsTa)b hoile (and mobile as
well), /'f:A/ ('f;[ -a)a ('f;Ta)b fertile /'mIsA/ ('ms)a ('msa)b miile
(and reptile erile] But we have: /'ds -aE/a /'dOUsaE, 'dsaE/b docile and /'GUu-
vn, -naE/a /-naE, -n/b juvenile. We also nd: /'p<-/a /'p<OU-/b pro-: /'p<sEsI,
--/a /'p<OU-, 'p<-/b procei, /'p<g<Es, -s/a /'p<OU-, 'p<-/b progre (but: /'p<OU-
g<m/ programa, -mmeb).
To end with, let us see: /sIvl'zEISn -laE-/a /-laE-, -l-/b civilization, /'kpl<i/a
/k'pIl<i/b capillary, /'kO:<l<i, 'k-/a /k'<l<i/b corollary, /'lb<tK:<i/a /l'b<-
t<i/b laboratory, /'nOUbdi, -b-, -b-/a ('n&bDi, -&bADi, -bDi)a /--, --/b ('n&b-
Di, -bDi)b nobody (and aJ- every- some-] Let us also notice this (seemingly
curious) word: /'aEn/a ('an, 'a;<n)a /'aEn/b ('an)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 rst. 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 // (A)a ()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 /, , I/: (m'nA[ns)a (-'nT-)b /m'nns/
monotonous, (ph<'vn[[])a (ph>'vnT)b /p<'vnd/ prevented, ('Dm)a (--
T)b /'Edz/ editors, ('mnsT)a (-sT)b /'mInst/ minier 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: ('khMf[b, -fT-, -f- -mT-)a (-MfT-, -fT-, -f- -mT-)b /'kn-
fb, -mf-/ comfortable, ('mA;<g<T, -T)a ('mA;g>T, -T)b /'mA:g<t/ Margaret,
(&jun'v;[i, -Ti)a (-'v;sTi, -sTi)b /jUun'v:si/ universi
However, not all unstressed syllables have vowel reduction or fall: ('khAmnT)a
('kh-)b /'kmEnt/ comment, ('sfO;T, -A;T)a /'sfO:t, -A:t/ (-T, -;T)b /'sft,
-O:t/ as#alt, ('khUupAn 'khj-)a ('khupn)b /'kUupn/ 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 weak forms) 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] ([bbn'Dn:, -Db-, -n- -n- Th-)a (T-, -n-, -Iin-, -n:)b it had been
done [it'd] ('m;n g'gO:n, D-, -A;n)a (-;n)b `e man had gone (Si'h T-
'sT;I, -T) /Si'hd t'stEI/ e had to Z (+ (-[ -)a (-T -)b /- -/);
has\ (hzi'gO:n, h-, -A;n)a (-;n)b has he gone? (Siz'Dn:, Siz- Sihz-)a (-n:)b e
has done [e's] (ts'bn:, [z- thz-)a (ts-, -I;in, -n:, Tz-)b it has been [it's]
('G:<G 'khm:, s- - h-)a ('G:G, -m:)b George has come (hi'h T-
'sT;I, -s) he has to Z
have\ (hvj'sIinT, h-, -j-) have you seen it? (Iv'gO:n, -A;n I- -h-)a
(-;n)b `\ have gone [`\'ve] (wi'khm:, -f'k- wi- wih-)a (-m:)b we have
come [we've] (wi'h T'sT;I, -f) /wi'hv t'stEI/ we have to Z (j'SD,
j-, -, -&h) you ould have (j'SD 'Dn:, -*, -v, j-)a (-n:)b you ould
have done [ould 've] + (-'kh-, -'w-, -'msT-a/--b, -'mI-, -'ma[-a/-T-b) could
would mu mZ might
he\ (hi'wnT) he went ('wni 'sO;T, w-, w-)a ('s;T)b when he sY it ('hzi) has
he?
her\ (h'a;)a (h>'-)b her \es ('Thl)a (-)b tell her ('gv 'hT)a (->-, - h-)b
give her her hat (Th'm, Thh-, Thw-)a (Thh'm, Th-, Th-, Thw-)b
to her mo`er
him\ (a'sO;m -m)a (-;m)b I sY him ('l[m 'n: -m -)a ('lT-)b let him in
his\ (h'phn:) his pen (hi'Thk z'bk) he took his book (hZ'ju) his you`
I\ (a'sI;i, -)a (a-, -)b I see (a'ThIk, ,- -- -- -w-)a (a-, - -- --
-w-)b I will take [I'll] ('ha; kDa'sT;I, -- --)a (-a-, -- --)b how could I
Z?
if\ (fa'm;I, f-, -) if I mZ (fj's;I&s, f-, -, --)a (-)b if you sZ so (f'nA
f'j;u, f-, -, -T)a (-n f'-)b if not for you
in\ (n'lnDn)a (--)b in London (m'ph<s)a (->s)b in Paris ('khnD)a
(-)b in Canada (hi's[ n'ch;, -, nn -T - -T n-)a (-T, -;)b he
sat in `e >air ('b<Ik[ n'ThU;u -T -)a ('b>-, -T, -;u)b break it in o (a-
mn'h;i, - -mn- am-)a (->i, -)b I am in a hurry [I'm]
is\ (z'j;u, -T'j- -) is it you? (ts'mI;i [z-)a (ts- Tz-)b it is me [it's] ('khm
'hI; -m -)a ('h;)b Kim is here [Kim's] ('s z'fa;n 'D;I, 's '- 's s'-) `is
is a ne dZ, ('<;z z'nas 'wmn, '<;z '-, '<;z z'-)a ('>;z)b Rose is a
nice woman, (s'DS z'waT, '-, s'-, s-) `is di is white
it\ (z'Th<U;u, z-, , -T'T-)a (-'h>;u, -'-)b is it true? (ts'O: '<aT, ts'-, s'- [z-)a
(-: '>- Tz-)b it is all right [it's] ('js ['; j'sT;)a (T'; j'sT;)b yes it is
('ThIkT, -T)a (-T)b take it ([bi'g; ':, T- pw- Tw-)a (T-)b it will
be a good `i [it'll] ([bbi'nas, Tw-, -Db- pw- Tw-)a (T-, )b it would
be nice [it'd]
i_\ (ts'Th;) i_ tail (T'h;D ts'ThIi 'b<kn, ts-)a ('b>-)b it had i_ tee`
broken
ju\ (avGs[T]'sIinm, -Gs-) (-G-, -G-)b I have ju seen him [I've] (Siz'Gs-
gO:n 'w;I, -s-, -A;n)a (--, --, -;n)b e has ju gone YZ [e's] (Gs'lak
'sn:, Gs-, -T'l-)a (Gs-, -n:)b ju like his son
maJ\ ('mni 'b;) maJ boys ('ha; mni'm:<, mni-, mni-)a (-:)b how
maJ more?
2. english 81
-m, -m, -m)b `e r time I went `ere ('<I;i&Tham 'f:< 'ThwE:, -T-,
-T-, -T-)a ('>-, 'f:> -, --, --, --, -:)b `ree times four are elve
to\ (Th'lnDn)a (--)b to London (Th'sku;U) to (ool (Th'gln, 'Thw-) to
Eland (Th'j;u, Th'j-, Thi'j-) to you (Th';n) to Ann ('khmTT, -TwT)a
('khm-)b come to it (j'hT, -fT, j-) you have to (Th'g;) to give (Th-
'wn:, Th-) to win (Th'IiT Th-)a ( Th'-)b to eat (Th'O;f, -'A- Th-)a ( Th-
'f)b to oer for to before consonants (, ) can be very short; and, be-
fore voiceless consonants, they are often devoiced: ('Thn T'fa;) ten to ve;
up\ ('mIkp j'ma;n, jU- -p-)a (-pj-, -j;- -p-)b make up your mind (z-
'wn p'; p- z-)a (z'wn p'; p- >z-)b `ere is one up `ere
[`ere's]
upon\ ('wn &pO;n'n, &pA-, &p-, p-)a ('wn &pn'n, p-)b one upon
ano`er ('la;n pO;n'la;n, -pA-, -p-, -p-)a (-p-, -p-)b line upon line
us\ ('Thls) tell us (lts'g;, ls-, ls-)a (-;)b let's go! but: ('l[s 'g;)a ('lT-
s 'g;)b let us go
was\ (awz'<O:, -A;)a (-'>;)b I was wro (hiw'f<n:)a (->-)b he was a
friend
we\ (wi'm;I) we mZ ('A;<nTwi, -mpwi, -mwi)a ('A;n-, 'A;m-)b aren't we?
were\ (Iw':)a (-w>-)b `\ were ill (wj';, -j-)a (w-, -;)b were you
`ere?
what\ ('sI;i wTjv'Dn:, wA-, w-, -TS-, -c-, -v- hw- W- -j- jh-)a (w-, w-,
)b see what you have done! [you've] (hi'nU;u w[i'wAn[[], wA-, w-, -'wO;n-
hw- W-)a (-'nj;u wTi'wnT, w-, )b he kn[ what he wanted (&wmj-
'sI, -j, &wA-, w- hw- W- -;)a (&wT-, )b what are you sZi?
[what're] (&wTDj 's;I, -TG, -TS, -c, -G, -, &wA-, w- hw- W-)a (&w-, )b
what do you sZ? [d'you] (&wTDj'DU;u, -TG, -TS, -c, -G, &wDj-, &wDj,
-, &wA-, w- hw- W-)a (&w-, )b what do you do? [d'you]
when (not interr.): (mwna'sO;T, w-, -n'-)a (-;-)b and when I sY it
(&sw~j'g ';, -w-, -T-)a (&s-, -;)b so when you get `ere
where (not interr.): ('phlIs wiw'fa;n, w-)a (w>i-, w>i-)b `e place
where he was found ('khnT<i w'phIip 's:, w-)a (-N>i w-, w-)b a
country where people si
who\ ('m;n 'DDT, h-) `e man who did it
will\ ('[ 'DU;u 'pw -Tw)a (-T 'D;u, )b `at will do [`at'll] (j'sI;i,
j-, ju- j jw) you will see [you'll] (wlp'w;k, -'w-, -T'w- wl-)a
(-;k)b will it work? ([bi'mI;i pw- Tw-)a (T-)b it will be me [it'll]
('GA;m wbi'hI;, -n bi-)a ('G;-, -;)b John will be here [ John'll] ('ch;c
[w]bi'f:, w-)a (-;c)b `e >ur> will be full
would\ (wDpbi'g;, -b-, -Tb-) would it be good? (ID'DU;u ID- -wD-)a
(-;u)b `\ would do [`\'d] (hi'khm:, -'k- hi hi- hiw-)a (-m:)b
he would come [he'd] ([bbi'nas, -Db- pw- Tw-)a (T-)b it would be nice
[it'd], ('GI;im wd's;I&s, -n d-)a (-)b Jean would sZ so
you\ (fj'DU;u, -j-)a (-;u)b if you do (a'Thj, -j, ,- a- aw-)a (-j,
-)b I will tell you [I'll] ('kj, -j)a (-j)b `ank you ('A;<~c, -)a ('A;~-
c)b aren't you? ('DDc 'sI;i, -c, -TS-) didn't you see? ('DDc 'khwT, -,
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>;bi, -b>i)b rYberry
/-b<i/ (in particular with monosyllabic roots, 2.3.5), ('khb)a ('khb)b cup-
board, ('wEkm)a ('w-)b welcome, ('snDI, -Di)a ('sn-)b sunday (often /-dEI/ in
an intoneme, but /-di/ in a preintoneme, /-dE/);
('phIim&b)a (-Tb, -b>)b Peterborough /-b<K/, ('Dm&b)a (-mb>, -b)b
Edinburgh /-b</, ('nU;u&b<i)a ('njubi, -b>i)b N[bury /-b<i/ (in particular
with monosyllabic roots, 2.3.5), ('lsT)a ('lsT)b Leicester, ('wsT)a ('wsT)b
Worcester, ('n;<fk)a ('n;fk)b Norfolk, ('Aksf)a ('ksf)b Oxford;
('khsm) Casham, ('D;m)a ('D>m)b Durham /;</ and ('khn&hm)a ('khn-
m)b Cunniham /-m/, ('glnD) Eland, ('phsmn)a ('phs-)b po-
man, ('Gn[[]mn)a (-nT-)b gentlemen, ('s<&mni)a ('s>mni)b ceremony /-mK-
ni/, ('phlm) PImou$;
('sO;s&phn)a ('s;spn)b saucepan /-pn/, ('nAnsns, -sns)a ('nnsns)b nonsense
/-sns/, ('j;<kSI, -S)a ('j;kS, -S)b Yorkshire /-SE/, ('hnsm) handsome, ('hnD-
sn)a (-D-)b Henderson, ('fksTn, -&sTn)a ('fksTn)b Folkestone /-stKn/.
Furthermore: ('la&b<<i, -b<<i, -bi, -b<i)a ('lab>>i, -b>i, bi, -b>i)b library
('khsT&m<i)a ('khsTm>i, -mi, -m>i)b cuomary (D'<kT<i, -Ti, -T<i, Da-)a
(D'>kT>i, -i, ->i, Da-)b directory ('D;<m&T<i)a ('D;m>i, -T>i)b dormitory.
Taxophonics
2.3.2.1. From the examples given thus far, the use of #onetic duration for the
various English phones will be suciently clear. However, we will summarize its
2. english 85
characteristics. In stressed syllables, the long vowels (/A:, O:, :/, and the possible
long ones from the diaphonemes /A;, ;, ;, O;, ;/) as well as the diphthongs (/Ii,
EI, aE, OE, aO, OU, Uu/) undergo a little shortening half-orteni, indeed: from
(:, ;) to (;, ) when they are followed, within the same word or rhythm
group, by at least one of the following three elements: (1) a voicele consonant, or
(2) an unreed vowel (and that changes them into diphthongs or triphthongs,
(, )), or (3) a whole unreed llable. e second elements of compounds
have secondary-stressed syllables; so they have no inuence on length.
erefore, we have: ('phl;I) /'plEI/ plZ, ('phl;I) /'plEIz/ plZs, ('phl;I) /'plEId/
plZed, ('phl;I&bk) /'plEIbk/ plZba$, but ('phlI) (or, possibly, ('phl;))
/'plEII/ plZi, ('phlI)a (-)b /'plEI/ plZer, ('phlIb) /'plEIb/ plZable, and
('phlIT) /'plEIt/ plate, ('phlIts) /'plEIts/ plates, ('phlI[)a (-T)b /'plEId/ plated,
('phlI[)a (-T)b /'plEII/ plati, ('phInT) /'pEInt/ paint
Equally: ('khA:<)a (-A:)b /'kA:/ car, ('khA:<)a (-A:)b /'kA:z/ cars, ('khA:<)a (-A:)b
/'kA:d/ card, ('khA:<D&b;<, -[b]&b-)a ('khA:D&b;, -b&b-)b /'kA:dbO:d/ cardboard, but:
('khA;<T)a (-A;T)b /'kA:t/ cart, ('khA;<bn)a (-A;b-)b /'kA:bn/ carbon, ('khA;<D)a
(-A;D-)b /'kA:dI/ cardi, ('kh[;]nT)a (-A;nT)b /'k;nt/ can't
Besides, also unstressed or half-stressed syllables shorten, as seen in cardboard (a
true compound, as to cupboard /'kbd/, that is crystallized, by now, so that a
more suitable spelling for the latter could certainly be cubbard]\ (phA;<'Ths-
pnT)a (also (ph-)a) (-A;'Ths-)b /pA:'tIspnt/ participant (phA;<'Ths&pIT)a (also
(ph-)a) (-A;'Ths-)b /pA:'tIspEIt/ participate ('m;I&DI) /'mEIdEI/ MZdZ Besides:
('wIspIp&b[;]skT)a (-&bA;skT)b /'wEIstpEIp-b;skt/ waepaper basket
ese degrees of length hold good both in intonemes and preintonemes.
2.3.2.2. For the English stressed vowels (/I, E, , , , U/), there is another inter-
esting fact about phonetic length. In syllables checked by nal 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 () to (;) ('l;) /'lId/ lid
(but ('lT) /'lIt/ lit], ('m;) /'md/ mad (but ('mT) /'mt/ mat] ('b;)a ('b;)b
/'bz/ buzz (but ('bs)a ('bs)b /'bs/ bus] ese degrees of length hold good even
in both intonemes and preintonemes.
On the other hand, when nal stressed syllables are checked by an isolated
voiced consonant ( not forming a diphonic pair that is /m, n, /), instead of
the vocoid, the contoid is a little lengthened (but only in intonemes, before paus-
es): ('Thn:) /'tEn/ ten, ('j:)a ('j:)b /'j/ you, ('b:) /'bU/ bull (but ('b&h;<n)a
(-;n)b /'bUhO:n/ bullhorn ('b '<;n)a ('>-)b /'bU '<n/ `e bull ran).
In both accents, though, there is an exception to the exception, insofar as /, /
are half-lengthened (in a preintoneme, too), instead of a following contoid:
('GA;n)a ('G;n)b /'Gn/ John, ('b;) /'b/ ba, ('h;n) /'hnd/ hand
In the sequences /I, E, U/, the rst element is half-lengthened, both in into-
nemes and preintonemes; this occurs before vowels as well, if nal in rhythm groups:
('hI;)a ('h;)b /'hI/ here, (';)a (';)b /'E/ `ere, ('phjU;, 'phj:)a ('phj;,
'phj:)b /'pjU/ pure ('hI; n';)a ('h;> n';)b /'hI< n[d]'E/ here and `ere
We should notice that, in American English, /I, E, U/ followed by vowels,
86 a handbook of pronunciation
within words or rhythm groups, become /I<, E<, U</: ('sp<T) both for /'spI<It/
('sp>T)b spear it and for /'spI<It/ ('sp>T)b spirit ('h<)a ('h>)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 (podevoicing), is very important: ('bA;)a
('b;)b /'bb/ Bob ('D;) /'dd/ dad ('G;)a ('G;)b /'GG/ judge ('v:)a
('v:)b /'v:v/ verve ('bA; 's;)a ('b;-)b /'bb 'sEz/ Bob sZs ('D; 'ThkT) /'dd
'tUkIt/ dad took it.
Of less importance is their devoicing after pauses or after voiceless consonants
(predevoicing), 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;)a ('b;)b /'bb/ Bob, .
A dot could be used even after aspirated /p, t, k, c/, but we will do that only
here, because (h) is sucient: (('ph&;I)) /'plEI/ plZ (('khwaT)) /'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): (('f;u)) /'fjUu/ f[ (('T))a
(('>-))b /'<Et/ `reat (('s2:<))a (('s2:))b /'snO:/ snore
Everyday-speech simplications
And there are combinations, too: ('lT<li, -[<-)a ('l>li, -T>-)b /'lI<li/ lite-
ralI (ph'Thkjlli, -kjli, -kli)a (ph-, -kjl- -kjl-)b /p'tIkjlli/ particularI
('ph<Abli, -bb-, -bb-, &ph<Ali)a ('ph>-)b /'p<bbli/ probabI
2.3.3.3. Some other cases: (w, w, w {,'Thj&wT, -})a ({-, --})b /wE,
w, w/ well, [I'll tell you what] (&nli'kh;, -ni-)a (--, -'-)b /IfOUnliaE'kUd/ if
onI I could ('skjuzmi, k'-) /Ik'skjUuzmi/ excuse me! (ba'ba;, b-, b-, &ba'ba;)a
(b-)b /baE'baE/ bye-bye (gb'ba;, gb-, g'-, g'- -D'b-, 'ba;) /gUd'baE/ goodbye
(gb'm;<n, gb-, g'-, g'- -D'm-, 'm;<n)a (-;n-, 'm;n-)b /gUd'mO:nI/ good
morni
Besides: ('js, 'jhs, 'jh, 'js, 'j, 'j, j, js), and ('j, 'j, 'ja:, 'jA:)a (-)b,
('j, 'jp, 'jp, 'j, j)a (-p, -, -)b /'jEs 'jE 'jEp 'jp/ yes! ('n;, 'n,
'np)a ('n-)b /'nOU/ no! (&O;'<aT, O;'-, -)a (&;-, ;-)b /O:'<aEt/ all right (khm'A;n,
-'O:n)a (-';n)b /km';n/ come on! (khm'n:) /km'In/ come in! (khm'I;, -'H-)a
(-;)b /km'hI/ come here!
More examples: (s'Da;n, s-, sT-) /sIt'daOn/ sit down (D';, aD'n;)a
(D';, aD'n;)b /aEdOUn[t]'nOU/ I don't know (D';, a&DD';)a (D-
';, a&DD';)b /aEdId[t]'nOU/ I didn't know ('spk&s, ak-)a ('spk&s)b
/aEIk'spEktsOU/ I expect so (a'sp, s-)a (-, s-)b /aEs'pOUz/ I suppose ('k
j'<aT, -)a (-, j'>-)b /aE'Ik j'<aEt/ I `ink you're right
nally: (agn'DUuT, - -&gn-, --, -O;-, -A-, -T)a (-, -uT -&gn-, --,
-;-, --)b /aEmgn'dUuIt -gOUn- -'gOUIt-/ I'm goi to do it [gonna do it] (hizg-
n'DT -&gn-, --, -O;-, -A-, -T)a (-&gn-, --, -;-, --)b /hizgnu'dIt -OU- -'gOUI-
t-/ he's goi to add it (wi&wAn's;I, -wO;-, -nT-, -n-)a (-w-)b /wiw;n'sEI/ we want
to sZ [wanna] (wi&wAn'IiT, -wO;-, -nT-)a (-nT-)b /wiw;nu'Iit/ we want to eat ('sm-
'Df<nT, 'smp)a ('sm-, ->nT, 'smp)b /'smI 'dIf<nt/ some`i dierent
88 a handbook of pronunciation
(f'f:, f'-, f'-, f'-) /fU'fI/ fulll (M'va<nmnT, -mm-, -m-, -a<m-, -,
-)a (->nm-, ->mm-, ->m-, -am-, -am-, -, -)b /In'vaE<nmnt, E-, -/ en-
vironment ('gvnmnT, -vm-, -vm-, -v-, -vm- -b- -mm-)a ('gvnm-, -vmm-,
-vm-, -v-, -vm- -b- -mm-)b /'gvnmnt/ government.
American English has, too: ('khnD&DIT, 'khn-, -DT) /'knddEIt, -dt/ candi-
date (nT'A;<kTk, n['-, n'-, -'A;<Tk) /n'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 rst //, although not
very frequently, even in stressed syllables: ('fA;<, 'fA;) /'fA:/ far`er ('f;,
'f;) /'f:/ fur`er (';<D, ';D) /'O:d/ order ('m;D, 'm;D) /'m;d/ mur-
der ('kh;<n, 'kh;n) /'kO:n/ corner ('mAmm, '-) /'mm/ `ermom-
eter ('f;<wD, 'f;wD) /'fO:wd/ forward ('f;<&w;D, 'f;&w;D) /'fO:w:d/ for[ord
(s'ph<a;, s'-) /s'p<aEz/ surprise ('gvn, -vn, -vn) /'gvn/ governor (ph-
'Thkjl, ph'-) /p'tIkjl/ particular
Also: ('khn[m]&b<i, -&b-) {('khnTbi, -b>i, -&b>i)b} /'knb<i/ Canterbury
('wO;m&b<i, -[-, 'wA-) {('w;Tbi, -b>i, -&b>i)b} /'wO;b<i/ Waterbury ('Am&b;n, 'A[-
-) /'b:n/ Otterburn ('b;n, 'b;-, b'nA:<, b-) /'b:nd, b'nA:d/ Ber-
nard ('khm&pl, -[-) /'kpIl/ caterpillar ('ED'b<i, -D&-) {('D&b>i, -bi, --
b>i)b} /'Edb<i/ elderberry ('<z&vwA;<, -z&- -&vw;<) /'<EzvwA:/ reservoir ('n;<-
<p, 'n;-) /'nO:<p/ Nor`rup ('sn, 'sn) /'sn/ Sou`erner ('m;D-
<, -D, 'm;D-) /'m:d</ murderer ('n;<n, 'n;<n, 'n;n) /'nO:n/
Nor`erner
Less systematically, dissimilation is possible even in rhythm groups: (h'hA;<T,
h'-) /h'hA:t/ her heart (j'm;i, j'-) /j'm:si/ your mercy (A;<'ph;<c, A;'-) /A:-
'pO:c/ our por> ('f:m, '-, '-, '-) /E'f:mz/ `eir rms ('D;[i, '-,
'-, '-) /E'd:i/ `\'re dirty (f'w:, f'-) /f'w:d/ for word (;<'hI;, ;'-, '-,
'-, -'H-) /O:'hI, '-/ or here
In addition to //, the following examples will show dissimilation for /</ (which
is used in British pronunciation too, due to a kind of simplication, even by anal-
ogy): ('la&b<<i, -b<<i, -bi, -b<i, -&b-, -b-)a ('lab>>i, -b>i, bi, -b>i, -b>i)b /'laE-
b<<i/ library ('fb<&<i, -b<<i, -bj-, -bj-)a ('fb>>i, ->>i, -b>i, bi, -b>i, -bj>i,
-bj-)b /'fEb<u<i, -bju-/ February ('sk<&T<i, -k-)a ('sk>>i, -&T>i, -k-)b /'sEk<-
t<i/ secretary.
In addition: (sT'nAg<f -g-)a (-g>- -g-)b /st'ng<f/ enogra#er (f-
'ThAg<f -g-)a (-g>f -g-)b /f'tg<f/ #otogra#er (ph<'fs ph- ph-)a
(ph>'fs ph-)b /p<'fEs/ profeor (ph<&nnsi'ISn ph- ph-)a (ph>&nnsi'I-
S ph-)b /p<nnsi'EISn/ pronunciation (ph<'ph<&T<i ph- 'ph<p-, 'ph<-
p<-)a (ph>'ph>>i, ph>- ph-)b /p<'p<tK<i 'p<Ep<tK:<i/ preparatory
All in all, we can see that dissimilation mostly occurs with: (';<) /'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: () // 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, /'h;<id/ hurried, /'bA:d/ barred, /'<bd/ rubbed,
/'GGd/ judged, /'plnd/ planned
/t/ (T) after voiceless consonants (except the very /t/): /'stpt/ opped, /'w;St/
waed, /'swIct/ ~ibed, /'l;ft/ laughed
/d/ (D)a (D)b after /t, d/, in order to be able to pronounce them: /'wEId/ wait-
ed, /'nIidd/ needed, /'stA: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/ ies, /'dEIz/ dZs, /'lEIdiz/
ladies, /'<bz/ rubs, /'wEIvz/ waves, /'plnz/ plans, /'Gnz/ John's, /'<Icdz/ Ri>-
ard's
/s/ (s) after voiceless consonants (except the grooved ones, /s, S, c/): /'tps/ tops,
/'<aEts/ writes, /'b<Es/ brea`s, /'GEfs/ Je's, /maE'pE<nts/ my paren_'
/z/ (z)a (z)b after /s, z S, Z c, G/, in order to be able to pronounce them: /'kIs-
z/ kies, /'<OUzz/ roses, /'dISz/ dies, /'swIcz/ ~ibes, /'ksz/ axes, /'cA:zz/
arles's
Stress
2.3.4.1. We know that (the position of) stress may be distinctive, in English:
('mp;<T)a (-;T)b import (noun, adj.), (m'ph;<T)a (-;T)b import (verb); ('ph<z-
T)a (->-)b present (noun, adj.), (p<'znT)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): ('smz 'bO;t '<I;i 'nU;u 'smO: 'blk 'khts)a ('b;t
'>I;i 'nj;u 'sm:)b Sam has bought `ree n[ small bla$ ca_ but we have: (-
w'lA:< 'kh<a;D *'phIip)a (-w'lA:)b and `ere was a large crowd of people,
(b[&ts*'g<I[sT m'ph;<Ts, -s)a (bT&-, -'g>IT-, -;-)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): (&v'sT&mIT)a (&v>'sT-)b overeimate, (&vz'bl[i)a
(-Ti)b visibili, (&ph;pn&Dkj'l<[i)a (&ph;-, ->Ti)b perpendiculari, ('<kg&na)a
('>-)b recognize, ('mks&k)a (-)b Mexico.
In compounds, the more frequent structure is ('&) (more rarely (&'): (&ff'ThI;in)
fteen). Sometimes, even ('') occurs, as in collocations (or occasional or free
compounds, which are, then, modiable): ('blk&b;)a (-;)b bla$bird (but ('blk
'b:)a ('b:)b bla$ bird), ('b&DA, -O;)a (-)b bulldog
Of course, there are also many instances like: ('glS&ThIic)a (-)b Eli tea>er
a teacher of English and ('glS 'ThIic)a (-)b Eli tea>er a teacher who is
English.
2. english 91
2.3.4.2. Let us now consider compounds such as r cla (noun and adverb)
and r-cla (adjective), and the collocation r cla in a sentence like `at was
`e r cla to be considered From a phonetic point of view, they are alike: ('f;[T]
'khl[;]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: /'f:st'kl;s/ (compounds: /''/) and /'f:st 'kl;s/ (collocation: /' '/).
Besides, patterns are exibly structured. As a matter of fact, we have: (&ff'ThI;in)
fteen and ('ph;I ff'ThI;in) page fteen but ('ff&ThIim 'ph;IG)a (-)b fteen
pages ('b<;n[D] 'nU;u)a ('nj;u)b brandn[ but ('b<;n[D]&nUu km'phjum)a
(-&nju, -T)b a brandn[ computer
Moreover: ('skn[D] 'h;n) secondhand but ('skn&hn 'khl;[])a
(-;[])b secondhand clo`es and ([]'O: &skn'h;n)a (>':)b `\'re all sec-
ondhand also (&fT'nU;un)a (&A;fT'n;un)b afternoon and (&gDfT'nU;un, gD&-,
gD&-)a (&gDA;fT'n;un, gD&-, gD&-)b good afternoon but ('fT&nU;un 'ThI;i)a ('A;f-
T&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 dierent degrees of intermediate stress, too, which (without an em-
phatic one, (")) are, in descending order: ('), (), (&), (), ( ). It is worthwhile observ-
ing nuances carefully: (&Dmn'sT<ISn 'ks&saz, 'Dmn&s-, -'sT<ISnks&saz)a
(-'sT>IS 'ks&saz)b demonration exercises ('l&vIm 'Ap&Im, 'l&vImAp&I-
m, 'lvIm&ApIm)a (-T> 'p&>IT)b elevator operator ('laT&has 'khIip, 'laT-
&haskhIip, 'laThas&khIip)a (-)b ligh`ouse keeper let us notice: ('laT 'has-
&khIip)a (-)b light housekeeper
Intonation
2.3.6. ere are several and quite varied queion tags, or tag queions in En-
glish, while, other languages generally have xed 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 dierent func-
tions: conrmations of somebody's suppositions (by means of conclusive into-
nemes), or actual queions, 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 conrmation, we will have: /Its'kOUd t'dEI. 'Is{}-
t./, /j<'mE<kn. 'A:ntjUu./, /EI'A:nt 'wE. 'A:EI./; while, for information: /Its-
'kOUd t'dEI. 'Is{t}t?/, /j<'mE<kn. 'A:ntjUu?/, /EI'A:nt 'wE. 'A:EI?/.
Other accents
f
f
('hT)i ('hT)a ('hT)b /'ht/ hut, ('hT)i ('hAT)a ('hT)b /'ht/ hot, ('s:)i ('sO:)a ('s:)b
/'sO:/ sY, (&scu'Sn)i (&sc'ISn)a (&sc'IS)b /sIcu'EISn/ situation
For /:/, we have (:): ('w:)i ('w:)a ('w:)b /'w:d/ word for /:<, ;</ (:<):
('h;<i)i ('h;i)a ('h>i)b /'h;<i/ hurry for //, in unstressed syllables, we nd ():
('b<[])i ('b<[])i (-[])a (-, -)b /'b<[z]/ bro`er[s]
For /I, E, U/ we have (;): ('h;)i ('hI;)a ('h;)b /'hI/ hear, (';)i (';)a
(';)b /'E/ `ere, ('ph;)i ('phU;)a ('ph;)b /'pU/ poor and for /I<, E<, U</ we
have (<, <, <): ('h<)i ('h<)a ('h>)b /'hI<I/ heari, ('m<i)i ('m<i)a ('m-
>i)b /'mE<i/ Mary, ('Dj<)i ('D<, 'D;-)a ('Dj>, 'Dj;>-)b /'dU<I/ duri
2.4.1.4. As to diphthongs, it is sucient to notice: /Ii, Uu, OU/ (ii, uu, ) (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, ('n;)i ('n;)a
('n;)b /'nOU/ no. Besides, for /aE, OE/, a second element like () is enough (
2.10, to avoid introducing/learning another vocoid; so much so that a pronunci-
ation with () is much better than foreign *(ai, Oi), and it is near the native
one): ('ga;)i ('ga;)a:b /'gaE/ guy, ('b;)i ('b;)a:b /'bOE/ boy Little has to be said
about /EI, aO/: ('D;)i ('D;I)a:b /'dEI/ dZ, ('na;)i (na;)a:b /'naO/ now
Neither the () taxophone of // near velars, nor other taxophones of /I, E, ,
U , Ii, Uu EI, aE, OE/ + /, / are necessary, but their use makes one's pronunciation
more authentic, although this international model is already denitely more
realistic than those oered in so many textbooks, even by English-speaking phone-
ticians (where, among other thing, /i:, u:/ are still indicated). us, taxophones
like ( #, #, #, # @, iI, uU a, ) (but (I, )) can really change one's pro-
nunciation from simple international (for foreigners) into native international
pronunciation. Instead, the use of (i:, u:) clearly indicates either a non native-like
pronunciation, or one which is regional or socially inappropriate.
96 a handbook of pronunciation
2.4.1.5. Both the and the American models have some problems of social
acceptability. , 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
articial aectation, which makes it disagreeable to many native speakers. is
generally associated with a high social position ( members of the aristocracy,
of the higher clergy or military ranks, Tory MP's, prestigious university professors,
), and a certain age group (over 50 years of age). Clothing, too, should be su-
ciently formal, to be suitable for the accent. If these conditions are lacking
for common natives the British neutral accent could prove to be denitely inap-
propriate.
It is to be said that, paradoxically, even an impersonation of an 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 nd 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) diculties, 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 international accent, which although it might seem to
be nobody's accent 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 nobody's accent, 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,
mediatic 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, 2.12-18 (which speak for themselves) must be carefully analyzed,
comparing them with the neutral ones, in order to capture the dierences, 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 as they can. 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 almost neutral (or, at least, less aected and less articial 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 unsubstantial neu-
tral kind of pronunciation It is no rare fact that some mediatic speakers uc-
tuate towards the neutral type (or away from it), for some words.
Here we wish to briey draw attention to some details, with reference only to the
respective (American or British) variant, for a direct (internal) comparison. In the
light of what has been seen so far, it will not be dicult to nd 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 /'s<i/ sorry ('sO:)a
/'s;/ so), all the more so because, for /O:/, some intermediate articulations be-
tween /O:/ and /A:/ are decidedly common: (A), as in ('s:, 's;)m;a /'s;/ so
and ('s:, 's;A)m;a /'sO:/ sY
Besides, we have // (x)m;a (//), not only when nal before a pause, but even
after a pause, even if near velar(ized) C, where in the neutral accent we nd
()a: (x'ph];) /'plaE/ appI (x's;) /'saEd/ aside (x'ThmpT) /'tEmpt/ attempt
(x'Th;In) /b'tEIn/ obtain (x'];sn) /'lOUn/ alone (x'g;) /'gOU/ ago (x'w;I) /-
'wEI/ YZ (x'<;n) /'<aOnd/ around (x'kh>:) /'k:/ occur
2.4.2.3. 2.13 gives the diphthongs. Let us observe the dierences for /Ii, aE,
aO, OU, [j]Uu, Uu/ (i, , , , [j]%, )m;a (Ii, a, a, , Uu, ju, Uu)a:
('b;i)m;a ('bI;i)a /'bIi/ bee ('g;)m;a ('ga;)a /'gaE/ guy ('D;n)m;a ('Da;n)a /'daOn/
down ('n;s)m;a ('n;)a /'nOU/ no ('Th%;)m;a ('ThU;u)a /'tUu/ o ('n;*)m;a ('nU;u)a
/'nUu/ n[ ('fj;)m;a ('fj;u)a /'fjUu/ fuel
further variants are fairly frequent, which produce: /I/ (i), /A:=, A:0/ (A<):
('khi:, 'sii)m;a for ('kh:, 's)a /'kI, 'sII/ ki sii and ('phA<k, 'khA<m>,
'DA<]i)m;a for ('phA;<k, 'khA;<m, 'DA;<l)a /'pA:k, 'kA:/ park carter, darli.
2.4.2.5. As far as C are concerned, keeping in mind that these observations are
general (and not absolutely obligatory for all speakers words) and that our
transcriptions are normalized, let us say that for /l[j]/ we have (][j]): (']T)m;a
('lT)a /'lIt/ lilt (']:)m;a ('l,:)a /'l/ lull ('mK]j'n)m;a ('mjn)a /'mIljn/ million
('nvsT)m;a ('nAvsT)a /'nvst/ noveli
Besides, /</ has a uvularized velar rounded articulation, which is darker (or
harder): ('<;>)m;a ('<I;)a /'<I/ rear ('<:<)m;a ('<:<)a /'<O:</ roar ('wn)m;a ('wn)a
/'wIn/ winner Please note that generally (<) exerts on // the same retracting and
raising eect of /k, g, , w/: (Th'<;In)m;a (Th'<;In)a /t'<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 nd syl-
lables with /{}/ + /mp, nt, k/, we very frequently have ({}) + (Mp, Mp)
(T, T, ) (k, k) (which are too often described simply as (0) la franai-
se): ('khMp)m;a ('khmp)a /'kmp/ camp ('sTMp)m;a ('sTmp)a /'stmp/ ump
('kh, -T)m;a ('kh[;]nT)a /'k;nt/ can't ('Ds, -T)m;a ('DnT)a /'dOUnt/
don't ('Ik)m;a ('k)a /'Ik/ `ink ('D<k)m;a ('D<k)a /'d<k/ drank
// behaves as in neutral American pronunciation, but often this use spreads to
/d/ as well, so that (']m>)m;a can represent either ('lm)a /'l/ latter or ('lD)a
/'ld/ ladder (&n'T;n, &nm>-)m;a (&nD'T;n)a /nd'stnd/ underand
e same goes for // = (`), 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 common peo-
102 a handbook of pronunciation
ple, when public people are concerned as politicians are the same pronuncia-
tion can be dened as Cockney, 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 2.16, /, , O:/ are higher, (E, O, oo). e last one
is doubled (since it moves upwards slightly), or diphthongized (downwards) when
lexeme nal, or followed by the /z, d/ (o[/]) grammemes. Also /E/ is higher,
though remaining within its own box, () (to be better distinguished from // (E)):
('[h]E)m;b ('hT)b /'ht/ hat ('<Ok)m;b ('>k)b /'<k/ ro$ ('lo;on)m;b ('l:n)b /'lO:n/
lYn ('lo;)m;b ('l:)b /'lO:/ lY ('mo;)m;b ('m:)b /'mO:/ more ('lo;)m;b ('l:)b
/'lO:z/ lYs ('pho;)m;b ('ph:)b /'pO:d/ pored In the vocogram, the diphthongs due
to the vocalization of // are marked in grey: /I, E, , , U/. Note the dierent
symbols: ('b;P[])m;b ('b:)b /'bI/ bill ('b;P[])m;b ('b:)b /'bE/ bell (';P[])m;b (';)b
/'/ Al ('khP[])m;b ('khT)b /'kt/ cult, ('bU;P[])m;b ('b:)b /'bU/ bull.
ose who systematically realize // as () are inuenced 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 dierent speakers, not the same, although in the same places.
Besides, /u/ is fronter, (%): (&sc%'ISn)m;b (&sc'IS)b /sIcu'EISn/ situation;
/, []|/ are lower (and fronter), (a): ('baa)m;b ('bT)b /'b/ butter ('b;a)m;b
('b;)b /'bI/ beer; whereas, /:, A:{}/ are diphthongized, (, A): ('f;[])m;b
('f:[])b /'f:[z]/ fur[s] ('khA;[])m;b ('khA:[])b /'kA:[z]/ car[s]
2.4.3.3. 2.17 gives us the positions of /I, I< E, E</. We have just seen
/I|/ (beer] instead, for /I</ (and /I<, I/ not before pauses) more often we have
(:): ('[h];<n, --)m;b ('h>)b /'hI<I/ heari ('[h]:< n':, --, -;)m;b ('h;> n-
';)b /'hI< n[d]'E/ here and `ere ('[h]: 'khamz 'san:, '[h];, &[h];'-, &[h]'-)m;b
('h; 'khmz 'sn:, &h'-)b /'hI 'kmz 'sn/ here comes `e sun. It is to be noted
that even /E<, E/ are more often long monophthongs than narrow diphthongs:
('m;<I, 'm-)m;b ('m>i)b /'mE<i/ Mary (beside `ere, just seen); /U/ practically
becomes /O:/: ('pho;)m;b ('ph:, 'ph;)b /'pU/ poor
Generally, in this accent, for // sequences we have (, a|): ('sIa)m;b
('sIi)b /'sIi/ seer (and also (A'DIa)m;b (a'D;)b /aE'dI/ idea), ('DT%a)m;b ('Du)b
/aE/ (A, Aa|), /aO/ (P, Pa|), /EI/ (I, Ia|), /OU/ (x, xa|), /OE/ (o, oa|),
/Ii/ (I, Ia|, I<), /Uu/ (T%, T%a|, T%<), /A:/ (A) + /A:/ (A, Aa|), /</ ([]<)
2. english 103
2.4.3.5. Even for mediatic British English, neutralizations are frequent be-
fore // (although less frequent than in Cockney, the typical and popular and less
educated dialect and accent of the East End of London). In fact, 2.19, we of-
ten nd ('f;P[])m;b both for ('f:)b /'fI/ ll and ('fi;I)b /'fIi/ feel in addition,
('[h]a;P[])m;b both for ('h;)b /'h/ Hal and ('h;)b /'hEI/ hail and both for
('ha;)b /'haO/ howl and ('h:)b /'h/ Hull, as well. Besides, we nd ('DO;U[])m;b
both for ('D;)b /'d/ doll and ('D;)b /'dOU/ dole and ('f;U[])m;b both for
('f:)b /'fU/ full and ('fu;U)b /'fUu/ fool (the last one is possible for ('fo;U[])m;b
('f:)b /'fO:/ 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
(), in all the cases indicated in 2.2.5.2; whereas we can say that, generally, ()
replaces (T) /t, / of neutral pronunciation. erefore: ('D<Op)m;b ('>p)b /'d<p/
drop ('D<Op)m;b ('>pT)b /'d<pt/ dropped ('phEk)m;b ('phk)b /'pk/ pa$
('phEk)m;b ('phkT)b /'pkt/ pa$ed ('<Ic, -S)m;b ('>Iic, -S)b /'<Iic/ rea>
('<Ic, -S)m;b ('>IicT, -ST)b /'<Iict/ rea>ed ('<Aa)m;b ('>aT)b /'<aE/ writer
('f&boU[])m;b ('fp&b;, --, -T-)b /'fUtbO:/ football ('g 'D;Pn)m;b ('gT 'Da;n, -)b
/'gEt 'daOn/ get down ('lIlI)m;b ('lITli, -li)b /'lEItli/ lateI ('hA)m;b ('ThA;T)b
/'tA:t/ tart ('hsa, -a)m;b ('ThsT)b /'tEst/ tester
e preglottalization of /p, t, k/ (and of /c/, as in neutral British English) also
occurs between V both within words and in sentences: ('phpa)m;b ('php)b
/'pEp/ pepper ('ba)m;b ('bT)b /'bE/ better ('bIka)m;b ('bIk)b /'bEIk/
baker ('l, -)m;b ('lTT, -)b /'lEIt/ let it In the /st, sc/ sequences, there is the
possible variant /s/ () (by further assimilating to the following contoid, which is
typical of broader accents): ('sOp, '-)m;b ('sTp)b /'stp/ op ('s<;I~, '<-,
'>-)m;b ('s>;I~)b /'st<EInG/ rae ('khwscn, -c-, -cn, -Scn)m;b ('khwsc,
-Sc)b /'kwEscn, -Sc-/ queion
For //, ([) can occur, too: ('ph<I, 'ph<[I 'hwnI, -n[I)m;b ('ph>Ti,
'ThwnTi)b /'p<Ii, 'twEni/ pret en Even () (in addition to (n, -)):
('khln, 'P[], 'b)m;b ('khlnT, 'T, 'b;T)b /'klIntn, 'Etn, 'b:tn/ Clin-
ton Elton Burton
Besides, we nd (n) for the grammeme >-i: ('<xln, '<OU-)m;b ('>l)b /'<OU-
lI/ rolli this example also shows the oscillation for /OU/ before /l/, under the in-
2. english 105
2.4.3.7. Our examples have already shown that one of the most evident charac-
teristics, which is socially stigmatized, is the zero realization of /h/ (= (`)):
('[h]A;)m;b ('ha;)b /'haE/ high. Consequently, even hypercorrections are frequent,
as happens with the name of the letter h (not without a certain internal logic):
('[h]Ic)m;b ('Ic)b /'EIc/. On the other hand, the reduced form of him is less fre-
quent: ('hP[]hm)m;b ('Thlm)b /'tElIm/ tell him equally for the reduced form
of a\ (I'mE;n)m;b ('m;n)b /'mn/ a man Also for /0n/, a less reduced form is
more frequent: (ph<&nansi'ISn, 'GEksn)m;b (ph>&nnsi'IS, 'Gks)b /p<nn-
si'EISn, 'Gksn/ pronunciation Ja$son
e sequences /t, d, n/ have the peculiarity of typically corresponding to /c,
G/ and (~): ('chT;%n)m;b ('Thj;un)b /'tUun/ tune ('GT;%n)m;b ('Dj;un)b /'dUun/
dune ('~T;%)m;b ('nj;u)b /'nUu/ n[ (in Cockney we actually nd ('n;%) = /'nUu/).
Occasionally, /, / can become /f/ and /d, v/ respectively (which is a typical
Cockney pronunciation, and can be heard even on the borders of the Estuary
area {and in further mostly metropolitan areas, which have been inuenced by
this accent}): ('<;I, 'f<-)m;b ('>I;i)b /'<Ii/ `ree ('s, 'D-)m;b ('s)b /'Is/ `is ('maa,
-va)m;b ('m)b /'m/ mo`er.
2.4.3.8. ere is nothing to say about /l/, while for // there is a typical vocal-
ization of (, ) which become (P) ( 2.19), after V or C as many examples have
already shown (particularly those of the neutralization before //). e lateralized
velar rounded approximant () (in brackets in the gure) 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 nd (), without lip-rounding, as a kind of halfway compromise.
Here are some examples, to complete the survey: ('lP[])m;b ('lT)b /'lIt/ lit-
tle ('phIpP[])m;b ('phIip)b /'pIip/ people ('DvP[])m;b ('Dv)b /'dEvz/ devils
('mP[]k)m;b ('mk)b /'mIk/ milk ('woP[]a)m;b ('w;T)b /'wO:/ Walter
('GnP[]mn)m;b ('GnTmn)b /'GEnmn/ gentleman. nally, (&akPl'-
106 a handbook of pronunciation
b, -P'h-)m;b (&k'h;bT)b /'k 'h:bt/ uncle Herbert where we can see the
normal kind of prevocalic l ((l)), and the intense one, again prevocalic: ('nOvs,
-)m;b ('nvsT)b /'nvIst/ noveli
2.4.3.9. In cases like the following, we often nd secondary stresses: ('DkS&n-
<I, 'sm&<I, 'Doom&o<I)m;b ('DkS>i, 'sm>i, 'D;m>i)b /'dIkSn<i, 'sEmt-
<i, 'dO:mtO:<i/ dictionary cemetery dormitory.
Regarding intonation, we have to say that the pitch of the rst 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 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 2.20,
where we show only the dierent type): (GspIk 'glS 'w;P[]12)m;b (Dj5spIik
'glS 'w:21, G-)b Do you speak Eli well?
/ / ( 2 2 2 2 ' 2 2 ' 2) // (2 2 2)
Text
2.5.0. e story e North Wind and the Sun (by Aesop) follows. It is given in
ve dierent (normalized) 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 rst
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: rst 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, uent 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 rst. 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 roer,
when a travel(l)er came alo wrapped in a warm cloak. \ agreed `at `e one
who r succeeded in maki `e travel(l)er take his cloak o ould be considered
roer `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 unications of vowels, with the merging of /O:,
O;, , ;/ into (:) and of /O:, O:<, <, ;</ into (:<) (('s;<i) /'s<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, (, ): ('nT, 'T)
// = //, // (, *, |)
// (), /</ (<, ), /:, :</ (':)
/E[]/ (, ),
/E</ (<),
/E/ ()
/[]/ (, )
/[]/ (, ), /</ (<), /;/ () /O:, O:<, <, ;</ (:<)
/A:/ (A:), /A:<, A:/ (A:<) /O:, O;, , ;/ (:)
/jUu/ (ju), /Uu/ (Uu)
/Ii[]/ (Ii, iI) /Uu/ (ju, Uu)
/[j/]Uu/ ([j/`]uU)
/EI[]/ (eI, e) /OU/ (oU)
/OE[]/ (, )
/aE[]/ (a, a, =, =) /aO/ (a, =)
/aO/ (, ), /aO=/ (x) /Ii, EI, aE, aO, OE, OU, Uu/ + // = //
/'naEt, 'aOt/ night out. ere are neutral and non-neutral taxophones for many vow-
els and diphthongs + //. e neutral ones which change phones are: /, , U
Ii, EI, aE, OE, Uu/; the others are /I, E/ and // with other variants; in addi-
tion to a possible insertion of () before (). As far as phoneme distribution is con-
cerned, certain words are pronounced with British vowel elements, others with
American 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 nd that hypercorrection can produce /jUu/ for /Uu/ [noon too do), due to
the fact that, for /Uu/, careful speakers prefer /jUu/, after /n, t, d/. In addition, we
nd the nasalization of /, / (, ), even in // (above all for //);
lastly, for /l/, we have (]), in these kinds of pronunciations.
/[]/ (E, )
/[]/ (, )
/[]/ (, ), /A:/ (a:)
neme // follows British use (although some young Australians, especially females,
who live abroad, present a uctuating and non-neutral use of (>)). Let us notice
(and very well too) the various taxophones + //.
We systematically nd /OUl/ (whereas in Cockney and in mediatic British
English minimal pairs occur such as /Ul/ /OUl/, as polar] Occasionally, we
can have /O:/, for /O:/ (either /Or/, or /or/). Except in neutral pronunciation,
for the /Ii, EI, aE, OE/ diphthongs, we have (, ); besides, /{}/ ({},
{}), without vocalizations (with (, ), after phones with labial component).
e diaphoneme // has ([), as in American English. For /0, 0/, neutral
pronunciation has (0, 0); thus, it has (l), 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 //).
/E[]/ (, ) /O:/ (: :,
f :0, o0)
/[]/ (, )
/[]/ (, )
f
/[]/ (, )
/A:/ (a:)
/Ii[]/ /Uu[]/
(Ii, iI) (u, Uu) /I[], I</ /[j]U[]/
(I>, I0, I|) ([j]P>,
/OU[/l]/ (T,
U, Ul) P0, P|)
/EI[]/ { /[j]O:/}
(EI, E) /OE[]/ /E, E</
(, ) (>,
0, |)
/aO/ (ax) /aE[]/ (a, a)
2. english 113
2.6.3. In addition to the neutral accent, with three vocograms (the rst three),
we present the mediatic accent (whose centering diphthongs {fourth and fth 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 unication of /E, E</ with
/I, I</, and the possible unication of /U, 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 (> >, ); whereas, for /I</, (I>, iI); for idea\ /aE'dIi/.
/</ (E> >) is always distinct (in all accents). Occasionally, we can nd /O:/,
for /O:/ (both /Or/, and /or/). e triphthongs are not attenuated into diph-
thongs. Even in neutral pronunciation, we have /I/ () (not (, )). In New
Zealand, /;, A;/ = /A:/, /;, O;/ = //, // = /, I/ ( /I/ is only an intentional choice);
ni /'fInIS/ ('fnS 'fnS), visit /'vIzt/ ('vzT, 'vzT).
e most typical characteristic consists in (, 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,
/:<, :/ is rounded, (:) (or only partially so, (+, +|), in the other pronunciation
kinds).
/i/ (i) /u/ ()
/I[], I/ (, , ) /U[]/ (, U)
/O:/ (o:)
/E[]/ (e, )
/:/ (:), // (, *, |)
/[]/ (E, ) /[]/ (, )
/A:/ (a:) /[]/ (O, )
/ / (2 2 2 2 ' 2 2 ' 2) // (2 2 2)
Mediatic-accent variants
/i/ (Ii) { /Ii/ (i, i)} /u/ (%)
/U[]/ (w, U)
/I[]/ (, )
// (, *, | )
/E[]/ (, []0, ) /O:[]/ (o, oU)
/:/ (+, +|)
/[]/ (O,
O)
#
/[]/ (, []0, ) /[]/ (, , )
/A:/ (a)
Broad-accent variants
/Ii[]/ (i, i) /Uu[]/ (%, u)
/OE[]/ (PI, P)
/OU[]/ (x%, U, Ul)
/aO/ (Ex)
/EI[]/ (a, a) /aE[]/ (, )
/m, mI, m/ besides, /EI, OU/ are (, P) before ('): vacation november; /-OU/ is
(P) in compounds when it is at the end of the rst lexeme (even if separated): win-
dow sill. For ar, /A/ (A;) is possible; even /O/ for /O:/. nally, it has (>)
for /<, </, and () for /i, i/: ('khA:, 'khA; 'mO; 'mE>, -R) /'kA:, 'mO:,
'mE<i/ car more Mary.
Triphthong attenuation is very frequent (but less than in aected pronuncia-
tion or in Cockney), even between word, also for /I/: ('f;, 'f: ~'GOT, ~-)
/'faE, aEIn'GOEIt/ re 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.
/I[], i, i/ (, ) /U[], u/ (, U)
f
// () {desinential as well)
/E[]/ (, )
// (, *, |), /:/ (:)
/[]/ (, ), /O:/ (O:)
/[]/ (, )
/A:/ (A:), // ()
/OE[]/ (O, O)
/aE[]/ (, ) /aO/ (AP)
2.6.5. It can be aunted by aristocrats and people of high social, religious, and
cultural standings. But it is generally thought of as too aected. It is characterized
2. english 117
by more peripheral /I, U/, (I, U) (instead of (, )); but they are more centralized
in /I, I< E, E< U, U</ (x, | x, | Px, P|); it presents /</ (x>),
and () for /i|/, as well: ('mx>) Mary.
en the phoneme /I/ is fully unstressed, it is (), as /i/ too (instead of (i)),
and, before pauses, it even becomes (): (&vz'blT, -T); it has /-I/ (-n, ) also
// is (), as many // are, as well.
Besides, /E, / are closer (e, E); when /I, E, / are nal, in stressed checked sylla-
bles ending in voiced C before a pause, they diphthongize as (I, e, E): ('bI,
'we, 'bE) big web bad.
/:, :< A:/ are backer, (x:, , :) (and the last one occurs more frequently
than in neutral pronunciation); for ar, /A/ (;) is possible. e diphthongs are
considerably narrow and have particular timbres, manly /Ii, Uu/ (ii, uu) and /aO,
OE, OU/ (Ax, @, +) (and (P), for (Ii, u, a, , ) and ()). For short vow-
els and for diphthongs with front rst elements, there are taxophones requiring
dierent symbols.
f
/U[], u/ (U, U)
/I[], i/ (I[|], , I), /I/ ()
/i/ (, |) /O:/ (o:)
// (, *, |), /:/ (x:)
/E[]/ (e[|], ) /zd/ (, x|), // (, |), // ()
/[]/ (E[|], ) /[]/ (, )
/A:/ (:), // ()
/Ii[]/ /[j]Uu[]/ /[j]U[], -U</
([j]uu, /I[], I</
(ii, ii) (x, |) ([j]Px, [j]P|)
f
F
[j]uu)
f
( (jx:) ( /[j]O:/
/OU+/ (P) = /j:/} (jx:) = /j:/}
/EI[]/
(, ) /OU[]/ (+, P) /O:zd|/
/OE[]/ /E, E</ = /O:[]/
/aE[]/ (@, @) (, |) (o:|, o:|,
(, ) /aO/ (Ax) o:[x]zD|)
/Ii/ /Uu/ (u:x,
(i:x, i:|) u:|) ( /U/,
( /I/
(Ix, I|)} not /O:/!
(Ux, U|)}
/EI/ /U/ (,
(, |) |)+/:/ /OE/
( /E/} (@x, @|)
/aE/ /aO/
(:, ) (A:, A)
It has no glottalization of /p, t, k, c/ ( neither (0), nor (); with the only pos-
sible occurrence of () for /t/, before sonants: ('skTln, 'skln) Scotland] /tj,
dj, sj, zj, j/ are fairly frequent, as in ('khwesTjn, -scn 'sjuup, 'suu-), queion
super (with (S Z, Sj Zj) only in colloquial or informal pronunciation, for /'kwEscn,
'sUup/ ('khwscn, 'sup)); /;f, -, -s, -ft, -st/ = /O:/ (o:), in addition to // () in
some words, today, but o maintains /O:/.
/</ (>, 'R, R, R, /</ (>, 'R, R, R, pR, bR, kR, gR) (also ('V), and even ('e,
118 a handbook of pronunciation
'=) in some frequent words: very terrible sorry tomorrow); non-written and
non-etymological /</ is frequent. Often, the aspiration of /p, t, k, c/ is very
weak (contrary to Cockney usage). In addition, it has (`) for /h/ hotel /w/ (w),
('wen) when unstressed my is /mI, m/ /-OU/ is (P) in compounds when it is at
the end of the rst lexeme (even if separated): window sill.
Triphthong attenuation is extremely frequent (as in Cockney), even between
words, also for /I/: ('f;, 'f: ~'G@T, ~-) /'faE, aEIn'GOEIt/ re 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.
/I/ (I, I,
I|, I>)* /[j]U/ ([j]oU,
-oU, -oU|,
/E/ (e, e, -oU>)* = /[j]O:/
e|, e:>)*
/aO/
/aE/
(X, X|)
(, |)
(, |)
(, |)
/I/
(, :, |, :>) /[j]U/
([j]o, [j]o:, [j]o|,
/E/ [j]o:>) = /[j]O:/
(, :, |, :>)
thong /aO/, quite often independently from context). For the grammeme /I/ we
have (In, n, ); and, for -`i (-fk) is frequent: ('m3n, 'samfk, 'n'I&fk)
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, (=), even (), in all cases where in the British
accent synglottalization is possible ( 2.2.6.1-2, 2.2.7.1), or where in mediatic
British English preglottalization occurs ( 2.4.3.6). Even for the phonetic reali-
zations we nd some dierences. In fact, in the most typical and broadest pronun-
ciations, /p, t, k/ are realized as the corresponding stopstrictives: (p, , w), even
aspirated (in the normal contexts expected for neutral pronunciation, too): (ph,
h, wh) (which can give the impression of stronger aspiration). However, the
most typical and broad element is the substitution of /t, / with () 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/, ('h, |h, 's, s)).
Examples: ('b) /'b</ butter ('woU) /'wO;/ water ('whO) /'kt/ cot-
ton ('s) /'sIt/ sitti ('bO) /'bm/ bottom ('b) /'b:t/ Burton ('P)
2. english 121
/'Et/ Elton ('whln) /'klIntn/ Clinton ('lP) /'lI/ little ('ph;na, -;-)
/'pA:tn/ partner ('sp>l) /'sEp<tli/ separateI ('lOs) /'lts/ lo_
More examples: ('phan, -n,, -n, -n[, -n) /'pEIn/ painter ('phan-
, -n,, -n, -n[, -n) /'pEInt/ paint it ('woU) /'wO:/ Walter ('oU,
-,, -[, -) /'hO:t/ halt it (';, -,, -[, -) /'stA:t/ art it (', ',,
'[, ') /'hIt/ hit it ('lP 'b 'b) /'lI 'bI v'b/ a little bit of butter
('ph 'p) /'pU 'p/ put up ('ph 'p) /'pU 'p/ put it up
In less broad pronunciations, an incomplete, attenuated stop is possible: (,),
which is less invasive; the vocoid preceding () can even be laryngealized, where-
as () can become zero, especially before another vocoid (adding, however, the
creaky phonation type), ( = , = , = ): ('lP 'b 'b). Generally, forms
such as lill ('lP) /'lI/ and little ('lP, 'lP;, 'lPU) /'lI/ maintain some dierences
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 () in elegant 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): ('b[, 'woU[, 's[n, 'b[m, 'l[P), ('phan[, -n
'phan[, -n), ('woU[, 'oU[, ';[, '[, 'l[P 'b[ 'b[, 'ph[ 'p, 'ph[-
[ 'p).
Other consonants can become (), especially /p, k/: ('O) /'stpI/ oppi
('fl 'l) /'fIlIp 'laEktIt/ Philip liked it ('whOn) /'kkni/ Co$n\ In a previ-
ous example, we have seen that typically /d/ becomes stopstrictive, (); besides,
commonly, /d/ is realized as (), when it is word-nal and followed by C or V
and in the grammeme sequence >-dn't /-dt/, as well: ('b> 'b) /'b<Ed 'b/
bread and butter ('g 'bo;) /'gUd 'bOE/ good boy ('{}) /aE'dIdt/ I didn't
For /st, st<, sc/, broad pronunciations have (, >, Sc): (';) /'stEI/ Z
('>O;) /'st<;/ ro ('wwhScn, -c) /'kwEscn/ queion In broad pronuncia-
tions, /, / become /f, v/; however, there are many intermediate nuances, includ-
ing the realizations of normal pronunciation: (f, v , , , , ). More of-
ten, // can be realized as (`, , d, D): (s'EXs z'm3n, s-, ds-, Ds-) /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: ('h{}) /'Iit/ eat. For /n, t, d/, the typical Cockney pronuncia-
tion has no //, but, in less broad pronunciations, mediatic-like types are possible,
as well: ('n;%, 'n;%, '~;%, '~T;%) /'nUu/ n[ ('h';n, 'hK;n, 'hK;n, 'h;n,
'chK;n, 'ch;n) /'tUun/ tune ('%w, '%w, '%k, 'T%k, 'G%k, 'GT%k)
/'dUuk/ duke.
Triphthong attenuation is extremely frequent, even between words, also for
/I/: ('f;, 'f; ~'GoT, ~-) /'faE, aEIn'GOEIt/ re I enjoy it. Substantially,
its intonation patterns correspond to the neutral ones. For [n]ei`er we generally
nd /Ii/.
3. Italian
resse (i compiti) she corrected the homework, with /E/ (), or in (se) fosse (vero)
if it were true with /o/ (), which diers from (le) fosse (&avate) the dug
graves, with /O/ (). erefore, there are seven Italian vowel phonemes: /i, e, E,
a, O, o, u/.
Let us continue our search 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 gure
of sixteen was reached by means of an arithmetical operation, using the alphabet,
instead of actual sounds.
Hence, if you take the Latin alphabet, take away the ve Italian vowels and
the ve non-Italian consonants [j k w x y), it seems logical to get to the sad
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 /ki*/ from ci /ci/, ro /'giro/ from
giro /'Giro/; there is even a vowel dierence between ho /O*/ and o /o*/. Incidental-
ly, we can but disapprove of the method 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 he/she has written, they have
studied. e poor (selsh and dishonest) result of managing to obtain correct
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 obscenities 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 dis-
covered them, not by graphic deduction, but by looking for opposites in mini-
mal pairs, that contain two similar words, but with one dierent phoneme, which
makes the meaning of the two words change, as seen with /ki*, ci/ and /'giro, 'Gi-
ro/ (respectively and ci, ro 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 dierent phonemic values, as in pre-
sento [una persona I introduce a person, /pre'zEnto/, from presentare] and presen-
to [un avvenimento I have a presentiment of an event, /pre'sEnto/, from presen-
tire pre-sentire) in the same way we have razza (ray, spoke: /'raQQa/) and
razza (race, breed: /'raqqa/)
We must not forget other spelling problems that derive from the fact that Ital-
ian represents a transformation and evolution of Latin, which had a certain num-
ber of phonemes (obviously dierent from Italian, both phonically and numerical-
3. italian 125
Vowels
also occur for /'e-o/: ('ve:do, -d) /'vedo/ vedo (but not for /'o-e/: ('do:ve) /'dove/
dove). at remains to be said, as can be guessed (and conrmation is always pre-
cious), is that this happens with a nal C too: ('ri:dR, 'ri:dn) /'rider/ rider(e) ri-
don(o), ('ip:siln, 'su:tR) /'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*/ pi ('kjE:do) /'kjEdo/ edo ('kwa)
/'kwa*/ qua ('bwO:no) /'bwOno/ buono. It is equally true that /ju, jE, wa, wO/ are
not diphthongs at all ( 3.1.2) but simply sequences of CV like /su, tE, va, nO/,
.
3.1. Italian monophthongs.
/i/ (i) /u/ (u)
Diphthongs
3.1.2. Italian grammar books put a lot of eort into complicating what is, in
fact, quite simple. Indeed, instead of the three very common structures, the re-
al diphthong ((', &, )), the hiatus ((', &)) and the heterophonic sequence
((0)), , (j), (w), and the like), they continue to consider only two of them:
diphthong (with fusion: syneresis) and hiatus (with separation: dieresis),
but with strained interpretations of medieval origin, of a graphic-grammatical and
graphic-metric nature. In fact, semi-vowels or semi-consonants do not exist:
they are merely an incredibly successful magic trick!
As a matter of fact (unless one expects to do magic in phonetics using graphic-
-grammatical categories), it is phonetically absurd to speak about a diphthong
for (j, w) (('pjE:no) /'pjEno/ pieno ('gwan:to) /'gwanto/ guanto] as only (i,
u) (('fai) /'fai/ fai ('pa;uza) /'pauza/ pausa] are real diphthongs, as any sequence
of (', &, ) (('a;uto) /'auto/ auto (au'tEn:tiko) /au'tEntiko/ autentico]
It is equally absurd to speak about hiatus for ('i, 'u), as only (i', u') are re-
al hiatuses, as any other sequence like (', &) ((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 diphthongs
the heterophonic sequences (/j, w/), and in hiatuses the real diphthongs (as
in ('ma;i, 'ca;o, 'bO;a) /'mai, 'cao, 'bOa/ mai ciao boa 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, , 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 dierent 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 dierent 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/ ascendant diphthongs, because they derive from the Latin /E, O/ (,
), and they nd they need to have to invent (as the best lawyer for the worst
criminals) ctitious realities, like semi-consonants and semi-vowels.
ese seem to be nearly cinematographic special eects, 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 (j, w) are natural-
ly part of /{0}0/ groups, as in /'fjanko, 'franko 'gwado, 'grado 'twOno, 'trOno
'gjande, 'grande, 'glande/ anco franco guado grado tuono trono ande gran-
de ande. Paradigmatically, /j, w/ are in opposition with /0/, clearly not with //
and can under no circumstance belong to vowel groups.
Even semi-vowels 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. 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, 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
m (M) (n) n (~) N ()
F pb t d k g
qQ c G
f v
_ s z S {Z}
j w
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 dierence, 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/ bao (peR-
'NOk:ki) /per'NOkki/ per oc ('toM:fo) /'tonfo/ tonfo ('dEn:te) /'dEnte/ dente
('fra~:Ga) /'franGa/ frangia ('fa:go) /'fango/ fango.
In neutral Italian, /N/ is self-geminant ( 3.3.1.5.): ('soN:No, loN'NO:mo) /'soNNo,
loN'NOmo/ soo lo omo. We can observe that the self-geminant /N/ is in opposi-
tion to geminates, as in ('soN:No) /'soNNo/ soo and ('son:no) /'sonno/ sonno, but
not simple C (('so:no) /'sono/ sono); the minimal pair is between the rst 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,
3.4 of MaPI] preceded by (kon) /kon/ con with, 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 briey 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 (&uba-
'bi:n, &uba'bi:n) (concentrating, here, only on the nasals, because the actual re-
gional dierences, doubtlessly, also regard other C, many V the type of phonation,
the syllabic structure and the intonation), for neutral (&umbam'bi:n) /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/ campaa. 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/
conguao. 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 ((, )): (&kjakkje-
'ra:Re) /kjakkje'rare/ acerare ('kik:k) /'kikko/ cco ('gE:ga) /'gEnga/ ghen-
ga.
e greatest regional problem concerning stops is provided in the gorgia tosca-
na ( Tuscan throat) and in the central-southern voicing, which will be exam-
ined in general terms (also because there are dierences between areas; but for
more precise details, various chapters of MaPI can be consulted). erefore, the
following gives just a general idea: (&ihaFi'a:ni) /ikapi'tani/ i capitani and ('ciwe
im'un:) /'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: ('datsa, -a, -a, 'da- 'pja{t}tsa, -a, -a d'zO:na, d'-, d'- ad'zO:-
to, ad'-, ad'-) (or, in broader accents, simply as constrictive: ('dasa, -a, -a, 'da-
'pjassa -a, -a 'zO:na, '-, '- a'zO:to, a'-, 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 /c/ to its corresponding constrictive, (S): ('pe:Se) /'pece/ pece (&diSe-
'Si:lja) /dice'cilja/ di Cecilia ('ka:So) /'kaco/ cacio. In Tuscany the same goes for /-
G/ = (Z): ('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 ocial 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, 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 /G/ = (GG): ('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) (
3.3.1.5.), with the voiced correspondent, but non-geminant ( 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
ame, (la) &ena (&aba'Zu:r, &aZi'go) /aba'Zur, aZi'go/ abat-jour 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 ( (sten'ta:to) /sten'tato/ stentato]
As far as VsV is concerned, modern neutral pronunciation resolves the problem
of the traditional one ( 3.3.5.2.); actually, every postvocalic intralexemic -s- (
in simple words, not compound] is voiced, /z/ (z): ('ka:za, 'ri:z, '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): (&pRndi'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 I jump again; whereas, the noun means to stand out and
is (Ri'zal:to) /ri'zalto/ Forms such as prosieguo, whose composition nowadays
is heard in a less denite way, vary: (pRo'sjE:gwo, -z-) /pro'sjEgwo, -z-/.
For /s, z/, in the north, an alveolar instead of dental articulation is frequent; ('O:-
ja) /'sOzja/ sosia (be warned that, too often, neutral Italian articulation, which is
dental {with lowered tip}, is dened alveolar, 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 ()), even in words like: *(&laza'lu:t, &laa-) /lasa'lute/ la salute, that
is (&lasa'lu:t).
In the center (except orence and Prato) we nd 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: (,
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/ acciaio ('wO:vo) /'wOvo/ uovo (kwa'lu:kw) /kwa'lunkwe/ qua-
lunque
In central-southern pronunciation, we sometimes nd /j/ = (jj): ('paj:jo, 'pa;jjo)
for ('pa:jo) /'pajo/ paio In the south we often have even (i'e:Ri, u'o:v) 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 nd 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:v) /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 French r, in Italian, that range from
uvular, like (K, , , 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, (, , ).
Let us take a detailed look at the Sicilian type (in Sicily, of course, and in
southern Calabria and the Salentina Peninsula as well) for /r, rr, tr, dr, str, sdr/.
Indeed, simplifying a little (even the transcriptions) for /r, rr/, we have the voiced
slit alveolar constrictive, (z), self-geminant if postvocalic: ({laz}'za:djo) /{la}'radjo/
(la) radio ('kaz:z) /'karro/ carro It is clearly not the postalveolar trill phone ((#),
cacuminal, ), as it has been described for generations, without proof.
Relative to this, we also have the bizarre pair =, which is persistently pre-
sented as (#, #), 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:g) /'drago/ drago
Lastly, the /str/ sequence (lexically quite common), is more often rendered by
the voiceless postalveo-prevelar constrictive (self-geminant, if postvocalic) (.): ('na.:-
.) /'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, (): ({la}'a:j) /{la}z'drajo/ (la) sdraio (much less commonly by (la{}-
'a:j)).
We will conclude by indicating the assimilation of /r0/, which is more typical
of eastern Sicily: ('pat:t) /'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/ foa.
We can observe how the self-geminant /L/ opposes to geminates, as in ('paL:La)
/'paLLa/ paa and ('pal:la) /'palla/ palla, not to (('pa:la) /'pala/ pala); the minimum
pair is between the rst 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/ paa ('O:Lo) /'Oljo/ olio (&paLLa'ti:v) /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 taa. 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, (, ), 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 modication, 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 ( 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 unrened publishers, of yielding to laziness of various forms
regarding the spelling of grammemes, : la una della, -re (in innitives) 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 universit, *voltare pagina for the more usual and denitely
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'universit, 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 dierences 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
linguistic Center).
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&Rttet'tren:ta), Dant'Alieri .
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 nal V of the articles and some pronouns {considered
by grammar, in front of even dierent 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'attivit l'esercitazioni a'italiani s'un argomento l'amica un'altra
l'inseo.
If one is not inuenced 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 inseo.
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
nd, propagated by schools, the (humiliatingly mechanical, without the slight-
est consideration, resulting in a damaging conditioned reex) use of the notorious
euphonic d well-sounding d, 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 cacophonous d, forming sequences such as /dd/).
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 dierent. ey must never be elided (or truncated 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 di andare, not da andare; in the same way, (u&nita'ljana, &unan'ti:k)
/unita'ljana, unan'tiko/ un'italiana un antico are nouns; hence, dierent from: (&u-
na&ita'ljana, &unoan'ti:k) /unaita'ljana, unoan'tiko/ una italiana uno antico, as in:
ne cerco (for instance: canzone, mobile) I'm looking for an Italian song/piece of
furniture.
A last reection, linked to spelling and school wrongdoings, has to be cast re-
garding the absurd complication of insisting on teaching the exception to the
rule, where the pronoun s is written with an acute (accent) in order to dier
from se (conjunction and also allotrope, or variant, of s 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 clear why, once the spelling rule is formulated, to distinguish s
from the other se, one must rack one's brains to excogitate the exception to s
stesso, s 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 nding (even in an ambiguous initial position) sen-
tences such as: Se stessi male, non potrei venire If I were ill, I couldn't come, which
is dierent from: S stessi malediranno azzeccagarbu della grammatica e
pettifogging lawyers of grammar will curse themselves; or: Se medesime compae
3. italian 135
di gioco perdono, vengono e&luse If equal playmates lose, they are excluded,
dierent from: S medesime comparano a tutte le altre ey compare themselves
to all the others.
erefore, eliminating the accent from s 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 s stante separate, apart.
Let us briey turn our attention to the improper use (decidedly incorrect
proudly lead by a fair number of intellectuals) of *e non *o non, as in: *italiani
e non. In Italian, the negative non /non/ not has to be followed by the term
which is denied, as in: giallo, non verde; cotto, non crudo; un lo, non due (li);
coa, non petto; essere, o non essere? Instead, the absolute negation (hence, nal in
the sentence) is only no /'nO*/ no (Eng. /'nOU/ ('n;, 'n;)) 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 To be, or not to be?, of course, and I love
you, but you don't). 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)
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/ jour Nevertheless, phonetically, in
the appropriate contexts, /z/ is lengthened: ('riz:ma) /'rizma/ risma.
ere are also ve 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/ bao ('laS:So) /'laSSo/ lao ('vOL:Lo) /'vOLLo/ vo-
o ('viq:qi) /'viqqi/ vizi ('viq:qj) /'viqqjo/ vizio (oQ'QO:no) /oQ'QOno/ ozono
e reason for this is to be found in Latin; as a matter of fact, these ve 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 ( 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 troublesome 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:m, loQ'Qa;ino, LiQ'Qi;i, LiN'NOk:ki) /unoS'Semo, loQ'Qaino, LiQ'Qii, LiN'NOkki/
uno &emo lo zaino zii oc (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 sucient 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/ aatto.
In the central-southern areas, there can be other self-geminant consonants (
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 nd 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, 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: (&pRtasa'po:ne) /pOrtasa'pone/ portasapone (pu&liSSis'kar:pe) /puliSSis-
'karpe/ puli&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 inuence: (&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 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: ({&}&&'). 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) societ, (ka&paci'ta) capacit, (pRo&babili'ta) probabili-
t, (be&ati'tu:din) beatitudine, (&Raqqjo&naliQ&Qabili'ta) razionalizzabilit, (aR&tifi-
&cali'ta) articialit, (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) oceanograco, (ka&RatteRiQ'Qa:bile) caratterizzabile, (in&telliGen'tis:sima)
intelligentissima, (oS&Sillo'gram:ma) ollogramma, (ef&feRveSSen'tis:sim) eerve&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) modicazio-
ne, (kom&poziq'qjo:ne) composizione, (ak&kjappafaR'fal:le) acappafarfalle, (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 rst, 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/ (&faRk-
'kwes:to), (fa&Rk'kwes:to), ('faRk 'kwes:to), ('fa;Rk 'kwes:to) far questo, /skoper'kjO
il'tetto = skoperkjOil'tetto/ (s&kopeRkjil'tet:to), (s&kopeR&kjil'tet:to), (s'kopeRkj il-
'tet:to), (s'ko;peRkj il'tet:to) &oper il tetto, /nonnepo'tep 'pju* = nonnepotep-
'pju*/ (&nonne&potep'pju), (&nonnepo&tep'pju), (&nonne'potep 'pju), (&nonne'po;tep
'pju) non ne pot pi, /bEn'kek k'reda = bEnkek'kreda/ (&bkek'kre:da), (b&kek-
'kre:da), ('bEkek 'kre:da) ben> creda, /me'tad 'dOze = metad'dOze/ (&metad'dO:ze),
(me&tad'dO:ze), ('metad 'dO:ze), ('me;tad 'dO:ze) met 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*/ ('vEgo &dopodi'te), ('vEgo do&podi'te) vengo dopo di te
/'karo pa'pa* = karopa'pa*/ ('ka;Ro pa'pa), (&kaRopa'pa), (ka&Ropa'pa) caro pap /lo-
'facco 'anke per'te* = lo'facco ankeper'te*/ (lo'facco &akepeR'te), (lo'faccoa&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 ama forse cos?
Intonation
3.3.4. 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;23) /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 dierences are found in the sus-
pensive intoneme, //, as can also be seen in the phonosyntheses of 16 of NPT/
HPh (and of 10-15 of MaPI).
which, at that time were dened as of Tuscan language, more often than of Ital-
ian.
e peculiarities of traditional pronunciation, therefore, are more rigid than
those of modern pronunciation (having less variations {and here, we relate above
all the strangest} and at the same time most capricious). The forms followed
by * are found only in some dictionaries or specic 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 (orthoepical) written ac-
cents: Ase annsso aurra* Bertldo* co ccca corner, notch, colltta con-
ntto crsima enrme esso flla Girgio gtta grppo lttera mccolo nsso
nrma* rgano* rcca dista, Rmolo sgmino site strpio strnna t&hio
as well as the endings: amarolo sttti stttero
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 ribizzo intiriz-
zito lazzi lezzo olezzo pettegolezzo ribrezzo rubizzo &orza arzo* sozzo*
3.3.5.2. e greatest dierence, however, concerns the use of single /s/, within
a word, between V (even if with /j, w/ before the second). e most signicant cas-
es (including derivatives) are: asino casa esi uso cosa cos 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/ inese (but /fran'ceze/ francese!) /ci-
nese'ria/ cineseria (but /borge'zia/ bor!esia!), /go'loso/ goloso /cellu'losa/ cellulosa
/kurjosi'ta*/ curiosit /'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/ rosico (but /ero'zjone/ erosione!).
However, in traditional pronunciation, we nd /z/ in cases such as: bisoo ca-
so esa 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 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
dierences 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 ( the end of 3.3.3.4).
Intonation does not change, except for the possible adoption (by some) of Tus-
can-type preintonemes ( the relative phonosyntheses in 16).
3. italian 145
Text
3.4.0. e story e North Wind and the Sun follows, given in eight dierent
normalized versions. We start with the (neutral) Italian pronunciations of (neu-
tral British) English this is the rst 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 10 12 of MaPI (in which there are 23 regional versions, in 10-14).
e respective (dialectal) phonosyntheses in 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 aected accent (snobbish, 3.4.9) is examined, as well, with a French
r (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 tenser
but also laxer V, at the same time ( 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 dierent opposite positions from the initial ones.
Furthermore, /c, G S, {Z}/ are postalveo-palatal over-rounded, (&, 1, , {});
there is also drawling 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 dierent from the conclusive
ones.
Lastly, we note a bigger pitch extension (which we will indicate paraphonical-
ly, with ()@, at the beginning of each paragraph) and creaky voice in intonemes
( tonic and posttonic syllables), which is represented by the specic diacritic: ().
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, uent 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 rst.
Speakers of American English could prepare their own version of the Italian pro-
nunciation of American English, with /a/ even for English // ( ('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
0.12).
Italian text
4.0. As well as the modern neutral accent, we will be presenting the inter-
national accent, as an alternative teaching proposal.
Lastly, we will also consider the mediatic accent ( from television), based on
current Parisian, and two types of accents which are very dierent from these, in
so far as they are structurally very dierent: southern pronunciation, represented
by Marseilles, and Canadian, from Qubec.
Vowels
4.1.1.2. 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 dieren-
ces. In our transcriptions, we use twenty-three elements (plus another less impor-
152 a handbook of pronunciation
tant one). e front series presents ve 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 dier slightly: (mi'_i) /mi'di/ midi (e'te) /e'te/ t ('fEt) /'fEt/ fte ('E) /'E/ e
note: (s'm) /sE'mwa/ c'e moi (v'+) /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 denitely fronter than in Spanish, Italian, German, : (b':\) /ba'gaZ/ bagage
('m) /'am/ me (traditional ('A:m), mediatic/Parisian (':m)); furthermore: (p-
'+i:) /paK'tiK/ partir (&p+i') /paKti'K/ partirons (p'l) /paK'la/ par l
e phoneme /A/ of neutral traditional pronunciation was mentioned, which
is no longer up-to-date; for this reason it is not in 4.1 (however, its tongue posi-
tion is slightly fronter than that of // (), in the vocogram of the international
accent, 4.4.1).
e back (and rounded; actually, back-central as can easily be seen in 4.1)
series is evidently dierent from that in most languages, although the phonemic
symbols are the same ((, , P, , ) for three phonemes /u, o, O/): (t'\:) /tu'ZuK/
toujours (dP'dP) /do'do/ dodo ('km) /'kOm/ comme note: (k'md) /kO'mOd/
commode (\'li) /ZO'li/ joli (k'se) /kuK'sje/ coursier (p'v) /puK'vu/ pour vo
(with unstressed /O/, and with unstressed /u/ in checked syllable in /K/).
e real dierence lies in the fact that normally, forms with unstable e are ex-
pressed with the zero phone and possible voicing assimilations, as seen in the sec-
ond transcriptions. e solution of the zero 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 solutions,
//, and /, +/ too, which are less improbable diaphonemically, to play safe we will
adopt // (#), before transforming it denitively into //, with the possibility of
indicating its presence or absence with /[]/ or //, in opposition to an un-
stressed stable e, represented by //.
en // is stressed, as in d-le le e, it actually becomes the phoneme // (re-
inforcing further what has just been said): (_i'l) /di'l/, (l#') /l'/ (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/ mre ('s:) /'sK/ sur
('f:) /'fOK/ fort
4.1.1.4. Modern neutral pronunciation keeps the opposition /'e/ >- -e -er -ez
V /'E/ >-s -et -Pe -a -P(en)t, which, however, is rather tiring, structurally, giv-
en that the others are neutralized: actually, there is no (longer) dierence between
p and pot they are both ('pP) /'po/, nor between pS and pSx\ ('p) /'p/, and
now, not even between bat and b\ ('b) /'ba/. We have, for example, ('ke) /'kle/
cl V ('kE) /'klE/ clPe ('pe) /'pKe/ pr V ('pE) /'pKE/ prs ('fe) /'fe/ fe V ('fE) /'fE/
fPt (v'le) /va'le/ valle V (v'lE) /va'lE/ vallet (p'le) /paK'le/ parler V (p'lE)
/paK'lE/ parla (s've) /sa've/ savez V (s'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: (&\#p'le, /p-) /ZpaK'le, Sp-/
je parlP (past: I spoke) V (&\#p'lE, /p-) /ZpaK'lE, Sp-/ je parla (imperfect: I
spoke) and (\#&pl#'e, /&p-) /ZpaKl'Ke, Sp-/ je parlerP (future: I will speak)
V (\#&pl#'E, /&p-) /ZpaKl'KE, Sp-/ je parlera (conditional: I would speak).
As far as -ai is concerned, we have: ('\e) /'Ze/ j'P (but ('\E) /'ZE/ j'Pe] ('e, 'E) /'ge,
'gE/ gP ('e, 'E) /'ke, 'kE/ quP ('bE -e) /'bE -e/ bP ('mE -e) /'mE -e/ mP ('vE -e)
/'vKE -e/ vrP furthermore, we nd ('mE) /'mE/ ma (with (m, me) /mE, me/ in a
preintoneme), as also for ('E) /'E/ tu [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'l, -e-) /ilE'la, -e/ il e l (m's2 'l:2) {(me-)} /mE'sa, a'lOK,/ {/me-/}
ma a, alors!
Nasalized vowels
4.1.1.5. Obviously, the four nasal vowels remain (which are more precisely de-
ned as nasalized] with their six taxophones, (', 3) //, ('^, }) /^/, () //, ()
//: (&3s't) /sEK't/ incertPn (}'b^) /^'bK^/ un brun (p'd) /p'd/ pen-
dant (m'n) /m'n/ mon nom in stressed and checked syllables, we have:
154 a handbook of pronunciation
('p:dX) /'pdK/ pRndre ('^:ba) /'^bl/ humble ('l:p) /'lp/ lampe ('n:bX) /'nbK/
nombre A funny example: (}'b 'v 'bl) /^'b 'v 'bl/ un bon vin blanc ese
phonemes can even be followed by N\ ('~i) /'ni/ ennui (n'nEtX) /n'nEtK/
non-tre (m'ne) /m'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): /, ^, , /; of
these, for modern neutral pronunciation, only /^/ can be used; the others reect
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 dierent today, it can still be found in various
regional pronunciations, even if, as we will see, /, ^, , / can be useful as
representatives of an international pronunciation, which is less tied to Paris and
to the neutral pronunciation of Parisian origin ( 4.2). Certain texts (and some
dictionaries), for the rst three, even use /e, J, /. In the transcriptions in the
French Dictionary (by R. Boch: Zanichelli, 1995), the present writer put /, ^,
, /, 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 /, ^, , /.
Modern neutral pronunciation rmly keeps these four nasalized vowel pho-
nemes, as do most regional pronunciations (even if with dierent, more traditional
timbres). Instead, for Paris (for the center and the west) /^/ merges with //, so,
there, /'bK/ corresponds to brin and even to brun (neutral ('b, 'b^), respective-
ly). Consequently, even the mediatic 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 tting 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 vowel ad-
justments: for /E/ ( the archiphoneme of /e, E/), we have (e) /e/, in an un-
checked pretonic syllable, if it is followed by a close(r) V ( /i, y, u e, , o /}:
(e'te) /e'te/ t (me'z) /me'z/ maon (le'z,) /le'zj/ l ySx (&epe'te) /Kepe'te/ r-
pter
On the other hand, we have () /E/, in an unchecked pretonic syllable, when it
is followed by an open(er) V ( /E, , O a , ^, /) and in checked syllables (in-
dependently of the context) too: ('tE) /E'tE/ ta ('t) /E't/ tant (l'zm) /lE-
'zOm/ l homm (&ep'tE) /KepE'tE/ rpta and (&evn'm) /evEn'm/ vnement
(mt's) /mEt's/ mdecin (p'te) /pEl'te/ pelleter (p'_y) /pEK'dy/ perdu (d'mi)
/Ed'mi/ et demi (t'p:) /tEK'pOK/ t reports (s's) /sE's/ s leons (&lsp'tka)
/lEspEk'takl/ l spectacl (dp'n) /dEp'n/ d pne For (initial) ex- esC- there is
4. french 155
a strong tendency to have /e/: ('z:pa, e-) /Eg'zpl, e-/ exemple (&s',e, &e-) /Eska-
'lje, e-/ caliers
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 nd the adjustment: (t'iba, te-) /tE'Kibl, te-/ terrible (s'e, se-) /sE'Ke, se-/ ser-
rer but (&p'E) /pEKO'kE/ perroquet furthermore, (&Pp#&m, 't:\) /opK'mjE
KE'taZ/ au premier tage (&}l'\ '~i) /^lE'ZE K'ni/ un lger ennui (despite (p#-
'm,e, le'\e) /pK'mje, le'Ze/ premier lger] ere are possible uctuations for -er +
V- and for the spelling (towards /e/): (&}le'\e '~i, &Pp#&m,e e't:\) and for P
Z (towards /E/): (pe'zi:, p-) /ple'ziK/ plair
4.1.2.2. On the other hand, the isolated forms often inuence the contextual-
ized ones, even in checked syllables, as in (&\evizi'tl n'vi:, -tel) /Zevizi'tEl na'viK,
-tel/ j'P vit le navire also for -ez: (&vl'v k'ny) /vula'vEK kO'ny, -eK/ vo l'avez
reconnu, and even for the proclitic -es monosyllables: (t'p:, te-) /tEK'pOK, teK-/
t reports (s's, se-) /sEl's, sel-/ s leons (&lsp'tka, les-) /lEspEk'takl, les-/ l
spectacl (dp'n, dep-) /dEp'n, dep-/ d pne also (d'mi, ed-) /Ed'mi, ed-/ et de-
mi
Lastly, it is evident that the eects of vowel adjustment, in unchecked syllables,
can go back further than the pretonic syllable, as long as syllables with dierent V
timbres do not intervene: (&epe'te) /Kepe'te/ rpt (&p't) /KEpE'ta/ rpta (&be-
e',e) /bege'je/ bgZer (&b'm) /bEgE'm/ bgPement
For // too, (not to be confused with (`), zero), in unchecked syllables (phon-
ically of course, and always independently of transcriptions easily found), we have
() // + a close(er) V and (#) // + an open(er) V\ (p#'v:) /pl'vwaK/ plS-
voir (p'e) /pl'Ke/ plSrer; whereas for /O/ in unchecked syllables, we general-
ly have () /O/, unless it is immediately followed by /z/, or by a syllable with /o/,
or it does not derive from /o/ and in -otion: (&mn'tn) /mOnO'tOn/ monotone (&s-
l'pEt) /salO'pEt/ salopette (b'se) /bO'se/ bosser but (\P'zEf) /Zo'zEf/ Jos#e (bP'bP)
/bo'bo/ bobo (P'ze) /Ko'ze/ rose (&gPs'te) /gKosjEK'te/ grossiret (&emP's) /e-
mo'sj/ motion.
Even for the (unstressed) V spelt , au, the more frequent pronunciation is /O/:
(&pi'tl) /Opi'tal/ hpital (':) /O'KOK/ aurore (m'is) /mO'Kis/ Maurice (-
'v:) /OK'vwaK/ au revoir In aujourd'hui the preposition combined with the
denite article [au +le] holds; instead, it is the second syllable that gives more,
due to /uK/, as well: (&P\'_i, -\-) /oZuK'di, -OK-/; but we can hear (&\-, &-
\-) /OZOK-, uZuK-/ too.
For bcoup [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) /bu'ku/. Even surtout and au fur et
mesure present us with the frequent colloquial pronunciation (s'tU, P&f#m-
'zy:), for (sY'tU, P&fyem'zy:) /syK'tu, ofyKeam[]'zyK/.
Likewise, for /, /, the base form is quite important: (d'z,Em) /d'zjEm/ dS-
156 a handbook of pronunciation
xime (b#'e) /b'Ke/ bSrrer In both cases, in checked syllables, we nd (#) //,
() /O/: (s#l'm) /sl'm/ sSlement (ps'te) /pOs'te/ poer
Above, we have already dealt with /^, / 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, , ) /i, y, u a , /.
Consonants
4.2.0. e table of 4.2 shows the French consonant articulations, which are
necessary for the right pronunciation of that language.
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 dierent 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/): (m'm) /ma'm/ maman (n'n) /na'n/ nanan (p'~,e) /pa'nje/ pa-
nier ('NP, '~,P) /a'No/ agn (&sN'm -~,#-) /sEN'm/ ensRgnement (m'tN
-~,) /m'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): (k'pi, -); or even pal-
atal, (-iN) (above all for the elderly); we also nd (-ig, -i, -iN) (a more autoch-
tonous version).
For (tautosyllabic) /nj, n/, we have (~,, ~): (p'~,e) /pa'nje/ panier (y'~,l) /y-
'njOl/ une yole ('~i) /'ni/ ennui. (Popularly, we nd that /nj/ and /N/ tend to
merge into (N): (m'~,E:) /ma'njEK/ manire becomes (m'NE:, m-), 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'pm) /yn'pOm/ une pomme (yn'b:) /yn'bk/ une ban-
que (yn'mE:) /yn'mEK/ une mre (&ynp'sn p's:t) /ynpEK'sOn pa'sjt/ une per-
sonne patiente (yn'fi,) /yn'fij/ une lle (&ynv'li:z) /ynva'liz/ une vale (yn'/E:z) /yn-
'SEz/ une >ae (yn'No:l) /yn'Nol/ une gnle (mn') /man'k/ mannequin (&yn-
'Es) /ynka'KEs/ une carse (yn'gt) /yn'gut/ une goutte and (&yn+y'lip) /ynty'lip/ u-
ne tulipe (yn&_,g'nl) /yndjagO'nal/ une diagonale (&yne'p:s) /ynKe'ps/ une r-
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, , ) and (n) + (k, g , ): (y~'No:l, m8', &y8'Es, yn'gt,
&yne'p: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 ('pis))
/'pKism/ prme
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/ (, ) (more appropriate symbols are ((, 8)), but it is
not absolutely necessary to use them), before front V (including /a, /), before /j,
/ and even at the end of syllables or rhythm groups, before a pause. In the other
cases, the velar articulation, (k, g) can also be prevelar, ((, )), but, again, it is
not necessary to use special symbols.
Let us take a look at some examples: (p'pe) /pu'pe/ poupe (be'be) /be'be/ bb
('tt) /'tut/ toute (d'dn) /du'dun/ doudoune (+y'_i) /ty'di/ tu d ('+) /'tj/ tiens
(k'_i:) /k'diK/ conduire (e'te e'+e) /e'te/ t ('d '_) /'d/ dSx (T'/U)
/kaut'Su/ caout>ouc (&D\'+if) /adZEk'tif/ adjectif (&s3m'its) /smO'Kits/ SPnt-
-Moritz (pid'z) /pid'za/ pizza (pi'ni) /pik'nik/ pique-nique ('k) /'kOk/ coq (-
') /ga'ga/ gaga ('l:) /'lg/ langue ('gP) /'gKo/ gros ('gi) /'gKi/ gr ('k) /'klak/
clac! Before tautosyllabic C (k, g) remain unchanged ({pro}velar), as can be seen
in some of these examples.
We nd complete voicing assimilation to the second element (if diphonic), in
cases such as: (&n'dt) /anEk'dOt/ anecdote (&T/P'se) /KEdSo'se/ rez-de->ase
(mt's) /mEd's/ mdecin (&ps'ly) /absO'ly/ absolu (&sd'dm) /sEt'dam/ cette da-
me ('kb d#/'pN) /'kup dS'paN/ coupe de >ampagne (/g'\:) /Sak'ZuK/ >a-
que jour (&v'v) /avEk'vu/ avec vo.
In these cases, a slowing down of the rate of speech can lead to only partial assi-
milation: (&n'dt, &/P'se, m's, &s'ly, &s'dm, 'k d#/'pN, &v'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: (&v'n) /avEk'vu/ avec no
(&v'1i) /avEk'li/ 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 /J, /, the more legitimate 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 ('s) /'s/ cent ('v:z) /'vaz/ ve ('/) /'Sa/
>at ('p:\) /'paZ/ page
Voicing assimilation (to the second element) also concerns diphonic pairs of
constrictives: (nv'z) /nuf'z/ no faons ('P;s pfy'me) /'Koz paKfy'me/ rose
parfume ('v\ 'v,E,) /'vaS 'vjEj/ va>e viRlle (/'sE) /Z'sE/ je sa but we have (/f) /Sv/:
(/'fl) /S'val/ >eval (/'fe) /aS've/ a>ev Slowing down, we can even have (n'z,
4. french 159
'P; pfy'me, 'v? 'v,E,, ?'sE /'l, /'e). On the other hand, speaking quickly, we
can nd cases such as: (/'p) /Z{n}sE'pa/ je (ne) sa p.
About /j/, it must be said immediately that, more than a real constrictive, it is
a semi-constrictive, (,), indeed, it is placed half-way between the approximant,
(j), and the truly (voiced) constrictive, (J) (denitely rarer in world languages):
(',E:, i',E:) /'jEK, i'[j]EK/ hier (',e) /ka'je/ cahier ('pe) /'pje/ pied (,P',P) /jo'jo/ yo-
-yo (&t#',e) /at'lje/ atelier (f'mi,) /fa'mij/ famille (s'lE,) /sO'lEj/ solRl (f#'t)
/fj't/ fSilleton Before a pause, we often have (): (f'mi, s'lE). French phono-
tactics, contrary to that of other languages, also has /Sj, Zj/: ('/) >ien (e'\,) r-
gion
Sequences such as //0Ki', 0li'// are given with (i',), and therefore even the
most appropriate and most modern phonemic transcription gives /i'j/: (pi',e)
prier (pi',:) priSr (pi',) plia (&sbli',e) sablier Hence, (bi',) stands for
both brillant and Briand but we could possibly have (bi') /bKi'/ for Briand
in controlled pronunciation. Furthermore, we have: (pe'i, pe',i) /pe'[j]i/ pZs (&be'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): ('si) /'si/ su ('~i) /'ni/ nuit ('i) /'li/ lui ('li) /'lwi/ Lou
('m) /'mwa/ mo ('p#) /'pwa/ po ('s#:) /'swaK/ soir ese examples also
show assimilation of voicing and place of articulation, as well as the important fact
that // and /w/ are two dierent phonemes, dierent 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 //'i, i'//, which normally become /'ji, i'j/, 4.2.3): (&psty'e)
obruer (gly') gluant (k'e) cloue (t') trouant
Laterals
4.2.5. ere is only one lateral phoneme (today, 4.2.1), (l) /l/, which assimi-
lates for voicing (and, before /j, /, for place of articulation): ('lyn) /'lyn/ lune
('p#l) /'pwal/ poil ('le) /a'le/ aller ('bl) /'bl/ blS ('ke) /'kle/ clef ('fy) /'fly/ ux
('p) /al'p/ alpin ('z:pa) /Eg'zpl/ exemple (':ka) /'kl/ oncle (':ga) /'gl/ on-
gle ('sfa) /'sufl/ soue Slowing down, or enunciating, we can also have (l): (':ga
-l), between a voiced C and a pause. Sometimes, we can hear something in be-
tween, with (): ('k, ':g). In /lj, l/ the articulation is prepalatal: (s',e) /su'lje/
soulier (',) /'lj/ liS ('i) /'li/ lui. Often, in non-neutral pronunciation, /lj/ and
/j/ tend to merge into /j/: (mi', mi',) million (mi', mi',) miliS therefore (s-
',e) soulier and (&fyzi',e) filier can correspond to (s',e) souiller (&fyzi',e) filler
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 faons); whereas, with regard to phone groups or connected speech the
phenomenon of liaon is typical ( 4.3.3.1-3).
e (unstable) // phoneme
debatable. More often than not, it seems that // (#) is introduced, in pronuncia-
tion, to avoid long sequences of C which turn out to be dicult to pronounce.
erefore from a phonemic transcription such as //msj, dd, at'lje//, , we could
actually have: (m#'s, d#'d, &t#',e) monsiSr dedans atelier as isolated forms;
in connected speech, the most common forms are, for instance: (}'s) /^m'sj/
un monsiSr (ld'd) /lad'd/ l dedans whereas atelier remains the same.
We could, on the other hand, start from a full or isolated 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 //, without complicating the pronunciation with dicult
or impossible groups: //bOn'tKi// bonnetrie //ZtlK'di// je te le red for the
normal (bn'ti, &/t#l#'_i &\#t#'_i). Generally, in phonemic transcriptions in
dictionaries, the // 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 // 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/, , as in: (ss'tis) /sOls'tis/ solice (&z's)
/Egzak'sj/ exaction (&ksk'me) /Ekskla'me/ exclamer (&kspi'me) /EkspKi'me/ ex-
primer (ks'p) /Eks'plwa/ exploit (p'k#) /puK'kwa/ pourquoi (p'sit)
/puK'sit/ poursuite (&syps't) /sypEKs'tKa/ superrat (&ipmne'zi) /ipEKmne'zi/
hypermnsie.
It is generally possible to increase the number of C in contact, if when dropping
//, constrictives, approximants, laterals and nasals (but also stops) are added
before or after: (is'sE;t 's &is#'s-) qui se sert de a? (\#n&l#_i'p, \&n#l#-) je ne le
red p (&ynp+it'fi,) une petite lle (&+yns'p) tu ne ser p ([&i],&bPkd'm:d)
il y a bcoup de monde (&np'lE #t's) on ne parlPt que de a (&s#&\#s'vE,
z&\#-) e-ce que je le sava? (\m'fi/) je m'en >e!
Furthermore: (/&k'b,) je cro bien (s&np'sy:) ce n'e p sr (s&i[]t#'fP,
sit'fP) ce qu'il te faut (s&t3'l) ce trPn l (/&sik'td l'v:) je su content de la
voir [de l'avoir] (/t#ld'm:d, /&t#ld#'m-) je te le demande (&s\#n&t#ldv'p, -d#v-,
s/&t#l) c'e que je ne te le deva p (lve'v, &l#-) levez-vo! (m&nem'l, m#&ne-)
menez-moi l! (\'_i kem'+iv dl'fn sp'b, -+if) je d que l motifs de l'enfant
ne sont p bons (&sYl#'b, syl-) sur le banc (si&\#nt#l_i'p, &si/t#l-) si je ne te le d
p (il&m#ldmt'p, in&m-, &inm#l&d#-) il ne me le demande p.
thin a word, we can nd: (bn'ti) bonneterie (s'l) cserole (m'ne)
amener ('pe) appeler (\'e) jouerP (pn'm) plRnement but: (&gl#'tE:) An-
gleterre (&mk#'_i) mercredi (&pl#'m) parlement (&ft#'m) fortement (&-
p#'m) amplement For parle-m'en, we often have (pl'm), which we can also
hear for parlement in fast or slow speech.
4.3.1.3. It is important to bear in mind the dierence that exists in French, for
the /0/ + /m, n, K, l/ + /j/ sequences, that would be too heavy, hence, a stable //
is inserted: (&sm#',e) sommelier (&ns#'m,) no semions (&st#'~,e) centenier
162 a handbook of pronunciation
Taxophonics
4.3.2.2. en, in connected speech, a word ends in /0K, 0l/ and is then fol-
lowed by another which begins with /0/, in slow and wary pronunciation, // is
inserted; but normally, even /K, l/ are dropped: ('t 'fm 't#) /'katK 'fam/ qua-
tre femm (&ynPt'f# y&nPt#-) /ynotK'fwa/ une autre fo (&mddP'tEl 'mEt# d-)
/mEtKdo'tEl/ matre d'htel (l#'pP;v b'nm, 'p- l#'pP;v#) /l'povK bO'nOm/ le
pauvre bonhomme (i's;p #'n &ilm#'s;bl#) /ilm'sbl k'n/ il me semble que
non (&3p'sib d#'fE: -ibl#) /pO'sibl dl'fEK/ impossible de le fPre ('pb _it'li
l#'pp#) /l'ppl dita'li/ le pSple d'Italie
In cases such as /vOtKp'n/ votre pnS apart from a slow and wary (&vt#p-
'n), we also have (vp'n, -t'n, vtp'n), and (&vtp#'n), considered rather un-
educated because of the dierence with its spelling. Decidedly uneducated (and
intentionally joky) is (&s#'pE, -e) for (ks'pE) exprs whereas (s'pE, -e) is rather
fast colloquial.
As seen (in the previous section), we have to have /0j/ in a word (// refers
to sonants: /m, n, K, l/), but it is not valid in sentences: (sl'm,) c'e le mien (&i[]-
fPl'~,e, fP-) il faut le nier (&i[l]z'+En ',) ils en tiennent liS (in'vl ',, &iln#-) ils
ne valent rien
However, one can denitely also have: (&d#n#', 'fE:) de ne rien fPre (&\#nd#-
'm;d# ',) je ne demande rien and even: (&sl#'m,), (&i[]fPl#'~,e, fP-), (&i[l]z'+En#
',), (in'vl# ',, &iln#-). Let us also consider: (&bnpt#'m '/P) bon appartement
>aud (&bn'pt '/P, -t#) Bonaparte man>ot If the context does not create
any ambiguity, we can easily have bon appartement >aud (&bnpt'm '/P).
ter is at times considered less advisable; similarly, for the second forms in: (\l#'v,
\#l'v) je le vSx (\#'+, \m#-) je me tiens (/&tv'_id v#'ni;, -'_i d#v'ni:) je t'a-
va dit de venir (spelling inuences the choice of preferably retaining word-in-
ternal //).
e second forms here are, thus, also uneducated: (/'k #/'k) je cro
(\m'fU &#\-) je m'en fo (l'mE d#l'y, l'mE #dl'y) l mecs de la rue (&dek-
'vi; l#'m:d, &dek'vi; #l'm:d) dcouvrir le monde ('pd l#me'tP, 'p d[]#l-)
prendre le mtro (i'mEt #'pi, i'mEt []#'pi) y mettre le prix (if it could be confused
with ils mettent le prix (i'mEt #'pi), the /K/ does not drop), (i[]'/E/ s#i[l]'v, i[]-
'/E/#s i[l]'v) il >er>e ce qu'il vSt actually, the second forms are often more
natural, but alas dierent from the spelling!
In the case of forms such as (&b'm) aboiement (\n't#) je nettoie (i[l]'v)
ils voient (/pe'e) je pZerP [pPerP] (i[]'s#) qu'ils soient (&#+y'E, #'+E, #'tE)
que tu P forms such as (&b,'m, \n't#,, i[l]'v,, /p,'e, i[]'s#,) and (&#-
+y'E,, #'+E,, #'tE,) are certainly uneducated.
ing) is considerable and typical: (}n&mii'lP) /^dmiki'lo/ un demi kilo (n'm) /ad-
'm/ demPn (tn'mEm) /tud'mEm/ tout de mme (&t3nm'\e) /tKdm'ZE/ en
trPn de manger (yn'g;n me'z) /yn'gKd me'z/ une grande maon (ln'm)
/ld'm/ lendemPn (d'g;n 'dm) /dE'gKd 'dam/ d grand dam (l'/;m d-
'mi) /la'Sb[K] da'mi/ la >ambre d'am (k'm,) /k'bj/ combien (∫'p, &it-,
il&n#-) /ilntb'pa/ il ne tombe p (l'l; m'dEn) /lE'lg mO'dEKn/ l langu
modern (yn'l; 'E:) /yn'lg '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'n,
-n'n, -'n, m3'n) /mt'n/ mPntenant (v3n'd, -n'd, -'d) /vt'd/ vingt-dSx
('b;~ pe&ife'i, -) p-) /'bk peKife'Kik/ banque pri#rique (\#n&m#tm'p, --
'p, -)'p) /ZnmtKp'pa/ je ne me trompe p
Lastly, this assimilation can occur even before a V\ (p'n) /p'd/ pendant (tl-
&mn'l) /tulmdE'la/ tout le monde e l as even before a pause: (tl'mn) /tul-
'md/ tout le monde (m'l:) /ma'lg/ ma langue It can also be found between a
non-nasalized V and N\ (&mnm'zEl) /madmwa'zEl/ mademoelle (&mnm'zEl)
/mEdmwa'zEl/ mdemoell (&#nm'de) /Kdm'de/ redemander (&nmi'e) /ad-
mi'Ke/ admirer (&_,Nns'+i, -n-) /djagnOs'tik/ diagnoic (&\m'm) /Zb'm/
enjambement and adverbial -ment: (fn'm) /fKwad'm/ froidement (v'm)
/vag'm/ vaguement (&kpn'm, -n'm, -'m) /kplEt'm/ compltement
Let us also note cases such as (li'Em p'/, -mp p-) /lwi'kEnd pKO'S/ le week-
-end pro>Pn
4.3.2.6. In uent, 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'tm) /sE'tOm/ cet homme (st#'fm) /sEt'fam/ cette femme (s't:) /asE'tK/ cet-
te hSre (st'_i:) /sEta'diK/ c'e--dire ({m}p&tt, -tX) /{mE}p'tEtK/ [ma] pSt-tre
(&:) /a'vwaK/ avoir ('l, v'l) /vwa'la/ voil (vlPT'/P:z, &l-) /vwalaotK'Soz/
voil autre >ose (&s, )'s-, p's-) /m'sj/ monsiSr (&kte'm) /ekute'mwa/ cou-
tez-moi (syz'm) /Ekskyze'mwa/ excez-moi (s&pE) /silvu'plE/ s'il vo plat
In a preintoneme, /swa'st/ soixante is currently reduced to ('s#;t), especially
in compounds: (s#t'sis, &s#d_iz'nf, -n_-) 66 79
Furthermore: (ps'#, &ps#, &ps, &s, s#, s, sk) /paKs[]k/ parce que (&s-
#, s#) /Esk, sk/ e-ce que {(p'k#s #&+y_i's) pourquoi e-ce que tu d a?]
(si'si) /sli'si/ celui-ci (&pi) /'pi/ pu (&pis#, -s, -sk) /'pisk/ puque ([]'b)
/[E]'b/ [eh] bien (bin bi'n would be better), ('p) /'ply/ pl (possibly: p'] (m-
&f) /mE'f/ ma enn [m'enn] (m&l;) /mEa'lOK/ ma alors [m'alors] (&b) /'b/
bon! (ban! would be better), (&n) /'n/ non! (nan! would be better), (', -E, -e, -)
/*'wE, -e/ oua! (&k;) /'kOK/ encore (st#&m, /t#-) /Zyst'm/ juement
Other examples: (dm&~,E;, t-) /dtutma'njEK/ de toute manire (tf&s) /d-
tutfa's/ de toute faon (&k#) /puK'kwa/ pourquoi (+yl&m) /natyKEl'm/ natu-
rellement (s&) /sl'm/ sSlement (D&\:) /tu'ZuK/ toujours (s&p, &p) /nEs'pa/
166 a handbook of pronunciation
Liaison
4.3.3.1. For the all-important liaon, it must be said that it concerns, to dier-
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 nd current conversation (real
language). Liaon 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]\
(&lez'mi) l am ( m t s d c] (&d'z:) dSx hSr (&+il'le) e-il al-
l? (&}g'tm) un grand homme (&pet'tE:) pied--terre (&ne'te) en t (&mn-
'mi) mon ami ( ton son once (&mn-) was neutral, but today it is no longer so,
although this pronunciation is still widely used), (&n't) on attend (&}nP'tEl) un
htel ( aucun htel] (&,3n'fE:) rien fPre (&b,3n'se) bien sez
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):
(&bn'mi) bon ami (&p'nE:) en plRn Pr (l#&m,'n:\) le Moyen-ge
We never nd liaon after et, or before the names of numbers or a disjunctive
h, (or aspirated h, because it was pronounced like that centuries ago!) and gen-
erally, before w y\ (e'l:) et alors (ie'El) lui et elle (&is':z) ils sont onze (l#'^)
le un (l#'it) le huit (dee'P) d hros ('P) en haut (&}is'i) un whky (}',P:t,
}',t, }',) un ya>t and (}'i) un oui, as well.
Even in current conversation, it is obligatory to link a noun with the determi-
nants that precede it: (le'zm) l homm (&sez'zP) c ox (te'z,) t ySx
(le&zPt#'zm, &lezPd'zm) l autr homm (&l#z'mi) lSrs ami (&dz'mi) dSx
am (°'zm) d grands homm (}>'mi) un grand ami
Subject pronouns and verbs are linked: (&nz'v) no avons (&vz'le) vo al-
lez (il'zEm, i'zEm) ils Pment ( (i'lEm) il Pme] (&n'vE) on avPt (\&nep'le) j'en
P parl (&+ilv#'ny, &+iv'ny) e-il venu? (&+il'vy, &+i'vy) ont-ils vu? (p't) pSt-on?
4.3.3.2. ere is a case in which pronunciation does inuence 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: v-y (v'zi), pens-y (&ps#'zi), mang-en
(&m\#'z)
As well as forms such as va-t'en! (with elision of the pronoun te) we nd analog-
ical euphonic t's in questions whith subject-auxiliary inversion: e-il? but P-
me-t-on? viendra-t-elle? convPnc-t-il?
Similarly, the adverbs trs tout bien are linked to adjectives (or adverbs) modi-
ed by them: (&tzy'+il) trs utile (&tt'+e) tout entier (b,3&nt'+if) bien attentif
Instead, p pl moins trop fort sez jama can link or not; but in normal
conversation they do not usually do so: (&p[z]'k:) p encore (&tPett',
t&pe-) trop troitement In xed expressions they obviously link; for instance,
pl as in (&pyz'm) pl ou moins which is always so.
(Monosyllabic) prepositions and conjunctions link to the forms that follow
them: (&ni'vE:) en hiver (&dz}'m) dans un mo (s'zEl) sans elle (/e'z) >ez
Sx (/e&z}n'mi) >ez un ami (k&+ilv'ny) quand il e venu (but linking is only
possible with the interrogative adverb, as in (k&[t]silv'ny) quand e-ce qu'il e
venu? but not in ('k +ilv#'ny, +iv'ny) quand e-il venu? so as to avoid /tt/).
If the prepositions and conjunctions are polysyllabic, a liaon is equally only possi-
ble: (&p&v/'te, &pz&v/'te) aprs avoir >ant (d#&vynme'z, -+y-)
devant une maon ere is linking with dont en, too: (l#/'fl d+i&lp'le) le >e-
val dont il a parl and en y have a pre-liaison: (&p#n'z) prenons-en (&nzi'sm)
no y somm
th auxiliaries and semiauxiliaries, liaon, 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 all\
(s&t3p'siba) c'e impossible (i[]&st'le) ils sont alls (&i[l]l'+) ils l'ont S (e ils l'ont
tu). ere are even some xed expressions which require linking: (le&/zeli'ze) l
amps-lys (&leze&tzy'ni) l tats-Un (&viz'vi) v--v (d#&tz't) 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 dierences. We
normally nd: (&m\e'k:) mangez encore (yn&p#+it'fi,) une petite lle (&lmit-
'pE:) l'ami de Pierre unless there are two rhythm groups (to emphasize the rst
part, for a particular reason): (m'\e 'k:, &ynp#'+it 'fi,, l'mi d#'pE:). Let us al-
so note such as cases: (&lm'sy:) la morsure (l&m'sy:, l'm; 'sy:) la mort
sre (&sde'gt) on s'en dgote ('s de'gt) on sent d goutt (&lk'l:) la
coulSr (&_yv3':\) du vin rouge (&lk'l; _yv3':\) la coulSr du vin rouge
(\v'dE _y'v ':\) 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 precise, text-book; it is
actually, a kind of mid-way between two kinds of transcriptions for the examples,
seen above, respectively with one or more intonemes: (m&\e'k:, ynp#&+it'fi,, l-
&mit'pE:). In reality, as we have seen, there are certain rhythmic needs, for which
the use of secondary stresses is quite dierent.
Intonation
4.3.5. 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 ve. Indeed, for partial questions, as well
as the normal (echo-like) anticipation of the interrogative rise, we nd a general
falling movement, from half-high (similar to that of an imperative preintoneme,
except for the obvious and correct dierence of echo anticipation).
A more structured solution, rather than adding a fth preintoneme, would be
a split in two (something like // and //, or even // and //, or //, or
//), leaving the task of clarifying the oddity to observation explanations.
We prefer a more concrete approach ( 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 signicantly raised, and the auditory eect 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 mediatic accent, such prominence is increased by the
half-lengthening of the V in the pretonic syllable (again, without any particular
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 mid parenthesis, as well ( 13.24 of NPT/HPh):
/./: (\v'le Psi%nem3 3) /Zva'le osine'ma./ Je vSx aller au cinma.
/?/: ([&s#]v&p%le'b, fsE1 1) /[Esk]vupaKle'bj fK'sE?/ [E-ce que] vo par-
lez bien frana? (p&le%v'b, fsE1 1) /paKlevu'bj fK'sE?/ Parlez-vo bien
frana? (&1km tlev3 3) /&kO'm tale'vu./ Comment allez-vo?
//: (&pi%'le v5+y:1 1 %'pe3 3) {(-%,-, %-)} /pia'le vwa'tyK ua'pje./ On
pSt y aller en voiture, ou 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 dierent char-
acteristics.
International accent
4.4.1.1. For teaching purposes, the international accent could be even more
appropriate than the neutral accent seen so far, inasmuch as it is considerably less
inuenced by Paris and, therefore, undoubtedly nearer to many other varieties of
pronunciation, once any marked peculiarities have been eliminated.
erefore, an international 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 dierent types (including the usual interference, not only phonological,
but due to spelling).
e vocogram in this section gives the international vowel articulations
which, as can be observed immediately from the comparison with 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, (, ^, , ). In the gure there are two grey markers, for /E, , O/ ( un-
stressed a third would be for /, ^/, which, however, coincides with //); 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 ((,
, #, }, ), which we prefer to use); or the higher-low timbre ((E, , , ^, 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 /K/: (&oZu'di) aujourd'hui
avoiding devoicing too.
Let us look at some examples, only for the cases in which there is a phonetic
dierence with the neutral (given in round brackets): ('lu) {('l)n} loup ('o) {('P)n}
4. french 171
('nOt) {('nt)n} note ('la) {('l)n} l (d#'d) {(d#'d)n} dedans ('bj) {('b,)n}
bien (bj'to) {(b,3'tP)n} bientt (pa'ti) {(p'+i)n} parti
Mediatic accent
It also manifests some uctuation both towards and away from neutral; how-
ever, we show it here in its most typical form, indicating some internal dierences
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 denite-
ly gradations (comparing it to g 4.1, of neutral); here, we bring your attention
to the major dierences. e fronting of /u, O/ (%, @) (and /O/ (), unstressed)
is evident; the (anti-clockwise) rotation of /, , / (a, _, ), (and // ()), is just
as evident, as is the absence of /^/ which merges with // (a), or varies between
the two, even with an intermediate realization (not shown explicitly in the voco-
gram) (8), with slight rounding. (In the uctuations towards the neutral accent,
we have (), as well; or (), with a slight derounding).
e /EK, K, OK/ sequences often shift back one box: ([:], @[:], [:]
). Furthermore, note in the vocogram the values for a\ (even if with individ-
ual uctuations) the distinction is generally maintained between /a/ and /A/,
which traditional neutral had adopted. However, the timbres (of Parisian and) of
the mediatic accent are: /a/ (), but () for /aK[0], wa/, and /'A[0]/ (), but /A/
(A) (in the banliSe we have /wa/ (A, '), often /aK/ (:) ar(C) even more
often /aj/ (:,), but (s) for /Asj/ -ation of traditional neutral).
Some examples follow (where t indicates traditional neutral): ('t) {('tU)n} tout
('n@t) {('nt)n} note (b'n@:) {(b'n:)n} bonhSr (&P\T'_i, --) {(&P\'_i)n}
aujourd'hui ('b,a) {('b,)n} bien (\'t_) {(\'t)n} j'entend ('b) {('b)n} bon (P'a,
-8) {(P'^)n} aucun
Others: ('p:) {('pE:)n} pre ('p@:) {('p:)n} pSr ('p:) {('p:)n} port ('p@
's:) {('p 's:)n} Paul sort (p'i) {(p'i)n} Par (p'+i:) {(p'+i:)n} partir
(v'l) {(v'l)n} voil ('s#:) {('s#:)n} soir (':) {(':)n} car Also: (p'p)
{(p'p)n} papa (pA's:\) {(p's:\)n} psage ('g) {('g)n, ('gA)t} gr ('g:s)
{('gs)n, ('gA:s)t} grse ('_,:ba) {('_,ba)n, ('_,A:ba)t} diable
For the banliSe accent; (vA'l) {(v'l)n} voil ('s#A:) {('s#:)n} soir ('k:)
{(':)n} car ('p:,) {('p,)n} pPlle (t'v:,) {(t'v,)n, (-A:,)t} travPl (p&ns-
's) {(p&ns's)n, (-sA-)t} prononciation
ere is a strong tendency to merge /e, E/ 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)
{('y)n} rue ('t%) {('t)n} trou ('tx) {('tX)n} quatre we must add that on an
uneducated level, the palatalization of /t, d k, g/ is much more evident, with ar-
ticulations going from stops to stopstrictives, (, , k): (y'i) {(+y'_i)n} tu
d ('mEk) {('mE)n} mec ('id) {('id)n} guide; /w/ is provelar rounded, ():
('m) /'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): (%p;'i) {(%p'i)n} Par (&p-
%i;'z,a) {(&p%i'z,)n} parien (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 mediatic one there is also the half-lengthen-
ing, seen above). e tonogram gives the intonation of the mediatic 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 international and mediatic 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 //. 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 //), where we nd (, #, ).
For /a/ there is a central timbre, (a). e articulation of // is unrounded cen-
tral, (), 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, /, ^, , /, 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
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/ (, ), /j/ (j), /w/ (), /N, nj/ (~j), /Ni/ (ni), /lj, l/ (j, ), /'[0]/+/j,
, w/ ([0])+(i', y', u'), /K/ (r) (even (), especially before a consonant; and,
in less broad accent, also (K, )}: (M'v;r) {('vE:)n} envers ('bak) {('b:)n} ban-
que ('ri) {('i)n} Henri ('r) {(\#'/E/)n} je >er>e (&trava'je) {(&tv',e)n}
travPller e banlieu accent has /t, d/ (+, _) + /i, j, y, /, and /K/ tendentially voice-
less (3) (and (x, ', )).
Furthermore: (mon'ta~j) {(m'tN)n} montagne (&ma~je'rizm) {(&m~,e-
'is))n} manirme (je'zo) {(,e'z)n} liaon (ki'zi;n) {(i'zin)n} cuine
(&sy'i) {(\#'si, /'si)n} je su (ly'i) {('i)n} lui (lu'i) {('li)n} Lou (bi'e)
{('b,)n} bien ('ra;r+, 'a;+) {(':)n} rare Hence, (j) is maintained well only be-
fore a non-nal V; indeed, we generally have: ('pi) {('pE,)n} pZe (pe'i) {(pe',i)n}
pZs
Consonant clusters are simplied, as in: (di'r;k) {(_i'Ekt)n} direct (di'r;t) {(_i-
'Ekt)n} directe (os'ky;r) {(p'sy:)n} obscur (&aek'ti;f) {(&D\'+if)n} adjectif (se-
'tambr+) {(sp't:bX)n} septembre (a've) {('vE)n} avec but: (s) 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. Liaon 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 disjunc-
tive h, hence, the use of linking and elision, : (le'z;m) {(l'zm)n} l homm
(&meza'mi) {(&mez'mi)n} m am (&noza'mi) {(&nPz'mi)n} nos am (&lrza'f;r+)
4. french 175
(&pA-)t} ptsier we also have an expansion of its frequency, to practically all cas-
es of /a|, wa|, av|, az|, as|, aZ|, aj|, a|, aK|, aK/ (with possible uctuation, especially
for /av|, az|/ and -ation which have both /A/ and /a/). ere is actual neutraliza-
tion too, as for: ('k) which goes for both (') {('kA)t} c and (') {(')t} k how-
ever, the contextual distribution with /A/ in an intoneme leads to situations
such as: ('s '\E 's) a j'Pme a
ere is uctuation between (w|) and (w|) (the latter is generally considered
to be more uneducated), up to (w), decidedly dialectal: ('mw 'mw 'mw) moi
e old opposition between /E/ and /E:/ is stable, we have for example: ('mEt5)
/'mEtK/ mettre ('mEt5) /'mEtK/ {/'mE:tK/t} matre ('fEt) /'fEt/ fPte ('fEt) /'fEt/
{/'fE:t/t} fte ('sEn) /'sEn/ sPne ('sEn) /'sEn/ {/'sE:n/t} scne (which now, in neutral
European French, is only: ('mEtX, 'fEt, 'sEn)n); whereas for the new opposition
constituted by /e:/, we have examples such as: ('neI\) /'ne:Z/ nRge ('bReI)
/'bKe:k/ brque (or brake] (/'nEZ, 'bKEk/n ('nE:\, 'bE)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 ('R\, 'R\) /'KaZ/ rage (('li:vX, 'py:, ':\, ':\)n). e
same occurs for /E:, e, , o, A/ in an intoneme for any checked syllables (without
repeating examples): ('fYt5) /'ftK/ fStre ('\oUn) /'Zon/ jaune (('f:tX, '\P:n)n), and
even in a preintoneme for unchecked syllables (again, without repetition): (&Re-
'te) /aKe'te/ arrter (neI'\e) /ne'Ze/ nRger (fY'te) /f'tKe/ fStr (\oU'nt5) /Zo-
'natK/ jauntre (f'/e) /fa'Se/ f> ((&e'te, ne'\e, f'te, \P'ntX, f'/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 rst vocogram of 4.6.3 shows (e',
#', o')): (&qIi'Re) /ati'Ke/ attirer (\Yy\'mA) /ZyZ'm/ jugement (&epUu'ze) /epu'ze/
poe (te'Re) /tE'K/ terrPn (b#'Re) /b'Ke/ bSrre (Y'R) /'K/ hSrSx
(&Rlo'\ER) /OKlO'ZEK/ horlogre (&ek'ze, --) /ekKa'ze/ cr or if they are /e, ,
o/ (even when non-derived or in syntagms): (leI'pe) /le'pje/ l pieds (dY'me)
/d'm/ dSx mPns (soU'f) /so'fa/ sofa
Nasalized V diphthongize in an intoneme ( both in tonic and pretonic sylla-
bles), but they are short in a preintoneme: ('fe) /'f/ n (se'qYyR) /s'tyK/ cRnture
(o'}J) /o'k^/ aucun (}J'\UuR) /^'ZuK/ un jour (m9'n9) /m'n/ mon nom (\A-
'tA) /Z't/ j'entends (&fi'ni) /fi'ni/ inni (&}n'mi) /^na'mi/ un ami (&mn'mi)
/mna'mi/ mon amie (&ntA'dA) /nt'd/ en entendant (('f, s3'+y:, P'^, }-
'\:, m'n, \'t &3fi'ni, &}n'mi, &mn'mi, &nt'd)n).
k.s'qI) /akus'tik/ acouique {(fi'lip, y'ni, k'zin &pimi'+i:v, &\yi'_i &minis'tE:, &-
ks'+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: (&kfN'qYyR) /kfi'tyK/ conture (&qNs'qI) /aKtis-
'tik/ artiique (&QIsp'te) /dispy'te/ dput (&deku'p\) /deku'paZ/ dcoupage
{(&kfi'+y:, &+is'+i, &_ispy'te, &dek'p:\)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/): (&pfe's#R, pf's-) /pKO-
fE'sK/ profsSr (&Ynivsi'te, &Ynvs'te) /ynivEKsi'te/ universit (&bili'te, &bI'te)
/abili'te/ habilit (&mNI'fI, m'-, mJ'-, m9'-) /maNi'fik/ magnique (bI'zR, b'z-)
/bi'zaK/ bizarre (p&pyl's9, &pp-) /pOpyla'sj/ population (e&pyR's9, &e-
p-) /epyKa'sj/ puration (&buj'ne, bj-) /bujO'ne/ bouillonner (&pin'je, pn-)
/pina'je/ pinPller (&Re'je, R'je) /OKe'je/ orRller (&lk'mOd, l'Od) la commode
{(&pf's:, &yni&vsi'te, &bili'te, &mNi'fi, bi'z:, p&pyl's, e&py's, &b,'ne,
&pin',e, &e',e, &lk'md)n). Even // drops more often than in European French:
('lY s+p'mEn, 'lYks p-) Luc se promne {('ly s#p'mEn)n}, and presents even dis-
tributions which are often considered uneducated, like: ('fER+ '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} Pm 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/): (p'~je) /pa'nje/
panier (A'~i) /'ni/ ennui {(p'~,e, '~i)n); /N/ varies between (N) and (J):
(m9'tN, -'tJ) /m'taN/ montagne (&sN'mA, &sJ'-) /sEN'm/ ensRgnement
178 a handbook of pronunciation
4.4.4.5. e groups with nal C are often simplied, above all in less controlled
language, even in the case of /0K, 0l/ (before a pause too): ('wEs) /'wEst/ oue (&-
/i'tE) /aKSi'tEkt/ ar>itecte (&ss'lis) /sOsja'lism, -ist/ socialme -ie ('mYs)
/'myskl/ mcle ('9, '9) /':gl/ ongle ('Es) /OK'kEstK/ or>ere ('9b, '9m)
/'bK/ ombre on the other hand, especially on an uneducated level, we can have
/`/ = /t/: (\i'jEt, \I-) /Zi'jE/ juillet (qi'Et, qI-) /ti'kE/ ticket (d+'bUt) /d'bu/ de-
bout ('pEt) /'pKE/ prt ('~It) /'ni/ nuit ('lIt) /'li/ lit (k9'pEt) /k'plE/ complet
(bRy'nEt) /bKy'nE/ Brunet (&mRi'sEt) /mOKi'sE/ Morset (tl'bOt) /tal'bo/ Talbot
Cases such as (p'qi t'mi) /p'ti ta'mi/ petit ami and (p'qI t'mi, p'qIt 'mi) /p'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 dierent
normalized versions. We start with the (neutral) French pronunciation of (neu-
tral British) English this is the rst 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 international accent, followed by the mediatic/Parisian 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, uent 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 rst.
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 0.12).
French Text
Canadian pronunciation
4.6.0. We would like to add some further information: about the aected Pa-
risian accent, in the form of phonosynthesis; and about the Parisian variants of the
banliS compared to mediatic Parisian ( 4.4.2.1-2 4.5.2.3); lastly about
Canadian variants ( 4.4.4.1-5 4.5.2.5). In NPT/HPh, you can nd the pho-
nosyntheses of Acadian French (acadien east Canada, 21.10) and cadien, Cajun
(from Louisiana, 21.16).
Aected 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 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/ (w, w, w) (dialectal (w)); as also /EK0/ = /aK0/
is dialectal; /u/ (, , , P, P), /O/ (, , ). We can observe the use
of (, T, ) (considerably centralized), which is independent from the moder-
ate systematic advancement of back phonemes (of type). We also indicate the
uvulo-postalveolarized velar rounded realization of /K/ (= (<)) and its vocaliza-
184 a handbook of pronunciation
tion through (, +, P , , ) (as the last element at the end of the diphthongs
e triphthongs), in /iK, yK, uK EK, K, OK aK/, given in the rst 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 (OP) of /O(:)0/). In rural accents, and above all for speakers
who are not so young, /S, Z/ (/, \) have dierent, peculiar and broad realizations,
as well as (S, Z): (, S, q , ', h, ', H). On an uneducated level, before a
rounded V /f/ = (). e French may have diculties in understanding Canadian
French, especially if the accent is very broad and uneducated.
/u/ (, , , P, P)
// (9)
// (, / a3), /^/ (^}) /O/ (, , +, , )
/E/ (), /E:0/ (, a)
/O(:)0/ ()
/a(:)0/ ()
// (, , As0, _, ) /A(:)0/ (, , )
5. German
Vowels
5.1.1. 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 dierence too (except for /a, a:/, whose tim-
bric dierence is now regional). As can be seen from 5.1, too, long vowels are
realized by closer vocoids, and a modern phonemic transcription must obviously
reect this characteristic: /i:, I e:, E a:, a/, /y:, Y :, /, /u:, U o:, O/.
But we have to add another inconvenient 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 5.1 the other variants are shown as separate ones.
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)
// (, , , )
/E:/ (:, ;) {(E: e: e;)}
/O/ (O, )
/E/ (E, ), // (, #)
// () { ()}
/a/ (a), /a:/ (a:)
Let us now see some examples: ('mi:t) /'mi:t/ Mte ('mIt) /'mIt/ Mitte ('ve:k)
/'ve:k/ Weg ('vEk) /'vEk/ weg ('Sta:t) /'Sta:t/ Staat ('Stat) /'Stat/ Stadt ('hy:t) /'hy:t/
Hte ('hYt) /'hYt/ Htte ('h:l) /'h:l/ Hhle ('hl) /'hl/ Hlle In addition,
(': 'E:-) /'E:K/ hre ('e:) /'e:K/ Ehre and ('th:l 'thE:-) /'tE:l/ Tler ('thEl)
/'tEl/ Teller nally, ('zOn) /'zOn/ Sonne ('thUk) /g'tKUk/ getrunken (-
'a:d) /g'Ka:d/ gerade ('Unz) /'UnzK/ unsere ('Unz) /'UnzK/ unserer
e last example (as well as a previous one Tler] shows () // -er which is
more traditionally rendered phonemically as //K// (and even as //r//); but // is
more convenient, since its most normal and frequent realization is as an intense
faryngealized uvular approximant (). It is a fact that a vocalization is also pos-
sible, as () (and this symbol is certainly more appropriate than (), which how-
ever is found in many publications), that is represented by a broken-line white
marker ( 5.1); but it does not occur in a systematic way. As a matter of fact, more
often we nd (), or even (1, &); occasionally, even something like ((, , )) (
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, (, #, ) ( 5.1): (nt'SUldIgU,) /Ent'SUldIgU/ Enuldigu (k-
'tho:b) /Ok'to:b/ Oktober ('phk#lt) /'pKklt/ Prklt Again, in unstressed
syllables, also /i, e y, u, o/ are to be added, even in phonemic transcriptions:
(&miko'fo:n) /mikKo'fo:n/ Mikro#on (&eali'th:t -E:t) /Keali'tE:t/ Realitt (&yna-
'mi:t) /dyna'mi:t/ Dynamit (&kolo'gi:) /kolo'gi:/ kolog (&uni&vzi'th:t -E:t)
/univEKzi'tE:t/ Universitt (&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, ) (except when word-nal): (&mIk'fo:n, &alI'th:t
{-E:t}, &Yna'mi:t, &#kl'gi:, &UnI&vzI'th:t {-E:t}, &fnl'gi:), and (nt'SUldI&gU,,
n-), but: ('Umi) /'gUmi/ Gummi ('zi:mi&le, -I-) /'zi:mile/ Simile ('pha: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): (a'ga:v) /a'ga:v/ Agave (a'me:i&ka, -I-) /a'me:Kika/ Ame-
rika ('a:fi&ka, -I-) /'a:fKika/ Afrika (&afi'kha:t, -I-) /afKi'ka:t/ Arikate Instead,
in certain transcriptions of traditional neutral pronunciation, we may nd /A/,
which somehow corresponds to /A:/; but there is not much coherence at all a-
mong dierent dictionaries and textbooks: (a'ga:v, A'gA:- a'me:i&ka, A'me:i-
&kA, -I- 'a:fi&ka, 'A:fi&kA, -I- &afi'kha:t, -A:t). Except for some loanwords, the
systematically gives /A/ ( /A/) for unchecked grapho-syllables (a strange
criterion indeed, to be frank), whereas Langenscheidt's dictionaries still using /A:/
mark /a/.
5.1.3. 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 (('al) /'al/ alle) and the long (('a:l) /'a:l/ Ahle) ones. e
three genuine diphthongs are (ae, ao, OY) /ae, ao, OY/ (often badly represented as
5. german 187
/ai, au, oyoi/): ('aen) /'baen/ BRn ('ael) /'ael/ Eile ('mae) /'mae/ MP
('aom) /'baom/ BQm ('aog) /'aog/ Auge ('thao) /'tao/ TQ ('lOYt) /'lOYt/ lT-
ten ('OYl) /'OYl/ 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, 'lIdi) /'leIdi/ Lady ('e:, 'I) /'geI/ gay (&no:'hao,
&nU'hao) /noU'hao/ Know-how ('So:, 'SU) /'SoU/ Show
e broken-line white markers show two possible realizations of /ae, ao/, in re-
duced forms ( 5.3.1), (e, o), as in: (en) /aen/ Rn (os) /aos/ Qs (also (Y)
/OY/: (Y) /OY/ S>). e marker corresponding to () is also valid for reduced
forms with /a/, as a possible variant: (s) /das/ das. We have intentionally not
placed it in the monophthong vocogram ( 5.1), also to avoid confusion with the
quite frequent, but not recommendable use of () to indicate our (, ) (or,
at best, (), see 5.1 and the rst part of 5.1, and the second part of 5.2).
5.2. German diphthongs.
Consonants
5.2. 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
m (M) (n) n (~) (,)
F pb 47 kg ()
q c
(6) f v
_ s z S Z () (,) /x/ (X )
/j/ () h
([) /K/
(R) l ()
+(, , n) /, / (, , , , , ? :, !, , , , /, , |, ) ( 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 ('thIk) /'tKIk/
trinken ('am&bi;t) /'anbi:d/ anbten ('fYMf) /'fYnf/ fnf ('Uns) /'Uns/ uns (strict-
ly (('Us)), but (n) is sucient), (i'thant) /di'tant/ d Tante (strictly ((i'4han4))),
('vYnS) /'vYnS/ wn(en, (khn'a:d) /kOn'Ka:d/ Konrade, ('mY~) /'mYn/
Mn>en ('an) /'an/ aeln ('lEst) /'lEst/ l ('lE) /'lE/ ler ('zI)
/'zI/ sien ('qhU) /'qU/ Zue ('dIU,) /b'dIU/ Bediu ('qhaetU,)
/'qaetU/ ZRtu ('qhaetU,?) /'qaetU/ ZRtuen (For /U, U/ we have
(U,, U,?).)
Please, note that the sequence /nj/ and /lj/ as well has two dierent realiza-
tions in stressed syllables, for tautosyllabic (after /:/) or heterosyllabic (after //)
elements: ('li:n,) /'li:nj/ Lin (Sam'pha~,) {((-a~-,))} /Sam'panj/ ampagner.
In lexical compounding, even for the negative prex 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 (), labiodental (), velar (n):
('Um&bndI, 'U-) /'UnbEndI/ unbndig ('UM&vIkI, 'U-) /'UnvIKklI/ unwirk-
5. german 189
5.2.1.3. In the previous section, we have shown up to six taxophones for the
two intense nasals, /, /. At rst 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 without reason, even nine could be useful, indeed:
(:, !, , , , /, , |, ). (All this without considering a denti-alveolar ((~)), for
which a simple () is sucient, as in (&aen'gu:t 'thak) ((&aen'gu:4~ '4hak))
/aen'gu:t 'tak/ 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 /, / and the tautosyllabic C before
them, /0M/, determine their exact place of articulation, (, , , , , ?), just
as in the examples seen above. Even the initial C of following syllables aect 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
uent 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 /0/ (0). We can see that
when they are followed by labiodentals (/, f, v/), the coarticulation of () and
(, f, v) produces an intense labiodentalized bilabial ((:)): (('zi:b: 'he:7)) /'zi:b
'e:Kd/ sben Pferde (('zi:b: 'fE4)) /'zi:b 'fEt/ sben Vetter (('zi:b: 'va:g)) /'zi:b
'va:g/ sben Wagen
en /0/ (0) is followed by laminal C, /t, d, q, z, n, l/, we have a denti-al-
veobilabial ((!)): (('zi:b! '4ha:g)) /'zi:b 'ta:g/ sben Tage (('zi:b! '7e:&mak)) /'zi:b
'de:maKk/ sben DM (('zi:b! 'qhIm)) /'zi:b 'qIm/ sben Zimmer (('zi:b! 'nE4))
/'zi:b 'nEt/ sben N>te (('zi:b! 'lEn7)) /'zi:b 'lEnd/ sben Lnder
nally, for /0/ (0) + /k, g, K/, we nd an intense velarized bilabial nasal (()):
(('zi:b 'khaq)) /'zi:b 'kaq/ sben Kaen (('zi:b 'gE)) /'zi:b 'gE/ sben
Ge (('zi:b ':7 'E:-)) /'zi:b 'KE:d/ sben Rder
5.2.1.5. As far as /0/ (0) -em is concerned (which is not derived by assimila-
tion, but is original, for //0m//), three groups deserve careful attention. In fact,
for /0/ (0) + /, f, v/, we nd a labiodental (): (mIt&i;z'he:t) /mItdi:z-
'e:Kt/ mit dsem Pferd (In&di;z'valt) /Indi:z'valt/ in dsem Wald
For /0/ (0) + /t, d, q, z, n, l/, once again, we have a denti-alveobilabial
((!)): ((qhu&7aen!'qhIm)) /qudaen'qIm/ zu dRnem Zimmer ((bae&7i;z!'lIt))
/baedi:z'lIt/ bR dsem Li>t For /0/ (0) + /k, g, K/, a velarbilabial (()):
((fo&maen'khO)) /fo:Kmaen'kO/ vor mRnem Ko ((mI4&aen'I)) /mItdae-
n'KI/ mit dRnem Ri
5.2.1.6. For /0/ (0) + /p, b, m/, for the third time we nd a denti-al-
veobilabial ((!)): (('u:4! 'phaes)) /'gu:t 'pKaes/ guten PrRs (('u:4! 'bal)) /'gu:t
'bal/ guten Ball (('u:4! 'mOg)) /'gu:t 'mOKg/ guten Morgen
For /0/ + /, f, v/, we have a labiodentalized alveolar ((/)): (('u:4/ 'hle:g))
/'gu:t 'le:g/ guten Peger (('u:4/ 'fa)) /'gu:t 'fa/ guten Fa (('u:4/ 'vaen))
/'gu:t 'vaen/ guten WRn For /0/ + /k, g, K/, we nd a velarized alveolar (()):
(('u:4 'khEl)) /'gu:t 'kEl/ guten Keller (('u:4 'gas4)) /'gu:t 'gast/ guten Ga
(('u:4 'o:4&aen)) /'gu:t 'Ko:tvaen/ guten RotwRn
Besides, for /0/ (0) + /p, b, m/, we again find a velarbilabial (()): ((&e;g-
'pho:l)) /ge:g'po:l/ gegen Polen ((&e;g'bElg,, -j)) /ge:g'bElgj/ gegen Belgn
((&e;g'mEksi&ko)) /ge:g'mEksiko/ gegen Mexiko
For /0/ (0) + /, f, v/, we nd a labiodentalized velar ((|)): ((&e;g|'han4))
/ge:g'ant/ gegen Pfand ((&e;g|'fak&ae)) /ge:g'fKakKae/ gegen FrankrR>
(('e:g|&va4)) /'ge:gvaKt/ Gegenwart
For /0/ (0) + /t, d, q, z, n, l/, we have a denti-alveolarized velar (()): (('e:g-
&4hael)) /'ge:gtael/ GegentRl ((&e;g'7ae)) /ge:g'dKae/ gegen drR ((&e;g'qhae))
/ge:g'qvae/ gegen zwR ((&e;g'zEks)) /ge:g'zEks/ gegen se7 ((&e;g'nOYn)) /ge:g-
'nOYn/ gegen nSn (('e:g&laes4U,)) /'ge:glaestU/ GegenlRu
5. german 191
5.2.1.7. In the case of /0/ (0) + /, f, v/, obviously, things do not change:
(mt'Saf 'fo:tos) /mIt'SaKf 'fo:tos/ mit (arfen Fotos equally, for /0/ (0) + /t,
d, q, z, n, l/: (mt'Saf 'qh:n 'qhE:-) /mIt'SaKf 'qE:n/ mit (arfen Zhnen
Instead, for /0/ (0) + /k, g, K/, we have a velarized labiodental (()): ((mt'Sa-
f 'khI)) /mIt'SaKf 'klI/ mit (arfen Klien
For /0/ (0) + /p, b, m/, for the fourth time we nd a denti-alveobilabial
((!)): (('m:+!&pnz,o&na;4, -a-, --, --, -s,- 'mE:+-)) /'mE:tpenzjona:t, -a-, -sj-/
Md>enpensionat besides, + /, f, v/, we have a labiodentalized alveolar ((/)):
(('m:+/&fOYntI 'mE:+-)) /'mE:tfKOYntlI/ md>enfrSndli> then, + /k, g/, we
have a velarized alveolar (()): (('m:+&khas 'mE:+-)) /'mE:tklas/ Md>enklasse
For /0/ (0?) + /p, b, m/, we nd for the third time a velarbilabial (()):
(('qhae4U &publi'qhi:?, -, -,, -n)) /'qaetU publi'qi:K/ ZRtuen publiz-
ren
For /0/ (0?) + /, f, v/, we again have a labiodentalized velar ((|)): (('qhae4U|
f'khaof)) /'qaetU f'kaof/ ZRtuen verkQfen
Lastly, for /0/ (0?) + /t, d, q, z, n, l/, once again we nd a denti-alveolarized
velar (()): (('qhae4U 'le:z)) /'qaetU 'le:z/ ZRtuen 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 /n/ + /[0], [0]/ becomes (n,
n, ) (with progressively stronger nasalization, up to a complete dropping of /n/):
(Ins'khi:no, Is-) /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-nal position (and in postnuclear position, but with some oddities
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 aspirated. Besides, /t, d/ are denti-alveolar (and rigorously the most appropri-
ate symbols would be ((4, 7)), but it is sucient 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/ Papr ('phaq) /'plaq/ Pla
('SpE) /'SpKE/ spre>en ('halp) /'halp/ halb ('apt) /'apt/ Abt (I'phak) /I-
'paKk/ i> parke ('Unt) /'Unt/ und ('lu:p) /'lu:p/ Lupe ('Eb) /'Eb/ Ebbe ('It)
/'bIt/ bitte ('tho:t) /'to:t/ Tod ('the:t) /'tKe:t/ treten ('Sta:s) /'StKa:s/ Strae
('laet) /'laet/ LRd (s'thIkt) /s'tKIkt/ es trinkt ('laed) /'laed/ lRder ('ak)
/'dak/ danken ('khInt) /'kInt/ Kind ('khaes) /'kKaes/ KrRs (s'khi:no) /das'ki:-
no/ das Kino ({ae}k'nal) /{aen}k'nal/ (Rn) Knall ('bIg) /g'bIKg/ Gebirge
('I) /'gI/ gi
('thaphth, 'Ythl), for ('thapt) /'tapt/ tappte ('Yt) /'KYtl/ rttle 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 exploded (('4hap$4$, 'Y4$)); but there
is certainly no need to mark such things.
By assimilation, in /pf, pv/, we have (]) + /f, v/: ('a]&fal) /'apfal/ Abfall (dier-
ent from () //) ('a]&e;t) /'apve:Kt/ abwerten In addition, as we have seen
for /n/ ( 5.2.1), even /t, d/ assimilate to a following C: ('mIk&Ift) /'mItgIft/ Mit-
gi ('mIk&km) /'mItkOm/ mitkommen ('mIp&I) /'mItbKI/ mitbrien
('mIp&nS) /'mItmEnS/ Mitmen( ('o:k&khaof) /'bKo:tkaof/ Brot kQfen
('o:p&I) /'bKo:tbKI/ Brot brien Note that we have (+) in: ('m:+ 'mE:+-)
/'mE:t/ Md>en
5.2.2.3. Let us also consider: ('ap&aot) /'aplaot/ AblQt ('apa&ti;f) /'aplati:f/ Ab-
lativ ('ap&u;f) /'apKu:f/ Abruf (at'Ep) /at'vEKp/ Adverb (&ati'a:l) /atmi'Ka:l/
Admiral ('Sto:f&aen) /'Sto:flaen/ StovlRn ('li:p&o;s) /'li:plo:s/ lblos ('e:k&o;s) /'Ke:k-
lo:s/ reglos ('m:kI) /'m:klI/ mgli> ('ni:tI) /'ni:tlI/ ndli> ('hantU,) /'han-
tlU/ Handlu ('y:pI) /'y:plI/ bli> ('va:knIs) /'va:knIs/ Wagnis but: ('laebnIq)
/'laebnIq/ LRbniz ('e:dn) /'Ke:dn/ Redner ('e:gn) /'ge:gn/ Gegner ('va:gn)
/'va:gn/ Wagner ('a:dl) /'a:dl/ Adler ('a:dlI) /'a:dlI/ adlig ('a:dlIg) /'a:dlIg/
Adlige ('OdnU,) /'OKdnU/ Ordnu ('e:gl) /'Ke:gl/ Regler (zIg'na:l) /zIg'na:l/
Signal (&zIgna'thu:) /zIgna'tu:K/ Signatur ('e:bn) /'e:bn/ ebne ('y:bI) /'y:bKI/
brig
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 // -e-]\
reden gegen Wagen Adel Regel Orden eben ber On the contrary, in the rst
group we have combinations (which may be clear or opaque) with grammemes
( prexes or suxes), so we will consider only three cases: Ab-[lQt] [b]-li>
from ben and [Wag]-nis
Words of lofty or foreign origin generally maintain a voiced stop before sonants:
/bl/ Double moblren 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 Pygme Segment Sigma Stigma /gn/ Agnoiker
Diagnose ignorren indignrt kognitiv Magnat Magnol Magnus Physiogno-
m prgnant Stagnation and /gn, n/ inkognito Magnesit Magnesium magne-
ti( Signal signikant /gl/ eglomisren Hrogly#e Iglu Neglig and /l/ e-
li( Jli lli( and // Sile /gK/ agrari( Agronom Allegro Dia-
gramm Emigrant Migrne Regress Saria Telegraf
But we have: /tm/ Admiral Kadmium In addition: /p, b/ Abort [= Abtritt] ab-
oral abortren 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 glottal stop
too (or glottal catch, but laryngeal stop is better) (): (&yb'al, &Y-) /y[:]b'al/ -
5. german 193
berall ('Ok) /g'OKk/ Georg (more frequently, ('e:k) /'ge:OKk/), ('ael) /b-
'ael/ beRlen (the'a:t) /te'at/ eater (f'aen) /f'aen/ VerRn ('vael []I'aos-
&abaet) /'vael I'aosaKbaet/ wRl i> QsarbRte ('Is 'aoX &aen[]'a) /'Is 'aox
aen'a/ iss Q> Rnen Ael
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, () may be missing but
only in completely unstressed syllables (and we indicate it here as ([])), provided
composition is not evident.
As can be seen, it also occurs within lexemes, in the context /'/ ('), but not
if the stress is weak: ('af) /'af/ Ae ('ap&a;t) /'apa:Kt/ Abart (f'aXt) /f'axt/
vera>ten ('aXt) /b'axt/ bea>ten (ds&avu'i:?, -, -n, -,) /dEsavu'i:K/
desavouren (e'EMfazIs) /de'EmfazIs/ Deem#asis (ant'a:qidUm) /ant'a:qidUm/
Antacidum (&ant[]a'qhi:t) /anta'qi:t/ Antacid ('aos≤z) /'aosle:z/ Qserlesen
('mUnt&a;t) /'mUnta:Kt/ Mundarten us we have: (thea'tha:lIS) /tea'tKa:lIS/
theatrali( but (&thea-) only in a kind of theatrical pronunciation.
In addition, we nd () even if the V is deep or assimilated, as in () //, for
//K//, provided it is an initial syllabic nucleus: ({qhu}'fa:?, -, -n, -,) /{qu}-
'fa:K/ [zu] erfahren
But we have: (ae'nand) /ae'nand/ Rnander with non-morphemic syllabica-
tion, which is acknowledged by the new orthography as well.
Further useful examples: (&'a&baet) /K'aKbaet/ erarbRten ('aof&Ste;U,)
/'aofKSte:U/ Auferehu ('vi:d&o;b,, -n) /'vi:dKo:bn/ wdererobern
('lat&aes) /'glataes/ GlattRs (la'ne:z&aes) /la'ne:zaes/ LaneseRs ('a&a;t)
/'gaa:Kt/ Gaart ('qhOl&h;U,) /'qOlh:U/ Zollerhhu (khlo'a:k) /klo'a:k/
Kloake (&ide'a:l) /ide'a:l/ ideal
(0), although the phonemic transcriptions do not use //, 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 () or //, especially for important cases.
A possible variant of ({:/0}, ) is (, :, , 0): (&I'a&baet, &I'-),
(&I0'a&baet, &I'a-) /IK'aKbaet/ i> erarbRte (I'ant&t), (I'act-) /I'ant-
vOKt/ i> antworte (&ide'a:l) /ide'a:l/ ideal erefore, we have () 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
{ 17.36 of NPT/HPh} entails a distinctive use of creaky voice for voiced pho-
nemes, (, ); however, certain speakers change these phones into (, 0) sequen-
ces, with () in second place.)
Stopstrictives
Constrictives
But let us see some examples: ('fYMf) /'fYnf/ fnf ('vas) /'vas/ was ('aes) /'Kae-
s/ rRen ('aez) /'Kaez/ rRsen ('zo:n) /'zo:n/ Sohn ('S:n) /'S:n/ (n ('Staen)
/'Staen/ StRn ('thIS) /'tIS/ ( (Ze'ni:) /Ze'ni:/ Gen (e'tha:Z) /e'ta:Z/ Etage ('u:X)
/'bu:x/ Bu> ('y:) /'by:/ B>er ('aX) /'bax/ Ba> ('E) /'bE/ B>e ('u:I)
/'Ku:I/ ruhig ('aoIs) /'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 committed variants
as well (which are more demanding), because they are not really necessary, al-
though very frequent and typifying. It would not be useless to acquire them to
obtain a smoother pronunciation: (&, , ) /f, , x/: ('ao&&Sib, 'aof&Sib)
/'aofSi:b/ Qf(ben ('h:st, 'h:st, -t) /'h:st/ h> ('aoX&Spk, 'ao&Spk)
/'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 (&, , ) /f,
, x/, irrespective of the context just seen.
Quite rightly, German pronouncing dictionaries clearly indicate both () //
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 dierence between ('thao) /'tao/ TQ>en
[->en] small rope and ('thaoX?) /'taox/ tQ>en to plunge, or ('khu:) /'ku:/
Kuh>en [->en] calf and ('khu:X?) /'ku:x/ Ku>en cake.
5.2.4.3. e sux -ig is (I) /I/, even when followed by C (except for those be-
longing to suxes which contain ()): ('kh:nI) /'k:nI/ Knig ('aenIst, -st,
-t) /'aenIst/ Rnig (f'aenIt) /f'aenIt/ verRnigt ('qhanqIst, -st, -t)
/'qvanqIst/ zwanzige but: ('kh:nIkI) /'k:nIklI/ knigli> ('kh:nIk&ae)
/'k:nIkKae/ KnigrR> ('kh:nIgIn) /'k:nIgIn/ Knigin
However, it is possible to nd ('kh:nI&ae), and even fairly systematically
(Ik) /Ik/: ('kh:nIk, 'aenIkst, f'aenIkt, 'qhanqIkst), which is the only pronun-
ciation really accepted in southern Germany, Austria, and Switzerland, in initial
position too: ('e:mIS, 'k[h]-) >emi( ('i:na, 'k[h]-) ina
e graphic sequence 7 has two dierent phonic structures: within lexemes
we nd (ks) /ks/: ('vEksn) /'vEksn/ we7eln ('vaeks) /'vaeks/ WR7el ('Oks)
/'Oks/ O7e ('zEks) /'zEks/ se7 ('fUks) /'fUks/ Fu7 ('daks) /d'daks/ der Da7
badger, while we have: (Xs) and (s, s, ) /xs, s/, when s is a grammeme: (s-
'aXs) /ds'daxs/ des Da7 of the roof, (s'phEs, -s, -) /ds'pEs/ des Pe7
(black) pitch.
/s, z/ are mostly articulated with lowered tip of the tongue; but, for /s/, we have
(): ('I) /'bIs/ bis(en and, except in slow speech, for /sS, sZ/, we have (SS,
S): ('aeS&Sak) /'aesSKak/ Eis(rank ([I]US'SnEl 'vEk) /ImUs'SnEl 'vEk/ i> muss
(nell weg (&aSe'ni:) /dasZe'ni:/ das Gen. /S, Z/ are with raised tip (((x, ))).
In addition, in the /kv, qv, Sv/ sequences, /v/ is more often realized as bilabial,
(6): ('khac) /'kvac/ Qua ('qhae) /'qvae/ zwR ('SEst) /'SvEst/ S>weer 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 generic 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, (), 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 nd (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 () /K/: ('aoIs) /'KaoKIs/ RQris ('I) /'bKI/ brien.
After tautosyllabic V instead, we will always transcribe () (voiced pharyngeal-
ized uvular approximant) for /K/, which is intense after C or diphthongs, () //:
('vi:) /'vi:K/ wir ('vI) /'vIK/ wirr ('mae) /'mae/ MRr (like MRer Meyer] ('Es)
/'bEs/ besser (h'fo:) /h'fo:K/ hervor.
Many phonemic traditions have considered //K//, mostly in parallel with //n,
m, l//, which we prefer to transcribe and interpret as (, , ) /, , /, although
we are perfectly aware that, in slower or emphatic speech, (n, m, l) are possible,
and ( K ) as well even (R), or rather (R, R) (in Switzerland) and (, ,
) (in Austria).
Contrary to what we do and Kohler (1977, 1995) and others as well several
authors distinguish between () and (:) ( /r/ and /:/, 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 ( 5.1.1 too).
For the prexes er- her- ver- zer- we will transcribe ({0}) /{0}/, since it is the
normal and most recommendable pronunciation, although in pronouncing dic-
tionaries (except the ) we still nd /{0}Er, {0}E/. 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}/.
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 ('haem) /g'haem/
gehRm ('da:&ha;b/vI, -H-) /'da:ha:bvi:K/ da haben wir ('Un&hael, -H-) /'Unhael/
UnhRl ('vIlhlm, -H-) /'vIlhElm/ Wilhelm Notice, however, cases like: ('e:) /'ge:/
gehen ('u:I) /'Ku:I/ 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 Dehnungs-h): ('ne:m) /'ne:m/ 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 briey say something about other possible variants for /0j, y,
u', o'a/: ('fe:,, -j) /'fe:Kj/ Fern (na'qho:n, -'qho:n, -qi'o:n) /na'qjo:n/ Na-
tion (qhya'no:z, qha-) /qya'no:z/ Zyanose (&manu'El, -u'El, -'njEl, -'nwEl) /ma-
nu'El/ manuell (thoa'lEt, thja-, thwa-) /toa'lEt/ Toilette
Laterals
5.2.6. In German neutral pronunciation we nd only (l) /l/ (with the corre-
sponding intense one, () //) and the variant (), before (, ,) (the last one is het-
erosyllabic only): ('lant) /'lant/ Land ('fOl) /'fOl/ voll (fa'mi:l,) /fa'mi:lj/ Famil
(va'nI,) {((-nI-,)) and (-Il)} /va'nIl[j]/ Vanille ('mI) /'mIl/ Mil> ('hElt) /'hElt/
Held ('e:z) /'e:z/ Esel ('u:dn) /'du:dn/ 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: &hats'hOYt 'nIq 'aof≥b) der Lehrer hat uns hSte ni>_
Qfgegeben
('thIS 'StElvI IndI'Ek2| 'Sak dI'and3 3) den ( ellen wir in d
E$e, den S>rank in d andere
('thIS dI'Ek&Ste;t) der (, der in d E$e eht
('thIS MvIdI'Ek g'StElt&ha) den (, den wir in d E$e geellt 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 modications given, in order to grasp the mechanism.
It is obvious that the most normal forms, in this list, are not the rst given,
but some of the succeeding ones, in suitable conditions.
aber ('a:b, &a;b, &ab, &ab, &a6) 'an, &an, &an, a, an:, n:)
also ('alzo, &alzo, &alzo, &azo) an'r an der ('and, &and, &and, 'a-
an ('an, &an, an, n), (-m, -M, -~, -, n, &an, &an)
-,) an die ('andi, &a-, &a-, -ani, -I, -)
am an dem ('and, &and, &ad, auf ('aof, &aof, &aof, aof, of)
'a, 'am, &am, am, m) aufs -f's -f das ('aofas, 'aofs, &ao-
ans an's an das ('ands, &ands, &an- fs, &aof#, &aofs, &aofs, aofs, aofs,
ds, 'ans, &ans, ans, ns) ofs)
an'n an den ('and, &and, &and, aufm auf'm -f dem ('aof, 'aof,
5. german 201
ly after V] -gegen )
die ('i:, &i;, &i, i, I, ), (d-), ([-) er ('e:, &e;, &e, e, , ) (() only
(only after V] when not preceded by (, , ) /, K,
die (rel.) ('i:, &i;, &i, i, I), (d-) / the most reduced forms do not
die (dim.) ('i:, &i;, &i, i), (d-) occur in nal position)
diese ('i:z, &i;z, &iz, &Iz), (d-) er- ('-, '-, '-) [-arbRten -bri-
diesem ('i:z, &i;z, &iz, &Iz), gen )
(d-) ('-, '-, '-) [-klren -saufen )
diesen ('i:z, &i;z, &iz, &Iz), (d) es 's ('Es, &s, s, s, s, #, s) (the
dieser ('i:z, &i;z, &iz, &Iz), (d-) most reduced forms occur in nal
dieses ('i:zs, &i;zs, &izs, &Izs), (d-) position too, in many cases)
dir ('i:, &i;, &i, I, ), (d-), ([-) eu> ('OY, &Y, Y, Y, Y,
(only after V the most reduced Y, Y)
forms do not occur in nal posi- euer ('OY, &Y, Y, Y, Y)
tion) eure ('OY, &Y, Y, Y, Y)
do> ('OX, &X, X, X), (d-) eurem ('OY, &Y-, &Ym, Y-,
du ('u:, &u;, &u, u, U, ), (d-), ([-) Y, Y-, Y-)
(only after V the most reduced euren ('OY?, -, &Y-, &Y,, -n,
forms do not occur in nal posi- Y-, Y-, Y-)
tion) eurer ('OY, &Y-, Y, Y, Y)
dur> ('U, &U, U), (d-) eures ('OYs, &Y-, &Ys, Ys, Y-,
dur>s -h das ('Us, 'Us, Y-)
'U#, 'Us, &Us, Us, -, fr ('fy:, &fy;, &fy, fY, fY, f)
-s), (d-) frs -r's -r das ('fy:ds, 'fy:ds, 'fy:-
dur>'n -h den ('U, &U, d#, 'fy:s, &fy;s, &fys, fYs, fYs, fs)
'U, &U), (d-) fr den -r'n ('fy:d, 'fy:?, 'fy:,,
dur> die ('Ui, &-, -I, -), (d-) &fy;,, &fy,, fY,, f,, fYn), (-m, -M,
eben ('e:b, &e;b, &eb, e, -n, -~, -)
em, em) fr die ('fy:di, &fy;-, &fy-, &fY-, -I, -)
Rn 'n ('aen, &aen, aen, n, ), (-m, gar ('a:, &a;, &a, a, a, ), (g-)
-M, -~, -, -,) gegen ('e:g, -g, ≥-, &ge-, e), (g-)
Rne 'ne ('aen, &aen, &aen, &n, &n, gegens -n's -n das ('e:gds, 'e:gds,
, n) 'e:gs, -gs, ≥-, &ge-, es), (g-)
Rnem R'm 'nem ('aen, &aen, gegen die ('e:gdi, -n-, --, ≥-, &ge-,
&aen, &aem, &n, &m, aem:, -I, -), (g-)
aem:, aem, em, m, nm, m, ) geworden ('vOd, &vd, -
Rnen R'n 'nen ('aen, &aen, &aen, &v,), (g- -m, -M, -n, -~, -)
aen:, aen, aen, en, n, nn, n, guten Abend! ('u:t 'a:bt, &u;t'-,
), (-m, -M, -~, -, -,) &ut-, &Ut-, &Ud-, Un'-, '-, '-,
Rner 'ner ('aen, &aen, &aen, &e- '-, [-]'a:mt, [-]'amt), (g-)
n, &n, &n, , n) guten Morgen! ('u:t 'mOg, &u;t-
Rnes 'nes ('aens, &aens, &aens, &e- 'm-, &ut-, &Ut-, &Ud-, -'m-, Un'm-,
ns, &ns, &ns, s, ns) -m'm-, 'm-, 'm-, 'm-, 'm-, 'm-,
Rnmal ('aen&ma;l, &aenmal, -l, -, [-]'mO,, [-]'mO,), (g-)
-mm-, &m-, &m-) guten Tag! ('u:t 'tha:k, &u;t'tha:k,
ent- (nt-, nt-, nt-, t-) [-wi$eln &ut-, &Ut-, &Ud-, Un'th-, 'th-, 'th-,
5. german 203
wie ('vi:, &vi;, &vi, vi, vI) wrde ('vYd, &vY-, vYt, vTt)
wie viel ('vi:fil, vi'fi:l, &vifIl, vI&fil, &vif) wrdest ('vYdst, &vY-, -s)
wieder ('vi:d, &vi;d, &vid, &vId) wrdet ('vYdt, &vY-)
wieder- (unstressed) (&vid-, &vId-) -zehn (-qe:n, -qe;n, -qen, -qn, -qn,
will ('vIl, &vIl, vIl, vI) -q), (-m, -M, -~, -)
wir ('vi:, &vi;, &vi, vI, v, v, v) -zehnte (-qe:nt, -qe;n-, -qen-, -qn-,
(the most reduced forms do not oc- -qn-, -q-)
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:d, &qu;d, &qu-
tic after verbs) d, &qU, 'qhUm, &qUm, qUm, qm,
wird ('vIt, &vIt, vIt, vt, vt, vt) q)
wirst ('vIst, &vIst, vIst, vst, vst, zur zu der ('qhu:d, &qu;d, &qud,
vst, -s) 'qhu:, &qu;, qu;, qu, qU, qw,
wo ('vo:, &vo;, &vo, vo, v) q)
wo- (vo'-, v'-, v'-) [-bR -her -hin zu'n zu den ('qhu:d, 'qu;d, 'qud,
-vor -zu ) 'qhu:, 'qhu:n, &qu;n, qu;n, qun,
(vo'-, v'-, v'-), ('vo:&-) [-ran -rauf qUn, qn, q), (-m, -M, -~, -, -,)
-raus -rber -runter ) -zu- (-qu-, -qU-, -q-) [abholen )
worden ('vOd, &vd, vn, -,), zusammen (qhu'zam, qu-, qU-, q-)
(-m, -M, -~, -) zwar ('qha:, &qa;, &qa, qa,
wurden ('vUd, &vUn, vUn, -,), qa, q)
(-m, -M, -~, -) -ren -r'n (after short or long
wurde ('vUd, &vU-, &v-) stressed V where /K/ (?, ) be-
wurdest ('vUdst, &vU-, &v-, -s) comes /Kn/ (,), through /n/)
wurdet ('vUdt, &vU-, &v-) (-'[:]?, -'[:], -'[:],, -'[:]n),
wrden ('vYd, &vYn, vYn, vTn, (-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 uent, but less clear 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'apS&nIt 'o:z) d abge(nittenen Rosen but: ('nIm
dI'apS&nItn) nimm d abge(nittenen
hab' es gesehen ('ha:bI, -I) /'ha:b[]I/ hab' i> ('le:zI, -I) /'le:z[]I/ les' i> and
('fnI) /'fn[]I/ n' i>, ('a:tI) /'a:tm[]I/ atm' i> but: ([I]'fn) /I'f-
n/ i> ne, ([I]'a:t) /I'a:tm/ i> atme; in the past tense, -e () // may be
dropped, before an enclitic i>, provided the verb does not end in -ete (t) /t/:
('maXtI) /'maxt[]I/ ma>t' i> ('UftI) /'dUKft[]I/ dur' i> but: ('fnt&I)
/'fntI/ nete i> ('a&baett&I) /'aKbaettI/ arbRtete i> Even -e () // be-
longing to nouns may often be dropped while maintaining the C voiced or half-
-voiced: ('li:bUn 'laet, --) /'li:b[] Unt'laet/ Lb' und LRd
We have already dealt with the assimilation of place of articulation to a follow-
ing C for /t, d, n, , s/ ( 5.2.1-2 5.2.4). We have also mentioned some alterna-
tions, especially for approximants and /K/ ( 5.2.4-5).
5.3.2.4. Frequently, the sequence /gn/ assimilates as well (and also /bn, dn/,
even if preceded by N with a subsequent reduction): (#'e:gnt, #'e:nt) /s'Ke:g-
nt/ es regnet ('agns, 'ans) /'agnEs/ Agnes (mag&nifi'qhEnq, ma&n-, -IfI-) /mag-
nifi'qEnq/ Magnizenz (zIg'na:l, zI'na:l) /zIg'na:l/ Signal, ('va:gn, 'va:n) /'va:g-
n/ Wagner ('e:bn, 'e:mn) /'e:bn/ ebne ('le:bt, -mt) /'le:bt/ lebend ('e:dn,
'e:nn) /'Ke:dn/ Redner ('ve:d, -n) /'ve:Kd/ werden ('fInd, 'fIn) /'fInd/ n-
den ('hEmd, -mb, -m) /'hEmd/ Hemden ( above).
Even sequences of N + a voiced stop can undergo nasal assimilation: (no'vEmb,
-mm) /no'vEmb/ November ('Umb&halt, 'Umm-) /'Umbhalt/ umbehalten
(qhUm'bae&Spi;l, -m'm-) /qUm'baeSpi:l/ zum BRspl ('Unds&ak, 'Unns-) /'bUn-
dsbak/ Bundesbank ('vand, 'vann-) /'vandK/ Wanderer ('vUnd&ba; 'vUn-
n-) /'vUndba:K/ wunderbar (qhU'mIndst, -Innst) /qu'mIndst/ zuminde ('U-
g&vIs, 'U-) /'UngvIs/ uewiss ('ag≥b, 'a-) /'agge:b/ aegeben
('a≥b, 'a&e;-) /'age:b/ aeben ('Uga, 'Ua) /'UgaK/ Uar
5.3.2.6. In fast speech, between vowels and in unstressed syllables, /b, g/ can be
realized as (6, ): ('ha:b, -6) /'ha:b/ habe ('le:g, -) /'le:g/ lege /t/ can become
([): ('It, 'I[) /'bIt/ bitte Under the same conditions, voiceless stops and con-
strictives may become half-voiced or totally voiced: (s&hat'dOX 'maXt, --, -d-)
/dashat'dOx g'maxt/ das hat er do> gema>t (S'Sa:fI 'OX&nIt, --, -v-, 'S-) /das-
'SafI 'dOxnIt/ das (a' i> do> ni>t (s&UsI'aX?, --, -z-) /dasmUsI'max/
das muss i> ma>en (s'aXI 'OX&nIt, --, --) /das'maxI 'dOxnIt/ 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/ Wn (s'as) /das'vas/ das Wasser ('qhanqI) /'qvanqI/ zwanzig ('zi:-
b) /'zi:b/ sben ('a:t&sa;m, -t'-) /'Ka:tza:m/ ra_am (&ZUna'lIst) /ZUKna'lIst/ Jour-
nali (t'U) /t'ZU/ D(uel (',a:) /'ja:/ ja (s'a:) /das'ja:K/ das Jahr ('o:t)
/'Ko:t/ rot ('Sak) /'SKak/ 6rank After voiceless stops, /K/ may become (X)
(while phonotactically remaining /FK/): ('phaes, 'phXaes) /'pKaes/ PrRs ('the:t,
'thXe:t) /'tKe:t/ treten ('khaes, 'khXaes) /'kKaes/ KrRs
Voiced stops are realized as half-voiced, both after pauses and after voiceless C:
('Uk) /'bUKk/ Burg ('mEs&a;) /'mEsba:K/ messbar ('e:nU,) /'de:nU/ Dehnu
(&asu'Et) /dasdu'Et/ das Duett ('u:t) /'gu:t/ gut ('aos&e:) /'aosge:/ Qsgehen
-m:, -m) /'e:b/ eben ('nEn, -n:, -n) /'nEn/ nennen ('hE, -:, -) /'hE/ hen
('le:g, -, -:, -) /'le:g/ 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:bs&e;t, -m-, --, -m-) /'li:bsve:Kt/ lbens-
wert ('UndsinIs&te;,Um, -nn-, -n-, -mI-) /'bUndsminIste:KjUm/ Bundesminie-
rium ('ands, -nn-, -n-) /'ands/ anders ('ag≥b, --, --) /'angge:b/ ae-
geben In addition to ('Sto:&m) /'StKo:mE/ Strohmee we can also have
('Sto:m&m, -&m-) /'StKo:mmE/ Strommee 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: ('ap&Ilt, 'a&Ilt) /'apbIlt/ Abbild ('a:t&aM, 'a:&-) /'Ka:tdam/ Raddam-
er ('mIt&thael, 'mI&th-) /'mIttael/ mittRlen ('mIt&qh;l, 'mI&qh-) /'mItqE:l/
mihlen ('vEk&khm, 'vE&kh-) /'vEkkOm/ wegkommen ('vEk&e:, 'vE&-) /'vEk-
ge:/ weggehen e same holds for grooved constrictives: (s'sElb, s'-, 's-)
/das'zElb/ dasselbe (s'SIf, S'S-, 'S-) /das'SIf/ das 6i ('aes&Sak, -S&S-, -&S-) /'aes-
SKak/ Eis(rank Instead, the /s/ sequence, as we have already seen, has a couple
of possible variants: ('h:st, -st, -t) /'h:st/ h> whereas /s/ is only slightly
modied: ('I) /'bIs/ bis(en
5.3.2.8. From the above, a phonation-type dierence can become almost func-
tional in three ways: ('ap&Ilt, 'a&Ilt) /'apbIlt/ Abbild in comparison with ('a:p)
/'a:p/ Apel and ('a:b) /'a:b/ Abel Let us also consider: ('mEnd&zo;n) /'mEndzo:n/
Mendelsohn ('mEnds&so;n, -s&-, -&s-) /'mEndszo:n/ Mendelssohn
If the geminates are produced by assimilation of place (of articulation), they can
be reduced (provided the syllable of the rst element is unstressed): (&Immu'ze:,
&Imu-) /Immu'ze:/ in Museen (mIk'khal, mI'kh-) /mIt'kaKl/ mit Karl (&magnifi-
'qhEnq, &man-, &ma-, &ma-, -IfI-) /magnifi'qEnq/ Magnizenz (qhUm'bae&Spi;l,
-m'm-, -'m-) /qUm'baeSpi:l/ zum BRspl (qhu&mInds'as&a;, -Inn-, -In-) /qu'mIndst
'dasda:/ zuminde das da
We have, though: ('am&mld) /'anmEld/ anmelden ('mIk&khm) /'mItkO-
m/ mitkommen ('o:k&khaof) /'bKo:tkaof/ Brot kQfen ('a:k&fa:?, -,
-&fa;,, -n) /'Ka:tgfa:K/ Rad gefahren ('am&bInd, 'am&m-) /'anbInd/ anbinden
('agns, -n-, --) /'agnEs/ Agnes ('a≥b, 'a&-, -e;m, -e;, -e;m) /'ange:b/
aeben ('Unds, -nn-) /'bUnds/ Bundes ('vand, -nn-) /'vandK/ 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 reection:
('ap&aez) AbrRse ('th:kI 'thE:k-) tgli> ('mUnt&at) Mundarten ('Un&aof-
ItI, -f&I-) unQfri>tig (f'ap&e;dU,) Verabredu ('a:p, -t) beat-
men
It is obvious then that German has very complex consonant sequences; here we
will give just one example: ('hEps[t]<) Herbbltter Please, note carefully even
the spelling sh /sh/: ('lands&haoshalt) /'landshaoshalt/ 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 rst lexeme and re-
duced stresses on the ex-prominent syllables of succeeding lexemes.
Here are some examples of two-lexeme compounds: ('SIf&fa;t) 6ifahrt ('aM-
&SIf) Dam(i ('a:n&ho;f) Bahnhof ('aez&ba;n) Eisenbahn ('ho:X&OYc) Ho>-
dS ('fy:&StYk) Frh$ ('aoto&ba;n) Autobahn (fa'mi:l,&na;m) Familnna-
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):
('Sif:faRt, 'dam:Sif, 'ba:nof, 'Ok:dic). 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 (Su'ma:keR), up to an ugly hybrid, which is neither German nor Romance: ('Su:-
maxeR).
5.3.3.4. ere are even cases where secondary stresses actually determine the
meaning of certain compounds: ('aMSIf&fa;t) Dam(ifahrt steamboat voy-
age, ('aM&SIffa;t) Dam(ifahrt steam navigation. In cases such as these,
of course, rhythmic uctuations are less frequent, although context must (and can)
help very much.
We also nd some compounds which do not bear primary stress on their rst
lexeme, as: (&,a;'qhe:nt) Jahrzehnt (&o;st'mo:n&ta;k) Oermontag (&Saq'aes)
(warzwR (&o;t'khOYq&Sst) RotkrSz(weer (&za;'bYk) Saarbr$en
Others have more than one primary stress: ('lao 'gy:n) blQ-grn, ('Um 'StOlq)
dumm-olz ('Se:sI 'hOlStaen) 6leswig-HolRn however, they can become
(&lao'gy:n, &Um'StOlq, &Se;sI'hOlStaen), mostly in preintonemes.
Let us observe: ('lao 'gy:n) blQ-grn blue and green ( two colors put along-
side, as in a ag or sports -shirt) and ('lao&gy:n) blQgrn bluish green, water
green ( two colors blended together, as painters do).
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 ()): ('aotoba;n
'anStalt 'alt&ho;XOYc 'haopt&aofa;b) besides, ('aezba;M&fa;pha;n, -fa;-
pha;n &Spi;l&va;?'aos&hands&zlSaft, -zlSaft &ho;X&akuUm'lIt&o;g&Slq-
o;f, -Slqo;f).
ere are frequent cases of (homographic) compounds with (stressed or un-
stressed) prex grammemes, which (in their unitary non-inected forms) is the
most important element of semantic dierentiation: berseen ('y:b&zq) to
pass over, (&yb'zEq, &Y-) to translate, umfahren ('UM&fa;?, -, -a;,, -a;n)
to run down, (UM'fa:?, -, -a:,, -a:n) to drive/sail round.
Generally, compounds with unstressed monosyllabic suxes have /I, , , , ,
212 a handbook of pronunciation
/, 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): ('vtI) w>entli> (('vt&I, -I) w>entli>en].
5.3.3.8. Cases such as the following are interesting too: 'ma>en S d 'Re>-
nu &fertig 'halten s d 'Psse 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, innitive
5. german 213
ones with um zu: s 'kann 'ni>t &kommen, &wRl s 'morgen &sehr 'frh 'Qf&ehen
&muss (except when dependent innitives are more than two: s i &sehr 'mde, &wRl
s, 'hSte 'Morgen, 'sehr &frh hat 'Qf&ehen &mssen i> 'hoe, &dass er d 'Re>nu
&hat be'zahlen &knnen i> 'hoe, &dass s &Rne 'nSe 'Wohnu hat 'nden &knnen]
i> 'fr>te, &dass er 'Re>t &hat der 'Arzt &wollte 'wissen, &ob er 'ber ge&habt &habe s
'sagt, dass &ihre 'Kinder im 'Garten &splten i> 'glQbe, &dass s 'ni>t 'hr &i i>
'wR, dass er vor &zwR 'Jahren &ope'rrt &worden &i i> 'brQ>e 'ni>t zu &antworten
es i 'immer 'besser, 'pnktli> &anzu&kommen um 'pnktli> &anzu&kommen, &mssen
wir d 'U-&Bahn &nehmen e constructions with modals and innitives in the last
position are to be noticed too: du 'sollte zum 'Arzt &gehen
With inected 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 'Kam {Qf das 'Da> 'sprit {ber
Let us add some typical examples, since the concepts expressed by some words
can (or must) come out: 'Anna darf 'ni>t &mitkommen der 'Brf &wurde von 'mir
ge&(rben. More: &kann man 'hr 's>wimmen? (two concepts), however: 'knnen
S Kla'vr &splen? something like to piano; 'morgen &muss i> 'gen i> muss
'morgen &gen [ 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'nren? er 'fuhr
in d 'Fern, um &si> zu er'holen er 'l das ReQ'rant, 'ohne zu be'zahlen 'bitte,
'blRben S &do> 'sien 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 &knnte 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. 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 ( 13.24 of
NPT/HPh), which is also typical of German and French. Here is another example:
und, &wenn es 'mgli> zu 'ma>en &wre, Others may be found in the transcrip-
tions of the text ( 5.5):
/./: (I'su:X aem'm:bg&Sft3 3) /I'zu:x aen'm:bgSEft./ I> su>e Rn Mbelge-
(.
/?/: ('khOmzi aos':st&ae1 1) /'kOmzi aos':stKae?/ Kommen S Qs er-
rR>?
214 a handbook of pronunciation
Other accents
North-eastern-Germany German
5.4.1.1. e rst 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 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 (suciently so to use dierent phonet-
ic symbols: (, , #, )).
We just give some examples to directly show the dierence, thanks to phonetic
transcriptions: ('fi;il) /'fi:l/ vl ('ze;e) /'ze:/ sehen ('Sp;t) /'SpE:t/ spt and:
('a;an) /'ba:n/ Bahn ('zo;o) /'zo:/ so ('u;ut) /'gu:t/ gut ('thy;y) /'ty:K/ Tr ('S;n)
/'S:n/ (n
5. german 215
e two realizations of // are slightly lower: ('Unz, -ndz-, -nQ-) /'UnzK/ un-
sere whereas, the possible vocalization of /K, / is closer: ('vas, -x, -x, -x, -x) /'va-
s/ 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/): (a, aP, +). In
addition, especially the rst element of /ao/ is slightly backer and the rst of /OY/
is slightly higher, and indeed it crosses the edge towards the upper box: ('as) /'aes/
Eis ('laP) /'blao/ blQ ('n+) /'nOY/ nS ere is little dierence for the two xeno-
phonemes /eI, oU/; while, for the possible rst-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 ( 5.3.2.9). nally, before pauses, /k/ can be realized as a (pre)-
velar stopstrictive (w, ): ('khi;ik, -w, -) /'kKi:k/ Krg ('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
western neutral one, approaching what may be called international.
/oU/* (o;o, U)
/eI/* (e;e, I) /OY/ (+) {(+)}
{/a/ ()}
/ae/ (a) {()} /ao/ (aP) {(P)}
Austrian German
5.4.2.1. is accent is the mediatic 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.
Long 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 (:), but ().
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
[], #, ) (in less broad accents we nd (I, Y, U E[E], , O), with the unstressed
variants (, #, ), slightly closer than in neutral pronunciation): ('fiS) /'fIS/ (
('fyMf) /'fYnf/ fnf ('hunt) /'hUnt/ Hund ('lf) /'Elf/ elf ('Spt) /'SpE:t/ spt ('q6#lf)
/'qvlf/ zwlf ('ft) /'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 nd (a, aA) (and (, ) in rened, or aected, ones),
whereas in broader ones we have: (, ) and (, ) 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 nd 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\ /'bAk/ ('Ak) bench, /'bak/ ('ak) bank, gambling table.
A very broad accent has /e:, :, o:/ (e, #, o): ('vee, 've) /'ve:/ weh (',
'#) /'b:/ B ('soo, 'so) /'zo:/ so
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), (i,
u) /i, u/: ('iks, 'iiks) Dx ('uq, 'uuq) 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 dierences in phonemic
length too, as in: ('Alt&tum) /'alttu:m/ Altertum ('Spiil&a) /'Spi:lba:K/ splbar
('hAq) /'ha:Kq/ Harz ('AS) /'ba:KS/ Bar( ('Sust) /'Su:st/ 6uer ('Sloos) /g-
'SlOs/ Ge(loss/-o
/ / (2 2 2 2 2 2 2) /./ (2 ' 2 3)
/ / ( 2 2 2 2 2 2 2) /?/ (2 1 2)
/ / ( 2 2 2 2 2 2 2) // (2 ' 2 1)
/ / ( 2 2 2 2 2 2 2) /,/ (2 ' 2)
/K-/ (-)
/e:/ (e, ee), /:/ (#) /o:/ (o)
/OY/ (O, O,, Y, , @, )
/ae/ (, E, ) /ao/ (A, , )
// (, x, a) /a[:]/ ([], [], [])
218 a handbook of pronunciation
e last two vocograms show both less broad (()) vowel realizations (up to
some aected ones, ()), and on the contrary some of the broadest realizations
(()).
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 nd velar (,
), in the sequences /kl, gl k, g/. In cennese pronunciation again, in addition
to () (and ()), we often nd it vocalized in (,) (as in the dialect): ('Spii, -,, 'ee-
s, -s,) /'Spi:g, 'e:z/ Spgel Esel.
Swiss German
5.4.3.1. Even for this accent we will add, when relevant but without further
gures, 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 rst vocogram shows the short and long monophthongs; actually, in this
accent long vowels are diphthongs with narrow movements: ('fiil) /'fi:l/ vl
('f5yy) /'fKy:/ frh ('uut) /'gu:t/ gut ('eet) /'be:t/ Beet ('l) /':l/ l ('oo) /'vo:/
wo ('SpEEt) /'SpE:t/ spt (which is very stable and typical) ('StaAt) /'Sta:t/ Staat e
last two have non-neutral broad variants: (E) and () (or (A) as well, which is
typical of Bern, and (), typical of Zurich). (e phonosynthesis in 17.20 of
NPT/HPh shows the characteristics of Zurich 6wyzert {which is Aleman-
nic}, with still more dierent realizations.)
Among short V, /I, Y, U/ have closer realizations (i, y, u): ('fiS) /'fIS/ ( ('fyMf)
/'fYnf/ fnf ('hun) /'hUnt/ Hund /E, , O/ are relatively open: ('Elf) /'Elf/ elf
('q6lf) /'qvlf/ zwlf ('Oft) /'Oft/ o (with unstressed taxophones (, #, )). For
the spelling we can also nd a broad non-neutral realization (): ('fElt, 'flt)
/'fElt/ fllt /a/ is back-central (A) (which, in broad marked pronunciation, is
rounded too, (), or fully back, ()): ('StAt) /'Stat/ Stadt
220 a handbook of pronunciation
5.4.3.2. For // we have (), in every context, since /K/ is not uvular: ('5aA)
/g'Ka:d/ gerade ('unsR) /'UnzK/ unsere ('unsR) /'Unz/ unser Besides, we gen-
erally nd (-n, -m, -l) for /, , /: ('faARn) /'fa:K/ fahren ('loom) /'lo:z/ lo-
sem ('eel) /'e:z/ Esel Only in very careful pronunciation, can we nd (, , ),
and (, , , ) as well, even for speakers lacking /K/ ().
Neutral Swiss diphthongs are /ae, ao, OY/ (a, Ao, OI): ('a) /'aes/ Eis ('lAo)
/'blao/ blQ ('nOI) /'nOY/ nS in the broadest accents we nd: /ae/ (EI, I), /ao/ (U,
U), /OY/ (I, o). Xenophonemes are decidedly closer than in neutral German: /eI,
oU/ (ee, ei oo, ou).
/ / (2 2 2 2 2 2 2) /./ (2 2 3)
/ / ( 2 2 2 2 2 2 2) /?/ (2 1 2)
/ / ( 2 2 2 2 2 2 2) // (2 2 2)
/ / ( 2 2 2 2 2 2 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), (i,
y, u) /i, y, u/: ('it) Dth ('5y) Regg ('hup) Huep Except in less marked
5. german 221
or aected accents, often /y:, y, y/ become /i:, i, i/. In the same gure, possible real-
izations of reduced forms of /ae, ao, OY, a/ are given: (, ao, , ).
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], ('jA) /'ja:kt/
Jagd ('nun) /'nu:n/ nun ('E5) /'e:Kt/ Pferd ('fO5) /'fo:K/ vor ('E5n) /'ve:Kd/
werden ('E5) /'e:Kd/ Erde ('uks) /'vu:ks/ Wu7 ('qE5tliX, -) /'qhE:tlI/ zrt-
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 aspirated (unless aspiration 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, () is lacking, except in less marked and less typical accents: (te-
'aAtR) /te'a:t/ eater (iX'Ess, i-) /I'Es/ i> esse as can be seen, though, lexeme
and grammeme boundaries are respected; only in very broad pronunciations resyl-
labication is possible: (i'XEss, i'-).
Voiced C of diphonic pairs are actually half-voiced, (, , , , ), even before
pauses or when syllable-nal (where, in neutral German, they are neutralized and
change to /p, t, k f, s, S/). erefore, we generally nd /s/ for /z/, after pauses or C\
('uu) /'bu:b/ Bube ('uu) /'bu:p/ Bub ('uu) /'du:/ du ('un) /'Unt/ und ('5ee-
n) /'Ke:d/ reden ('oon) /'bo:g/ Bogen ('taA) /'ta:k/ Tag ('in) /g'vIn/ Ge-
winn ('leen) /'le:z/ lesen ('saAn) /'za:g/ sagen ('ins) /'bInz/ Binse ('paA)
/'pa:Z/ Page
5.4.3.4. Often, /f/ occurs instead of (neutral) /v/: (no'fEmR) /no'vEmb/ No-
vember (fn'tiil) /vEn'ti:l/ Ventil (ful'kaAn, -w, --, -k-) /vUl'ka:n/ Vulkan
For sp internal or nal we have /sp, st/, but in broad accents we nd /Sp, St/:
(k'nOsp, -Sp) /k'nOsp/ Knospe (ist, -iSt) /bIst/ bi ('EstRn, -St-) /'gEstn/
geern Generally, for internal /S/ (after stressed short V, like for the other C] we
have (SS): ('muSSl) /'mUS/ Mu(el for 7 /ks/, in marked pronunciation it is fair-
ly possible to have /xs/: ('fuks, -Xs, -s) /'fUks/ Fu7
Only in less broad speech, can we nd // (but articulated as postpalatal (), or
prevelar ()); whereas, normally, we have /x/ (X, ): ('milX, - -, --) /'mIl/
Mil> ('AoX, -) /'aox/ Q> -ig is regularly /Ik/ ( /Ig/; while /I/ is decidedly
rare and only voluntarily used): ('q6Anqi) /'qvanqI/ zwanzig and // is normal-
ly /k/: ('kiinA, 'w-, '-, 'k-) /'i: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 rst 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 [:], #, ) (in less broad accents we nd
(I, Y, U E[:], , O), and unstressed (, #, ) only slightly closer than in neutral
pronunciation): ('fiS) /'fIS/ ( ('fyMf) /'fYnf/ fnf ('hunt) /'hUnt/ Hund ('lf) /'Elf/
elf ('Sp:t) /'SpE:t/ spt ('qV#lf) /'qvlf/ zwlf ('ft) /'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 (, :) (which will certainly be not-
ed at once; 16.15 of NPT/HPh, where the phonosynthesis of the South-Tyrolese
5. german 223
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), (i,
uX) /i, u/: ('tit, 'iit) Dtl ('wuXns, 'ku:ns) Kuens In the same diagram, the
possible realizations of /ae, ao, OY, a/ (, P, , ), 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 dierences.
ere are dierent 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:/: ('e:st,
'he:t)).
224 a handbook of pronunciation
/ / (2 2 2 2 2 2 2) /./ (2 ' 2 3)
/ / ( 2 2 2 2 2 2 2) /?/ (2 ' 2 1)
/ / ( 2 2 2 2 2 2 2) // (2 ' 2 2)
/ / ( 2 2 2 2 2 2 2) /,/ (2 ' 2)
(, , ) in less broad accents. However, between voiced phones, they are half-
-voiced, (, , ) (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:, -b) /'bu:b/ Bube Half-voiced realizations are possible before internal /n,
l/, too: ('Onu, 'm:li) /'OKdnU, 'm:klI/ Ordnu mgli>. (In Bolzano, we
can nd (B, , ) /b, d, g/: ('u:B).)
ere is no (), 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, (,)):
(i's, ,I',Es) /I'Es/ i> esse (te'a;At, te',a:-) /te'a:t/ 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;A) /'za:g/ sagen ('also) /'al-
zo/ also ('as) /'Kaez/ rRsen (which becomes identical to ('as) /'Kaes/ rR-
en] In initial position, before C /s/ practically never occurs (substituted by /S/);
and, as far as internal and nal /st/ is concerned, in broad accents, as in the dia-
lects, we typically nd /St/: ('li:pst, -St) /'li:pst/ lben ('snst, -nSt) /'zOnst/
son However, in case they are heterolexemic, we have /st/: ('sams&taAw) /'zams-
ta:k/ Samag (as in the dialects: ('sams&tiw, 'sms&tik) /'samstik/ samig]
For (tautosyllabic) /0v/, we have (0V): ('SVst) /'SvEst/ 6weer ('qVa)
/'qvae/ zwR ('kVl) /'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 v against neutral phonemic rules, in cases such
as positiv\ ('vi:q, &posi'ti). In addition, /j/ is approximant (j): ('ja;A) /'ja:/ ja; gener-
ally, /x/ is velar (x): ('na;Ax) /'na:x/ na>; // is maintained (even before /K/, contrary
to what happens in Austria): ('tu) /'dUK/ dur>; but, for /-I/ we have /-Ik/ (ex-
cept in less broad accents) ('qVanqik, -w, -k) /'qvanqI/ zwanzig for // we syste-
matically nd /k/: ('ki:na, 'w-, 'k-) /'i:na/ ina However, in the dialects, except
in Bolzano, () // does not exist at all.
e most widely used articulation for /K/ is a uvular constrictive, (), 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 nal position, after stressed V, in
the dialects and in the broadest accents we have (R): ('a;A -a;A -a;AR -:R)
/'Ka:K/ rar ('hi: 'hi: 'hi:R) /'hi:K/ hr
Text
5.5.0. e story e North Wind and the Sun follows, given in some dierent
(normalized) versions. We start with the (neutral) German pronunciation of
(neutral British) English this is the rst 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 dierent, if
we had used the special (denti-alveolar) symbols ((4, 7, )), and instead of (, )
(for /K, /) respectively the following possible variants: () (neutral), (x) (north-
-eastern), (, 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, uent 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 rst.
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 0.12).
German text
5.5.2. Ein ritten si> Nordwind und Sonne, wer von ihnen bRden wohl der
Strkere wre, als Rn Wanderer, der in Rnen warmen Mantel gehllt war, des Weges
daherkam. S wurden Rnig, dass derjenige fr den Strkeren gelten sollte, der den
Wanderer zwien wrde, sRnen Mantel abzunehmen.
Der Nordwind bls mit aller Ma>t, aber je mehr er bls, deo feer hllte si>
der Wanderer in sRnen Mantel Rn. Endli> gab der Nordwind den Kam Qf. Nun
erwrmte d Sonne d Lu mit ihren frSndli>en Strahlen, und (on na> weni-
gen Augenbli$en zog der Wanderer sRnen Mantel Qs. Da muse der Nordwind zu-
geben, dass d Sonne von ihnen bRden der Strkere war.
Hat dir d Ge(i>te gefallen? Wollen wir s wderholen?
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 dierences 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, , g/, in contexts where they are realized as (B, , ).
Placing /B, , / (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 inec-
tion of lemmata things do not remain constant.
Although, even for Spanish, the number of speakers with American accents is
decidedly more consistent than for the Iberian ones, we prefer to consider the lat-
ter primary, 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
national accent as well as regional variants. A more in-depth study of Spanish
pronunciation will give the macro-koins (at least seven for America), to be consid-
ered neutral, as well as more specic, 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 ve vowel phonemes realized as shown in the rst vo-
cogram in 6.1 (i ', e a ', o u) /i, e, a, o, u/. For e, o in the neutral Iberian-
-American accent, in stressed syllables, (, ) are normal, but, in unstressed sylla-
bles, we nd (e, o).
We therefore have examples such as: ('i)i ('si)a /'si/ s ('kRi;i)i (-sis)a /'kRisis/
cr ('mil) /'mil/ mil (eR'nR)i (s-)a /eR'neR/ cerner (pa'pl) /pa'pel/ papel ('tjR-
r:a) /'tjeRr:a/ tierra ('i)i ('sis)a /'seis/ se ('r:to)i ('r:s-)a /'r:esto/ rto ('t;Xa)i
(-xa)a /'texa/ teja (de'XaR)i (-x-)a /e'xaR/ dejar ('jmpRe)i ('s-)a /'sjempRe/ siempre
(a'tnto) /a'tento/ atento (u't)i (us-)a /us'te/ ted ('p)i (-s)a /'pe/ pez ('d-
e)i (-z-)a /'ese/ dde ('k;o)i (-so)a /'keso/ o
And: ('p;Co) /'peco/ pe>o (kom'pR) /kom'pRe/ compr ('anda) /'anda/ anda
(pa'ta;ta) /pa'tata/ patata ('gRr:a) /'goRr:a/ gorra ('i) /'oi/ hoy ('r:;a)i (-sa)a /'r:o-
sa/ rosa (';Xa)i (-xa)a /'oxa/ hoja (mo'XaR)i (-x-)a /mo'xaR/ mojar (fa'BR) /fa'boR/
favor ('l)i ('s-)a /'sol/ sol (r:a'n)i (-s-)a /r:a'on/ razn ('glpe) /'golpe/ golpe
('p;Lo)i (-,o)a /'poLo/ poo (La'm)i (,a-)a /La'mo/ am (de'k;Ro) /e'koRo/ deco-
ro ('tu;Bo) /'tubo/ tuvo (r:u'mR) /r:u'moR/ rumor
tion (for speakers words; the second vocogram of 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)), (/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, (, ), is normal and
safe, used by real natives (despite the variations). In unstressed syllables, in the
same contexts, instead of ((E, O)), we obviously nd (, ).
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 (('Ei))i (('sEis))a /'seis/ se (('r:Eto))i (('r:Es-))a /'r:es-
to/ rto (('tE;Xa))i (-xa)a /'texa/ teja ((d'XaR))i (-x-)a /e'xaR/ dejar (('jempRe))i ('s-)a
/'sjempRe/ siempre ((a'tento)) /a'tento/ atento ((u'te[]))i (us-)a /us'te/ ted (('pe))i
(-s)a /'pe/ pez (('dee))i (-z-)a /'ese/ dde (('ke;o))i (-so)a /'keso/ o
And: (('pe;Co)) /'peco/ pe>o ((kOm'pRe)) /kom'pRe/ compr (('gORr:a)) /'goRr:a/ gor-
ra (('Oi)) /'oi/ hoy (('r:O;a))i (-sa)a /'r:osa/ rosa (('O;Xa))i (-xa)a /'oxa/ hoja (m'XaR)i
(-x-)a /mo'xaR/ mojar ((fa'BOR)) /fa'boR/ favor (('Ol))i ('s-)a /'sol/ sol ((r:a'On))i (-s-)a
/r:a'on/ razn (('gOlpe)) /'golpe/ golpe (('po;Lo))i (-,o)a /'poLo/ poo ((La'mo))i (,a-)a
/La'mo/ am ((de'ko;Ro)) /e'koRo/ decoro ((r:u'mOR)) /r:u'moR/ rumor
Also for /i, a, u/: ((en'tIR)) (en'tiR)i (s-)a /sen'tiR/ sentir, ((&aI'naR)) (&ai'naR)i (-s-)a
/asig'naR/ ignar, (('I;Xo)) ('i;Xo)i (-xo)a /'ixo/ hijo, (('mIRr:)) ('miRr:a) /'miRr:a/ mirra,
(('r:I;ko)) ('r:i;ko) /'r:iko/ rico; (('ka;N)) ('ka;Na) /'kaNa/ caa, (('ma;Co)) ('ma;Co) /'ma-
co/ ma>o, (('ma;,o)) ('ma;,o) /'maJo/ mayo, (('ka;Le)) ('ka;Le)i (-,e)a /'kaLe/ cae, (('bai-
le)) ('baile) /'baile/ baile; (('qlto)) ('alto) /'alto/ alto, (('mq;X)) ('ma;Xa)i (-xa)a /'maxa/
maja, ((kq'XOn)) (ka'Xn) /ka'xon/ cajn, (('kqu)) ('kaua)i (-sa)a /'kausa/ caa, ((q-
'un)) (a'un) /a'un/ an, ((bIl'Bqo)) (bil'Bao) /bil'bao/ Bilbao, ((q'O;R)) (a';Ra) /a'oRa/ a-
hora; (('tuRko)) ('tuRko) /'tuRko/ turco, (('Xunt)) ('Xunta)i ('x-)a /'xunta/ junta, (('lu;Xo))
('lu;Xo)i (-xo)a /'luxo/ lujo, ((tuR'r:On)) (tuR'r:n) /tuR'r:on/ turrn, ((r:u'mOR)) (r:u'mR)
/r:u'moR/ rumor.
6.1.1.3. After /'i, 'u/, /e|, o|/ remain (e, o), even if at times we can nd (, )
(in neutral pronunciation too): (fe'li;e)i (-ses)a /fe'lies/ felic ('r:i;ko) /'r:iko/ ri-
co Even in an unstressed syllable, above all next to /x, r:/, we can have the (, )
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 (w) (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 ()), through pro-
gressive shifts: ('pw;Blo, 'pw;- 'pw;- 'pj;- 'p;-) /'pweblo/ pueblo ('nw;Be, 'nw;-
'nw;- 'nj;- 'n;-) /'nwebe/ nueve (&ata'lw;o, -w;- -w;- -j;- -;-)i (&as-)a /asta-
'lwego/ hta luego (kwe'tjn, -w- -w- -j+- -+-)i (-s-)a /kwes'tjon/ cuestin. 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 6.1).
6. spanish 233
Diphthongs
Consonants
6.2.0. 6.2 gives the consonant articulations, of the two neutral accents, which
are necessary for satisfactory Spanish pronunciation.
Instead, 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.
6.2. Table of Spanish consonants.
velar round.
labiodental
postalveo-
prevelar
palatal
alveolar
bilabial
palatal
uvular
dental
velar
m (M) (n) n (~) N () (,)i
F pb t d k g
C {} ()
f i ()i /J/|(,) xa () (X)i
_ sa (z)a ()i ()i ()
(B) () j|() (F) w|(j)
R|r:
(l) l () Li
+ (N, M, M, , ) text
Nasals
6.2.1.1. ere are three nasal phonemes, /m, n, N/, with various taxophones for
/n/ (m, M, n, ~, N, , , ,) ( 1.9.1-2; () is semi-provelar without full contact; we
could add dental (()), before /t, d /, for which, however, (n) is sucient; below,
we add ve more taxophones, (N, M, , M, ), for nasal heterorganic sequences):
('ma;no) /'mano/ mano ('ni;No) /'niNo/ nio (um'pRr:o) /um'peRr:o/ un perro (im-
'bjRno) /im'bjeRno/ invierno (iM'fjl) /in'fjel/ inel (&konten'dR) ((-te-)) /konten-
'deR/ contender (kon'to) ((-'-))i (-s-) ((-'s-))a /kon'ebto/ concepto (kon';Xo)i
(-s;xo) ((-'s-))a /kon'sexo/ consejo ('nr:a) /'onr:a/ honra ('a~Co) ((-nC-)) /'anco/ an-
>o ('kNuXe)i (-xe)a /'konJuxe/ cnyuge (uN'uke) /un'Junke/ un yune (&koN-
Le'BaR)i (-N-, -NJ-)a /konLe'baR/ conevar (u'w;Bo) /un'webo/ un huevo ('bako)
/'banko/ banco ('tgo) /'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/ bao ( Italian ('baN:No) /'baNNo/ ba-
gno]
As for /nw/, even with the prexes cons-, ins-, trans-, the most normal and suit-
able articulation is with (): (&kotRu'jn)i (-stRu'sj-)a /konstRug'jon/ construc-
cin (i'tante)i (-s-)a /ins'tante/ instante (&tRafoR'maR)i (-s-)a /tRansfoR'maR/ trans-
formar obviously forms such as (&iRku'tanja)i (&siRkus'tansja)a /iRkuns'tan-
ja/ circunstancia are included in this case.
For the rst elements of /mn, nm/ sequences, we nd some further taxophones,
6. spanish 235
Stops
Stopstrictives
Constrictives
6.2.4.1. ere are ve constrictive phonemes; four are voiceless: (f) /f/, (i, sa)
//, (i, 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, (), or approximant, (F)): ([&uM]fa'BR
[&um]a- [&um]Fa-) /[un]fa'boR/ (un) favor
6.2.4.3. For /s/ the place of articulation changes, from one accent to another,
as /s/ is (apico-)alveolar in Iberian Spanish, ()i, but, (lamino-)dental in American
Spanish, (s)a: (e'ta;o)i (es'ta;os)a /es'taos/ tados ('l)i ('s-)a /'sol/ sol ('pi;o)i
(-so)a /'piso/ po (After a stressed V as well as the normal and more appropriate
(';)i (';s)a, one can also have (')i ('ss)a: ('pio)i ('pisso)a.) Before voiced
diphonic heterosyllabic C (/b, d, g/), /s/ becomes voiced, ()i (z)a: (lo'BuRr:o)i
(loz'BuRr:os)a /los'buRr:os/ los burros ('de)i ('dze)a /'ese/ dde (di'uto)i
(-z'us-)a /is'gusto/ dgto In the /st, s/ sequences, in neutral Iberian pronunci-
ation, (, ) 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/ cena (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 dierent, even if complementa-
ry: in the Iberian accent voicing prevails, (, , ); whereas in the American accent
voicelessness prevails, (s, , z) (even if we transcribe only the rst realization for
each accent): ('mimo)i (-s-)a /'mismo/ mmo ('ila)i (-s-)a /'isla/ la (lo'w;o)i
(-s'w;sos, -'sw-)a /los'wesos/ los huos ('d ',Rr:o)i ('d)a /'os 'JeRr:os/ dos hierros
e last example shows the normal articulation of the postalveo-palatal kind, ()i
()a, before /J/, as also before /L, N/: (la'Lu;Bja)i (la',u;Bjas)a /las'Lubjas/ l uvi
(lo'N;No)i (lo-, -os)a /los'NoNos/ los oos.
If the voiced C that follow are tautosyllabic ( when they are part of the same
syllable), /s/ remains voiceless, as even before V /s/ (being initial in the phono-
-syllable): ('jta)i ('sjs-)a /'sjesta/ sita ('w;lo)i ('s-)a /'swelo/ suelo (&loa'mi;o)i
(-sa'mi;os)a /losa'migos/ los amigos (&mieR'ma;no)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), dened as aspirate(d)
/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 /h/. For exam-
ple: (lah'kwhtah) /las'kwestas/ l cut ('dHe) /'ese/ dde ('ihlah, 'iH-) /'is-
las/ l (&loha'mi;oh) /losa'migos/ los amigos (in neutral pronunciation: (la'kw-
ta)i (las'kwstas)a, ('de)i (-z-)a, ('ila)i ('islas)a, (&loa'mi;o)i (-sa'mi;os)a]
Such pronunciation can even get to transform (h, H) into (`), zero.
les/ tr bail (laz'B;tas laH'B;tah la'B- la'- la'F- -a) /las'botas/ l bot
(loz'a;,os loH'a;,oh lo'a- lo'xa- lo'a- -o) /los'gaJos/ los gaos (tuz'a;tos
tuH'a;toh tu'a- tu'xa- tu'a- -o) /tus'gatos/ t gatos ((=) and derivatives have
an intermediate type of phonation between (h) and (H)). Neutral pronunciation
is: (mi';o, &tRe'Baile, la'B;ta, lo'a;,o, tu'a;to)i (-z-, -s)a
Otherwise, before son(or)ants (/m, n l r:/), /s/ can change into (0, =0): ('miz-
mo, 'mi-, 'mis- 'miH-, 'mi+- 'mi-, 'mi)-) /'mismo/ mmo ('azno, 'a-, 'as- 'aH-,
'a- 'an- 'a-) /'asno/ no ('izla 'i-, 'is- 'iH- 'i+- 'i- 'ia-) /'isla/ la (&ir:a'l, &i-5) /is-
r:a'el/ Israel Neutral pronunciation: ('mimo, 'ano, 'ila)i (-s-)a (&ir: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'mn)i (xa-)a /xa'mon/ jamn (Xe'miR)i (xe-)a /xe'miR/ gemir ('di;Xe)i (-xe)a
/'ixe/ dije when word-nal it is weaker, as it becomes an approximant of the same
place of articulation, or even laryngeal: ('b, -h)i ('b, -h)a /'box/ boj it is cur-
rently lost in: (r:e'l[], -[h])i (r:e'l[], -[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 (()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 't-) /'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;o) /'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 /[]bw, []gw/, in familiar pronunciation, there can be a simplication,
through a velarized bilabial approximant, (), for the rst case; or a constrictive,
()) (or (), 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'-, a'w-) /a'bwelo/ abuelo (la-
'Bwlta, la'm-, la'-, la'w-) /la'bwelta/ la vuelta ('a;wa, 'a;)a, 'a;ma, 'a;wa) /'agwa/
a%a (la&wape't;na, la&)a-, la&ma-, la&wa-) /lagwape'tona/ la %apetona
Instead, for /w, w/, in familiar pronunciation, a more vigorous realization is
more frequent (compared to the phonemic transcription): (');Bo, 'm-, '-) /'we-
bo/, (&a)e'kaR, &ame-, &ae-) /awe'kaR/. Even in the /nw/ combination ( 6.2.1.1),
we can have this more vigorous articulation: (u'w;Bo, u')-, u'm-, u'-) /un-
'webo/ un huevo In all these cases, we can hear even syntagmatically divided reali-
zations, (w, 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, ) (/j, w/, /b, /), 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 /0e, 0o, 0a/; therefore we
have, (semi-palatal) /e/ (), (semi-velar rounded) /o/ (j) and (semi-prevelar) /a/ (F):
('pR) /pe'oR/ peor ('ta;tRo) /te'atRo/ teatro ('pj;ta) /po'eta/ poeta (&kjau'laR)
/koagu'laR/ coa%lar (u'nF;Xa)i (-xa)a /una'oxa/ una hoja ('lF;tRa pa'Ra;a) /la'otRa
pa'Raa/ la otra parada In slower, or more solemn speech, we doubtlessly have
(pe'R, te'a;tRo, po';ta, &koau'laR, &una';Xa, la';tRa pa'Ra;a)i (-xa)a, as the phone-
mic transcription indicates.
Furthermore, there are pronunciations which are currently considered to be un-
educated: ('pjR, 'tja;tRo, 'pw;ta, &kwau'laR); there is a dierence, and it is more
than enough to distinguish the three dierent realizations, even if the dierence
between (, 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: ('tjRr:a)
/'tjeRr:a/ tierra ( Italian ('tEr:Ra) /'tErra/ terra] (laR'r:a;jo) /laR'r:ajo/ la radio ( 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 dierences, for vowels too): ('ka;Ro) /'kaRo/
caro ('kaRr:o) /'kaRr:o/ carro ('p;Ro) /'peRo/ pero ('pRr:o) /'peRr:o/ perro (&ene'RaR)i
(-s-)a /ene'RaR/ encerar (&eneR'r:aR)i (-s-)a /eneR'r:aR/ encerrar
In our phone(ma)tic analysis, (Rr:) /Rr:/ are heterosyllabic sequences, constitut-
ed by (syllable-nal) (R) /R/ or (word-nal) (R) /R/ + (r:) /r:/, with one + three (or
four, to give more emphasis) tappings. ey are not simple segments such as /R,
r/ (or r r 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;o) /'lao/ lado (ka-
'lR) /ka'loR/ calor (kol'aR) /kol'gaR/ colgar ('alto) {((-lto))} /'alto/ alto (el'i;ne)i {((el-
'-))} (el'si;ne)a {((el's-))} /el'ine/ el cine (ko'Cn) ((-L'C-)) /kol'con/ col>n (eL'N;-
No) /el'NoNo/ el oo (eL'uke, eL'J-) /el'Junke/ el yunke ('kla;e)i (-se)a /'klase/
cle ('d;Ble) /'oble/ doble (e'la;wa) /e'lagwa/ el a%a (La'maR)i (,a-)a /La'maR/
amar ('ba;Le)i (-,e)a /'baLe/ vae (&koNLe'Bando)i (-N-, NJ-)a /konLe'bando/ con-
evando 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 //), 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 (-l, -l, -lO)): ('l)i ('s-)a
/'sol/ sol (lau'Rl) /lau'Rel/ laurel
Structures
Taxophonics
ruido (al'da;no &alde'a-) /alde'ano/ aldeano (lal'ta leal-) /leal'ta/ lealtad ('li;na
-nea) /'linea/ lnea ('tja;La to'a-)i (-,a)a /to'aLa/ toaa (&kjau'laR ko&a- &koa-) /koa-
gu'laR/ coa%lar ('r:al r:e'al) /r:e'al/ real
It must be remembered that there is a dierence between normal pronunciation
(not slow): ('ta;tRo, &empo'RaR, 'pj;ta, &kjau'laR) {/te'atRo, empeo'RaR, po'eta, koa-
gu'laR/ teatro empeorar poeta coa%lar] and uneducated pronunciation: ('tja;tRo,
&empjo'RaR, 'pw;ta, &kwau'laR), in addition to slow pronunciation: (te'a;tRo, &empeo-
'RaR, po';ta, &koau'laR, ko&a-).
Except in particular cases, for rhythmic reasons (as will soon be seen), the pro-
nunciation with unauthorized diphthongs is decidedly uneducated (although it
is very widespread, especially in Latin America): (pa'i, 'pai)i (-s)a /pa'is/ pas (ma-
'i, 'mai)i (-s)a /ma'i/ maz (ba'ul, 'baul) /ba'ul/ bal e transformation from
/ae, ao/ to /ai, au/ is equally uneducated: ('kaen 'kain) /'kaen/ caen (tRae'Ran
tRai-) /tRae'Ran/ traern (bil'Bao -au -aU) /bil'bao/ Bilbao (&baka'lao -au -aU)
/baka'lao/ bacalao in cases such as (&e[]tRa&oRi'na;Rjo, e[]&tRaoR-)i (-s-)a /e[k]stRaoR-
i'naRjo/ extraordinario we also have the possibility of: (&e[]tRFoR-, e[]&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 denitely more probable near a stress and
at the end of a word: (r:e'fRio)i (-s-)a /r:es'fRio/ rfro (&r:efRi'aR)i (-s-)a /r:esfRi'aR/ r-
friar but: (&eMfRja'mnto) /enfRja'mento/ enfriamento (and also: (r:e'fRja;o)i (-s-)a
/r:es'fRjao/ rfriado] (flu'tuo) /fluk'tuo/ ucto (&flutu'aR) /fluktu'aR/ uctuar
(and also: (&flutu';o)i (-so)a /fluktu'oso/ uctuoso as well as suntuoso virtuoso
which have only (-'tw;o)i (-so)a /-'twoso/) but: (&flutwa'jn)i (-'sj-)a /fluktwa-
'jon/ uctuacin
e innitives with /i, u/ (which are prevalently short, and their derivatives,
even when there is no longer the simple, original form) are: (re-, mal-)criar ar
rfriar enfriar liar (ex-)piar (-, -)triar %iar (d)viar enviar reenviar puar
ruar uctuar concluir excluir incluir ocluir recluir (a-, in-, re-)uir diluir
(re)huir ()muir (re)construir instruir tatuir (re)constituir dtituir instituir
rtituir stituir intuir For ui the pronunciation with /'wi/ is frequent, as well.
Other cases of /i, u/ appear for short or compound forms: (&aBi'n) /abi'on/ a-
vin (bi';njo) /bi'enjo/ bienio (biu'ni;Boko) /biu'niboko/ biunvoco (tRi';njo) /tRi-
'enjo/ trienio (tRi'agulo) /tRi'angulo/ trin%lo (&djei';Co, &dj-)i (-si-)a /jei'o-
co/ diecio>o (&beinti';Co, &bi-) /beinti'oco/ veintio>o (gi'n) /gi'on/ %in (i'a;-
to) /i'ato/ hiato (pRi'R) /pRi'oR/ prior (kRi'ana)i (-sa)a /kRi'ana/ crianza (fi'ana)i
(-sa)a /fi'ana/ anza (fi'a;o) /fi'ao/ ado (fi'ambRe) /fi'ambRe/ ambre (pi'a;no)
/pi'ano/ piano but: (&tRjagu'laR) /tRjangu'laR/ trian%lar (kRja'tu;Ra) /kRja'tuRa/ cria-
tura (fja'R) /fja'oR/ ador (koM'fjana)i (-sa)a /kon'fjana/ conanza (fjam-
'bR;Ra) /fjam'bReRa/ ambrera (pja'nita)i (-sta)a /pja'nista/ pianta Let us note,
logically: ('Cja;pa)i (-s)a /'cjapas/ iap (with no dropping of /j/ after /c/).
We also have: (u'i;a) /u'ia/ huida (as huir] (di'uRno, 'dju-) /i'uRno, 'ju-/ diur-
no (but: (dju'tuRno) /ju'tuRno/ diuturno] (Xu'i;jo, 'Xwi-)i (xu'i;sjo, 'xwi-)a /xu'ijo,
'xwi-/ juicio (but: (Xwi'j;o, 'Xw)i (xwi'sj;so, 'xws)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 cuta ruido ruin ruina arruino
However, in fast speech, forms with /u', i'/ easily become /'w, 'j/: (&detRu'iR,
de'tRwiR)i (-s-)a /estRu'iR/ dtruir (&kotRu'i;o, ko'tRwi;o)i (-s-)a /konstRu'io/
construido (fi'a;mo, 'fja;mo)i (-s)a /fi'amos/ amos (&atu'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:;lo,
lFaR-, laR-) /laaR'r:eglo/ la arreglo (lee[]'pli;ko, le-, le-)i (-s-)a /lee[k]s'pliko/ le expli-
co (&loolBi', &ljo-, &lo-) /loolbi'e/ lo olvid ('agulo o'tuo) /'angulo ob'tuso/ n-
%lo obto (&impla'ka;Ble e'k;no, &impla'ka; Ble-) /impla'kable en'kono/ implaca-
ble encono (la'pR;a 'a;e u'na~CoR r:e'mano, la'pR; 'a;e)i (-sa, -se, -so)a /la'pResa
'ae u'nancoR r:e'manso/ la pra hace un an>o remanso (e'laiRe 'entRa il'Bando, e-
'lai 'Ren-)i (sil-)a /e'laiRe 'entRa sil'bando/ el aire entra silbando (&ma'B2 'kwa;tRo ';-
Xo2 ke', &kwa'tR;-)i (&maz-, -xos, -s)a /mas'ben 'kwatRo 'oxos ke'os/ ms ven cua-
tro ojos e dos
If the V are dierent, we have: (loaR'r:;lo, ljaR-) /loaR'r:eglo/ lo arreglo (la&o-
eR'B, &lFo-)i (-s-)a /laobseR'be/ la observ (loe[]'pli;ko, lje-)i (-s-)a /loe[k]s'pliko/ lo
explico (lae[]'pli;ko, lFe-)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 ('t;o a'k;Lo,
't; ja-)i (-,o)a /'too a'keLo/ todo aeo ('tRite o'ka;o, 'tRi to-)i (-is-, -so)a /'tRiste
o'kaso/ trte oco (&entReilu'j;ne, &entRi-)i (-sj;nes)a /entReilu'sjones/ entre ilio-
n ('pu;o auen'taRe, 'pu; jau-)i (-sen'taRse)a /'puo ausen'taRse/ pudo aentarse
(';Roe iNmoR'tal, ';Rje, i-, i-, '; Rji-) /'eRoe inmoR'tal/ hroe inmortal (pa'la;jo
au'uto, pa'la; jjau-)i (-sjo, -sto)a /pa'lajo au'gusto/ palacio au%sto (e'un e&a-
no'ta;o, ano-)i (s-, s-)a /se'gun seano'tao/ segn se ha notado (de'ambo 'm;o,
'dam-)i (-s)a /e'ambos 'moos/ de ambos modos
Obviously, the function of the vocalic elements that come into contact also
count. In fast speech, a grammeme-nal 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 zero, even with dierent V\ (la&aBRi-
'R, &lFa-, &laB-) /laabRi'Re/ la abrir (lo';jo, 'lj;-, 'l;-) /lo'ojo/ lo odio (mi'i;Xo,
'mi;-, 'mi;-)i (-xo)a /mi'ixo/ mi hijo (lae'p;a, lFe-, le-)i (-s'p;sa)a /laes'posa/ la
posa (la';Ra, 'lF;-, 'l;-) /la'oRa/ la hora (lai'Xi;ta, lFi-, li-)i (-x-)a /lai'xita/ la hijita
(lo'u;niko, 'lju;-, 'lu;-) /lo'uniko/ lo nico (lou'nj;Ron, lju-, lu-) /lou'njeRon/ lo unie-
ron (me'i;Ba, 'mi;-, 'mi;-) /me'iba/ me iba (me&ima'Xi;no, &mei-, &mi-, &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 Toms, but retranscribed and completed): (&eka'l;Ra aR'r:i;Ba, -'l; RaR'r:i-)i
6. spanish 245
(&eska-)a /eska'leRa aR'r:iba/ calera arriba (a'ma;a e'p;a, a'ma; Fe-, e-)i (-s-)a
/a'maa es'posa/ amada posa (o'fRta i,'Xuta, o'fR tFi,-, -ti,-)i (-'x-)a /o'feRta in-
'xusta/ oferta injta (pa'la;BRa o'a;a, pa'la; BRFo-, BRo-)i (-s-)a /pa'labRa o'saa/ pala-
bra osada ('ka;a u'milde, 'ka; Fu-, u-)i (-s-)a /'kasa u'milde/ ca humilde ('kj;Re a-
'BlaR, 'kj; Ra-, -Ra-) /'kjeRe a'blaR/ iere hablar ('pw;e ekRi'BiR, 'pw; e-, e-)i
(-s-)a /'pwee eskRi'biR/ puede cribir
Also: ('nmbRe i'lutRe, 'nm bRi-, bRi-) /'nombRe i'lustRe/ nombre iltre ('tj;ne
oR'u;Lo, 'tj; no-, no-)i (-,o)a /'tjene or'guLo/ tiene or%o ('Xnte u'milde, 'Xn tu-
tu-)i ('x-)a /'xente u'milde/ gente humilde ('ka;i apa'a;o, 'ka; ja-)i (-si, sj-)a /'kasi a-
pa'gao/ ci apagado (mi&epe'Rana, &mie-, &mje-)i (-es-, -sa)a /miespe'Rana/ mi
peranza ('ka;i impo'i;Ble, 'ka; im-)i (-si, si-)a /'kasi impo'sible/ ci imposible
(mio&Blia'jn, mjo-)i (-'sj-)a /miobliga'jon/ mi obligacin (&niuna'B, ni&u-,
&nju-)i (-s)a /niuna'be/ ni una vez
And also: ('gRi;to a'u;o, 'gRi; tja-) /'gRito a'guo/ grito a%do ('p;ko e'fwRo,
'p; kje- ke-)i (es-, -so)a /'poko es'fweRo/ poco fuerzo ('n;Ro iM'fjRno, 'n;
RjiM-, RiM-) /'negRo in'fjeRno/ negro inerno ('kwaRto o'ku;Ro, 'kwaR tjo-, to-)i
(-s-)a /'kwarto os'kuRo/ cuarto oscuro (e'ga;No u'ma;no, e'ga; Nju-, Nu-) /en'gaNo
u'mano/ engao humano (&uami'ta, &wa-)i (&s-, &s-)a /suamis'ta/ su amtad ('im-
petu &epan't;o, -&tu e-, -pe &twe-)i (-es-, -so)a /'impetu espan'toso/ mpetu panto-
so ('tRi;Bu i'gRa;ta, 'tRi; Bwi-) /'tRibu in'gRata/ tribu ingrata (&poRuo'nR, -wo-)i
(-s-)a /poRsuo'noR/ por su honor (e'pi;Ritu u'ma;no, -Ri tu-)i (-s-)a /es'piRitu u'mano/
pritu humano ('L;a aao'RaR, 'L;a ao-, 'L; ao-)i (',-)a /'Lega aao'RaR/ ega
a adorar ('i;Ba aenen'dR, 'i;Ba en-)i (-s-)a /'iba aenen'deR/ iba a encender
Further examples still: (be'gana ai'Ra;a, -an Fai-, -an ai-)i (-sa, s-)a /ben'gan-
a ai'Raa/ venganza airada (e'ta;Ba ao'a;a, -e'ta; BFao-, Bao-)i (es-)a /es'taba ao-
'gaa/ taba ahogada (r:o'a;a au'R;Ra, r:o'a; Fau-, au-)i (-s-)a /r:o'saa au'RoRa/
rosada aurora ('kulta eu'R;pa, 'kul tFeu-, teu-) /'kulta eu'Ropa/ culta Europa (a'pRn-
de aa'BlaR, a'pRn daa-, da-) /a'pRende aa'blaR/ aprende a hablar ('auRea e'pa;a, 'au-
Ra, &au RFe'pa;a)i (-s-)a /'auRea es'paa/ urea pada ('mwRte ai'Ra;a, 'mwR tai-,
tai-) /'mweRte ai'Raa/ muerte airada (pRe'u;me aon'daR, pRe'u; maon-)i (-s-)a /pRe-
'sume aon'daR/ prume ahondar ('fRnte au'uta, 'fRn tau-, tau-)i (-s-)a /'fRente au-
'gusta/ frente au%sta (biR'Xi;neo e'kanto, -njo, -no)i (-x-)a /biR'xineo en'kanto/
virgneo encanto
'taRlo, -Bo a't-, 'bwl Bja't) /'bwelbo aa'taRlo/ vuelvo a atarlo ('bgo aempe'aR,
'b gjaem-, gjFem-, gjem-)i (-saR)a /'bengo aempe'aR/ vengo a empezar ('kwRpo
ai'R;o, 'kwR pjFi-, pjai-) /'kweRpo ai'Roso/ cuerpo airoso (di'pwto ao&Bee'R,
di'pw tjFo-, tjao-)i (-s'pws-, -sR)a /is'pwesto aobee'eR/ dputo a obedecer
('fauto au'pi;jo, 'fau tjFu-, tjau-)i (-st-, -s'pi;sjo)a /'fausto aus'pijo/ fato a-
picio
And also: (an'ti;wa alti'B, an'ti; wal-)i (-s)a /an'tigwa alti'be/ anti%a alti-
vez ('lgwa e[]'tRa;Na, 'l gwFe-)i (-s-)a /'lengwa e[k]s'tRaNa/ len%a extraa (e-
'ta;twa iN'm;Bil, i-, i-, e'ta; 'twFi-)i (es-)a /es'tatwa in'mobil/ tatua inmvil ('a;-
wa olo'R;a, 'a; wFo-) /'agwa olo'Rosa/ a%a olorosa ('fa;twa u&mani'a, 'fa; twFu-)
/'fatwa umani'a/ fatua humanidad ('mu;two a'mR, 'mu; twja-) /'mutwo a'moR/
mutuo amor ('aRwo em'p;No, 'aR wjem-) /'aRdwo em'peNo/ arduo empeo (peR-
'p;two im'p;Rjo, peR'p; twjim-) /peR'petwo im'peRjo/ perpetuo imperio (kon'ti;-
nwo e'l;Xjo, kon'ti; nwje-)i (-xjo)a /kon'tinwo e'loxjo/ continuo elogio
Lastly: ('fa;two oR'u;Lo, 'fa; twjoR-, twoR-)i (-,o)a /'fatwo oR'guLo/ fatuo or%o
('mntRwo u'ma;no, 'mn tRwju-, tRwu-)i (-s-)a /'monstRwo u'mano/ monstruo
humano ('r:;Xja au&toRi'a, 'r:; XjFu-, Xjau-)i (-x-)a /'r:exja autoRi'a/ regia autori-
dad (koR'r:j aepe'RaRlo, Fe-, e-)i (-s-)a /koR'r:jo aespe'RaRlo/ corri a perarlo (pa-
'la;jo au'uto, pa'la; jjau-, jjFu-)i (-sjo, -us-)a /pa'lajo au'gusto/ palacio au%s-
to (i'ni;kwo au'u;Rjo, i'ni; kwjau-, kwjFu-) /i'nikwo au'guRjo/ inicuo au%rio (em-
'bi;jo aeu';Bjo, em'bi; jjaeu-, jaeu-, jFeu-, jjeu-)i (-s-)a /em'bijo aeu'sebjo/
envidio a Eebio
do and el ejido (&ele'Xi;o)i (-x-)a /ele'xio/, hele>o and el he>o (e'l;Co) /e'leco/, he-
leno and el heno (e'l;no) /e'leno/, la sab and l av (la'a;Be)i (la'sa;Bes)a /la's-
abes/, ena% and en a% (e'na;wa)i (-s)a /e'nagwas/, enojo and en ojo (e'n;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 (';0),
often have ('00): ('u ;mo, "ummo) /'umo/ humo ('a;na, "anna) /'ana/ Ana ('ba;le,
"balle) /'bale/ vale ('X;fe, "Xffe)i (x-)a /'xefe/ jefe ('di;e, "die)i (-se, -sse)a /'ie/
dice ('ka;a, "kaa)i (-sa, -ssa)a /'kasa/ ca ('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 dierences with respect
to Portuguese or Italian.
e position of stress is distinctive (and we can also nd threefold phonemic
contrasts, the rst of which are nouns, indicated graphically too), as in: ('a;nimo)
/'animo/ nimo (a'ni;mo) /a'nimo/ animo and (&ani'm) /ani'mo/ anim; (kon'ti;-
nwo) /kon'tinwo/ continuo (&konti'nuo) /konti'nuo/ contino and (&konti'nw)
/konti'nwo/ continu; ('li;mite) /'limite/ lmite (li'mi;te) /li'mite/ limite (&limi't)
/limi'te/ limit (de'p;ito)i (-s-)a /e'posito/ depsito (&depo'i;to)i (-s-)a /epo'sito/
deposito (de&poi't)i (-s-)a /eposi'to/ deposit
For stress oscillations, some examples follow: acn acne aerstato aerostato
ambrosa ambrosia atmsfera atmosfera atriaco atraco balatre balastre
cartomancia cartomanca (but only farmacia] cclope ciclope conclave cnclave
dinamo dnamo elixir elxir etope etiope (but only miope] gladolo gladiolo
bero ibero mil msil olimpiada olimpada orga orgia smos osmos pel-
cano pelicano pensil pnsil pentagrama pentgrama perodo periodo policro-
mo polcromo polgloto poligloto radar rdar reptil rptil reuma rema ter-
mostato termstato utopa utopia varic vric
6.3.2.2. Lexical compounds and adverbs ending in -mente keep quite a strong
stress even on the rst element (according to the scale: ('), (), (&)): (kRta'Bla)i
(-sas)a /'koRta'bolsas/ cortabols ( (&koRta';Ra) /koRta'oRa/ cortadora), (ika'pj)
/'inka'pje/ hincapi (entRe')i (-s)a /'entRe'os/ entreds ( (&entRe' mu'Ca;Co)i
(-os)a /entRe'os mu'cacos/ entre dos mu>a>os), (r:e&al'mnte) /r:e'al'mente/ real-
mente (&beRaRa'mnte) /beRa'eRa'mente/ verdaderamente (fail'mnte)i (-s-)a
/'fail'mente/ fcilmente Also: (ta'Bia) /'toa'bia/ todava (a&i'mimo)i (a&si-
'mis-)a /a'si'mismo/ immo (as can be seen in the last example, if a rst /'/ is imme-
diately followed by another /'/, it becomes (&); on the other hand, however, if the
form were with only one /'/, /asi'mismo/, we should have (&ai'mimo)i (&asi-
'mis-)a).
To begin to acknowledge the dierent stress patterns, compared for instance
to Italian, we give some useful examples: (nal stressed) acento amplo varo
(stressed on the last but one syllable) abdico altero animo celebro certico com-
puto convoco denomino deposito dputo edico indico integro interrogo lim-
ito medito modico modulo penetro signico /signi'fiko/, termino venero violo
/'bjolo/, vito /bi'sito/, limpio /'limpjo/. Furthermore: (a'n;mja) /a'nemja/ anemia
(bi'a;mja) /bi'gamja/ bigamia (&kaRam'b;la) /kaRam'bola/ carambola (di'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'lmbRe) /e'lombRe/ el hombre (&elal-
'kalde) /elal'kalde/ el alcalde (la'ka;a)i (-sa)a /la'kasa/ la ca (la'Ci;ka)i (-s'Ci;kas)a
/las'cikas/ l >ic (&una'mi;o) /una'migo/ un amigo (&una'tRinta peR';na)i
(&unas-, -'s;nas)a /unas'tReinta peR'sonas/ un treinta person
We then have forms (plurisyllabic too), such as salvo excepto mediante duran-
te and phrases such as rpecto a junto a encima de delante de\ (a'ma;laa) /a'ma-
laga/ a Mlaga (ko'n;La)i (-,a)a /ko'neLa/ con ea (&empa'Ri)i (-s)a /empa'Ris/ en
Pars (&inom'bR;Ro)i (&sinsom-)a /sinsom'bReRo/ sin sombrero (&ante't;o) /ante'to-
o/ ante todo (&baXolo'aRBole)i (-xolo's-, -es)a /baxolo'saRboles/ bajo los rbol
(&kontRa&mia'mi;o)i (-sa'mi;os)a /kontRamisa'migos/ contra m amigos (&dee-
'i)i (-z-)a /ese'oi/ dde hoy (&entRelo';Xo)i (-'s;xos)a /entRelo'soxos/ entre los o-
jos
Furthermore: (&ata'ki, -Fa-, -aa-)i (&as-)a /astaa'ki/ hta a (&ajael'pwRto,
-jFel-)i (-sj-)a /ajael'pweRto/ hacia el puerto (&paRa&nwetRo'wpee)i (-stRos'ws-
pees)a /paRanwestRos'wespees/ para nutros husped (&oBRel';BRe)i (&soBRel's-)a
/sobReel'sobRe/ sobre el sobre (e&eto')i (-&seto's)a /egebto'os, ekepto-/
excepto dos (&Xuntoala'pwRta, -tja-)i (&xu-)a /xuntoala'pweRta/ junto a la puerta
Also: (en&ima&ela'ka;a)i (-si-, -sa)a /enimaela'kasa/ encima de la ca (de&lan-
tee'mi) /elantee'mi/ delante de m Even in more or less long series: (&paRa&entRe-
no';tRo, pa&RFen-)i (-'s;tRos)a /paRaentReno'sotRos/ para entre nosotros (&poRen&i-
mae't;o)i (-'si-)a /poRenimae'too/ por encima de todo (&poRe&lante&elXaR-
'in)i (-x-)a /poRelanteelxaR'in/ por delante del jardn (&deepo&RentRelo'aRBo-
le)i (&dez-, -'saRBoles)a /esepoRentRelo'saRboles/ dde por entre los rbol
Generally, conjunctions too are destressed (even in phrases, such as en cuanto
(e) puto e suputo e), except with ora ya bien (disjunctive), (consec-
utive), no obstante con todo fuera de (adversative), en efecto por tanto por consi-
%iente e (consecutive), apen an no no bien ya e luego e dpus
e en tanto e (temporal), a no ser e dado e con tal e (conditional), por
ms e a par de e mal e ya e (concessive), (&pwetoke&noloa'Bia)i (-sto-,
-sa-)a /pwestokenolosa'bia/ puto e no lo saba At the beginning of elliptic ques-
tions y is stressed: ('i tupa;Re21)i (-'pa;-)a /'i tu'paRe?/ Y tu padre?
ko so'noRo/ eco sonoro (&aLe'a;o)i (&a,-)a /aLe'gao/ ha egado and ('a;No 'ma;lo) /'a-
No 'malo/ ao malo (&eto'ma;o)i (&es-)a /esto'mao/ tomado and ('to 'pi;o)i
('s-)a /'esto 'pio/ to pido (o&nami'Ra;o)i (s-, -s)a /sonami'Raos/ son admira-
dos and (' naRmo'nj;o)i ('s-, -so)a /'so naRmo'njoso/ son armonioso
Object pronouns are destressed, la le lo l l los me nos os se te\ (me'pa;Ro
ami'RaR, me'pa; Rja-) /me'paRo ami'RaR/ me paro a mirar (o'Bi;mo Be'niR)i (oz'Bi;-
moz)a /os'bimos be'niR/ os vimos venir (e&lepeR'j)i (se&les-)a /selespeR'jo/ se l
perdi instead, subject pronouns and indirect pronouns are stressed:(' ')i ('s)a
/'Jo 'se/ yo s ('tu 'pw;e)i (-s)a /'tu 'pwees/ t pued ('l 'di;e)i (-se)a /'el 'die/ l
dice (poR'mi) /poR'mi/ por m (&paRa'ti) /paRa'ti/ para ti
Possessive adjectives are destressed: (&mieR'ma;no)i (-se-, -os)a /miseR'manos/ m
hermanos (tu';Xo)i (-'s;xos)a /tu'soxos/ t ojos (u'ma;Re)i (s-)a /su'maRe/ su
madre (&nwetRo'tjmpo)i (-s-)a /nwestRo'tjempo/ nutro tiempo
Generally demonstrative adjectives, are destressed (but, from a pragmatic point
of view, they can be considered potentially stressable): (&ete'li;BRo, -)i (-s-)a /este-
'libRo/ te libro (&ee'pRr:o, -) /ese'peRr:o/ e perro (a&keLamu'XR, ak-)i (-,amu-
'x-)a /akeLamu'xeR/ aea mujer
6.3.2.6. e forms of tratamiento, don doa fray sor san santo santa are al-
ways destressed before a name: (do'kaRlo)i (-s)a /on'kaRlos/ don Carlos (&doNao-
'l;Re)i (-s)a /oNao'loRes/ doa Dolor (&antoo'migo)i (&s-)a /santoo'mingo/
santo Domingo vocative forms are destressed too, seor seora seorita padre ma-
dre hermano hermana to ta\ (e&NoRmaR'ti;ne)i (s-, -s)a /seNoRmaR'tine/ seor
Martnez (e&NoRama'Ria)i (s-)a /seNoRama'Ria/ seora Mara (&paRean'dR, -Ran-)i
(-s)a /paRean'dRes/ padre Andrs (tio'Xwan)i (-x-)a /tio'xwan/ to Juan however,
when not used in forms of tratamiento, we have the regular: (&una'anta mu'XR)i
(-s-, -x-)a /una'santa mu'xeR/ una santa mujer ('bi;no ele'NR maR'ti;ne)i (-s-, -s)a /'bi-
no else'NoR maR'tine/ vino el seor Martnez (ae'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 (&pataaR'r:i;Ba) /patasaR'r:iba/ pat arriba
In numbers formed with y the rst element is destressed: (&tReintai'iko)i (-'s-)a
/tReintai'inko/ treinta y cinco (kwa&Rentai'j;te)i (-'s-)a /kwaRentai'sjete/ cuarenta y
siete also cien(to) before mil is destressed: (jeN'mil, -M'm-, -'m-, -M'm-, -'m-)i
('s-)a /jen'mil/ cien mil (&oCo&jenta'mil pe';ta)i (-&sj-, -as-, -'s;tas)a /ocojen-
tas'mil pe'setas/ o>ocient mil pet Simple numbers are stressed, even if mono-
syllabic and near a stress: ('d 'a;to)i (-z 'a;tos)a /'os 'gatos/ dos gatos ('tR
ka'Ba;Lo)i (-s ka'Ba;,os)a /'tRes ka'baLos/ tr cabaos ('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: ('kmpRame&lo) /'kompRamelo/ cmpramelo (a-
'Na;ae&le)i (-s-)a /a'Naasele/ adele (&entRe'andoe&lo)i (-s-)a /entRe'gandoselo/
entregndoselo
Words such as aune cone pore sino are normally /'aunke, a'unke 'kon-
ke 'poRke 'sino/, however, in sentences, their composition can also bring to stress
patterns such as: (&auke, a&uke, au&ke, au'k &koke, ko&ke, ko'k &poRke,
poR&ke, poR'k &ino, i&no, i'n)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'l, 'da;me'l, &dame'l) /'amelo/ dmelo!
6.3.2.9. e words (a';Ra) /a'oRa/ ahora (a'i) /a'i/ ah (a'un) /a'un/ an/aun
when they are not in an intoneme and are linked to what follows, have the fre-
quent variants ('aoRa, 'ai, 'aun): (a';Ra 'Bj;ne, 'aoRa) /a'oRa 'bjene/ ahora viene (a'i
e'ta, 'ai)i (-s-)a /a'i es'ta/ ah t (a'um mi 'pa;Re, 'aum) /a'un mi'paRe/ aun mi pa-
dre (a'un no'ale, 'au)i (-s-)a /a'un no'sale/ an no sale
In the case of /'j[s]/, in an intoneme (more often when followed by a pause), it
is quite frequent to nd, even in the neutral accent, a pronunciation which, often,
seems as if it were /'i[s]/, while, it is generally only /i'[s]/: (paR'tj, &paRti')
/par'tjo/ parti (a'j, &ai')i (-s)a /a'jos/ adis (lim'pj, &limpi') /lim'pje/
limpi (em'pj, &empi') /em'pje/ en pie however, occasionally, we can doubtless-
ly have the structure ('i&), even in neutral pronunciation: (paR'ti&, a'i&, 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/ a frequently become ('ja) /'ja/: (al'dia i'jnte, al'dja, al-
&djai'-) /al'dia si'gjente/ al da si%iente ('p;ko 'ia e'pw, 'ja, 'p;ko&ja)i
(-z, -z, -s-, -s)a /'pokos 'ias es'pwes/ pocos d dpus (u'tia o'l;Re, u'tja, u&tja-
o'-)i (s-, -s)a /su'tia o'loRes/ su ta Dolor (&doNama'Ria maR'ti;ne, &doNama'Rja, &do-
Nama&RjamaR'-)i (-s)a /doNama'Ria maR'tine/ doa Mara Martnez (gaR'ia u'tjR-
r:e, gaR'ja, gaR&jau'-)i (-s-, -s)a /gaR'ia gu'tjeRr:e/ Garca Gutirrez (&noeR'Bia
paRa'na;a, -'Bja, -&BjapaRa'-)i (-s-)a /noseR'bia paRa'naa/ no serva para nada
Other examples: (ea&Bia'pwto e'pj, ea&Bja'-, a-)i (s-)a /seabia'pwesto e'pje/
se haba puto de pie (&eta'Ria kan'ao, -'Rja, -&Rjakan'-)i (&es-, 'sa;os)a /esta'Rian
kan'saos/ taran cansados (&nopo'RiaN Le'a; Ra'tjmpo, -'RjaN, -&RjaNLe'-)i (,-)a
/nopo'Rian Le'gaR a'tjempo/ no podran egar a tiempo
Intonation
6.3.3. 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'Bi ma'Na;na poR&lama'Na;na3 3)i (me'Bi ma'Na;na poR&lamaNa;na3 3)a /me-
'boi ma'Nana poRlama'Nana./ Me voy maana por la maana.
Text
6.4.0. e transcribed passage, e North Wind and the Sun is given in four
normalized versions. We start with the (neutral) Iberian pronunciation of (neu-
tral British) English this is the rst 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, uent 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 rst.
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 0.12).
Spanish Text
6.4.2. El viento norte y el sol poraban sobre cul de eos era el ms fuerte, cuan-
do acert a par un viajero envuelto en an>a capa. Convinieron en e ien ant
lograra obligar al viajero a itarse la capa sera considerado ms poderoso.
El viento norte sopl con gran furia, pero cuanto ms soplaba, ms se arrebujaba
en su capa el viajero; por n el viento norte abandon la empra. Entonc bri el
sol con ardor, e inmediatamente se dpoj 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 rst examples (for the time being, let us ignore the dia-
phonemic undersigned dots): (i'fi;si)b /e'fisi/ dicil ('e;]i)b /'ee/ ele ('E;])b /'E-
a/ ela (ba'ta;t)b /ba'tata/ batata ('nO;v)b /'nOva/ nova ('no;vu)b /'novu/ novo
('u;v)b /'Suva/ >uva However, it is worthwhile reecting upon (A): ('pAu) /'pau/
pau (Au'das)b {(Au'aS)l] /Au'as/ audaz ('mA)b {('mA)l] /'ma/ mal (A'tu;R)b
{(A'tu;R)l] /A'tuRa/ altura
It is important to distinguish: seca ('se;k)b /'seka/ dry (f.) and ('sE;k)b /'sEka/
dries, drought, nuisance, cerco ('seku)b /'seKku/ siege and ('sEku)b /'sEKku/
I enclose, lobo (']o;bu)b /'obu/ wolf and (']O;bu)b /'Obu/ lobe, fosso ('fo;su)b
258 a handbook of pronunciation
7.1.1.3. Notice that nasalization slightly changes some timbres according to the
diaphonemic symbols used: in particular /a/ (A) (for which, as already seen,
() is possible as well), and /'e, ', 'o, '/ (, 9) (when stressed, some-
times, (e, ) may be heard too).
Besides, in word-nal position, we have real diphthongs for what too often and
too hurriedly is transcribed simply as /e, /, /ein, oun/ ('I, eI '9U, U).
Instead, in the same position, /an/ opposes the diphthong /aun/: ('OfA) /'OKfan/
rf ('OfAu) /'OKfaun/ rfo 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/ ama ('bA;Nu)b /'baNu/
bao ('fi;nu)b /'finu/ no
ereas, in traditional neutral Brazilian pronunciation, nasalization occurs on-
ly in checked syllables (in N] either stressed or not; but a troublesome occur-
rence of nasalization is possible (or not), in unchecked stressed syllables followed
by /N/ (and with /'a/ (') as well): (&ama'NA)b;t /Ama'Nan/ ama ('bA;Nu,
'b;Nu)b;t /'baNu/ bao ('fi;nu)b;t /'finu/ no
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 troublesome occurrence of nasalization is possible even
in /N/ ( an unchecked, unstressed syllable, followed by /N/): (&amA'NA, &a-
ma-)b;m /Ama'Nan/ ama ('bA;Nu)b;m /'baNu/ bao ('fi;nu)b;m /'finu/ no
7. portuguese 259
Lusitanian peculiarities
7.1.2.1. Let us now consider 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 g-
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;], -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 /, / or, deviating from ocial symbols, they should rather
be /, /; but, more traditionally rendered as /, /) for (, '), (', ) in
addition to () /A/ and (A) /'a/, with occasional oppositions between them
(that, in a Lusitanian way, we ought to mark as /a, /).
Besides, (, a, ) /, A, / are needed, above all, but not only, in some pre-stress
syllables (by diachronic fusion, which is no longer visible in present-day spelling:
// = //, deriving from a prior /0/ structure, with C dropping), in addition
to /, , A; K, K, AK/, as we will see (and /Au/, and /Ai/ as well). Fur-
thermore, /-e2 -2 -e, -/ are special cases, as in: (ab'd9;meI)b (B'O;mI,
-n)l /ab'me2/ abdmenb -menl ('ipsi]U)b (-]U, -n)l /'ipsi2/ psilon
(&ik]u'zi;vi)b (&i~k9u-, -v[], -v)l /inku'zive, -/ inclive
7.1.2.3. Here are some other remarkable cases of /A/, in Lusitanian pronuncia-
tion, by contraction: (a) /A/ [a a] (a'ke;]) /A'kee/ ele (ka'vIR) /kA'veiRa/ ca-
veira ((a, a'ke;]i, ka'veIR)b] after CC (simplied or not in their pronunciation
and in spelling): (a'sAu)l /A'saun/ a(c)o (fa'tu;R)l /fA'tuRa/ fa(c)tura ((a'sAu, fa-
'tu;R)b] Also (ka'm9iS)l /kA'moins/ Cames.
Before looking at the two further phonemes, let us consider the Lusitanian
pronunciation of the examples already given in Brazilian pronunciation: (d'fi;si)l
/i'fisi/, ('e;])l /'ee/, ('E;])l /'Ea/, (b'ta;t)l /ba'tata/, ('nO;v)l /'nOva/, ('no;vu)l /'no-
vu/, ('Su;v)l /'Suva/ ('pAu)l /'pau/, (Au'aS)l /Au'as/, ('mA)l /'ma/, (A'tu;R)l /A'tu-
Ra/ ('se;k)l /'seka/, ('sE;k)l /'sEka/, ('seku)l /'seKku/, ('sEku)l /'sEKku/, (']o;Bu)l
/'obu/, (']O;Bu)l /'Obu/, ('fo;su)l /'fosu/, ('fO;su)l /'fOsu/.
Furthermore (for both accents): (ka'da;ve)b (k'a;vR)l /ka'davK/ cadver ('sO;-
Ro)b (-R)l /'sORK/ sror (a'ma;ve)b ('ma;v)l /a'mav/ amvel ('Akjo)b ('A-
k)l /'akw/ lcool (vo'ta)b (v'taR)l /v'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 dierence in the real-
ization of /e, o, / (e) and (o).
7.1.2.5. Here are some specic examples to show the dierences 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 (), which is close to the realization of the
Lusitanian phoneme // (but we must lay great stress on the fact that they are
not alike {even if, occasionally, for some words or specic speakers, the dierence
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 oddity consists of the change of /i/ (i) -i- sequenc-
es (not /i/ (i)) into //, /e/, in adjacent syllables (except for the last one,
whether stressed or not, and the rst one, if absolute initial, with no C-]\ (&dv'iR,
dv'-)l (&ivi'i)b /eve'iK/ dividir (m[]'niStu)l (mi'nistRu)b /me'nistRu/ mintro
(&in[]'mi;u)l (&ini'mi;gu)b /ine'migu/ inimigo (p&vi][]'Za;u)l (pRi&vi]e'a;du)b /pRe-
vi'Zjau/ privilegiado (in the very last example, the sequence (-]'Z-) /-'Z-/ results
from what has just been said previously).
However, we nd 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[]&vi'RimS)l (i&vii'RiAmus)b /evei'Riamus/ dividiramos and in superla-
tives: (&df'si;]imu)l (&ifi'si;]imu)b /efi'siimu/ diclimo
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 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, U, 9U)b (I, I, 3I, I oU, 9U, U 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 rst element is closer, as occurs however for
/a/ (4) as well).
rst of all, we list them phonetically and diaphonemically: (eI)b (I, I)l /ei/,
(Ei, i) /Ei/, (ai) /ai/ (ui) /ui/, (oi) /oi/, (Oi, i) /Oi/ (iu) /iu/, (eu) /eu/, (Eu, u)
/Eu/, (Au) /au/, (oU, o)b (o, oU)l /o/ ((o) unstressed //).
Here are some actual examples: ('eIs)b ('IS)l /'Keis/ re (]eI'9is)b (]I'S9iS)l
/ei'Soins/ Leix ('Eis)b ('EiS)l /'KEis/ r (&i'zi;tus)b (&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/ s
(iRikA'm~Ci)b (iRik'mnt)l /i'ROika'mente/ heroicamente ('viu) /'viu/ viu (and
vi-o as well, since the supposed dierence between the two is not at all phonic in
neutral pronunciation but derives exclusively from a wish to keep dierent mor-
phological forms separate), ('seu) /'seu/ seu (&apu'zi;Nu)b (&Spu'zi;Nu)l /SapEu'zi-
Nu/ >apeuzio (a'pEu)b (S-)l /Sa'pEu/ >apu ('mAu) /'mau/ mau ('voU, vo)
/'vo/ vou Lusitanian pronunciation prefers the reverse order, ('vo, 'voU), for the
last one.
Instead, the simplication of ('eI)b ('I, I)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'fuu/ inuo ('mou) /'mou/ moo (']u)b (-)l /'ua/ lua ('i)b ('i)l /'Kia/ ria
('bo)b (-)l /'boa/ boa including combinations as in the true triphthongs resulting
from /i+ei, i+au/: ('vieIs)b ('viIS)l /'vieis/ veis, ('ieIs)b ('iIS)l /'Kieis/ reis, ('ieI)b
('iI)l /'Kiein/ riem, and ('iAu)b ('i4u)l /'Kiaun/ riam.
ger and avoid people believing that, in unstressed syllables, in Lusitanian pronun-
ciation they can be reduced to /u, i/ (as some texts say).
Let us briey 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 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), ((E, O)) in-
cluding intermediate shifts (however, again of the opening type, although limit-
ed), that is ((Ep, OQ)).
Consonants
7.2.0. 7.3 is the table of the Luso-Brazilian consonants, which we will exam-
ine systematically, according to manners of articulation. Instead, 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
m (n) n (~) N () ()
F pb t d ( )b kg
(C )b
f v ()l ()l
_ s z ( )b Sl Zl
(B)l ()l /j/() /w/(j) ()b
R /K/
(])| ()b L
/b/ (b, Bl), /t, d/ (t, Cb d, b, l), /k, g/ (k, b g, b, l), /K/ (b, l)
Nasals
7.2.1.1. Portuguese has three nasal phonemes: (m, n, N) /m, n, N/: ('kA;ms)b
('k;mS)l /'kamas/ cam ('kA;ns)b ('k;nS)l /'kanas/ can ('kA;Ns)b ('k;NS)l /'ka-
Nas/ ca
In addition, there are some taxophones for /0, , /, 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 ('Ki, 'esu, 'A, 'bRAku,
's, 'us/'uS) 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/: ('tmpu)b
('tpu)l /'tempu/ tempo ('s9mbR)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, ), /t, d/ + /i, e, j/): ('A~-
Cis)b ('Ant?S)l /'antes/ ant ('vnd)b ('vnd)l /'venda/ venda ('9~i)b ('9nd)l /'on-
de/ onde () + /k, g/: ('nuk)b ('nu~k)l /'nunka/ nunca (']igj)b (']igj)l /'in-
gwa/ ln%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/: ('p9mb)b (-)l
/'pomba/ pomba ('kAg)b ('kAg)l /'kanga/ canga ('tnd)b (-)l /'tenda/ tenda
('Ci;midu)b ('ti;miu)l /'timiu/ tmido (sAu'da;i)b (sAu'a;)l /sAu'ae/ saudade
('Ca;tRu)b ('ta;tu)l /'tjatRu/ teatro Before front V or /j/, /k, g/ are realized as pre-
7. portuguese 267
velar, ((, l)), by coarticulation, but it is not necessary to use these symbols, ex-
cept for Brazilian pronunciation before /i, e, j/, where we nd (, ): ('intu)b ('kin-
tu)l /'kintu/ into ('tO;i)b ('tO;k)l /'tOke/ toe (iN'E;tu)b (i~'kE;tu)l /in'kjEtu/ in-
ieto ('i)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 /d/ (d), with an api-
cal contact (since both of them are homorganic articulations), or for emphasis or
precision: ('b9U) /'boun/ bom ('da) /'a/ d ('ga;tu)l (-u)b /'gatu/ gato ('kAdu)l
(--)b /'kadu/ caldo (p'i;u)l (pe'i;du)b /pe'iu/ pedido with emphasis: (p"di:-
du)l
7.2.2.2. In all other cases, continuous realizations occur, (B, , ): two approxi-
mants and a constrictive respectively. e diaphonemic transcription shows /b, ,
g/, just to underline the dierence 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, , g/ is po-
tential, since it corresponds to (b, d, g) or (B, , ) according to actual contexts,
not in absolute terms. In addition, we must admit that in the case of /b, g/, and
of /Rb, R, Rg/, and of (more) scholarly consonant clusters as well, stop realizations
are not rare, (b, g Rb, Rd, Rg), even without emphasis or desire for precision.
Some examples: ('a;B)l ('a;b)b /'aba/ aba (u'BRi~ku)l (u'bRiku)b /u'bRinku/ o
brinco (]iZ'Bo, ]Z-)l (]iz'bo)b /ia'boa/ Lboa ('ABu, 'Ab-)l ('Abu)b /'abun/
lbum (R'Bi;tu, R'b-)l (a'bi;tRu)b /aK'bitRju/ arbtrio (&Bi'kaR, &bd-)l (&abi-
'ka)b /abi'kaK/ abdicar (&um'a;)l (&umade'da;d)b /umae'aa/ uma deda-
da (p'RAu)l (pa'dRAu)b /pa'Raun/ padro ('aRju, 'aRd-)l ('adju)b /'aKwu/ r-
duo (dZ'3I)l (dez'dI)b /ea'ein/ ddm
More: ('a;u)l ('ga;gu)b /E'gagu/ gago (A'OS, A'g-)l (A'gOs)b /A'gOs/ algoz
('veZu)l (-zgu)b /'veagu/ vgo ('Ra;s)l (a'gRa;s)b /a'gRasa/ a graa (&R'ma;s,
&Rg-)l (&agA'ma;s)b /aKga'masa/ argamsa
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 b, Sl, b, Zl) /f, v s,
z S, Z/: ('fa;su)b (-su)l /'fasu/ fao ('po;vu) /'povu/ povo ('su)b (-)l /'su/ sul (te-
'sAu)b (--)l /ten'saun/ tenso ('vAs)b (-s)l /'vasa/ valsa ('pEs)b ('pEs)l /'pER-
sa/ persa (pa'se;u)b (p's;u)l /pa'seju/ pseio ('va;zu) /'vazu/ vo (a'za)b ('zaR)l
/a'zaK/ azar (u'zi~us)b (u'zindS)l /u'zindjus/ os ndios (e'Res)b (S'ReS)l /Se'Res/ xe-
rez ('Eis)b ('SERS?S)l /'SEKSes/ Xerx ('o;i)b ('o;Z)l /'oZe/ hoje
In neutral Brazilian, the diaphonemes /s, a/ behave as /s, z/, with distribution:
/s|, s[]=, z[], z/, while, in neutral Lusitanian, they correspond to /S|, S[]=, Z[],
z/ (as well as in the Carioca accent, of Rio de Janeiro, but of course realized as
(|, []=, [], z)): ('tRas)b ('taS)l /'tRas/ trs (us'pais)b (uS'paiS)l /us'pais/ os pa (i-
'sEsu)b (?S'sEsu)l /is'sEsu/ excelso (bas'ta)b (bS'taR)l /bas'taK/ btar (pes'ka;d)b
(p?S'ka;)l /pes'kaa/ pcada (az'mAis)b (Z'mAiS)l /aa'mains/ m ('azm)b
('aZm)l /'aama/ ma (az']As)b (Z']AS)l /aa'ans/ ls (further examples occur in
other parts of this chapter).
Approximants
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 ( ('ka;u)b (-u)l /'kaKu/ car-
ro] ('bRa;su)b (-su)l /'bRasu/ brao ('fRiu)b ('f-)l /'fRiu/ frio
Besides (as a realization of /K/) (R) occurs, in Brazilian pronunciation, in word-
-nal position too, when followed by a subsequent word-initial V with resyllabi-
cation and behavior as if in word-internal position ( /K/=/R/), while in Lu-
sitanian pronunciation it also occurs in nal position, even before a pause or a C\
(']e 'RAgu)b (']e 'RAu, -gu)l /'eK 'agu/ ler algo (']e 'poUku, 'po-)b (']e 'poku, 'poU-)l
/'eK 'poku/ ler pouco (nAu']e)b (n4u']eR)l /naun'eK/ no ler
Laterals
7.2.6.1. ere are two lateral phonemes, in both accents, /, L/. For the former
we prefer the symbol //, to a more generic /l/, since even before V it is realized as
a semi-velarized alveolar, (]) (or () 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 () (with the possible uvularized variant, (l)):
(']eICi)b (']It)l /'eite/ leite ('k]a;Ru)b ('k9a;-)l /'kaRu/ claro (a'zu ]is'ku;Ru)b ('zu ]ZS-
'ku;Ru)l /a'zu is'kuRu/ azul curo ('sAtu 'sAutu)b ('sAtu)l /'satu/ salto (e'RA
-Au)b (Z'RA)l /Ze'Ra/ geral ('ta;Lu) /'taLu/ tao
ile in Brazilian pronunciation, as already said, /, 0/ may become (u); in
Lusitanian pronunciation, // 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 dierent 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'peu[s]/ capelo(s) but (ka'pE;][s])b (k'pE;][S])l /ka'pEa[s]/ capela(s)
7. portuguese 271
However, not all feminine forms have /E/; actually, very often it is not so: ('ze;-
bR)b (-BR)l /'zebRa/ zebra ('se;d)b (-)l /'sea/ seda on the other hand, not all
masculine forms have /e/ either, so we nd: ('bE;]u) /'bEu/ belo compared to ('ne;-
gRu)b (--)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 ('poku)l /'poKku/ porco while we have ('pOkus)b ('pOkuS)l /'pOKkus/ porcos
and ('pOk[s])b ('pOk[S])l /'pOKka[s]/ porca(s) as well; besides: ('no;vu) /'novu/ no-
vo but ('nO;vus)b (-S)l /'nOvus/ novos and ('nO;v[s])b (-[S])l /'nOva[s]/ nova(s) (fo-
'mo;zu)b (fR-)l /foK'mozu/ formoso but (fo'mO;zus)b (fR-, -S)l /foK'mOzus/ formo-
sos and (fo'mO;z[s])b (fR-, -[S])l /foK'mOza[s]/ formosa(s) however, we nd: (is'po;-
zu[s])b (?S-, -u, -S)l /es'pozu[s]/ poso(s) and (is'po;z[s])b (?S-, -[S])l /es'poza[s]/ po-
sa(s) but (is'pO;zus)b (?S-, -S)l /es'pOzus/ posos for the couple.
Nevertheless, we also nd several cases with no variation, such as: (a'do;bu[s])b
('o;Bu, -S)l /a'obu[s]/ adobo(s) ('godu[s], -[s])b (-Ru[S], -[S])l /'goKu[s], -a[s]/
gordo(s) -a(s) Of course, there are cases with /O/ as well: ('mO;du[s])b ('mO;u, -S)l
/'mOu[s]/ modo(s) ('fO;ku[s])b (-ku[S])l /'fOku[s]/ foco(s) and even feminine forms
with /o/: ('go;t[s])b (-[S])l /'gota[s]/ gota(s) ('fos[s])b ('fos[S])l /'foKsa[s]/ fora(s)
7.3.1.4. For verbs the endings which cause closing are -o -a - -am /-u, -a, -as,
-aun/ (for second-conjugation forms with stress on the stem): ('de;vu, 'mo;vu {-,
-s, -Au})b ({-, -S, -4u})l /'evu, 'movu {-a, -as, -aun}/ devo movo
In checked syllables in N the eect is neutralized: ('vndu, 'v~i)b ('vndu,
'vnd)l /'vendo, 'vende/ vendo vende however, in unchecked syllables, Lusitanian
pronunciation maintain the dierence (contrary to Brazilian pronunciation due
to nasalization): ('t;mu, 't;mi)b ('te;mu, 'tE;m)l /'temu, 'tme/ temo teme ('k9;mu,
'k9;mi)b ('ko;mu, 'kO;m)l /'komu, 'kme/ como come
For the same reason, nouns behave in the same way: ('s;no)b ('sE;nR)l /'s-
njK/ snior/snior ('k9;miku)b ('kO;mku)l /'kmiku/ cmico/cmico (with far from
unquestionable consequences on spelling, which is still overestimated).
Lastly, we have the pronouns ('e;]i[s], 'E;][s])b (-, -S, -[S])l /'ee[s], 'Ea[s]/ ele(s) e-
la(s) ('esCi[s], 'Est[s])b ('eSt, -?S, 'ESt[S])l /'este[s], 'Esta[s]/ te(s) ta(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) ( 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'gRi, &a]i-, &a]I-)b (&]'Ri)l /ae'gRia/ alegria (so'i;zu, su-, sU-)b (s'i;zu)l /su'Ki-
zu/ sorro (ve']u;du, vi-, vI-)b (v']u;u)l /ve'uu/ veludo
However, the same vowels may be realized as (, ) when the stress is on open-
er V\ (e']O;u, -)b (']O;Zu)l /Ke'OZju/ relgio (ko'E;tu, k-)b (k'E;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 , ) would
certainly produce something excessive, and strange or foreign, since lofty or rare
words must remain unchanged. Furthermore, we have for instance: foia /fo-
'LiNa/ (fu'Li;N)b (f'Li;N)l calendar, but (fo'Li;N)b (f'Li;N)l small leaf, which
is felt as a derivative, contrary to the other.
ni'gI)b (-3I)l /kounnin'gein/ com nin%m (']A A~'Ci;g, ']A;~ 'C-)b (']A 4n'ti;,
']A;n 't-)l /'an an'tiga/ l antiga (kU'As)b (kU'As)l /koun'ansja/ com nsia (']A
a'zu, ']A; 'z-)b (']A 'zu, ']A; 'z-)l /'an a'zu/ l azul (kU'i;su)b (-su)l /koun'isu/ com -
so (u'9;meI 'Atu, '9;-)b (u'O;mI 'Atu, 'O;-)l /un'mein 'atu/ um homem alto (se-
'qE;])b (s'qE;])l /sein'Ea/ sem ela (nA'E)b (n4'E)l /naun'E/ no
dierence as to ('kai) /'kai/ cai] In Lusitanian pronunciation, again, the rare in-
stances of /i/ oscillate between /i/ (which is more modern and closer to spelling)
and (the more traditional) /e/: ('taksi, -s)l ('taksi)b /'taksi/ txi forms like /'ZuRi/ j-
ri and /'ZuRe/ jure (which are alike in Brazilian pronunciation: ('u;Ri)b], in the Lu-
sitanian one, may either be alike: ('Zu;R)l, or dierent: ('Zu;Ri)l (-R)l (respectively).
7.3.2.5. en, at word boundaries, /KK, / come together, they normally sim-
plify: ('ma[] 'A)b ('ma[R] 'A)l /'maK 'Kja/ mar real ('ma[] umo'Ro;zu)b ('ma[R]
umu'Ro;zu)l /'maK Kumo'Rozu/ mar rumoroso (kA'nA[] ']impu)b (k'nA[] ']ipu)l
/ka'na 'impu/ canal limpo (kA'nA[] ]imi'ta;du)b (k'nA[] ]m'ta;u)l /ka'na emi-
'tau/ canal limitado
e same occurs to /ss, az, sS, aZ/, when they are alike: ('dE[s] 'sE;ku]us)b ('dES
'sE;k]S)l /'Es 'sEkuus/ dez sculos ('dEs 'a;]is)b ('dE[S] 'Sa;]S)l /'Es 'Sais/ dez xal
(a[z]'z9;ns)b (Z'zo;nS)l /aa'zonas/ as zon (&azA'nE;]s)b (&[Z]Z'nE;]S)l /aaZa'nEas/
janel however, we nd (Ss)l (s)b /ss/ in cases such as: (&?Ss']nt)l (&ise']~Ci)b
/isse'ente/ excelente (kZS'seR)l (kRe'se)b /kRes'seK/ crcer
7.3.2.6. In the Lusitanian accent, the sequences /0, 0R/ are often realized as
(0], 0R) (even if it is better to avoid such a pronunciation): ('f9oR f']- f']-)l
('f]o)b /'foK/ or ('k9a;Ru k']- k']-)l ('k]a;Ru)b /'kaRu/ claro ('g]O;R g']- g']-)l
('g]O;R)b /'gORja/ glria on the other hand, the sequences /0e, 0eR, 0o, 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'i 'ki)l (ke'Ri)b /ke'Ria/ eria (ko'Ro k'Ro 'ko)l
(ko'Ro)b /ko'Roa/ coroa (f'ROS, f'OS, 'fOS)l (fe'ROs)b /fe'ROs/ feroz (&mR'seR, m'seR)l
(&meRe'se)b /meRe'seK/ merecer in current speech it is normal to reduce the preposi-
tions combined with the denite article pelo(s) pela(s) to /p, p/, also in Brazil-
ian Portuguese [p'lo po]\ (p]A'mAi, &pe]A-)b (p9'mAi, &p]-)l /p[e]a'main/ pela
me ('p]ja, p]u'a, &pe]u-)b ('p9jaR, p9u'aR, &p]u-)l /p[e]u'aK/ /p[e]u'aK/ pelo ar
Compare also: pra ('pa;R)b (-)l /'paRa/ and para (&paRa, pRa0, pa0, paR,
pR)b (&pR, p0, p0, pR, p)l /&paRa, pRa0, pa0, paR, pR/: (&paRa'ka,
pRa'ka &paRa'i, pRa'i)b (&pR'ka, p'ka &pR'ki, p'ki)l /para'ka, pRa'ka para'ki,
pRa'ki/ para c para aqui; besides, cada is generally (&kada)b (&k)l /&kaa/. 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}, a{s}, na{s}/, and even for mas (mas)b (mS)l /mas/
(with a possible secondary stress for rhythmic reasons, (&mS)l).
In Lusitanian pronunciation again, also /0es, 0es/, mainly in the ending - may
be quite reduced, up to becoming intense syllabic (however the dropping of
the V is an uneducated feature): ('f9o;RZS, -R -S)l ('f]o;Ris)b /'foRes/ or; combining
what we have just seen above, in a fairly uneducated type of Lusitanian pronun-
ciation, we may also nd (f']o;S)l), (u'mAs?S, -s)l ('mAsis)b /Ko'manses/ ro-
manc (&StAu'RAnt?S, &t-, -nt)l (&estAu'RA~Cis)b /Kestau'Rantes/ rtaurant
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'fItu)l may stand for (pe'feItu, pRe'feItu)b /peK-
'feitu, pRe'feitu/ perfeito prefeito (with additional variants (p'fItu, p'fItu)l].
7.3.2.7. In both accents, the preposition com /koun/, followed by the articles,
is reduced to /kon-/ (k), and to (ku, kj, k) as well: (ku'pai, ku-, kju-)b (ku-
'pai, ku-, ku-)l /kon-u'pai/ com o pai (k&umAmu'LE, ku-)b (k&ummu'LER,
ku-)l /kon-umamu'LEK/ com uma muer (kaz'mAus, kaz-, kjaz-)b (kZ-
'mAuS, kZ-, kZ-)l /kon-az'mauns/ com mos (kuz']o;bus, kuz-)b (-Z']o;BS,
k-)l /kon-uz'obus/ com os lobos
Even em is considerably reduced, up to /in/: (eIm&putu'ges, im-)b (I&pt-
'eS, i-)l /impuKtu'ges/ em portu%s (eI'ka;z, i-)b (I~'ka;z, i~-)l /in'kaza/ em
casa (&eqAu'si;]u, iAu-, q-)b (qAu'si;]u, iAu-, q-)l /in-Au'siju/ em auxlio 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, /kon, in/. Otherwise, we could use //, as a diapho-
neme (but in a systematic way, then, for all cases of //); however, this would
make the transcription heavier and the phonemic analysis more complicated.
In rhythm groups, sequenze // = (0): (u&mejA'mi;gu)b (u&mej'mi;u)l
/umeua'migu, umewa-/ o meu amigo.
7.3.2.8. In the neutral Lusitanian accent, as the various examples have shown,
we systematically nd (in addition to (, n, ~, , , 9, , , ), 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&pf-
'so;RZS &pt'e;zZS)l (us&pRofe'so;Ris &potu'ge;zis)b /uspRofe'soRes poKtu'gezes/ os prof-
sor portu%
In Lusitanian pronunciation, /e, u/ 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'kup, d?S'kup)l (is'kupi)b /es'kupe/ d-
culpe (&tvi'zAu, &tv-, &t]-)l (&te]evi'zAu)b /teevi'zaun/ televo (p'tE;s, &p'tE;s)l
(&ape'tE;si)b /ape'tEse/ apetece ('pOt, ?S'pOtuS)l (is'pOtus)b /es'pOKtus/ d-
portos
More: (S'paR, &?Sp'RaR)l (&espe'Ra)b /ispe'RaK/ perar (St'tA, &?S-)l (&ista'tA)b /ista-
'ta/ tatal (Sk'seR, &k-, &?S-)l (&iska'se)b /iska'seK/ cser (&'B;Nu, )l (e'bA;Nu)b
/Ke'baNu/ rebao (&supuRi'a[] u'sO, &sp&Ru-)l (a&supeRoRi'da[i] du'sO, -'dad
du-)b /asupeRjoRi'ae u'sO/ a superioridade do sol (u'fin s';n, s-, ds-)l (u-
'fi~ ise'mA;n)b /u'fin dese'mana/ o m-de-semana (p'RE;s um'Bo i';, p'RE;s
u-, p'RE;s m-, p'RE;s mu-)l (reading: (pa'RE;semi uma'boa i'dE;, -se mqu-)b} /pa-
'REseme uma'boa i'Eja/ parece-me uma boa ideial/idiab
7.3.2.9. We will now consider some examples of /0u/ = () (the symbols for
/0e/ are more normal, as in ('knt)l ('k~Ci)b /'kente/ ente]\ ('a;t[[])l ('a;Ci-
mu[s])b /'atimu[s]/ timo(s) ('p9;[])l ('p]A;nu[s])b /'panu[s]/ plano(s) ('b;[])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 rst element, uctuating between a primary and a strengthened sec-
ondary one (according to the scale ('), (), (&)); so we have a rst /'/ followed by an-
other which is realized as () (or (&), if adjacent to the primary stress): (gjada-
'oUp, -o;-)b (-R'o;p, -oU-)l /'gwaKa'Kopa/ %arda-roupa (agja'fOCi)b (aj-
'fOt)l /'agwa'fOKte/ %a-forte (ACi'baiu)b (At'BaiSu)l /'ati'baiSu/ altibaixo (&mA-
'kRa;du)b (&mA'ka;u)l /'ma'kRjau/ malcreado (pAu~i']O)b (pAund']O)l /'paun-
de'O/ po-de-l (]uzobRazi']eIRu)b (-BRzi']I-)l /'uzbRazi'eiRu/ lo-brileiro
ere are even some particular compounds, still considered as such (rather than
crystallizations), which maintain the two elements fairly independent; rst of all,
we nd adverbs in /-'mente/ -mente: (ustA'm~Ci)b (ZuSt'mnt)l /'Zusta'mente/ j-
7. portuguese 277
7.3.3.3. In addition, nouns with the diminutive inx /-z-/ -z- follow the same
pattern: (mu&L'zi;N)b (mu&LR'zi;N)l /mu'LEK'ziNa/ muerzia (ku&L'zi;N)b
(k&LR'zi;N)l /ku'LEK'ziNa/ coerzia (fA'zi;N)b (f4'zi;N)l /'OKfan'ziNa/ or-
fzia (avena'zi;N)b (Savn'zi;N)l /'Savena'ziNa/ >avenazia (a&v'zi;N)b (-
&v'zi;N)l /a'vO'ziNa/ avozia (a&vo'zi;Nu)b (&vo'zi;Nu)l /a'vo'ziNu/ avozio (&pai'zi;-
Nu)b (-'zi;Nu)l /'pai'ziNu/ paizio (9meI'zi;Nu)b (mI'zi;Nu)l /'mein'ziNu/ ho-
menzio (9meIza'Au)b (mIz'Au)l /'meinza'Kaun/ homenzarro (a-
&pa'ze;Lu)b (&pa'z;Lu)l /Ka'pa'zeLu/ rapazeo (]e&s'zi;tu)b (]e&s'zi;tu)l /en'sO-
'zitu/ lenolzito (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 signicant and certainly not free from serious doubts
when consideredmerely from an orthograc 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/ inuir,
/in'fluu/ inuo, /in'flui{s}/ inui(s), /in'flwi/ inu, /in'flwiu/ inuiu, /in'flwia{s}/
inua(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; /i-
'zia/ dizia, /i'Ria/ diria, /i'Rei/ direi.
Now, let us observe very carefully the following: /sa'iK/ sair, /sa'iu/ sado, /sa'in-
du/ saindo, /'sai{s}/ sai(s), /sa'i{s}/ sa(s), /'saju/ saio, /'saja{s}/ saia(s), /sa'iu/ saiu,
/sa'ia{s}/ saa(s), /sa'iaun/ saam, /'saein/ saem, /sa'iRaun/ saram, /sai'Riaun/ sairiam,
/sai'Raun/ sairo, /sai'Ria{s}/ sairia(s), /sai'Rei{s}/ sairei(s), /sa'iReis/ sareis, /sai'Rieis/
saireis.
Lastly, we nd inxed futures and conditionals (which are quite odd {for
foreigners or Brazilians, too}): (&se']jI)l /'se'wei/ s-lo-ei (t&ta'sa)l /tRa'taK'sja/
tratar-se- (&diR'LjaS)l /'iK'Lwas/ dir-o-s (kn&ta']jAu)l /kon'ta'waun/ cont-lo-o
(&faR'L[]i)l /'faK'Ljia/ far-e-ia (pu&e's[]i)l /po'eK'sjia/ poder-se-ia (&di']jiS)l /'i-
'wias/ di-lo-i (kn&ta']i4)l /kon'ta'wian/ cont-lo-iam (reading: (&se']jeI, tRa&ta-
'sa, &i'Ljas, kn&ta']jAu, &fa'L[]i, po&de's[]i, &i']jis, kn&ta']iA)b).
Intonation
7.3.3.4. 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:
/./: (t;Nu Ami;gu muintu simpa;Ciku3 3)b ('t;Nu 'mi;u muintu sipa;ti-
ku3 3)l /'teNu un-a'migu 'muintu sim'patiku./ Teo um amigo muito simptico.
278 a handbook of pronunciation
/ / (2 2 2 2 2 2 2) /./ (2 3 3)
/ / ( 2 2 2 2 2 2 2) /?/ (2 1 2)
/ / ( 2 2 2 2 2 2 2) // (2 2 2)
/ / ( 2 2 2 2 2 2 2) /,/ (2 ' 2)
International accent
7.4.3. As shown in the gure, the vowels are (i e 'E, a 'O, 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, &psu'daR, 'o;ku, 'u;va) /'vi, 'ee, Ea'Eve, 'gata, 'pOsu,
pOsu'aK, 'oku, 'uva/ vi ele ela deve gata posso posso dar oco uva; including their
nasalized taxophones (i, e, A, , u): ('si, pen'denti, 'kAntu, 'knta, '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 troublesome
for foreiners, including the other thorny vocalic problem metaphony otherwise,
we would really have a foreign pronunciation. Written e o, in unstressed sylla-
bles, always correspond to /e, o/ (even for esC-); while, only for nal -e(s) -o(s), we
have /e, u/: ('lentis, 'mu;Rus) /'entes, '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, i,
ui Au): ('bei, 'mAi, 'pi, 'muintu, 'nAu) /'bein, 'main, 'poin, 'muintu, 'naun/
bem me pe muito no. Certainly, () 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 genuine, might easily renounce the nasalization caused by // 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 // sequences, in most languages without distinctive/phone-
mic nasalization. Normally, this slight degree of nasalization is not indicated.
7.4.8. Also the approximants, more naturally, are such, (j, w) /j, w/ (not semi-
-approximants, (, j)): ('glO;Rja, 'ligwa) /'gORja, 'ingwa/ glria lngua As to /R/,
of course, we have (R) (even in word-nal position before a pause or C, where in
neutral Bazilian we nd () /K/). ile, for /K/, a more natural (r) is better (af-
ter C\ /n, l, s/, diaphonemically /n, , s/): ('nra, &ura'pas, 'biuru, ra'laR) /'onKa, un-
Ka'pas, 'biKu, Ka'aK/ 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 dier-
ent normalized versions. We start with the (Brazilian and Lusitanian) neutral
pronunciations of (neutral British) English this is the rst 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 reect, according to the occurring segments.
In the text, for the Lusitanian accent, we indicate (|) /a/, before a pause (even
though in the isolated examples given we simply put (), in order to better show
the dierence 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 uent
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 rst.
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 0.12).
Portuguese text
7.5.3. O vento norte e o sol poravam sobre al dos do era o ma forte, an-
do sucedeu psar um viajante envolto numa capa. Ao v-lo, pem-se de acordo em co-
mo aele e primeiro conse%se obrigar o viajante a tirar a capa seria considera-
do o ma forte.
O vento norte comeou a soprar com muita fria, m anto ma soprava, ma
o viajante se embruava na sua capa, at e o vento norte dtiu. O sol briou
ento com todo o plendor, e imediatamente o viajante tirou a capa. E sim o ven-
to norte teve de recoecer a superioridade do sol.
Voc gostou da htoria? Queremos repet-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 rst, 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, $, , p, c, t, y, x.
Vowels
8.1.2. In stressed syllables, the six vowel phonemes are realized as the black
markers shown in 8.1. Obviously, those with a white center also correspond to
certain unstressed contexts that we will see later on: (t'i;) /t'ri/ tp$, ('F;Rx) /'era/
<pa, ('kFak) /'kak/ kak, ('tjt) /'tot/ tot, ('jum) /'um/ ym, ('bFt) /'bt/ vt; be-
sides, we also have: ('qF) /'qex/ zex, ('SFak) /'Sak/ wag, ('qjuk) /'quk/ yk
As some examples have already shown, there are some interesting dierences in
our phonetic transcriptions in comparison with phonemic ones, apart from sym-
bol variations such as () /e/, () /o/. ile (i, I) /i/ remain unchanged (except in
the sequences (i, I) /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 dierent C from /j, c, SS, ZZ/).
erefore, the V which are dierent from /i/ are realized as sequences of (F) + (, a,
), or (j) + (, 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 evident than fully prevelar, or velar rounded, approx-
imants, (j, w); (w) occurs in English ('wT, 'khwT, 'w:{<}) /'wIt, 'kwIt, 'wO:/ wit,
quit war. It is to be noted that (F, j) are quite dierent from a phonic zero, (`)
(although nave Russian speakers would swear they cannot hear anything but (`)),
as in English ('T, 'khT, ':{<}) /'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 nd normal C, which are palatalizable\
/m, n r l/ (m, m n, ~ R, , ), /p, b t, d k, g/ (p, p b, b t, d, k, g, ),
/f, v s, z x/ (f, f v, v s, z, = , ). en there are the palatal ones, in the gener-
al sense that they have a real palatal component, in spite of their actual realiza-
tions: /j, c, S[S], ZZ/ (j, C[], [], [] ). nally, we have non-palatal C\ /q S,
q/ (q S, q), which are never palatalized (although spelling sequences like >z$, we,
exist: (S'ju;) /S'ju/ w). We must always keep in mind, then, that */qi Si, qi/
do not occur (any longer) but are substituted by /q S, q/, in spite of historical
writings with >z, w, followed by >$, e, r, ,
8.1.3. After all, /i, j/ (and /c S, Z/) would be sucient 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 and are real-
ized closer fronter, as indicated by the grey markers, (e, , P, ) (again in 8.1):
('je1) /'jelj/ el, ('p) /'pjatj/ rt, ('P;x) /'tjotja/ tetr, ('fak) /tju'fjak/ t-
frk
e advancement and raising of /[0]iJ/ is decidedly more limited, (i) (and, in
the vocogram, there is no special marker dierent from the one for normal (i), so
as not to render the gure too heavy): ('bi) /'bitj/ $t
On the contrary, // does not occur in this context; and this phenomenon has
cultivated the myth of the complementarity of (i, ). So they are grouped in the
same phoneme, /i/, under the delusion that a better phonological system has
been found while it is exactly the contrary. Deviant schooling is hard to re-
8. russian 289
move, especially in regard to ocial writing (with its known and less known prob-
lems) and respelling, which is even used in schools in addition to in scientic
works, with phonemic ambitions. ey would like to nd almost alchemist's
solutions, by using complex formulae and esoteric iconograms (which we will be
careful not to reproduce).
In 8.1, three potential phonetic diphthongs, (, a, ), are also given,
which can substitute (;, a;, ;) (with an interpalatal variant, (), for (;)). ey
occur mainly in intonemes, in unchecked (nal or internal) syllables: (';, '
-;, -) /f'sje -jo/ bce, ('F;x, 'Fx) /'exa/ <xo, ('dFa;, 'dFa) /'da/ da, ('Fa;tx, 'Fa-
tx) /'atam/ atom, (';x, 'x) /'djadja/ drdr, (d'nj;, d'nj) /d'no/ dno,
('gj;xs, 'gjxs) /'golas/ goloc ey also occur for emphasis, in which case they
are longer; thus we have in unchecked syllables (; a;, ; ;): ("; -;)
/f'sje -jo/ bce, ("F;x) /'exa/ <xo, ("dFa;) /'da/ da, ("Fa;tx) /'atam/ atom, (";-
x) /'djadja/ drdr, (d"nj;) /d'no/ dno, ("gj;xs) /'golas/ goloc; and in checked
syllables ( a, ): ("~t) /"njet/ net!, ("qFa1) /"qalj/ al!, ("p) /"pjatj/
rt!, ("dj) /"dom/ dom!
8.1.5. Russian also has some ocial diphthongs, all with a front second ele-
ment /i/ (i). Here we will consider those occurring in stressed syllables: ('ii)
/'kii/ k$=, ('peiI) /'pjeiti/ e=te, ('mFai) /'mai/ ma=, ('Ci) /'cai/ ua=, ('bji)
/'boi/ o=, (z'Pi) /za'rjoi/ ape=, ('djuiI) /'duiti/ dy=te, ('i) /'djuim/ d=m,
('vFidu) /'vidu/ bv=dy. As can be seen, the orthography uses >= for diphthongs
consecrated by grammar. However, mainly in loanwords, others occur, as >ay:
('fFaust) /'faust/ Fayct
In unstressed syllables, we have (again with (i)): ('ju;ii) /'uljii/ yle=, (Cii'kj-
sii) /cii'kofskii/ Ua=kobck$=, (ti'gFa;) /tai'ga/ ta=ga, (';xiI) /'djelaiti/ de-
la=te, (ui'dju;) /ui'du/ y=dy, (i'mjkx) /djui'mofka/ d=mobka, ('b;i)
/'bjeli/ elv=
Instead, purely graphemic diphthongs like >$$ $r e$ ee ae ar a oe ve
do not correspond at all to phono-diphthongs, except when the semi-approximant
(which is shown in round brackets, ([])) falls completely. In this case, howev-
er, if the last vocalic element is /i/, the phone (I) is maintained (instead of having
(i), as in phonemic diphthongs, unless it is followed in sentences by (J) or /i/):
(vI'tFa;i[]I, v-) /vi'taliji, v-/ b *tal$$, (vR'i[;]I) /vras'siji/ b Pocc$$, ('i;i[]x) /'lil-
ija/ l$l$r, (mu'=e[;]I) /mu'zjeji/ mye$/-ee ( (mu'=ei) /mu'zjei/ mye=], ('e~~i[]I)
/a'sjenniji/ ocennee, (6'ma[;]I) /v'maji/ b mae, (&~IRu'SF;mxx) /niru'Smaja/ nepyw$-
mar, (z'nFa;u) /z'naju/ na, ('nj;vx[]I) /'novaji/ noboe ('juqn[]I) /'juqnji/ -
nve
e following examples are bisyllabic: (p'juk) /pa'uk/ ayk, (m'i;) /ma'ji/
mo$, (p'Ft) /pa'et/ o<t, (du'F1) /du'elj/ dy<l
out a pause), (', , x) /a/, (u) /u/, (', ) //: (CI'sF;) /ci's/ uacv, (pi'i;) /pi'ti/ r-
t$, (&Iz'kFa;) /jiz'ka/ rvka, ('bF;Cii) /a'bcii/ ovua=, (&jIR'FaS) /jira'laS/ epa-
law, (&piimi'i;) /pirimi'li/ epemel$, (&IIk'ti;CIstvx) /ilik'tricistva/ <lektp$uect-
bo, ('pj;I) /'poli/ ole (&gx'vFa;) /gala'va/ goloba, (&v'gFat, -n'g-) /avan'gart/
abangapd, (k'RFasnxx) /k'rasnaja/ kpacnar (u'Rjk) /u'rok/ ypok, ('dj;mu) /'do-
mu/ domy, (u'dFa;) /sju'da/ cda, ('u;kx) /sju'sjukatj/ cckat (q&viI'zFa;-
qx) /qvili'zatsja/ z$b$l$az$r, (q'nFa;) /q'na/ ena, ('dF;R) /'dr/ dvpv
For >r in unstressed syllables (and for >a too), it is important to explain that
neutral pronunciation has only /i/, whereas a transversal use of (, ) (or even (x,
) /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 inected
forms with >r+C consonantal endings (inuenced by the forms ending in -r, with
/ja/, (Jx, x)).
us we have: (pIt'nj;, pt-, pt-) /pit'no/ rtno, (jI'zFk, j-, jx-) /ji'zk/ r-
vk, (&jIz'i;, &j-, &jx-) /jiz'ki/ rvk$, (&pxgi', --, --) /pagli'djel/ oglrdel,
('nFa;CIt, -Ct, -Cxt) /'nacit/ nauat, ('bF;Cii, -Ci, -Cxi) /a'bcii/ ovua=,
('kjRmIt, -mt, -mxt) /'kormit/ kopmrt, ('e;I, -, -x) /'djetim/ de-
trm, ('zFa;~I, -, -x) /'zanil/ anrl, (&bIspi'i 'p, &bsp'i 'p, -p'i) /bis-
pi'ti 'pjatj/ e rt$ rt. (Please note the variants of -$k, as well: (s'pju+~Ik,
-~k, -~xk) /s'putnik/ cytn$k.)
Normal traditional pronunciation, for (pretonic) >e', r', a' ( occurring in
syllables immediately before the stressed or tonic one in a intoneme) has /e/ (',
'+J), which correspond to modern pronunciation /i/ (I, iJ): (&bIsp'i 'p)t, as
also happens in: (bi';zx)m (b-)t /bi'rjoza/ epea.
In vowel sequences which contain /a/, we generally nd an extension of the use
of (): (&vb-R'qFa) /vaabra'qatj/ boopaat, (nu'gFat) /nau'gat/ naygad, (&pd-
n'mju;) /paadna'mu/ o odnomy, (&nstR'vFa) /naastra'vax/ na octpobax, (u&d-
n'vj;) /uadna'vo/ y odnogo, (su'Cas[t]vxvx) /sau'cas[t]vaval/ coyuactbobal,
(st'vs[t]vxvx) /saat'vjes[t]vavatj/ cootbetctbobat.
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:
('bjp) /'bop/ o, ('mjsk) /'mosk/ mog, (t'RjupI) /t'rupki/ tpyk$, (s'dju;)
/fsa'du/ b cady. In addition, no voiceless C occurs before a diphonic voiced C (/v/
is a bit particular, 8.3.1.7): (z'bF) /z'btj/ cvt, (zg'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) /'qr/ $p, ('tFa;) /r'ta/ pta, ('kFatx) /'karta/ kap-
ta, ('bjb, -;b) /'bobr/ op, (_'~p5, -;p) /d'njepr/ Dnep.
e table in 8.2 shows the consonantal articulations of Russian, which are
necessary for an appropriate pronunciation of this language.
Instead, 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 dierent 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
m (m) (M) () (n) n (~) (2)* (~) (N) ()
F p b (p b) t d (+ _) ( ) k g
q (Q) C () ( )
( 6) f v (f v)
_ s z {} S q ( =) (,|J) (y)
() j|() (F) () (j) {H}
R ()
() (L)
/c S, Z x r l/ (C , R ) /m, n r l/+(), 5 , ) + ((, , , , 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 specic symbols like ((, n, )) are not necessary, but as we will shortly
see (, , n , ) 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, ('Fapx) /'lampa/ lama, ('njR) /'norm/
nopm, ('it), 'i;tM) /'ritm/ p$tm, ('i), 'i;M) /'rifm/ p$fm, (m'nji) /m'noi/
mno=, ('Fa;) /m'xa/ mxa, (mg'Fa;) /mg'la/ mgla, ('mik) /'mik/ m$g, ('m)
/'mjatj/ mrt, ('e)) /'sjemj/ cem, (Im'bi) /im'birj/ $m$p, (M'vjn) /am'von/
ambon, ('FafxRx) /'amfara/ amfopa. /m/ can occur befor heterorganic C: (m'gFa;)
/m'gla/ mgla, (m'nji) /m'noi/ mno=, ('Fa;) /m'xa/ mxa; however, for /mj/, and
/m/ + /i/, we regularly have (m): ('mi) /'mir/ m$p, ('e)) /'sjemj/ cem.
For /n/: (&jmbi't, &j-) /ombi'rjot, on-/ on epet (kM'vFa;, -'v-) /kan'va/
kanba, ('fFas, -) /an'fas/ anfac, ('njs) /'nos/ noc, (n'RFaf) /n'raf/ npab, (k-
'mFa~I) /ka'mandi/ komande, ('~i) /'nitj/ n$t, ('kj8) /'konj/ kon (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, , _ k, , g, ). e actual articu-
lation of palatalized /t, d/ is prepalatal which are stopstrictives, but stops before ho-
morganic (or similar) C. Sometimes, we nd (+) before pauses as well): ('pjt) /'pot/
ot, ('jp) /'lop/ lo, ('peC) /'pjec/ eu, ('tjp) /'topj/ to, ('gj;up) /'golupj/
goly, ('bFt) /'bt/ vt, ('bF;tx) /a'betam/ o <tom, ('bi) /'bitj/ $t (t'Rjut)
/t'rut/ tpyt, (q'qFa;) /at'qa/ otza, (t'kju;) /t'ku/ tky, ('vit) /'vit/ b$d, ('mFa;)
/tj'ma/ tma, ('fju;) /tj'fu/ tfy!, (pi+'a;, pIt-, 'pe+-x) /pit'lja, 'pjetlja/ etlr, (R-
'bj+~Ik) /ra'botnik/ paotn$k, (C'Ct) /at'cot/ otuet, ('djup) /'dup/ dy,
('tjdq) /'todq/ tot e, ('e8) /'djenj/ den, (_'~a;) /d'nja/ dnr, ('pj_I) /'podli/
odle, (d've, '-) /d'vjerj/ dbep, ('kFak) /'kak/ kak, ('pFakt) /'pakt/ akt, ('k)
/'ljok/ leg, ('kF;txmu) /'ketamu/ k <tomy, (Is'j) /kias'kjor/ k$ockep, ('ipii)
/'gipkii/ g$k$=, ('gjt) /'got/ god, (g'nju) /g'nutj/ gnyt, (n'i;) /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/,
(, ), which we have already seen in 8.2.2). Besides, before a stressed V dier-
ent from /i/ /c/ is (C): ('qFk, -%) /'qkl/ z$kl, ('q) /a'tjeq/ otez, (b'RFaqqx)
/b'raqqa/ patza, (pQ'dFaR) /plaq'darm/ lazdapm (u'Ci;i1) /u'citilj/ yu$-
tel, ('Cas) /'cas/ uac, (pi'C;) /pli'co/ leuo, ('juC) /'luc/ lyu, (C'n) /c'ljen/
ulen, (n'if) /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, v, v, 6) /f, v/, (s, z, =) /s, z/, (S, q) /S, q/; apart from the palatalized taxo-
phones (of the rst 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): ('fjn) /'fon/ fon, ('Rjf) /'rof/ pob, (&t'mFat) /afta'mat/ abto-
mat, (f'fimI) /f'filjmi/ b f$lme, ('vf) /'vjerfj/ bepf, ('vjk) /'volk/ bolk,
(S'kFa6q) /S'kavq/ wkaf e, ('sjn) /'son/ con, (R'sFa;) /ra'sa/ poca.
More examples: ('~k) /s'njek/ cneg, ('j) /'osj/ oc, ('zjup) /'zup/ y, (z'ji)
/z'loi/ lo=, (z'b;) /z'bjelm/ c elvm, (I'=u) /i'zjum/ $m, (&bi=i'ei) /biz-
di'tjei/ e dete= ('SF) /'Sestj/ wect, ('jS) /'loS/ lo, ('qFst) /'qest/ ect,
(q'dFa) /q'datj/ dat, ('qju;) /xa'qu/ xoy
Lastly, there is another quite particular voiceless constrictive, ([]) /S[S]/, which
is lengthened before //, and has a traditional variant constituted by the corre-
sponding constrictive followed by the (sometimes very weak) stopstrictive, (C).
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 dierent from /i/), a () is inserted: ('it)m ('Cit)t /S'Sit/
$t, (p'a;dx)m (pC'-)t /paS'Sada/ oada, ('bj)m (-C)t /'borS/ op.
e possible sequence (C) /Sc/ is dierent but quite normal indeed as in:
(' 'C) /S'Sem/ c uem ( 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, ('j), or in initial posi-
tion in a rhythm group before a (stressed or unstressed) V, (j). Instead, /j/ is real-
ized as a palatal semi-approximant after a (stressed or unstressed) V before anoth-
er unstressed V\ () (and also in ('i), for /'ji/ >V$\ (m'i;) /ma'ji/ mo$] after
C it changes the sequence into a palatalized C, /0j/ (J) ( 8.1.2-3): ('j) /'jatj/
rt, (ji'vi) /ji'vitj/ rb$t, (&mxI'kjsii) /maji'kofskii/ Markobck$=, (b1'SFa;-
x) /balj'Saja/ olwar
For emphasis, or precision, we can have (j) = (,) (semi-constrictive) and, respec-
tively, () = (j). However, it is more important to note that, in non-slow speech,
the /ij, ji/ sequences are realized as (i, i): ('FaRmix, -mix) /'armija/ apm$r,
(m'i;, m'i;) /ma'ji/ mo$ Sometimes, even in /j/ () sequences (without /i/),
it is not easy to adequately perceive (), especially in fast speech: (b1'SFa;x, -a[;]x)
(above all in preintonemes).
tionaries and teaching texts): ('qF) /'qex/ zex, ('Sjk) /'Solk/ welk, (&pxR'Sjut)
/para'Sut/ apawt, ('qFk) /'qrk/ z$pk, ('qF) /'qtj/ $t, ('jS) /'loS/ lo
However, they can be followed by /j/ (j) inserted before the V (and in writing
> appears too): (S'jt) /S'jot/ wet, (S'ju;) /S'ju/ w, ('jqju) /'loqju/ lo
Besides, the palatal approximant can occur after (J), too, then it is phonemical-
ly geminated /jj/, and may have a distinctive value: (s'j;) /srj'jo/ cvpe, (p'ju;)
/pj'ju/ , ('ja;vx) /dj'javal/ drbol, (v'jut) /vj'jut/ bt, (IL'jiC) /ilj'jic/ *l$u
Even /sjj, zjj/ can occur, but indicated with > (although not always, as in ce-
$tcr, given at the end of this section): ('j;x) /sj'jexatj/ cexat, (I='j) /izj-
'jatj/ $rt
e same occurs for /Jj/: (C'ja;) /c'ja/ ur ( ('Cat) /'cat/ uad, where the dier-
ence is clear also because of the dierent syllabic structure). e /ljj, njj/ sequences
are realized as (Lj, Nj) (or even (j, ~j), which are already suciently dierent):
(L'jt) /lj'jot/ s/he pours let ( ('t) /'ljot/ ice led or ight let] (L'ju;)
/lj'ju/ l, (vRN'j;) /vranj'jo/ bpane, (sviN'ja;) /svinj'ja/ cb$nr
ere are even cases where a (0) remains separated and uninuenced by a fol-
lowing /j/; thus, if no stress sign is already present, it is necessary and sucient to
introduce a hyphen in the phonemic transcription (as the orthography uses >):
(b'j;x) /ab'jexatj/ oexat, (s'j;qqx) /s'joqqa/ ce$tcr, (d&vu'ja;Rusni)
/dvux'jarusni/ dbyxrpycnv=, (tRansjI6R'peisii) /trans-jivra'pjeiskii/ tpance-
bpoe=ck$=, (&mq'ja;Rusni) /mjeq'jarusni/ merpycnv=
Trills
8.2.6. In the Russian phonemic system we nd only one trill phoneme, which
is realized either as an alveolar, (R), or a prepalatal, (), 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 ( 8.3.1.7): ('RFat) /'rat/ pad, ('pFak)
/'park/ apk, ('pt5, -;t) /'pjotr/ etp, ('is) /'ris/ p$c, ('at) /'rjat/ prd, (f'nFa) /fa-
'narj/ fonap, (i~'ab, -;bh) /sin'tjabrj/ centrp, ('vi[, -;) /'vixrj/ b$xp, ('jb-
Rxs) /'obras/ opa, ('Fad-Is) /'adris/ adpec
8. russian 295
Laterals
8.2.7. ere is only one lateral phoneme, /l/, with three taxophones, (, , L)
(independently of devoicing 8.3.1.7); () is a velarized alveolar, but it be-
comes (velarized) dental (which is possible to transcribe with (())), by assimilation
before /t, d q s, z/: ('F;q) /'lq/ lv$, (s'j;vx) /s'lova/ clobo, ('dju) /'dul/ dyl,
('ig, 'i;g) /'igl/ $gl, (s'mFs, -;s) /s'msl/ cmvcl, ('i;qx) /'liqa/ l$za, ('qFa1) /'qalj/
al, ('tj1kx) /'toljka/ tolko, ('pFamx) /'paljma/ alma, ('Rjub1, -;b) /'rublj/
pyl, ('mF, -;) /'mslj/ mvcl, (&pj'it-Rx) /pol'litra/ ol-l$tpa (here, lexical
compounding shows how it prevails against phonology, since /l+l/ remain separat-
ed, (), instead of fusing into ()), (L'jut) /lj'jut/ lt Some speakers can have a
velarized dental (or denti-alveolar) () 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 dierences between the
phonic structure and writing, which inevitably presents surprises, since pronun-
ciation changes, whereas orthography remains unchanging, except for ocial re-
forms (which however are always insucient 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 dierent 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 exi-
ble, for instance: /sis'tra/ (Is'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/, 8.0.3).
In the case of graphic geminates, for instance, we have: An-na /'anna/, but pho-
netically ('Fannx).
/'kjaxta/ Krxta, (t'jt) /t'kjot/ tket, (s'kj;I) /fs'kori/ bckope, (I'nu) /fti-
'nutj/ btrnyt, (&pt'fe) /part'fjelj/ optfel, (v'v) /v'vjol/ bbel, (v'jt) /vj'jot/
bet, ('ju;gu, 'kju-) /'kjugu/ k gy, (k'~i;gx) /k'niga/ kn$ga, (6z'gat) /vz'gljat/
bglrd 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'gat)m (-at)t /vz'gljat/ bglrd
More examples: (s'RFa;zu) /s'razu/ cpay, (s'pRFa;vx) /s'prava/ cpaba, ('zFatRx)
/'zaftra/ abtpa, (zR'Ck) /zra'cok/ pauok, (';ii) /f'sjakii/ bcrk$=, ('j) /sj'-
jel/ cel, (6'=) /v'zjatj/ brt, (='iqqx) /z'liqqa/ l$tcr, (=';x) /z'djelatj/
cdelat, (dRu='ja;) /druzj'ja/ dpyr, (S'ju;) /S'ju/ w, (&tS'gFa) /atSa'gatj/ otwa-
gat, (q'dFa) /q'datj/ dat, (q't) /q'djot/ det, (q'qFa) /q'qatj/ cat,
(C'Ctnx) /c'cetna/ tetno, ('PI) /x'ljoSSi/ xlectue, (IS'Sjkx) /iS'Solka/ $
welka, (i'i;) /sti'xi/ ct$x$, ('vjst) /x'vost/ xboct, (Rs'qvt) /ras'qvjet/ pacz-
bet, (C'n) /c'ljen/ ulen, (kp'C;ni) /kap'coni/ kouenv=, (CIz'mRnx) /criz-
'mjerna/ upemepno, (&CI'vj;) /atci'vo/ otuego, ('mFa;) /tj'ma/ tma
Some further examples: (m'gFa;) /m'gla/ mgla, (m'nji) /m'noi/ mno=, (m'Fat-
Si) /m'latSi/ mladw$=, (m'~e;~i[]I) /m'njeniji/ mnen$e, ('e;~i[]I) /m'Seniji/
men$e, (m'RFa;mx) /m'ramar/ mpamop, ('Caqqx) /m'caqqa/ muatcr, (I-
'iz, -;z) /ria'lizm/ peal$m, (If'RFag, -;g) /diaf'ragm/ d$afpagm, ('ve,
-;) /'vjedjm/ bedm, (d'RFa), -;M) /d'raxm/ dpaxm, ('me;Iq) /'mjesiq/ mecrz,
('pm) /f'prjamj/ bprm, ('kFa=8, -;=) /'kaznj/ kan, ('pe}, -;) /'pjesnj/ ec-
n, ('bFa;) /l'ba/ la, ('gFa) /l'gatj/ lgat, ('ig, 'i;g) /'igl/ $gl, ('RF, -;%) /'rxl/
pvxl, (b'dni) /b'ljedni/ lednv=, (S'a;px) /S'ljapa/ wlra, ('tFa;) /r'ta/ pta,
(R'qF;) /r'q/ p$, (R've;~i[]I) /r'vjeniji/ pben$e, ('~g, -;g) /'njegr/ negp, (pi't)
/fpi'rjot/ beped, ('ja;nx) /rj'janastj/ prnoct, (6'njut[, -;t) /v'nutrj/ bnytp
nally, let us also observe: (spk'nju) /fsplak'nutj/ bclaknyt
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
((00); only here we will use the special diacritics): (k'Rjp$kx) /ka'ropka/ kopo-
ka, (p$'i;qx) /p'tiqa/ t$za, (kup$'qju;) /kup'qu/ kyzy, ('jt$pusk) /'otpusk/
otyck, (t$'kFaC) /t'kac/ tkau, ('t$kx) /S'Sotka/ etka, (d$'gFat$kx) /ad'gatka/
otgadka, (s'vFa$bx) /s'vadjba/ cbada, (ig$'dFa;) /fsig'da/ bcegda, (g$';) /g'd-
je/ gde, (pd$'mt$kx) /pad'mjotka/ odmetka, (zt$'me;~i[]I) /zat'mjeniji/ atme-
n$e, ('ib$nu) /'gibnutj/ g$nyt, (k$'~i;gx) /k'niga/ kn$ga, (i$'mji) /sidj-
'moi/ cedmo=, (d$'ve, $'ve) /d'vjerj/ dbep (kg$'dFa;) /kag'da/ kogda, ('g$b)
/'tjogb/ tek v
More: (pIt'nj;) /pit'no/ rtno, ('bdni) /'bjedni/ ednv=, (&vxR+'~ik) /varat-
'nik/ bopotn$k, (pi+'a;, pIt-, 'pe+x) /pit'lja, 'pjetlja/ etlr, (fu+'a) /fut'ljar/ fyt-
lrp. In addition: (p'pFaC$kx) /ap'packatj/ oaukat, (t'tju;dx) /at'tuda/ ot-
tyda, (q'qFa;) /at'qa/ otza, (&I'nju) /atti'nul/ ottrnyl, ('u;dx) /at'sjuda/
otcda (also (t'u-), and even (t'su-) /at'su-/), (&CCI'vj;) /atci'vo/ otuego (pho-
nemic geminates and similar homorganic sequences are realized as phonetic
geminates with a limited duration of the second element: ((00))).
8. russian 297
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, ('vjtkx) /'votka/ bodka, (&bIskxI'ba;~ii)
/biskali'banii/ e kolean$= 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: (&~Ig'djt) /anig'dot/ anekdot, (&6gx~Is-
'tFan) /avganis'tan/ Afgan$ctan, (g'dj;mu) /g'domu/ k domy, ('Qb) /a'tjeq-
bl/ otez vl
However, >b /v/ alone is not sucient 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'qFn) /d'qn/ d$nn,
(k'Fas) /k'las/ klacc, (p'Fa) /p'laS/ la, (pRg'RFamni) /prag'ramni/ pog-
pammnv=, ('Rjusii) /'ruskii/ pycck$=, (g'Rjupkx) /g'rupka/ gpyka, ('mjni)
/'moSni/ monv=
thin lexemes, written geminate C can be realized as phonetically slightly
geminated C (00): (&ss'nFans) /assa'nans/ acconanc, ('vFannx) /'vanna/ banna,
('kFassx) /'kassa/ kacca, ('tjnnx) /'tonna/ tonna
However, more often CC are pronounced short: (&Is'tFat) /atis'tat/ attec-
tat, (b'jn) /ba'lon/ allon, (b'ein) /ba'sjein/ acce=n, (gR'mFa;Ikx) /gra-
'matika/ gpammat$ka, (d&iI'Rjkx) /drisi'rofka/ dpecc$pobka, (I';=i[]x) /i'ljuz-
ija/ $ll$r, (&kxIk'if) /kalik'tif/ kollekt$b, (&mii'mt5, -;t) /mili'mjetr/ m$l-
l$metp, (&ku'pFa;qx) /aku'paqja/ okkyaz$r, (&pxR'e1) /para'ljelj/ apallel,
(pI'Rjn) /pi'ron/ eppon, (&Iq') /riq'sjor/ pe$ccep, ('tF;~Is) /'tenis/ tenn$c,
(&iI'tj;i[]x) /tiri'torija/ tepp$top$r, (tR'eibus) /tra'ljeibus/ tpolle=yc, (I-
'nqx) /i'sjenqja/ <ccenz$r, (I'fkt) /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 nal ones after a
stress), especially near /r, l/: (px&mx', &pxm-, &pxm-) /pamala'djel/ omolo-
8. russian 299
8.3.1.11. Some personal pronouns have reduced forms: (i'ba;, i'a;, i&a, ,
x) /ti'bja/ ter, (i'b;, i';, i&, I) /ti'bje/ tee, ('vFas, &vas, vs, vxs) /'vas/ bac,
('vFa, &va, v, vx) /'vam/ bam, ('j i&b'dFa, 'j I'dFa) /'ja ti'bje 'dam/ r
tee 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): (&ii'i;, i'i;) /disi'ti/ decrt$, (-
'i;nxqqx, -nxqx, -nqx) /a'dinaqqatj/ od$nnadzat, ('i;nxqqxti, -nxqxti,
-nqxti, -nxqti) /a'dinaqqati/ od$nnadzatv=, (d'vFaqqx, d'vFa;qx, d'vFaq$)
/d'vaqqatj/ dbadzat, (d&vFaqq'i;, d&vFaq'i;, dvaq'i;, dvq'i;) /dvaqqa'ti/
dbadzat$, (&pii'at, &pii'-, pi[]i'-) /piddi'sjat/ rtdecrt, (pi'i;iiI, -'i;i-
I, -'i[;]iI) /pi'tidisiti/ rt$decrt$, (&S=i'at, &S=i'-, Si'-) /Szdi'sjat/ wectde-
crt, (S'i;iiI, -'i;iI, -'i[;]iI) /Ss'tidisiti/ wect$decrt$, ('emixt,
'e)xt) /'sjemdisjat/ cemdecrt, (i'mi;iiI, -'mi;iI, -'mi[;]iI) /si'midisiti/
cem$decrt$, ('vj;imixt, 'vj;i)xt, 'vj[;]i)xt) /'vosimjdisjat/ bocemde-
crt, (v'mi;iiI, -'mi;iI, -'mi[;]iI) /vasj'midisiti/ bocm$decrt$.
8.3.1.12. rst and patronymic names are usually reduced by frequent use, more
than by particular phonemic rules: ('j;IpxviC, -ipiC, -IpC, -pviC, -pC,
-pC) /'osipavic/ Oc$ob$u, (b'i;sxviC, -'iviC, -'i;iC, -'i;sC, -'iC, -'i) /ba'risav-
ic/ op$cob$u, ('pFavxviC, 'pFa;vviC, -aviC, -avC, -a;iC, -a;C, -a1C, -aC) /'pav-
lavic/ ablob$u, (&Ik'sFandRxviC, -andviC, -andviC, -andiC, -andRC, -a;~iC, -a;-
nC, &Ik'sFa8C, Ik'sFa8C, k'sFa8C) /alik'sandravic/ Alekcandpob$u, (v'i;mIRx6-
nx, -mIR6nx, -mIRxnx, -mIRnx) /vla'dimiravna/ Blad$m$pobna, ('f;dxRx6nx, -dxRx-
nx, -dR*nx, -dxRnx, -dnx) /'fjodaravna/ Fedopobna, (b'i;sx6nx, -'isnx) /ba'risav-
na/ op$cobna, (kR'~e[;]i6nx, -'~6nx) /kar'njejivna/ Kopneebna, (f'e[;]i6nx,
-'i6nx, -'6nx) /fa'djejivna/ Fadeebna, (p'fi-jI6nx, -'fi;I6nx, -'fi;Inx, -'finx)
/par'firijivna/ opf$p$ebna, (s've;LI6nx, -'ve;I6nx, -'ve;I6nx, -'ve;Inx, -'venx)
/sa'vjeljjivna/ Cabelebna
In addition: ('Fannx mI'Faix6nx, -*nx, -xnx, -nx) /'anna mi'xailavna/ Anna
M$xa=lobna, (&Ik'sFand RIk'sFandRxviC, &Ik'sFand RIk'sFandRC, Ik'sFand RIk-
'sFandRC, Ik'sFand RIk'sFa8C, Ik'sFan Ik'sFa8C, Ik'sFan 'sFa8C, 'sFan 'sFa8C) /alik-
'sandr alik'sandravic/ Alekcandp Alekcandpob$u, (&kxnst8'i; ns'Fa;kxviC, -'i;
nI-, &kxns'i; ~I'sFakC) /kanstan'tin sa'akavic/ Konctant$n *caakob$u, ('pFa;vI
'vFa;nxviC, -vi lI-, -vFa8C, 'pFa '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/ e$, (jIv'aqqx, Iv'aqqx, Iv'aq[q]) /jiv'ljaqqa/ rblrtcr, (&~i[]i'stvInnx, ~ii-
'svInx -'ev-) /nijis'tjestvinna/ neectectbenno.
In general, /ij/ = /i/ ( 8.2.5.1): (vi'i;x, -'i;x, -'ix) /vi'tija/ b$t$r, ('FaR-
mix, -mix, -mIx) /'armija/ apm$r, (&pi'iqqx, &pi-) /priju'tiqqa/ p$t$tcr,
(pi'j+~iI, pi'+~iI) /pri'jatniji/ p$rtnee, (&qi'j;, -i';) /qti'je/ $t$e, (&bi-
'ju;, &bi'u;) /'btiju/ vt$
Besides, usually /0jj/ = /0j/ and /ii0/ = /i0/: (&pi-jI'vji, &piI'vji) /pirj-
ji'voi/ epebo=, (b'CC-jI, -;CI) /b'cacjix/ vuau$x, ('kFa=N-ju, 'kFa=~-ju,
'kFa=~u) /'kaznjju/ kan, (u'bjv-ju) /lju'bovjju/ lob, (Svii'qFa, Svi'-, SvI'-)
8. russian 301
widespread (and also relatively less far away from spelling). We will now indicate
the dierences which mainly fall within traditional pronunciation, according to
the old Moscow pronunciation which was acquired orally. It used to include
some strange exceptions, which in some respect contributed to dierentiate 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 rename it and you'll get over it!.
8.3. Vowel elements of traditional (and alternative) Russian.
/'i, 0iJ, i'/ (i)t, /i/ (I)t /JuJ/ (%)t, /JuJ/ (T)t
/', '/ ()t, // ()t, /'/ ()t /'u, u'/ (u)t, /u/ (U)t
/...'/ (, )t:a /'..., / ()t:a {/u/ (, , )a}
/'JeJ/ (e)t /'JoJ/ (+)t
/'e/ () {/'e, "e/ ([])}t /'o/ {/o/} () {/'o, "o/ ([])}t
/'JaJ/ (, "[])t /a'/ (a)t, /a/ (x)t
/'a/ (a), /'a, "a/ (a[])t
>e /i/ ()t, >e, r, a /i'/ (', J'J)t >e /i[]/ ()a, /0iJ/ ()a, >< /i/ ()a
8.3.2.2. Modern pronunciation has () for //, whereas alternative (and often
traditional) pronunciation presents opener variants, (, ). Most frequently, we
nd ()a:t before a stress; ()a:t after it; and ()a ()t in absolute nal position: (&b-
ta'vji)t (&bt-)a (&bt-)m /bta'voi/ vtobo=, (&va'i)t (&v-)a (&v-)m /vxa-
'ditj/ bvxod$t, ('vF;m%)ta (-m%)m /'vml/ bvmvl, ('j;pt)t (-pt)a (-pt)m
/'opt/ ovtv (q'vjtnx)t (-nx, -)a (q'vjtnx[]I)m /q'votnaji/ $botnoe.
As we have already said, the timbres of // (, ) 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
dierence, so that they are rarely confused, even in spelling, except for evident cas-
es of poor education.
For >e, in traditional pronunciation we often nd ()t even for // (in addi-
tion to /i/, most surprisingly, as we have seen at the end of 8.3.2.1): ('juCS)m
(-)t /'lucS/ lyuwe ('ju;q)m (-)t /'xuq/ xye Here we may actually speak of pos-
sible neutralization between /i/ and //, which was more typical in former times.
For /u/, alternative pronunciation may have (0, 0, 0 J, JI, JJ, JiJ) (in
8.3, () is indicated by a thin broken line and glossed in brackets): (&kv'kFaq-
qx, &k-, &k-)a (&kuv-)m /kuvr'kaqqa/ kybvpkatcr, (&sm'tj;x, &s-, &s-)a
(&su-)m /suma'toxa/ cymatoxa, ('zFa;mq, -m-, -m-)a (-uq)m /'zamuqm/ a-
myem, (kx&m~i'j;, -mi-)a (-m~i-)m /kamjuni'kje/ kommn$ke, (I'me;&iix,
-[]i&-)a (-&-)m /i'mjejuSSiisja/ $me$=cr, ('dFa;, I-)a (-)m /sju'da/ cda
In traditional pronunciation, we can even nd: (mid've;mI)t (mId've;imI)m
/mid'vjedimi/ medbedrm$ (mi've;)t (mId've;i)m /mid'vjedim/ medbe-
dem, ('vF;~s)t (-Is)m /'vnis/ bvnec and, on the other hand, also: (i'vj_~I)t (I-
'vjd~x)m /si'vodnja/ cegodnr.
8.3.2.5. Traditional pronunciation has (C)t (or, as we have already said, (C)t or
(C)t) for (, , )m /S[S]/, but in Moscow the latter more agile pronunciation
has always prevailed: (Ci'kFa;)t (I-)m /SSi'ka/ eka, (ta'vFa;iC)t (t'vFa;i[])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: ('jI)
/S'Sastjji/ cuacte, (&pii'piIk) /piri'piSSik/ epe$cu$k (the -u$k sux is not inter-
preted as a separation element).
However, in modern pronunciation also (C) /Sc/ is normal, when a clear mor-
phemic boundary which is shown in writing too is present, with >cu u wu
u ctu du, never >: ('C) /S'cem/ c uem, (&ICI'vj;) /iSci'vo/ $ uego /SS/
or /Sc/ occurs in: (vIs'njuIti, -C-) /vis'nuSSiti, -Sc-/ becnywuatv=, (b'RjIti,
-C-) /ba'roSSiti, -Sc-/ opoduatv=.
But, between a lexeme and a sux, we generally nd () /SS/: (Iz'vjIk) /iz'voS-
Sik/ $bou$k, (&pii'beIk) /piri'beSSIk/ epeeu$k, ('qjI) /'qoSSi/ ectue Be-
tween a prex and a lexeme, (C) /Sc/ is most frequent: (bi&CI'veCni) /biScila-
'vjecni/ ecuelobeunv=, (&ICI'i) /iScir'titj/ $cuept$t, (&RxCI'aqqx) /raSci-
'xaqqa/ pacu$xatcr. In most commonly used words and when a prex is no lon-
ger thought of as something separated, we usually nd () /SS/: (Rx'skx) /raS'Sos-
ka/ pacuecka, ('jI) /S'Sastjji/ cuacte, ('t) /S'Sot/ cuet, (I's) /iS'Ses/ $cue
nally, the phoneme /ZZ/ () (which occurs in few words and only within lex-
emes) is more typical of traditional than modern pronunciation: ('t)t (q'qjt)m
/q'qot/ et, ('jeU)t ('jqqu)m /'jeqqu/ ey, (&U[]i')t (&u[]Iq'qFa)m /ujiq'qatj/
yeat, ('pjI)t ('pjqq)m /'poqq/ oe for derivative forms of ('djC)t
('dj, 'djS)m /'doS, 'doStj/ dod we have: ('djIk)t ('doZZik/ ('djqIk)m
8. russian 305
8.3.2.8. Please, note the dierences between modern and traditional pronuncia-
tion, as far as /rJ/ is concerned: ('RmIn)m ('emIn)t /'tjermin/ tepm$n, (I-
'pit)m (i-)t /tir'pit/ tep$t, ('Rbi[]x)m ('eb-)t /'sjerbija/ Cep$r, ('vf)m ('vef)t
/'vjerfj/ bepf, ('CRvI)m (-vI)t /'cervi/ uepb$, ('R~i[]I)m ('e~-)t /'tjerniji/ tep-
n$e, ('qF)m (-)t /'qertj/ epd, (u'Ri[]I)m (-ei[], -[])t /u'sjerdiji/ ycep-
d$e, ('Ri)m (-e-)t /s'tjerlitj/ cteplrd, ('vi[]x)m ('ve-)t /'vjersija/ bepc$r,
('vR=Ix)m ('ve=Isx)t /s'vjerzilsa/ cbep$lcr, (k'mCIsii)m (ka'meCiii)t /ka-
'mjerciskii/ kommepueck$=
Even in traditional pronunciation, as in the modern one, we nd (RJ) after ('0)
(with non-front V]\ ('pFai[]x) /'partija/ apt$r, ('jRI) /'ordir/ opdep, ('djuR-
~I) (-)t /'durni/ dypne, ('mFaRx) /'marlja/ maplr, ('kjuI) (-)t /'kursi/ kypce e
same occurs after unstressed V (even if they are front ones): (vI'e) /vir'tjetj/ bep-
tet, (IR'vis) /sir'vis/ cepb$, (vIR'=i;x) /vir'zila/ bep$la, (=IR'~Isti) /zir'nisti/ ep-
n$ctv=
C + () are not palatalized: (v'e;mx) /v'rjemja/ bpemr (f'skx) /f'rjeska/ fpecka
(p'tt) /par'trjet/ optpet (n'di) /xan'dritj/ xandp$t (pIz'it) /priz'rit/
pep$t
For reexive forms (in -cr -c] traditional pronunciation has /s/ (not /sj/), ex-
cept for nal-stressed gerunds: (ba'j)t (b'j)m /ba'jasj/ orc (against (&sxbi'RFa;-
xs)t (-bI-)m /sabi'rajas/ co$parc] On the contrary, in modern pronunciation re-
exive verbs have /sj/ (), except in the third person (in -tcr) and innitives (in
-tcr): (~i'qqx) /ni'sjotsa, -qqa/ necetcr, (k'djuqqx) /kla'dutsa, -qqa/ kla-
dytcr, (u'Ciqqx) /u'citsa, -qqa/ yu$tcr
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 >$) 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 >C$ as /0/: (v'tFa;i[]U)t /v-
'taliju/ b *tal$, ('kF;I)t /'kri/ k *pe, by distinguishing them from (vi'tFa;i[]U)t
/vi'taliju/ B$tal$, ('i;I)t /'kiri/ K$pe However, native speakers spontaneously
make things natural, by eliminating artices, thus today a palatalized 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, 'kri/ (where indicates a kind of in-
tentional pronunciation used because it should be used).
Some other particular cases: ('sjnq) /'sonq/ colnze, ('Custvx) /'custva/ uybc-
tbo, ('tF;iCx, 'tFx) /'tsica/ tvcrua, (i'Cas, 'as, as) /si'cas/ ce=uac (in the
sense of now, at once), (p'qFa;ustx, -a;stx, -astx, b'q-) /pa'qalusta, -l[]sta/ oa-
ly=cta, (z'dRFastvuiI, -astui-, -asI, -a, z&dRa, za) /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 ctd
ctl ctn ctc ctck (the last one is /sk/): wectdect nebctka cuactl-
bv= wectndzat $bctnv= wectct /Ss'sot/, typctck$=; >d = (`) in
>dk dn\ odka pdn$k dno
Besides, the sequences >tc dc correspond to /q/, in the pronominal endings
-t()cr: (smi'jqqx) /smi'joqqa/ cmeetcr, (smi'jaqqx) /smi'jaqqa/ cmertcr, and
when >t d are followed by the endings -ck$= -ctbo -ctb$e\ ('qii) /'djeqkii/
detck$=, (t'sjuqtvi[]I) /at'suqtviji/ otcytctb$e, (&pR[]Iz'vjqtvx) /prajiz'voq-
tva/ po$bodctbo, ('qtvi[]I, s'-) /s'ljeqtviji/ cledctb$e
Lastly, in the genitive endings -ogo, -ego, >g corresponds to /v/: ('b;xvx) /'bjela-
va/ elogo, (jI'vj;) /ji'vo/ ego us the oddity of (I'vj_~x) /si'vodnja/ cegodnr
today (= of this day) is only seeming.
8.3.2.11. Here we will present the dierences between modern and traditional
pronunciation, in reference to poststress-syllable V with // () for /i/ (>e general-
ly described as //, 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: ('pj;I)m (-)t /'poli/
ole, ('mj;I)m (-)t /'mori/ mope, ('ve;CI)m (-C)t /'vjeci/ beue, (q'iI)m (-C)t
/q'liSSi/ $l$e, (z'dFa;~i[]I)m (-[])t /z'daniji/ dan$e.
Genitive, dative, and locative singular of nouns and adjectives, -e-: ('i;~Ivx)m
(-~-)t /'siniva/ c$nego, ('i;~Imu)m (-~-)t /'sinimu/ c$nemy, ('i;~I)m (-~)t /f'si-
nim/ b c$nem.
Nominative and accusative, singular and plural, of neuter adjectives and pro-
nouns, -e: (z'lj[;]I)m (-[])t /z'loji/ loe, (z'F[;]I)m (-[])t /z'lji/ lve, (t'kj[;]I)m
(ta'kj[;])t /ta'koji/ takoe, (t'i[;]I)m (ta'i[;])t /ta'kiji/ tak$e, (i'ba;qjI)m (-j)t
/li'bjaqji/ lere, ('vjCjI)m (-j)t /'volcji/ bolue.
Collective numerals, -e\ (d'vj[;]I)m (-[])t /d'voji/ dboe, (t'Rj[;]I)m (-[])t /t'ro-
ji/ tpoe
Comparatives, -ee: (smi'e[;]I)m (mi'e;)t /smi'ljeji/ cmelee, (vIR'~e[;]I)m (viR'~e[;-
])t /vir'njeji/ bepnee.
Instrumental singular non-feminine and genitive plural, -em, -eb: (u'Ci;iI)m
(-)t /u'citilim/ yu$telem, ('pFa;CI)m (-)t /'placim/ lauem, (q'iI)m (q-
'i C)t /q'liSSim/ $l$em, (b'RFa;jIf)m (-f)t /b'ratjif/ pateb, ('SFL-jIf)m (-f)t
/'Sljjif/ w$leb.
Instrumental of plural nouns, -am$: ('kFap-imI)m (-mI)t /'kaplimi/ kalrm$,
('tju;CimI)m (-mI)t /'tucimi/ tyuam$, ('RjimI)m (-mI)t /'roSSimi/ poam$.
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 rst pre-
tonic 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: (&gx'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 rst 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); () because of its semi-attenuated timbre and a signicantly
dierent pitch from that of the tonic (syllable) which follows it, although not
marked by a particular symbol ( 8.4). Lastly, (vFa;) 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 specic handbooks, we come across statements which
claim that the pretonic syllable, (), would have the second degree of stress,
whereas all the others are weaker, above all those after the stress. However, in (s'tFa;-
Rxx) /s'taraja/ ctapar for the last two syllables, we have a timbre attenuation (as
for (&gx)) and a stress weakening (as for ()), 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 rmly 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 rmly maintained, with no neutralizations,
which are typical of unstressed syllables): (&sxmtxstR'je;~i[]I) /samaljotastra'je-
niji/ camoletoctpoen$e, (ktjfiik'pFakx) /kartofilika'palka/ kaptofeleko-
alka, (bjtpRxv_'~ik) /bortpravad'nik/ optpobodn$k, (spjtkRu'qjk)
/sportkru'qok/ coptkpyok, (FRxfjt'~i;mxk) /aerafotas'nimak/ a<pofo-
tocn$mok, (IktRxpFaRxpx&dxgI'vFa;i1) /iljektraparapadagri'vatil/ <lektpoa-
poodogpebatel, (gjsz'dFat, -sI-) /gos-z'dat/ Goc$dat, (&mjz'bFa~k, &mjs-,
-k) /moz'bank, mos-/ Mocank, (&ik'RFatkx[]I) /ik'ratkaji/ $ kpatkoe, (pjCI-
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: (&tst'Fa;) /atsta'la/ ot ctola, (&pxdk'nj) /padak-
'nom/ od oknom, (km'~;) /kam'nje/ ko mne, (b'RFa-jx i'stR) /b'ratjja i'sjostr/
patr $ cectpv, (~Iz'nFa;u) /niz'naju/ ne na, (sk'qF;kx) /ska'qka/ cka$-ka,
('j2q) /'onq/ on e, (pi'+-I) /pri'djotli/ p$det l$.
Also pronouns and monosyllabic adjectives can be destressed: (mjib'RFat)
/moib'rat/ mo= pat, (='jn, -xn) /z'djesj-on, -an/ dec on, (ku'dFa vbI'qF;I)
/ku'da vbi'qti/ kyda bv e$te, (&vii'j8, &ve-) /visji'junj/ bec $n also <to:
(k'tj Ft=';x, It-) /k'to etaz'djelal, i-/ kto <to cdelal, ('vj;njn, -xn) /'vo-
non, -an/ bon on, ('vj;tjn, -xn) /'voton, -an/ bot on, (vjt'tjut, v-) /vot'tut,
va-/ bot tyt, (vjn'tFa, v-) /von'tam, va-/ bon tam, (jnz'nFa) /onz'nal/ on
nal, (S&tjjn'tFa, S&tj-) /Stoon'tam, Stao-/ uto on tam, (&njjm'bFta,
-'b-, &nj-) /noom'bltam, -n'b-, nao-/ no on vl tam, (tk'vjt) /tak'vot/ tak
bot (conjunction, whereas the adverb tak does not reduce), ('jn d'ja;) /'on da-
'ja/ on da r (a conjunction again, whereas da, the adverb yes is generally stressed
('dFa;); nally: (I'mju xbS'tj;) /i'mu xadjbS'to/ emy xot 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, bec, da, kak, l$, mne, mv, ne, n$, o, on, ot, o, od, cem, to,
tv. e pronoun r I can be reduced immediately after a stressed V: (m'gu jx-
pxvI'dFa) /ma'gu japavi'datj/ mogy r ob$dat, (g'e jxm'gu pxu'Ci) /g'dje ja-
ma'gu palu'citj/ gde r mogy olyu$t, (kg'dFa jxnCI'nFa) /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: ('vj;Rxn) /'voran/ bopon
and ('vjR-jn) /'voron/ bop on, (&iI'sFa;) /tili'sa/ teleca and (&eI'sFa;) /tjeli'sa/
te leca, (jIs'~ei) /jis'njei/ rcne= and (js'~ei) /jas'njei/ r c ne=.
Intonation
8.3.4. 8.4 shows the preintonemes and intonemes of neutral Russian then
we just give some illustrative examples, to use for comparisons:
/./: (F;tx mjibi;mi &kxp=i;tx) /'eta moi-lju'bimi kampa'zitar./ >to
mo= l$mv= komo$top.
/?/: (vF; jI&vjvi;iI2 2) /'v jivo'vidili?/ Bv ego b$del$?, (&FtxIkkj;2 2, -k-)
/etalik'ko?, -x'ko?/ >to legko?
//: (jeI &j~~Itve;It2 2 &janxpI'Sju; jI&vjqqju;3 3) /'jesli on-niat'vjetit ja-na-
pi'Su jivoaq'qu/ Ecl$ on ne otbet$t, r na$wy ego otzy.
Elliptic questions ( those without a verb) have the pattern / /: (aIgzFa;mI-
n2 2) /aig'zamin/ A <kamenv?
8.4. Russian preintonemes and intonemes.
/ / (2 2 2 2 2 2 2) /./ (2 3 3)
/ / ( 2 2 2 2 2 2 2) /?/ (2 2 2)
/ / ( 2 2 2 2 2 2 2) // (2 2 2)
/ / ( 2 2 2 2 2 2 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, rst comes the English text
( 2.5.2.0) with a broad Russian accent, then the Russian translation, with neu-
tral pronunciation; then follows the traditional 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, uent 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 rst.
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 0.12).
Russian text
9.0.1. In this chapter, we will deal with the neutral modern pronunciation of
Arabic, corresponding to the supraregional language used in ocial 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-
cations, modications 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 nal, or followed by y w therefore, they are not always ay aw). Further-
more, we use (for /S , X, /, not digraphs: sh th dh kh gh {not even
for the last two}) and ` (instead of j for /h, Z/), while keeping =
(for /t, d, , /) and q ' as well (for /, , H/).
Vowels
9.1.1. Neutral Arabic has three short and three long vowels, with some remark-
able taxophones, due to the inuence 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 diphthongs 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 inuence of conso-
nants and syllable structure. e inuence 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 neutral pronunciation (not a regional one),
even if for V this usage is quite close to that of Levantine Arabic ( 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, U)); 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{:},
{:}, u{:}) (shown by the black markers in the vocogram of 9.1), we should note
well and constantly recognize the list of the Arabic realizations, which have the
following distributions (besides, /:/ in unstressed syllables = ()):
9.1.3. According to the distributions just seen, 9.1 shows the realizations of
the Arabic short and long vowels, (i[:], I[:], [:]) /i[:]/, ([:], a[:], A[:], [:]) {and ([:]),
Koranic variant, 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/ il ('IH -) /a'iH/ aqi ('Af:) /'aff/ a ('tA:h) /'ta:h/
=aaa ('baHda) /'baHda/ bada ('5a:HIn) /'ra:hin/ raahin ('H:a) /'ha:a:/ haaaa
('wld) /'walad/ walad ('b:b) /'ba:b/ baab ('U:) /'u:/ uuq ('Huna) /'huna:/
hunaa ('u:f) /'u:f/ uuf (U'u:m) /Xu'u:m/ uuum ('fUndU) /'fundu/ funduq
9.1 does not show the dierent realizations of /ai, au/, which result from the
combination (-, a-, A-) + (-i, -I, -) or + (-u, -U), according to context: ('bIt) /'bait/
bait ('ina) /'aina/ aina ('Haini) /'Haini:/ ainii ('AIl) /'ail/ qail ('fUz) /'fauz/
fauz ('lUn) /'laun/ laun ('HaUd) /'Hauda[h]/ auda-h (mU'Auwm, -Awwm)
/mu'awwam/ muqawwam (u'l:d) /au'la:d/ aulaad
9.2. Some Arabic vowel variants.
9.1.4. 9.2 shows some frequent realizations of /ai, au/: (e:, I o:, U), which
are very widespread outside neutral and Levantine Arabic. However, one's pronun-
ciation may still be considered neutral, although colloquial, even if it uses such
variants, provided all other articulations are appropriate; in fact, this pronuncia-
tion is quite common indeed: ('be:t, 'bIt) /'bait/ bait ('e:na, 'I-) /'aina/ aina
('He:ni, 'HI-) /'Haini:/ ainii ('e:l, 'Il) /'ail/ qail ('fo:z, 'fUz) /'fauz/ fauz ('lo:n,
'lUn) /'laun/ laun
9.3. Colloquial variants and neutralizations.
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
9.1), as happens for the variants of /ai, au/ as well. e white markers indicate un-
stressed realizations of /i, a, u/, (, , P), which are considerably centralized. e
9. arabic 319
9.1.5. Actually, not every single realization given in 9.2-3 is necessary for a
good neutral pronunciation of Arabic. Nevertheless, if they are rationed and used
in a natural way (speaking uently), they help to give greater spontaneity, simi-
lar to natives' speech again within a kind of colloquial neutral pronunciation. is
will depend on words themselves for instance, ([:]) is more likely to occur in
lofty words.
Furthermore, still within the neutral accent (but slightly more international),
there is another possibility taking a dierent 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: ([:]), 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 international pronunciation
are shown in 9.4. is clearer pronunciation generally uses: (i[:], [:] [:], A[:]
U[:]) (it may also have (:, :) for /ai, au/), but not necessarily (, , P) (and least of
all ()), seen in 9.1.4.
e systematic and complete shortening of unstressed long vowels belongs to
modern and international pronunciation (and to dialects as well, generally).
However, in Koranic pronunciation, /:/ (appearing in phonemic transcrip-
tions) are realized as half-long, (;). It is not dicult 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 ('H:a, -a;) /'ha:a:/ haaaa ('Huna, -na;) /'huna:/ hu-
naa
9.4. International vowels.
Geographic variants
9.1.6. Among the main variations and deviations from the neutral form (
9.1.3), in certain areas, we nd that /a:/ never has the (:) 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:,
U) ( 9.2), from Maghreb to the Persian Gulf, except in the Levant. However,
the diphthongs are kept, generally, when they are in absolute nal position or fol-
lowed by /j, w/: ('bIt, 'bIt, 'be:t) /'bait/ bait ('lUn, 'lUn, 'lo:n) /'laun/ laun ('HaIn,
'HIn, 'He:n) /'Hain/ ain ('sijId, 'sjjId) /'sajjid/ sayyid (mu'Suw, -ww-) /mu'Saw-
wi/ muawwiq ('ni, 'nj:, -J:) /'nai/ nayy ('lu, 'lw, 'l)) /'lau/ law
Especially in Maghreb, besides (unstressed) /i, u/, even /a/ may be dropped (with
possible consequent stress shifts): ('nta, 't, n't) /'anta/ anta ('Alm, 'Alm)
/'alam/ qalam (li's:n, 's:n) /li'sa:n/ lisaan (s'l:m, s'l:m) /sa'la:m/ salaam
Consonants
9.2.0. 9.5 gives the articulations of modern neutral Arabic. e main geo-
graphical variants of some phonemes are given in round brackets.
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.
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'ml)
/mu'maal/ mumaal (tm'Zi:d, -'Gi:d) /tam'Zi:d/ tam`iid ('Sms) /'Sams/ ams
9. arabic 321
Stops
9.2.2.1. Arabic has no /p/, but only /b/, which however is realized as (p) when
followed by voiceless C\ ('b:b) /'ba:b/ baab ('lbn) /'laban/ laban ('haps) /'habs/
abs On the other hand, neutral Arabic has no /g/, either; while it has two voice-
less stops, velar and uvular, /k, / (k, ) (in phonemic opposition): ('k[h]Uskusu,
kUs'k[h]usu) /'kuskusu/ kuskusu (mIk'a:5) /mik'a:r/ mikaar ('mlIk) /'malik/
malik (A'di:m) /a'di:m/ qadiim (A'd:m) /a'da:m/ aqdaam ('sU:) /'su:/ su-
uq (&AUR':n) /alur'a:n/ al-Qur'aan (sA':tA -':-) /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:
('klb) /'kalb/ kalb dog and ('Alb) /'alb/ qalb heart.
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 aspirated,
when at the beginning of a stressed syllable (but we will mark it only here). Often,
/t, d/ are denti-alveolar if nal before a pause (but it is not necessary to use ((4, 7))):
(t'd:wUl) /ta'da:wul/ tadaawul (Si't[h]:) /Si'ta:/ itaa' ('had:) /'hadd/ add (bA-
't:ts -'t:-) /ba'ta:tis/ ba=aa=is ('dA5t, 'd5t 'd5t) /'dart/ ar= ('dAZIR, -GIR) /'da-
Zir/ a`ir
Although neutral Arabic has no (g), this sound is frequent in dierent 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 Koranic and ancient (G)), or to
(g), is a part of this trend.
9.2.2.2. e last neutral Arabic stop is // (), 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: ('mi:n) /a'mi:n/ amiin ('s:-
Ih) /'sa:ih/ saa'i ('m:) /'ma:/ maa' ('znn) /'zanna/ zanna' (&Ru's:) /ru-
a'sa:/ ru'asaa' (l':n) /al'a:n/ al-aan ('5as) /'ras/ ra's ('bd) /'bad/ bad'
(s':l) /sa'a:l/ sa''aal
erefore, in Arabic, any word traditionally beginning with a V ( vowel), pho-
nically, begins with // (), because in this language all phono-syllables begin with
a C\ ('b) /'ab/ ab (i'd:n) /i'da:na[h]/ idaana-h ('Um:) /'umm/ 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 prexed; 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 nd (, I)); of course, in isolated
forms, // is prexed 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
//), therefore the two words are linked. If, instead, the preceding word ends in a
C then the V is added, but // is not. e reader is referred to grammars, where this
phenomenon (called wala) 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: ('Ibn, -b, -bn) /'ibn/ ibn ('ImRU)
/'imru/ imru' ('Ism, -s, -sm) /'ism/ ism (I'n:ni) /i'na:ni/ inaani\ (Ra'i-
tUb 'ni:) /ra'aitu b'ni:/ ra'aitu bnii ('b:bUl 'bIt) /'ba:bu l'bait[i]/ baabu al-baiti
Constrictives
9.2.3. Among the phonemes belonging to this articulation manner, we nd /f/
(f) (but not the corresponding voiced phoneme): (fa'5i:d) /fa'ri:d/ fariid (If's:d)
/if'sa:d/ ifsaad In addition, there are two diphonic pairs, which pose no problems,
/, s, z/ (, s, z): ('l:) /a'la:/ alaa ('mIR) /'mair/ mair ('m:)
/'amm/ amm ('da5s) /'dars/ dars ('za:5) /'za:r/ zaar ('knz) /'kanz/ kanz
However, there are two more diphonic pairs, with variations that may pose
some phonemic dilemmas; they are /, / ( /D) and /S, Z/ (S Z/G). We prefer (,
Z) for their voiced members, as they are more modern and more integrated in the
phonologic system than their more Koranic variants (D, G), which are consid-
ered more prestigious (even by those who do not use them), but have dierent
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'U:, -DU:D) /mah'u:/
mauu ('A:mI, 'D-) /'a:mi/ aami' ('URUR, UR'u:R) /'urur, ur'u:r/ ur-
u(u)r (Ra'I:) /ra'Xi:/ raii (IS'h:d) /iS'ha:d/ ihaad (S'j:) /aS'ja:/ ayaa'
(mu'SuwS, -wwS) /mu'SawwaS/ muawwa ('Sd:) /a'Sadd/ aadd (RaS'S:S)
/raS'Sa:S/ raaa (S'SaRu) /aS'Sarw/ aarw (Z'mi:l, G-) /Za'mi:l/ `amiil ('ZmaH,
'G-) /'aZmaH/ a`ma ('t:Z, -G) /'ta:Z/ taa`
In the pharyngeal place of articulation, we nd the voiceless constrictive /h/ (h)
(currently, the corresponding 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/ matuum (mU'hAd-
dR) /mu'haddir/ muair ('faRIh) /'farih/ fari (fah'ha:S) /fah'ha:S/ faaa
Arabic also has a diphonic pair of uvular constrictive trills, (, ) (as will be
seen), which phonemically may be represented with the ocial symbols /X, /.
Approximants
9.2.4.1. Let us rst consider the least peculiar ones (although there are rather
free occurrences), /j, w/ (j, w), even realized as (i, u), for /0j, 0w/ (and, possibly,
9. arabic 323
for /j0, w0/, in colloquial variants, as no doubt in the dialects) and (ij/jj,
uw/ww), for /jj, ww/): ('j:wIR) /'ja:wir/ yaawir (wA'I:j) /wa'i:ja[h]/ waii-
ya-h (wU'u:l) /wu'u:l/ wuuul ('sijId, 'sjjId) /'sajjid/ sayyid (nu'w:m, nw'w-)
/naw'wa:m/ nawwaam ('ni) /'najj/ nayy (mn'Hi:) /man'hi:/ manhiyy ('bu)
/'abw/ abw ( ('bu) /'abu/ abu).
e Koranic pronunciation prefers (-jj-, -ww-), as in the cases seen above. Let
us also consider these further examples, which show us the dierences between
modern pronunciation and traditional Koranic 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 ((:, :)).
is happens when they are in absolute nal position, and when in front of /j,
w/ or after /i:, u:/, respectively, or when nal, after C as the following examples
will show: ('ki, 'kj, -J) /'kai/ kay ( ('kij, 'kj:, -J:) /'kaij/ kayy), ('lijn, 'ljjn)
/'lajjan/ layyan (ju'mi:j, -J) /jau'mi:ja[h]/ yaumiiya-h ('Humi, 'Humj, -mJ) /'Humj/
umy ('lu, 'lw, 'l)) /'lau/ law ('duwaR, 'dww-) /'dawwar/ dawwar (Ha'du:w,
-u:)) /Ha'du:wa[h]/ aduuwa-h ('sHu, 'sHw, -H)) /'sahw/ sahw ('Hafu, 'Hafw, -f,
-f) /'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, (()), which is fairly easy to de-
tect by its lower intrinsic tonality and creaky voice): ('HaIn) /'Hain/ ain ('maHa)
/'maHa/ maa ('baHda) /'baHda/ bada ('naHnaH, naH'na:H) /'naHnaH, naH'na:H/ nana(a)
(&faHHa'li:j) /faHHa:'li:ja[h]/ faaaliiya-h ('slH) /'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, (), :
('na?na?, na?'na:?, 'nAnA, nA'nA:).
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: (&Itti'Z:h, -G-) /itti'Za:h/
itti`aah ('mUhtaR) /'muhtar/ muhtar ('h:iHi) /'ha:ihi:/ haaihii (mH'bu:l)
/mah'bu:l/ mahbuul (&InHi'z:m) /inhi'za:m/ inhizaam (hf'n:f) /haf'na:f/ haf-
naaf ('hija) /'hija/ hiya ('ka5h) /'karh/ karh ('bIh) /'bih/ bih ('AH-w) /'ahwa[h]/
qahwa-h (wh'h: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 (, h) /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, Koranic pronunciation, while the normal realization of -a is (a): ('hz-
z) /'hazza[h]/ hazza-h (movement) but: ('hzza) /'hazza/ hazza ((he) shook).
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 inuence 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 nal /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, / (, ): (ba'Si:S) /baX'Si:S/ baii ('aU) /'XauX/ au (fa'a:Ri) /faX-
'Xa:ri:/ faaarii ('adn) /'adan/ adan (A'i:R) /a'i:r/ aiir ('b:lI) /'ba:li/
baali (mS'u:l) /maS'u:l/ mauul (t'waUl) /ta'waul/ tawaul
Laterals
9.2.6. ere is one lateral phoneme, /l/ (l, , ) (the third taxophone occurs in
contact with /t, d, , , /; it occurs as a phonostyleme ( a stylistic phoneme),
too, in Allaah and derivatives (provided that it is not preceded by /i/): (&mut'l-
lI) /muta'lali/ mutala'li' (tl'bi:s) /tal'bi:s/ talbiis ('lIl) /'lail/ lail (m'j:n)
/mal'ja:n/ malyaan (zl'lA:A) /zal'la:a[h]/ zallaaqa-h (tA'I:h) /tal'i:h/ talqii
(&bIsmIl'l:h) /bismil'la:h/ bismillaah (A'A:h) /al'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, // is (D), as in traditional and Ko-
ranic pronunciations; elsewhere it is often realized as (), 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 nd
(g); elsewhere, even (), as in Luxor (southern Egypt), and (J).
For instance, /'Za:b/ `aab is ('Z:b) in the Levant (Lebanon, Palestine, and Syria)
and in Africa (except Egypt and Sudan, as we have just said); it is ('G:b) in Arabia,
the Persian Gulf, Jordan, Iraq and among the rural and the nomadic peoples in
Morocco.
e grapheme qaaf // (), very often becomes (), 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 'lb,
'Halb, 'klb, 'glb).
In the Levant, // is generally quite weak, therefore we could transcribe it as (,)
(instead of ()). In Maghreb, // 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-nal 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 nal V in polysyllables tend to drop (above all
grammemes, and /n/ as well, except for accusative /an/, which may become
/a:/, 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: ('h: ki't:b, 'h:k 't:b) /'ha:a: ki'ta:b[un]/ haa-
aa kitaabun (l'hIb-RU 'tAijIb, 'tAjjIb) /al'hibru 'tajjib[un]/ al-ibru =ayyibun (k-
'tptu&biHi, -tUb&Hi) /ka'tabtu 'bihi/ katabtu bihi ('k:n fIl'bIt) /'ka:na fi:albait[i]/
kaana i al-baiti (Ra'itUl 'bIt) /ra'aitu al'bait[a]/ ra'aitu al-baita (k'tpt) /ka-
'tabt[u]/ katabtu (&sk'ki:n) /saka:'ki:n/ sakaakiin
9.3.1.3. As far as pausal forms are concerned, let us take from Mitchell (1990)
some examples (which we will transcribe again appropriately), keeping in mind
that this is an orthological phenomenon people use to pronounce words in into-
nemes, including preceding words, semantically and syntactically linked (to the
exclusion of grammemes).
is category includes nal short V (with or without innitive -n) ('ktb)
/'katab[a]/ kataba ('jktUb) /'jaktub[u]/ yaktubu ('li~ 'jktUb) /'lian 'jaktub[a]/
li'an yaktuba (&fim'da:RIs) /fi:ma'da:ris[a]/ i madaarisa (fIl'bIt) /fi:l'bait[i]/ i al-
-baiti ('bIt) /'bait[un]/ baitun (fi'bIt) /fi:'bait[in]/ i baitin ('mUnu 'zmnI
w'Zi:z, -'Gi:z) /'munu 'zamanin wa'Zi:z[in]/ munu zamanin wa`iizin However,
(accusative) -an /an/ may be substituted by /a:/: (aH'tA:HU HIS'5i:n di'na:Ra) /aH'ta:-
hu HiS'ri:na di:'na:ra:/ a=aahu iriina dinaaran except for adverbs like: ('ha:ln)
/'ha:lan/ aalan ('dAn) /'aidan/ aian
Lastly, taa' marbuu=a [ -atin -atan -atun] is substituted by /a[h]/, which is
generally pronounced () (or (h) in a {more} Koranic style): (Z'mi:l)
/Za'mi:la[h]/ `amiila-h ((Z'mi:ltUn, &Zmi'ltUn, %G-) /Za'mi:latun/ `amiilatun]
('ftahat m'da:Ris li'li:j, &mda'5is) /'ftahat ma'da:risa lai'li:ja[h]/ fataat
madaarisa lailiiya-h ((li'li:jtn, &lili'jtn) /lai'li:jatan/ lailiiyatan] (&fIlZu'nin,
-Gu-) /fi:lZu'naina[h]/ i al-`unaina-h ((Zu'nin&ti, &Zuni'nti, %G-) /Zu'nainati/
`unainati]
9.3.1.5. In nal position, after C the sonants (/m, n, r, l/) may be realized in
dierent 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 (syllabic), or even with a short epenthet-
9. arabic 327
ic V (like (I, )): ('s, -s, -sm) /'ism/ qism ('ldn, -d, -dn) /'ladn/ ladn ('ftn,
-t, -tn) /'fatn/ fatn ('dUHn, -hn, -H, -Hn) /'duhn/ duhn ('bdR, -d{, -dR) /'badr/
badr ('mR, -, -{) /'mir/ Mir ('ft, -t, -tl) /'fatl/ fatl ('5At, -t, -tl) /'ratl/ ra=l
9.3.1.6. Here are some examples of typical Arabic phonotactics: (ta'bi:n) /ta-
'bi:n/ ta'biin (m'u:R) /ma'u:r/ ma'uur ('mtHUb) /'matHu:b/ matuub ('-
Al) /'aal/ aqal ('mdal) /'madXal/ madal (m'Hu:R) /ma'Hu:r/ mauur
('tzHu) /'tazhu/ tazhu ('msZId, -GId) /'masZid/ mas`id ('mSta) /'maSta/ mata
(mS'u:l) /maS'u:l/ mauul ('mA5aH) /'maraH/ mara ('AdZaHa, Ad'ZaHa) /'ad-
ZaHa/ a`aa ('AdHafa, Ad'Hafa) /'adHafa/ aafa ('Alma, A'lma, -D-) /'ala-
ma/ alama (maH'Zu:n, -Gu:n) /maH'Zu:n/ ma`uun (maH'5u:f) /maH'ru:f/ maruuf
(aH'm:l) /aH'ma:l/ amaal (ma'Su:S) /ma'Su:S/ mauu ('mAha) /'maha/ maq-
ha (ml':n) /mal'a:n/ mal'aan ('jisu, ji'su) /'jaiasu/ yai'asu ('uHma,
u'Hma) /'auhama/ 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 // stands for light syllables (/0/); /9/ for
heavy ones, consisting of heavy sequences like /00, :0, 0/, 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, // indicates a light or heavy sylla-
ble indierently with no direct inuence on stress assignment. e symbol //
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. /', '9, '9/,
3 syll. /', '9, '9, '9/,
4 syll. /', '9, '9, '9, 9'/,
5 syll. /', '9, '9, '9, 9', 9'/,
6 syll. /', '9, '9, '9, 9', 9', 9'/.
9.3.2.2. e solution to the formulae with variants, to read along columns, pro-
vides:
/'9, 9'/ /', '/, /', '/,
/', '/, /'9, 9'/, /'9, 9'/,
/'9, 9'/, /9', '9/, /9', '9/,
/9', '9/ /9', 9'/ /9', 9'/.
328 a handbook of pronunciation
In patterns with two possible stressings, the variants may be of two types: Egyp-
tian (Cairo), as: /'9, '9, '9, '9/; or literary ( dialectal, like
those of southern Egypt), with the following possibilities: /', 9'/, /',
9', 9'/, /', 9', 9'/.
For /9'/, we may nd the patterns /'9/, /9'/ as well (which
constitute loftier 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
nal stress on words ending in /:, / (which, in neutral pronunciation, are not
heavy enough to bear the stress): ('mInHuma, mIn'Huma) (and (&mInHu'ma:))
/'minhuma:/ minhumaa
9.3.2.3. Prexes do not inuence the application of the stress rule, in dierent
actual words; nor does the article even when it assimilates completely. But the rule
is modied 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 (s'f~Z, -~G) /sa'fanZ/ sa-
fan` (mu'5a:sIl) /mu'ra:sil/ muraasil (ta'5assUl) /ta'rassul/ tarassul ('HA:m, HA'-)
/'Ha:ima[h]/ aaima-h (&mut'fuw, -ww) /muta'fawwi/ mutafawwiq (mu-
'fa:RAA, &mufa'5AA) /mu'fa:raa[h]/ mufaaraqa-h ('md-Ra&s, md'5as) /'mad-
rasa[h]/ madrasa-h ('daRa&ki) /'daraki:/ darakii (&kt'bHu, 'kt&bHu) /kata-
'bahu/ katabahu (mU'tli&f, 'mUt&lif) /muX'talifa[h]/ mutalifa-h (&Z'lu-
HUm, 'Z&luHUm, 'ZlHUm, -G-) /aZa'luhum, a'Zalhum/ a`aluhum
Intonation
9.3.3. 9.6 shows the preintonemes and intonemes of modern neutral Arabic.
e continuative intoneme is not very dierent 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 ( 9.1.1 which
dier 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:
/./: ('l: 'aHRIf m&din'tkUm23) {(m'di:n&tkUm23)} /'la: 'aHrif[u] madi:na'ta-
kum./ Laa arifu madiinatakum.
/?/: ('hl lki't:b Z'di:d21) {(G-)} /'hal alki'ta:b[u] Za'di:d[un]?/ Hal al-kitaabu
`adiidun?
//: ('jUmkIn &nnA'u:m bi'Zul2 2) {(bi'Gu-)} /'jumkin[u] anna'u:m[a] bi'Zau-
la[h]/ Yumkinu an naquuma bi-`aulatin
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, uent 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 rst.
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 0.12).
Arabic text
9.4.2. is passage highlights the fact that the modern classical Arabic lan-
guage is a rather articial concept. In fact, the currently unwritten short V have
very uctuating realizations, due to both their presence or absence and to their
timbres (themselves) (/i, a, u/). As the number of recordings (of dierent speakers)
increases, the number of variations also increases (even for stressing and ortholo-
gy, the use of intonemes and pauses).
10.0.2. Nasalized vowels are marked with the tilde sign (i q ); 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 (a a a a).
Postalveolar consonantal articulations are indicated by a dot underneath (=
); postalveopalatal ones with an inverted circumex (or haek\ ; instead of
traditional, but ambiguous, c j ] for the uvulars of Arabic and Persian loanwords
we use particular signs (q ; 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 aspirated consonants (ph bh h). 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 ('mt-lb)
/'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. 10.1 shows Hindi vowels, which are ten: three short, (I, , U) /i, a, u/,
and seven long. ese are actually diphthongs: ve 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), (, ) /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 ('pR)
/'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 ('bR) /'bEEr/ baer ('booR) /'boor/ boor ('bR) /'bOOr/ baor We will
deal again with the geographic variants of /EE, OO/ ( 10.2.) again, only here,
('bR) /'bEEr/ baer ('bR) /'bOOr/ baor\ ('bR, 'bR) (east: Bihar, West Bengal),
('bER, 'bOR) (west: Rajasthan), ('bER, 'bOR) (northwest: Panjab)
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/ (', &, ) /OO/ (', &, )
/a/ (', &, 0, )
/aa/ ('aa, &a, a)
10.1.2. Each Hindi vowel may be (distinctively) nasalized: (t'hI) /t'hi/ thi ('H,
&H, H) /'h/ hq (b'H) /b'h#/ bh (k'h_9) /k'h/ kh let us consider, for
instance ('Rg) /'rag/ rag vein ('RAg) /'rg/ rg color. Currently, words such as
('Hs) /'hans/ hans tend to be confused with ('HAs) /'hs/ hs but, in neutral pro-
nunciation, they must be accurately distinguished, even if () has no full contact
with the alveolar ridge ( 10.2.1.1-2).
en ae ao are followed by /j, w/, they correspond to (I, U) /ai, au/: ('mIa)
/'maijaa/ maeyaa (tI'jaaR) /tai'jaar/ taeyaar ('HUja) /'hauwaa/ haowaa e same
should hold for Sanscrit words with ae ao (ai au); but, currently, these become
/EE, OO/ ( 10.2.1.2).
Intra-lexemic sequences /eeh0, ooh0/ may be realized as short vowels: ('seeH-Ra,
'seH-) /'seehraa/ seehraa ('mooH-lt, 'moH-) /'moohlat/ moohlat when there is gram-
memic derivation, no shortening occurs: ('leeH-j) /'leehja/ leehya ('mooH-na)
334 a handbook of pronunciation
Consonants
laryngeal
palatal
provelar
veolar
alveolar
bilabial
palatal
uvular
postal-
dental
velar
velar
(l) l ($) ()
/c, G/ (C, ), /S/ (), /j/ (j, ), /F/ (), /w/ (6, j), /r/ (R), /#/ ()
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/; () (()) before /s/; (n) ((q)) before /j/; and () {()} 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, , , n, ) are nasals with no full contact ( semi-nasals): semi-
-labial, semi-alveolar, semi-postalveo-palatal, semi-palatal, and semi-provelar, re-
spectively, 9.9 of NPT/HPh and () {()} as well, given in 10.2.1.2.)
336 a handbook of pronunciation
On the other hand, it is not necessary to use the specic symbols ((, n)):
('mooH) /'mooh/ mooh ('smbl) /'sambal/ sambal ('tiim ma'tae) /'tiin maa'taaee/
tiin maataaqq (sM'jaad) /sam'waad/ samwaad ('nbH) /'nabh/ nabh ('kNh)
/'kanh/ ka=h ('k~) /'kanG/ kan ('6) /'wanS/ wan ('n-j) /'anja/ anya
('pk) /'pank/ pa;k (&I,x'laab) /ina'laab/ inqalaab ('sIH[x]) /'sih[a]/ sinh(a)
In some words, we nd /m, n/ before heterosyllabic C\ (b'Ramda) /ba'raam-
daa/ baramdaa (m'Ruud) /am'ruud/ amruud ('gUmi) /'gumii/ gum=ii ('CmCa)
/'camcaa/ amaa ('Umka) /'Gumkaa/ humkaa ('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, ) occur, and
have specic graphemes, ;, but generally change into /n/. e more frequent
genuine realization of /N/ is (), a nasalized postalveolar ap, or even a nasalized
postalveolar approximant, (), so it is better to use the symbol of the diaphone,
(), 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): ('bam) /'baam/ baam ('ban) /baan/ baan ('ba, 'baN,
'ban) /'baaN/ baa (gx'ee, gx'Nee, gx'nee) /ga'NeeS/ gaee ('gN-j, 'g-j)
/a'gaNja/ agaya ('6amI, -m, -m-, -nm, -mm, -mm) /'waamaj/ wa;-
may ('ba[gx]) /'baang/ baa;g.
Furthermore, we nd the sequences /mh, nh/ (mH, nH), which (together with
/lh/ (lH) and, possibly, /h/, 10.2.1.1) have no ocial devanagari graphemes,
but combinations. In fact, they did not occur in Sanscrit, contrary to aspirated
stops and stopstrictives, including /#h/ (H), which comes from /h/ (H): (kUm-
'HaaR) /kum'haar/ kumhaar ('kanH) /'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): ('pl) /'pal/ pal (p'hl) /p'hal/ phal ('baala) /'baalaa/
baalaa (b'Haala) /b'haalaa/ bhaalaa ('tl) /'tal/ tal (t'hl) /t'hal/ thal ('daa6a) /'daa-
waa/ daawaa (d'Haa6a) /d'haawaa/ dhaawaa ('ppa) /'appaa/ =appaa ('hppa)
/'happaa/ =happaa ('iil) /'iil/ iil ('Hiil) /'hiil/ hiil ('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/ (, , ), /kh/ (xh, x), /g/ (, y, ), /gh/ (H). Before
front vowels (and before /j/), /k[h], g[h]/ are realized as prevelar, but no special sym-
bol (([], [])) is necessary. In nal position, the stops may have inaudible reali-
zations: ('naak, 'naak) /'naak/ naak ('b, 'b) /'ab/ ab this is contrary to what hap-
pens to aspirated sequences, which do not simplify, even if the laryngeal element
may be less evident, in this nal position (for other C too), but it may not be miss-
ing, because it is distinctive\ ('siikh, -kh) /'siikh/ siikh ('nbH, -bH) /'nabh/ nabh
(and: ('booH, -H) /'booGh/ booh ('baaH, -H) /'baa#h/ baah] Only within se-
quences like /0h0/, may /h/ be dropped ( 10.3.1.5). Also word-nal /h/ may be
attenuated (and even disappear completely): (b'jaaH, -aaH, -aa) /b'jaah/ byaah
Stopstrictives
10.2.3. We only nd the postalveopalatal diphonic pair, /c, G/ (C, ) (with the
corresponding aspirated sequences, /ch, Gh/ (Ch, H), and the possibility of
(H), as for the stops): (C'tUR) /ca'tur/ atur ('Raa) /'raaG/ raa ('Caal) /'caal/
aal (C'haal) /c'haal/ haal ('aal) /'Gaal/ aal ('Haal) /G'haal/ haal In various
contexts, there may be frequent attenuated realizations ( constrictives or approx-
imants), even after a pause (or, instead, sharper ones, stops), for /c/ (, T), /ch/
(h, h Th), /G/ (, D), /Gh/ (H, H DH).
Constrictives
/'bi/ bi; on the other hand, /s/ becomes (), before //: ('k) /'kas/ ka= (shown
in writing as well). In Persian, Arabic, and English loans, we nd /F, z/ (, z) too
(which currently become /ph, G/ (ph, )): ('iiR, ph-, -'kiiR) /Fa'iir, -'kiir/ fa-
qiir ('Ut, 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, H, h); (the semi-palatal) () occurs between V in unstressed syllables: ('jee)
/'jee/ yee (dH'jan) /dh'jaan/ dhyaan ('lIe, 'lie, 'lIe) /'lijee/ liyee en in unstressed
nal position (with an inherent a] its sound is (-, -); when preceded by i it
drops, often closing /i/ {(I) = (i)}, except in careful pronunciation: (s'm, s'm)
/sa'maj/ samay (k', -, k'-) /k'aj/ kay ('Caa, 'Caa) /'caaj/ aay (k't-RI,
kt'RI, -I, -i, k-) /k'atrij/ katriy
Generally, the /h/ phoneme is (H) while (h) occurs with voiceless aspirated
consonants. Near nasalized V /h/ becomes nasalized: ('Haathi) /'haathii/ haathii
('gRH) /'garh/ garh (&pH'laa) /pah'laa/ pahlaa ('teeRH, te'RH) /'teerah/ teerah
(HAs'naa) /hs'naa/ hsnaa ('bH) /'bh/ bh (kx'H) /ka'h/ kah
ere are two further approximants, for Persian and Arabic loans, for which the
ocial uvular constrictives symbols are often used, /X, / (or, even less precisely,
the velar ones, /x, /); 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'l, d'kl) /da'Xal/ daal ('aana, 'kaana) /'Xaanaa/ aanaa ( (k'haana)
/k'haanaa/ khaanaa] ('baar, 'baa, 'baag) /'baa/ baa ( ('baag) /'baag/ baag]
Trills
10.2.6. In this category, we have an alveolar tap, /r/ (R), and a postalveolar ap,
/#/ (), 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 (), similar to British English (>), but with
no lip rounding at all): (H'R) /ha'ra#/ hara (l'kaa) /la#'kaa/ lakaa ere is al-
so the sequence /#h/ (H, H): (b'Haa, -'Haa) /ba#'haa/ bahaa ( (b'aa, -'aa) /ba-
'#aa/ baaa]
Laterals
10.2.7. ere is only one lateral phoneme, /l/ (l, $, ); besides, there is the se-
quence /lh/ (lH): ('laat) /'laat/ laat (m'laal) /ma'laal/ malaal ('k C'loo) /'kal ca-
'loo/ kal aloo ('k j'H 'aao) /'kal ja'h 'aaoo/ kal yah aaoo ('aa$a) /'aalaa/
aalaa ('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 @, or, less positively, with tense voice @.
Taxophonics
10.3.1.1. As for V which oppose as short and long, or rather (very) narrow
diphthongs (generally represented by geminating symbols, 10.1.1), gemina-
tion is distinctive for C too: (p'taa) /pa'taa/ pataa ('ptta, pt'taa) /'pattaa/ pattaa
(b'Caa) /ba'Caa/ baaa ('bCCa, bC'Caa) /'baccaa/ baaa (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: ('st-j) /'satja/ satya ('gd-Ha) /'gadhaa/ gadhaa (bH'Rk) /abh-
'rak/ abhrak ('Uk-l) /'Sukla/ ukla (kh'aa) /akh'#aa/ akhaa. Obviously, in case
of isolated initial sequences, the syllabicity scale joins the elements in one syllable,
although some slight dierence is maintained; however, if they are internal, they
divide into two syllables, including the preceding V\ ('Hiil) /G'hiil/ hiil. e
geminates are (00) 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 dierence with geminates): ('bU:Ha) /'buhaa/ buhaa (6Id[:]-
'aal, -l, -l) /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: (st'jaaRth-
pR&ka) /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 ( 10.1.3, although here we indicate it simply as an audi-
ble oset, ($)), not only in complex consonant clusters, but also in not quick pro-
nunciation at the end of words, even after a single consonant: ('gR$) /a'gar/ agar
('uupR$) /'uupar/ uupar (Cl$'naa) /cal'naa/ alnaa ('phuul$) /'phuul/ phuul ('R$-
bt) /'Sarbat/ arbat (bd$'maa) /bad'maaS/ badmaa (l$'kaa) /la#'kaa/ lakaa
(khI$'kii) /khi#'kii/ khikii (bH'jaas$) /abh'jaas/ abhyaas (s'm$) /sa'maj/ samay
(Rakh$'naa) /rakh'naa/ rakhnaa (Ug$'naa) /ug'naa/ ugnaa ('uugH$na;) /'uughnaa/
ghnaa ('p$) /'pa#/ pa ('pH$) /'pa#h/ pah
/d'juut/ dyuut (dH'jni, -'nii) /dh'wani/ dwani (dH'RU6) /dh'ruw/ dhruw ('joo-
Hi, jo'Hii) /'joo#hii/ yoohii (C'jUt) /c'jut/ yut ('jooti) /G'jooti/ yooti (k'jaa)
/k'jaa/ kyaa (k'jaath) /k'waath/ kwaath (kh'jaab) /kh'waab/ khwaab (g'Ram)
/g'raam/ graam (g'lani) /g'laani/ glaani (gH'Ra, -N, -n) /gh'raaN/ ghraa.
More: ('jaal, kh'j-) /X'jaal/ yaal (H'Raas) /h'raas/ hraas (m'an) /Sma'Saan/
maaan ('Rant) /S'raant/ raant ('lth) /S'lath/ lath (s'kndH) /s'kandh/ skandh
(skhx'ln) /skha'lan/ skhalan (s'tR) /s'tar/ star (s'thl) /s'thal/ sthal (s'mIt) /s'mit/
smit (s'nan) /s'naan/ snaan (s'jaaH) /s'jaah/ syaah (s'jRg) /s'warg/ swarg (s'Ra)
/s'rasaa/ sra=aa (z'jaada) /z'jaadaa/ zyaadaa (6jt'haa) /wjat'haa/ wyathaa (6'Rt)
/w'rat/ wrat In words of Sanscrit origin, we nd /k/ as well (but people current-
ly change it into (k)): (k', k'N, k'n) /k'aN/ ka (k'iiR, k'-) /k'iir/ kiir
Furthermore, for initial sequences with /s0[0]/, current pronunciation prexes
an epenthetic vowel, generally (I, I) but (, ) is possible as well (and also (, ),
with the other variants seen, 10.1.3), before non-front vowels: (s'tRii, Is-, Is-)
/s'trii/ strii (sp'huuRti, Is-, Is-, s-, s-, s-, s-) /sp'huurti/ sphuurti (s'nan, Is-, Is-, s-,
s-, s-, s-) /s'naan/ snaan (also with /S/: ('look, I-, I-, -, -, -, -) /S'look/ look]
10.3.1.4. Short (I, , U) /i, a, u/, in nal position, have the peculiarity of fading
( 10.1.3), to a point in which they drop as well, as happens to /a/, ( , , , 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: ('R6i, R'6ii) /'rawi/ rawi ('tIthi, tIt'hii)
/'tithi/ tithi ('Iu, I'uu) /'SiSu/ iu ('6su, 6'suu) /'wasu/ wasu e change V =
VV 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 /'/ + /, , 0/
are realized as ('): ('Hani) /'haani/ haani ('kam) /'kaam/ kaam (ak'Rant)
/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'tae) /'tiin maa'taaee/ ti-
in maataaqq ('tiim p'piite) /'tiin pa'piitee/ tiin papiitee ('tiiN 'aal) /'tiin 'aal/ ti-
in aal ('tii kx'ml) /'tiin ka'mal/ tiin kamal ('tii 'gaa) /'tiin 'gaaj/ tiin gaay
('tii~ 'Caa) /'tiin 'caaj/ tiin aaj ('tii~ 'aal) /'tiin 'Gaal/ tiin aal ('tiin 'Raab) /'ti-
in Sa'raab/ tiin araab ('tiin 'jaaR) /'tiin 'jaar/ tiin yaar ('tiiM 'jR, -R) /'tiin 'war/
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): ('gbR, g'bR) /'akhbar/ Akhbar (tz'biiH) /tas'biih/
tasbiih ('CUb 'bho, 'CUb b'hoo) /'cup 'bEEhoo/ up bae=hoo ('Chb gx'jaa)
/'cap ga'jaa/ hap gayaa ('saad 'bho, 'saad b'hoo) /'saath 'bEEhoo/ saath bae-
=hoo ('p 'tk) /'ab 'tak/ ab tak ('Haad 'doo) /'haath 'doo/ haath doo ('aakpR) /'aag-
par/ aak par ('eeg 'dIn) /'eek 'din/ eek din ('pki) /'Gabki/ ab ki ('baa 'kaa, -k
'k-) /'baa 'kaa/ baa kaa
Stress
10.3.2.1. In Hindi, stress position is not distinctive; in fact, the same speaker,
in dierent occasions, may stress dierent syllables of the same word. Furthermore,
these uctuations 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 weights. is holds for isolated words as well, but always
with a certain exibility. 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 uctuating (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)
dierentiation and identication of stress strength on the dierent non-light sylla-
bles (but sometimes on the light ones too).
10.3.2.3. Stress assignment, in isolated words, is done on the basis of the identi-
cation of the heaviest syllable, as in: (U'paadHi) /u'paadhi/ upaadhi (U'pant)
/u'paant/ upaant (&ad-jo'pant) /aadjoo'paant/ aadyoopaant (kx'lii) /ka'lii/ kalii
(kx'man) /ka'maan/ kamaan ('kamna) /'kaamnaa/ kaamnaa (ko'aaRk, ko-
'NaaRk, ko'naaRk) /koo'Naark/ kooaark ('andani) /'Xaandaanii/ aandaanii
('jamIti) /G'jaamiti/ yaamiti (&tabe'daaR) /taabee'daar/ taabeedaar (&tIgU'naa)
/tigu'naa/ tigunaa (p&RIthji'Raa-Ra&so) /prithwii'raaGraasoo/ prithwiiraaraasoo.
More: (bn'duugba&zi) /ban'duukbaazii/ banduukbaazii (mR'dagi) /mar-
'daangii/ mardaangii ('mandHa&ta) /'maandhaataa/ maandhaataa (&st-jpR'kaa)
/satjapra'kaaS/ Satyaprakaa (st'jaaRth-pR&ka) /sat'jaarthprakaaS/ Satyaarthprakaa
(&sago'pag) /saangoo'paang/ saangoopaang (sa'man) /saa'maan/ saamaan ('Ha-
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): ('kUnt&la, &kUnt'laa) /Sa-
'kuntalaa/ akuntalaa (gx'eeRI&a, gx&e;RI'jaa) /ga'#eerijaa/ gaeeriyaa (b'HaadU&Ri,
b&HadU'Rii) /ba'haadurii/ bahaadurii (&mHab'HaaRt) /mahaab'haarat/ mahaa-
bhaarat (jUd'HIhIR, &jUdHI'hIR, --, -'-) /jud'hiir/ Yudhi=hir (s'HuulI&t, s&Hu-
lI'jt) /sa'huulijat/ sahuuliyat.
More examples: (a'iiR-jad, &aiR'jaad) /aa'Siirwaad/ aaiirwaad ('xl&mnd, &-
xl'mnd) /'almand/ aqlmand (&ana'kani, a&naka'nii) /aanaa'kaanii/ aanaakaanii
(dH'jni, -'nii) /dh'wani/ dhwani (pR'jRti, &pRjR'tii) /par'wartii/ parwartii
('bRtn, bR'tn) /'bartan/ bartan ('bs-ta, bs'taa) /'bastaa/ bastaa (b'Haala, bHa-
'laa) /b'haalaa/ bhaalaa (s'mIti) /sa'miti/ samiti (sa'lana, &sala'naa) /saa'laanaa/
saalaanaa (s~'Caaln, &s~Ca'ln) /san'caalan/ sanaalan (&HeRap'heeRi, He&Raphe-
'Rii) /heeraap'heerii/ heeraapheerii
10.3.2.4. e same stress pattern occurs in inected and derived words as well:
(&bd-HI'k) /badhi'k/ badhik (&lg-HU'tR) /laghu'tar/ laghutar (&UCI'tm) /Su-
ci'tam/ uitam (ms'leega, &msle'gaa) /mas'leegaa/ masleegaa ('laap&ta;, &lap'taa)
/'laapataa/ laapataa (&gxlI'jaaRa, gx&lIa'Raa) /gali'jaaraa/ galiyaaraa (pa'glpn, &pa;-
gl'pn) /paa'galpan/ paagalpan (sUn'dRta, &sUndR'taa) /sun'dartaa/ sundartaa
(bH'naapa, &bHna'paa, b&Hna'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'sooigHr) /Ra'sooiighar/ rasooiighar ('deenIka[a]la) /'deeSni-
kaalaa/ deenikaalaa ('6Ijko[o]) /'wiSwakooS/ wiwakoo ('duuRdRi[i]) /'duur-
darSii/ duurdarii ('dI-la[a]) /'dilGalaa/ dilalaa ('nImnlIkhIt) /'nimnlikhit/
nimnlikhit ('CndRka[]nta) /'candrakaantaa/ andrakaantaa ('CndRka[]nta-
sntti) /'candrakaantaasantati/ andrakaantaasantati.
More: (m'HaasbHa[a]) /ma'haasabhaa/ mahaasabhaa ('HIndumHa[a]sb-
Ha[a]) /'HIndumahaasabhaa/ mahaasabhaa ('tuu 'kRmHi[i]M j'jkti&H) /'tuu
'karmhiin w'jaktihEE/ tuu karm-hiin wyakti hae ('tuu 'kRmHi[i]&H) /'tuu 'karm-
hiinhEE/ tuu karm-hiin hae ('amja[]m) /'aamwaam/ aam-waam ('kan-
ka[]n) /'kaankaan/ kaan-kaan ('kmka[a]) /'kaamkaaG/ kaam-kaa ('-
sa6[]sa) /'EEsaawEEsaa/ 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. 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 10.4), are generally accompanied by falsetto (for all kind of voic-
es), which is marked with () after the intonemes, in phonotonetic transcriptions.
In addition, paraphonically, there is a peculiar type of Indian voice, with murur
344 a handbook of pronunciation
/ / (2 2 2 2 2 2 2) /./ (2 3 3)
/ / ( 2 2 2 2 2 2 2) /?/ (2 ' 2 1)
/ / ( 2 2 2 2 2 2 2) // (2 2 2)
/ / ( 2 2 2 2 2 2 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, uent 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 rst.
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 0.12).
10. hindi 345
Hindi text
10.4.2. Uttarii hawaa aor suura is baat par haga rahee thee ki ham doon
mqq zyaadaa balwaan kaon hae. Itnee mqq garam oogaa pahnee eek musaar udhar
aa niklaa. Hawaa aor suura doon is baat par razii hoo gayee ki doon mqq see
oo pahlee musaar kaa oogaa utarwaa deegaa wahii zyaadaa balwaan samhaa
aayeegaa.
Is par uttarii hawaa apnaa puuraa zoor lagaakar alnee lagii. Leekin wah aesee
aesee apnaa zoor bahaatii gayii waesee waesee musaar apnee badan par oogee koo
aor bhii zyaadaa kas kar lapee=taa gayaa. Ant mqq hawaa nee apnii kooi band kar
dii. Phir suura teezii kee saath niklaa aor musaar nee turant apnaa oogaa utaar
diyaa. Is liyee hawaa koo maannaa paaa ki un doon mqq suura hii zyaadaa bal-
waan hae.
Kyaa tumhqq yah kahaanii ahii lagii? Kyaa phir see sunaa?
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, //i, y, u// (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 () (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 tradition still follows the
phonic fantasies of the Swedish sinologist Karlgren, who thought that Mandarin
had two retroex (or apical) vowels, rendered with q, Q. Actually, the former
is simply a back-central vocoid, (M) (where neither retroexion 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
(syllabic) contoid, (). e same phonetic tradition also lists another rare
retroex (mid) vowel, rendered with & which in actual fact is just the sequence
(X) /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 illustration of
the IPA, a paper which still proposes q, Q although ipa-like masqueraded as ()
in both cases! by describing them as (apico-)laminal denti-alveolar (where no
groove is rightly mentioned) and apical post-alveolar approximants respectively.
e latter is correctly described ( can()), while the former is still not ( can());
there is a great (articulatory, auditory, and acoustic) dierence between () and
(M).
348 a handbook of pronunciation
11.0.4. Continuing, we have e (X, , , x, ) /X/, with its variant () /e/, in con-
tact with /j, , i/, and () in /ae/. But it is decidedly safer to consider /X/ and /e/ as
two separate phonemes, given the great phonetic dierence, 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 (, ) /o/, with its variant (U) /u/ which, on the
contrary, is a dierent 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 (X, ). 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, , A) /a/, with its variants (jEn, n),
for ian an
nally, this phonetics la chinoise even has a zero (graphic) vowel, when
pinyin smartly economizes on vowels, by writing un for /wXn/ (wn), and iu for
/jou/ (jPU), ui for /wei/ (wI), where a vowel is denitely 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/ (, A),
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 ocial romanization. Starting from the (conso-
nantal) initials, which are 23 (with a double occurrence of y with two dierent
values), we will not present them alphabetically, but by logical phonic groups, ac-
companied by their phonemic and simple phonetic transcriptions. e latter is
simple, 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)
// (), > /h/ (h), // (), r />/ ()
j /q/ (), q /qh/ (h), x /s/ ()
g /k/ (k), k /kh/ (kh), h /h/ ()
w /w/ (w), y /j/ (j) // ().
11. chinese 349
11.0.6. e nals are 39 (with a triple occurrence of i with three dierent 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 > ). Also the exclamations e
/e/ (), yo /jo/ (j) occur (with various tonemes, including zero). e three nals
which do not occur without an initial are marked with an asterisk (*):
a /a/ (a), ia ya /ja/ (ja), ua wa /wa/ (wa)
ai /ae/ (), uai wai /wae/ (wa) yai /jae/ (sole word, yi)
an /an/ (n), ian yan /jan/ (jEn), uan wan /wan/ (wn), an yuan (j/q/x +
uan) /an/ (n)
a /a/ (a,), ia ya /ja/ (ja,), ua wa /wa/ (wA,)
ao /ao/ (A), iao yao /jao/ (ja).
e /X/ (X, Xx|, x|), ie ye /je/ (j), e yue (j/q/x + ue) /e/ ()
ei* /ei/ (I) {(I)}, en /Xn/ (n), er /X>/ (X)
e* /X/ (x,), ue we /wX/ (w,) {(w,)}.
i /i, ji/ (i, ji), (s-/z-/c-) /M/ (M) > (with ], (-/-/>-) // () >
in yin /in/ (In, jIn) i yi /i/ (, j)
ui wei /wei/ (wI) {(wi)} > uei .
o /o/ (, x|), ou /ou/ (U) {(U)}, uo wo /wo/ (w, wx|) {(w, wx)}
o* /u/ (U,) > u , io yo /ju/ (jU,) {(U,)} > iu yu .
u wu /u, wu/ (u, wu) yu (j/q/x + u) /y, y/ (y, y)
un wen /wXn/ (wn) > uen ; n yun (j/q/x + un) /yn, yn/ ([]Yn) {([]In,
[j]Yn)}
iu you /jou/ (jPU) {(ju)} > iou .
Vowels
11.1.2. We rst nd the close front phoneme, /i/: /[j]i, [j]in, [j]i/ ([j]i, [j]In,
[j],) (C/y)i (C/y)in (C/y)i\ (5ji) /5ji/ yi ([~ii) /ni/ n` (4jIn) /jin/ yn (5In)
/5qin/ jin (7j) /6ji/ y (5,) /5qi/ ji
e close front-central rounded, /y/, has a fairly limited distribution, ju qu xu
yu n l (with or without -n), /[]y, []yn/ (y, Yn) [variants with -n (In, jYn)]:
(5y) /5qy/ ju ([yy) /qhy/ q (7y) /6sy/ x ([yy) /y/ y (5Yn, 5In)
/5qyn/ jun (4Yn, 4In) /qhyn/ qn (7Yc, 7Ic) /6syn/ xn (4Yn, 4In,
4jYn) /yn/ yn
More conveniently, pinyin could have used z c s y, by phonemically in-
terpreting these syllables as we do: (y, hy, y) /qy, qhy, sy/. It could also have
avoided introducing an allograph without dieresis, by using y for ocial 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 /jyn, in/). erefore, it could have been possible to write
z/c/siu, instead of the ocial j/q/xio for (jU,, hjU,, jU,) /qju, qhju,
sju/ (thus also -u instead of the ocial -o for (0U,) /0u/).
wn, x,, w,) e er en un/wen e ue/we. e last one has the variant (w,)
(which, if necessary, could be indicated as /wo/): (7$[*|]) /6X/ (5n) /5Xn/ en
(4X) /X>/ r (7fc) /6fXn/ fn ([wc) /wXn/ wn (7f*) /6fX/ f (5w, 5w,)
/5wX/ we (7w 7w) /6wX/ w
nally, we have /o/ ([x|]) o which can occur in the syllables /o, jo/ ([x|], j[x|]),
in addition to normal /mo, po, pho, [0]wo/ ([[x|], ([x|], (h[x|], [0]w[x|]), mo
bo po wo/Cuo\ (5[[x|]) /5mo/ mo (4([x|]) /po/ b (7([*|]) /6pho/ p ([w,
[w*|) /wo/ w (5w) /5wo/ uo (A,5j) /ha5jo/ hyo (7[*|]) /6o/ ! en
nal, 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 ([[x|], ([x|], (h[x|]), while maintaining /mou, phou/ (mU, phU) (/pou/
does not occur), by dissimilation.
11.1.5. Lastly, we nd the open vowel, /a/, with its distributions and various
taxophones, /[0]a, jan, an, [w/0]an, [j]a, wa/ (a, jEn, n, [w/0]n, [j]a,, wA,):
(5a&a) /5aqa/ aza (7ta) /6tha/ t (4jEn) /jan/ yn ([tjEEc) /tjan/ din (7c)
/6an/ yun (5n) /5qhan/ quan ([c) /an/ n (4pn) /phan/ pn (4pa,)
/pha/ p ([wc) /wan/ wn (5ja,) /5ja/ ya ([wAA) /wa/ w
Diphthongs
11.1.6. Moving to diphthongs ( 11.3), we have /[w]ei, [w]ae, [j]ao, [j]ou/ (I,
wI, , wa, ja, A, U, jPU) ei ui (u)ai (i)ao ou iu for /[w]ei, [j]ou/ there are
also the possible variants (shown in 11.4, which are not necessary for a good
pronunciation, and will be given only here), (I, wi, U, ju): (5fI 5fI) /5fei/ fei
(5twI 5twi) /5twei/ dui ([) /ae/ i (7wa) /6wae/ wi ([AA) /ao/ o ([AA)
352 a handbook of pronunciation
/5hX>/ heir (5tj) /5thja>/ tianr (4mj) /mja>/ mior (7qw$Y) /6qhwX>/ cur
(4twX) /thwX>/ tur (^m5pwX) /5an5phwX>/ anpor
11.5. Vowels with />/ () -r.
Vowel reductions
can be seen, () applies to /i[n/], y[n]/: (7tii), (-3d) /6titi/ ddi (5kw~&i), (-3B)
/5kwansi/ guanxi (7jay), (-3i, -3) /6sjaqy/ xiqu and (), to /M[>], u[]/: (7ji-
), (-3z) /6jisM/ ysi (7), (-3) /6>Xn/ rni (5u&u), (-v) /5ufu/ ufu
11.6. Reductions in syllables with zero tones (in quick speech, not in an intoneme; where-
as, in an intoneme, the distinct timbers of 11.1-3 are used).
/i, y/ () /M, u/ ()
/Xn, e/ (), /[w]ei/ ([j]) /X[], o/ (X), /[j]ou/ ([]X)
11.1.10. In addition, we have (), for /[j/]e, [w]Xn/, and () for /[w]ei/: ([j-
^), (-1, -1) /qjeqje/ jijie (7c), (-3c, -3c) /6ae>Xn/ iren (7mI-
mI), (-3m) /6meimei/ mimei then (X), for /X, [w]X, [w]o/, and ([j]X), for
/[j]ou/: (5kX&X), (-2gX) /5kXkX/ gege (t^2X), (-1"X) /tiX/ d`e (5jEN&x,),
(-2X,) /5sjanX/ xiane (7jm$), (-3m$) /6qjemo/ jimo (nw^j), (-1hjX)
/nwanhwo/ nunhuo (7UU), (-3d$) /6utou/ ytou (4px,&jU), (-2X)
/phXjou/ pyou
Lastly, we have: () in /a[n], [w]a, [w]ae/: (7paa), (-3b) /6papa/ bba (5ma&ma),
(-2m) /5mama/ mama (i^wan), (-1hjn) /sihwan/ x`huan (7ttha), (-3t)
/6thaethae/ titai and () in /a, [j]ao/: ([w^a,), (-1,) /wana/ wna
(5&A), (-2d) /5tao/ idao
11.7. Further reductions in syllables with zero tones (in even quicker speech, not in an
intoneme; then, in an intoneme, the realizations of 11.6 are also possible).
/i, y, M[>], u/ ()
/ae/ (), /ja/ (j) /ou, uo, ao/ ()
/X, -X- a, -a-/ ()
11.1.12. 11.8 shows the vowel realizations that are sucient for an interna-
tional pronunciation of Chinese, without taxophones (thus, without either atten-
uations or neutralizations such as those given in 11.6-7; equally, we do not even
mark creaky voice, which is superuous in an international kind of pronuncia-
tion). So we have only eight monophthongs (and four diphthongs, obtained by
combining ve 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/ yn (5in) /5qin/ jin
(7ji) /6ji/ y (5i) /5qi/ ji
en: (5y) /5qy/ ju ([hyy) /qhy/ q (7y) /6sy/ x ([yy) /y/ y (5yn)
/5qyn/ jun (hyn) /qhyn/ qn (7yn) /6syn/ xn (yn) /yn/ yn
Besides: ([wuu) /wu/ w (lu) /lu/ l (5ku) /5ku/ go (7u) /6u/
(j) /je/ y (5tj) /5tje/ die (7) /6e/ yu ([) /se/ xu and (7x) /6X/ (5xn)
/5Xn/ en (X) /X>/ r (7fxn) /6fXn/ fn ([wxxn) /wXn/ wn (7fx) /6fX/ f (5wx)
/5wX/ we (7wx) /6wX/ w
More: (5m) /5mo/ mo (p) /po/ b (7ph) /6pho/ p ([w) /wo/ w (5w)
/5wo/ uo (ha5j) /ha5jo/ hyo (7) /6o/ ! and (5a&qa) /5aqa/ aza (7tha)
/6tha/ t (jan) /jan/ yn ([tjaan) /tjan/ din (7an) /6an/ yun (5han)
/5qhan/ quan ([aan) /an/ n (phan) /phan/ pn (pha) /pha/ p ([waan)
/wan/ wn (5ja) /5ja/ ya ([waa) /wa/ w
11.1.13. And with />/: ([) // ` (5h) /5h/ >i () // (7) /6>/
r (5kx) /5kX>/ ger ([pxx) /pX>/ bnr (h) /hX>/ hr ([naa) /na>/ nr
(7pa) /6pa>/ bnr (5q) /5qa>/ zar (5hwa) /5hwa>/ huar (7mja) /6mja>/
minr and (pha) /pha>/ pir (7hx) /6qhX>/ qr (7wx) /6wX>/ r
(7qhx) /6qhX>/ cr ([qxx) /qX>/ z`r (5kwa) /5kwa>/ guanr (5kw) /5kwX>/
gor (5swxn&nx) /5swXnnX>/ sunnr (i&fwx) /sifwX>/ xfur (^ljx) /5-
ljX>/ ilir (jx) /qjX>/ jir (^ha2pa[thwxx) /5hapathwX>/ habatu`r (7tu-
j) /6tuqjX>/ djir (7jx) /6jX>/ ynr ([sxx) /sX>/ s`r (5hx) /5hX>/ heir
(5thja) /5thja>/ tianr (mja) /mja>/ mior (7qhwx) /6qhwX>/ cur (thwx)
/thwX>/ tur (^am5phwx, -n5p-) /5an5phwX>/ anpor
For diphthongs, we have: (5fi) /5fei/ fei (5twi) /5twei/ dui ([aa) /ae/ i
(7wa) /6wae/ wi ([aa) /ao/ o ([haa) /hao/ ho (ja) /jao/ yo (5u) /5ou/
Ou ([mu) /mou/ mu (5tju) /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 11.9, which gives the consonantal articulations of Mandarin
Chinese, to save space, we show as /05/ (whereas here we use /0[h]/ = /0, 0h/)
the consonants which are distinct phonemically. In fact, simple consonants oppose
aspiration sequences ( Stops, 11.2.2, and Stopstrictives, 11.2.3).
Instead, 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.
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
m ([) (M) (n) n (N) (~)* () ,
F p5 (b) ((5 {) t5 (d) k5 (g) ()
q5 (Q) 5 (") (5 )
f (v) (5 ) ()
_ s (z) () ( B)
j|() |(%) () w|(j) () h
l
// (,), />/ (), /h/ (, h, , )
Nasals
11.2.1.2. Here are some examples: (5mA) /5mao/ mao (5[) /5mo/ mo
([nAA) /nao/ no (4pn) /phXn/ pn (4px,) /phX/ p (c7$[*|]) /hXn6X/
hn (n[tii) /qhanthi/ qunt` (Ic[qAA) /qinqao/ j`nzo (0twn-
7ljEc) /6twan6ljan/ dunlin (m4mIn) />Xnmin/ rnmn (lwm5pn) /lwXn-
5pan/ lnban (0ka[faa) /6khanfa/ knf (^pjE~4i) /5pjanqi/ bianj (0I[)
/6qin/ jn` (0p4) /6phanqe/ pnju (pjE4y) /pjany/ biny
(k[c) /kan>an/ gnrn (pI[kwc) /5pinkwan/ bingun (0jE7wA)
/6qjan6wa/ jinw (ljE,5wn) /ljan5hwan/ linhuan
th //: (ji7lu p,5n) /ji6lu phi5an/ yl p'an (^fx,7mjEc) /5fX6mjan/
femin (kU7pu) /khu6pu/ kb (0la,7fI) /6la6fei/ lfi (^fx,[tii) /5fX-
ti/ fed` (kU7i) /khu6si/ kx (p,4y) /phiy/ py (^x,7) /5X-
6>/ er (^fx,7c) /5fX6an/ fen (p,[kw, -[kw*|) /phikwo/ p-
gu
In an international-like pronunciation, it is sucient 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 aspiration sequences, /0h/, with distinctive values. Before see-
ing some examples, we have to explain the nature and consistency of both voice-
lessness and aspiration.
In fact, except in slow (or very accurate) speech, /p, t, k/, are realized as (p,
t, k) only in fully stressed syllables (or after pauses), whereas in half-stressed sylla-
bles they are (, , ) or, in unstressed syllables, even (b, d, g). On the other hand
in a parallel way /ph, th, kh/, in fully stressed syllables (or after pauses), are
sequences of voiceless stops and a laryngeal constrictive (decidedly strong, (p,
t, k) (however, it is possible to nd a semi-constrictive as well, ((0)), which is
weaker). In half-stressed syllables, we have sequences with a laryngeal approximant,
(ph, th, kh); whereas, in unstressed syllables, we can even nd 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 aspiration and non-aspiration, however, is
guaranteed. In fact, we have: /0h/ ('=, &=h, =), /0/ ('=, &, ), 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 nd ((), which sounds rather dierent from the common (p) of
other languages.
11.2.2.2. Let us now see the following examples: (5pIn) /5pin/ bin (4() /po/
bo (7() /6pho/ p (4tj) /tje/ di (4kwI) /khwei/ ku (^p,5pa,) /5phi-
5pha/ pipa (0tjA[wuu) /6thjaowu/ tiow (5kx,) /5khX/ ke (7paa|,
7pa3b) /6papa/ bba (7tii|, 7ti3dI) /6titi/ ddi (5kX&X|, 5kX2g) /5kXkX/ gege (7-
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 aspiration as for Chinese
stops) we have /q, qh/ ('q, &, Q 'q, &qh, q) z c in addition to (', &, ',
&h, ) j q before /i, y j, / ( 11.2.0), and /, h/ (', &2, " ', &h, )
>
Here are some examples (expressly chosen with nal syllables without tonemes):
(5q) /5qae/ zai (7qwc) /6qhwan/ cun (5y) /5qy/ ju (4U, 4jU,)
/qu/ qi (7I) /6ei/ i (4) // > (4pi&M|, 4pi2Q) /piqM/ bzi
(4Yn&qh|, 4Yn2q 4jYn-) /ynqhae/ yncai (7tjEi|, 7tjE2) /6tjanqi/ din-
ji (5tjE~&hi|, 5tjE~) /5thjanqhi/ tianqi (4q&2|, 4q2") /qhae/ ci-
i (4pI&hn|, 4pI2n) /pheihXn/ pi>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, &, v) f, /s/ ('s, &, z) s and (', &, B) x, // (', &, ) \
(4fu) /fu/ f (4ma&n|, 4mavn) /mafan/ mfan (7swI 7swi) /6swei/ su (5fx,&u|,
5fx,2z) /5fXsu/ fesu (4&i|, 42B) /sesi/ xuxi (5u) /5u/ u (4lja,&|,
4lja,2) /lja/ lii Before /o/ (not /ou/) we nd (5) (as happens for other la-
bial phonemes): (45{x}) /fo/ f erefore, we could even pose /fwo/ instead of /fo/
(but /fou/ (fU), by dissimilation).
Approximants
mi-velar rounded (j): (5ja,) /5ja/ ya (4tj) /tje/ di (5j) /5qje/ jie (li^jEn|,
li1bEn, -1bn) /lipjan/ l`bian ([yy) /y/ y (7) /6qhe/ qu (7tiU|, 7ti-
3B%U) /6tisu/ dxio (4wA,) /wa/ w (5wI 5wi) /5hwei/ hui (tA-
^lwn|, tA1ljn) /thaolwXn/ tolun
nally, we nd /h/ (', &, h) (in addition to ('0, &0h, 0`), already seen for
stops and stop-constrictives): ([AA) /hao/ ho (5&A|, 5n2h) /5Xnhao/ en-
hao In theory, we could have two dierent phonemes in the inventory of conso-
nants: /x, h/, with /x/ = (', &, h); but it is better to have only one, with
the taxophones already seen (which are dierent 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 /05/ dierent 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 nd
just one phoneme, /l/ (l) (which remains unchanged even before /i, y j, /): (4l)
/lae/ li (4lU,) /lu/ l ([ljEEc) /ljan/ lin (7ly) /6ly/ l
Structures
11.3.0. We will deal, in particular, with the typical tones (tonemes as distinctive
elements), including a zero ton(em)e and their combinatory variants (taxotones),
in addition to degrees of stress, and nally 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 @, 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 nal 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.
(X) /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, ([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 // after /[h], , >/ /o/, only after /m, p[h], f, w/ (and in
362 a handbook of pronunciation
some particular forms after /j, l/ or with a zero segment); and /y/, only after /j,
l/, /n/ (~), or after /q[h], s/ ([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 /[h], / occur before /i, j y, / (or
some others), nor /k[h], h/ before /i, j y, / or /o, wX/. ere are some other more
specic limitations; but this is the spirit of Mandarin phonotactics /n, / are the
only, very frequent, syllable-nal 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): (4Xm, 4X, 4X2m, 4X-
2mX) /XnmX/ nme ([q$$C, q$1, q$1m, q$1mX) /qXnmX/ znme ([wC,
w, w1mn, -1mn) /womXn/ wmen (5tam, 5ta, 5tamn, -mn) /5tha-
mXn/ tamen Contrary to some other Chinese languages, in Mandarin there is a
somewhat exceptional occurrence of /m/ in nal position, as indeed that of (),
which we nd in : (4) /Xm/ really?/what??, (7) /6Xm/ yes/I understand.
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 dierent sylla-
bles, both for rhythmic reasons and prominence interplay, and even for communi-
cative purposes, even when they are not distinctive. us, for instance, /5akan-
u6jiX/ agny can have two frequent realizations, (5a,u0ji-
[$$, 5a,2u7ji2$); equally, even forms such as the following often oscillate:
(0ji[tjEEc, 7jijEc) /6jitjan/ ydin (0(4fI, 7(&I) /6pofei/ pfi (^qM7wI, 5qM-
0wI) /5qM6wei/ ziwi (0I[fuu, 7Iu) /6sinfu/ xnf (^In5tjEn, 5In^thjEn)
/5qin5thjan/ jntian (0[lii, 7li) /6aeli/ il` (^km5pI, 5km^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 zero tone which bear a
strong stress, verging on mid pitch (again independently of any intoneme): (4px,-
&jPU, &px,'jPU) /phXjou/ pyou (7t, 0') /6tan/ dni (4&U, &-
'U) /hou/ hou In addition, even forms bearing a tone (even written) are
weakened, as happens to the grammemes and locatives in the following examples:
(5w0a, -&a,) /5wo 6a/ uo (6ti0ja, -&ja) /6ti 6sja/ d xi (5wuli, 5wu-
&li) /5wu li/ wu l` (4t,li, -&li) /thi li/ t 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 tones. In fact, in
languages such as Mandarin, a dierent 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 ([m) /mae/ mi to buy and (7m) /6mae/
mi to sell. We now add some classical examples (giving some of the most fre-
quent meanings, among the possible ones, since transliterations may unify dier-
ent words {as phonic transcriptions do too}, represented by dierent characters,
or ideograms): (5ma) /5ma/ ma mother, (4ma) /ma/ m hemp, ([maa) /ma/
m horse, (7ma) /6ma/ m scold, curse
As some transcriptions have already shown, tonemes 3 and 4 are realized with
creaky (or laryngealized) voice, when their low taxophones occur: ([), () and (7)
(with secondary stress, (0)), but not when the high ones occur: () and (6). To-
nemes 1 and 2, which are high, (5), (4), always have normal voice quality.
A prudential warning is necessary about some diagrams which appeared in cer-
tain publications, not only in tourist-like 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 \
364 a handbook of pronunciation
(4wu) /wu/ w 3 ([) // >5 \ ([) /se/ xu 4 (7) /6/ >Z \ (7ja) /6qja/ qi
11.10 shows for the four tonemes the four basic tones, in addition to three
taxotones, which are as important phonetically (or rather tonetically): in addi-
tion to 3=2 ( 3+3 = 3+2) () // >5 ( a toneme 3 occurring before another
3 which becomes almost a tone 2 {although it is generally but falsely identied
with an actual tone 2}), we have a half 3, () // >5 ( a 3 before one of the oth-
er three), and a half 4, (6) /6/ >Z ( 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 rst half of their full forms.
In a kind of international pronunciation, toneme 2 can be realized as () (in-
stead of (4) ((-1))), as the ocial taxotone of // followed by another //. e pitch
characteristics of the other ton(em)es ought to be very similar to the ocial ones,
even for the zero toneme, as we have done in our transcriptions (although for
the latter, somehow, a realization of mid pitch {and weak stress} could be sucient,
(2), again in this type of international pronunciation).
11.10. Mandarin tonemes and tones.
1 /5/ (5) >1 2 // (4) >Q 3 // ([) >5 3 // () >5 3 // () >5 4 /6/ (7) >Z 4 /6/ (6) >Z
11.3.3.2. Obviously, although at rst 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 simplication, 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 11.11, the
allotones ( taxophones) of tonemes 3 and 4 present a simplication, by assimila-
tion.
11.11. Fundamental taxotones.
/ 5/ = ( 5)
/ / = ( 4)
/ / = ( [)
/ 6/ = ( 7)
/6 6/ = (0 7)
11. chinese 365
In fact, toneme 3, ([) = (), // >5 shortens (even as its segmental duration does),
when it occurs before tonemes which are dierent from itself: (w5X) /hwo-
5hX/ hu>e (t*4y) /tXy/ dy (kA7) /khao6/ ko on the other
hand, when it occurs before itself, / / >5 5 it has the taxotone (): (U[pjAA)
/oupjao/ ubio
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 dier-
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,7ja) /6fa6qja/ f-
ji 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 international accent {which is a simplication}).
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 [) = (5 [)
/ / = (4 [) = (4 [)
/ / = ( [) = ( [)
366 a handbook of pronunciation
11.3.3.3. As far as sequences of three tonemes are concerned, there are some
dierences (and they must be indicated and applied too) in comparison with tone-
mic (and graphemic) representations. 11.12 shows what happens to sequences
of /5 / followed by one of the other four tonemes: (5tU,^nM5fx,) /5tunan5fX/
donn-fe (5s~^~jE~4i) /5sannjanqi/ sannin-j (5jEN^~[jaa) /5sjan>Xn-
qja/ xianrn-ji (5i^U,7) /5sihu6/ xih.
11.13 shows sequences of / / followed by one of the other four tonemes:
(4mI^laM5fa,) /meilan5fa/ Mi Lnfa (4^mI4l) /haemeilae/ hi mi
li (4w~^hn[tUU) /wanqhantu/ wnqun d (4Y~^ja,7jEc) /syn-
ja6qjan/ xny-jin
nally, 11.14 shows the behavior of the sequence / / preceded by one of the
rst three tonemes: (5s~^jEn[j<<Y) /5sanjanqjX>/ Sanyn Jir (4N^u-
[pjAA) /hanupjao/ hnbio (w^j[jU) /wojejou/ w y yu
/5 / (5 &) / / (4 &) / / ( ) /6 / (7 )
/5 / (5 2 ) / / (4 2 ) / / ( 1 ) /6 / (7 3 )
11. chinese 367
/5 6/ (5 1 7) / 6/ (4 1 7) / 5/ ( 2 5) / / ( 2 4) / / ( 2 7)
Intonation
/ / ( ) (( )) /./ (13)
/ / ( ) (( )) /?/ (31)
/ / ( ) (( )) // (^)
/ /( ) (( )) /,/ (2)
368 a handbook of pronunciation
11.3.4.2. Besides isolated 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), 11.19 also shows the taxotones (given in grey) of the four
marked tonemes, as they are modied 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, 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 isolated 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 rst as a simple identication step, then as active drills as well).
Undoubtedly, they require a good deal of commitment and constancy. But any
eort will certainly be rewarded. Of course, paraphonics further modies things
depending on attititudes, feelings, .
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:
/./: (5ta0 4ji2gX X7c2dX 5ku&~ja,13) /5tha jikX hX6antX 5kunja./ Ta yge
hn de gunia.
/?/: (5ta0 4ji2gX X7c2dX 5ku&~ja,2ma31) /5tha jikX hX6antX 5kunjama?/
Ta yge hn de gunia ma?
//: (5ta0 4ji2gX X7c2dX 5ku&~ja,^ 7t3 w1bu 7tha13) /5tha jikX hX6an-
tX 5kunja 6tan wopu 6aetha./ Ta yge hn de gunia, dni w b i ta.
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 dierent
normalized versions. We start with the (neutral) Chinese pronunciation of (neu-
tral British) English this is the rst step of the phonetic method (the written text
is given in 2.5.2.0). e Chinese translation follows, in its neutral and interna-
tional versions.
Please, note the creaky vowels before pauses ( with creaky phonation type, or
laryngealization, (); 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 authenticity, 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 # @.
At the end, as always, there is the version which gives the English pronuncia-
tion of Chinese, by neutral British speakers, uent 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 rst.
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 0.12).
370 a handbook of pronunciation
Mandarin text
A short appendix
Vowels
12.1.1. Japanese has only ve vowels, which can be distinctively short or long
(or rather doubled, being realized almost as monotimbric diphthongs), and they can
combine into dierent kinds of sequences (as happens in a slow pronunciation of
Spanish in Saavedra or La Habana (aa'B;Ra, laa'Ba;na). However, too often
Japanese pronunciation is hastily described as having the vowels of Spanish and
the consonants of English. We will see that this is not the case at all.
12.1 shows the actual articulations of the ve vowels: (i, , , , M ii, , aa,
, MM) /i, e, a, o, M ii, ee, aa, oo, MM/ i e a o u ii ee aa oo uu Since long
vowels are phonemic sequences (and phonetic geminations), the marker of (short)
/a/ () 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 dierent thing.
Let us now see some examples for each phoneme: (3i') /ie ie (i2k) /gi-
koo/ ginkoo (3s'6i3ii) /sabisii/ sabiii (i2P) /eie/ eien ('3) /eao/ egao
(3k'm2t) /kaRate/ karate (i) /ai/ ai (Q2:i) /zassi/ zasi (3m't) /mato/ ma-
to ('n3k) /neko/ neko (3'i3mM) /okiRM/ okiru (3't2k) /otoko otoko (3sM'mi)
/sMmi sumi (2kM3m) /akMma/ akuma (3mM'i) /mMsi mui.
12. japanese 375
/a/ ()
/aa/ (aa)
12.1.3. Besides, neutral Japanese has the peculiarity of presenting vowel devoic-
ing. Indeed, there are two partially dierent degrees.
e rst 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 accented morae, after which pitch becomes low), and never in in-
terrogative post-intonemes (which shows a raising of the basic pitch): (3i'i) /kisi
kii (3k2i'6i2mM) /kMtibiRM/ ku>ibiru (3h'n2i) /hanasi hanai (3i't3q)
/hitotM/ 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 nd (, ), especially in the
rst syllable of words, when followed by syllables containing the same vowel (more
rarely /e/ () is devoiced too): (3k't2n) /katana katana (3t'k2m) /tokoRo/ toko-
ro (3k's2ni) /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 (=|). erefore, before a
pause, the last vocoid is (, , , , ) (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 rst part is voiced,
whereas the second one is voiceless: (()) 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 eect is almost that of a very short (h), ((h)) or a semiapproxi-
mant ((h)).
For emphasis, strictly speaking (or for other paraphonic implications connect-
376 a handbook of pronunciation
ed with states of mind), often an actual (h|) sequence may be heard: (s3d3s-
kh2) /soodesMka,/ 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'm2n) /nimono/ nimono and // () g (ve-
lar): (3'3k) /aakM/ agaku.
But we must add at once that, within words, // () g (which is sometimes tran-
sliterated as ) can be systematic only in the most neutral type of pronunciation,
after /, /. In fact, nowadays the oscillation between /g / (g ) is very
widespread, but with much uctuation 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
12.2.1.2. An even bigger phonic problem regarding nasals arises from a fourth
Japanese nasal phoneme, moraic n, // (sometimes transliterated as ] 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, h/ (s , z B
j, h F ), (2) before a vowel (which is heterosyllabic, of course), and (3) in
nal position before a pause (or, again, before continuous consonants, or vowels)\
(3P'sP) /esa/ ensan (h2j) /hasja/ hana (MN2jM) /MjM/ un'yu (kP-
2) /kaa/ kanwa (3ssP3i) /seesahi/ seisanhi (sP3j3k) /sehjakM/ sen-
hyaku (3g'sP3F) /gosehM/ gosenfu (tP3i) /tai/ tan'i
However, by assimilation, // has other realizations too. In fact, it is articulated
as a homorganic nasal (to a following consonant and is always intense): (1) (,
, ), before the correspondent stops /p, b t, d/ (p, b t, d) and /k, g/ (k, g, ):
(s3p) /sapo/ sanpo (i2bMP) /sibM/ inbun (M2tP) /Mte/ unten
(tN3i) ((-3i)) /teki/ tenki (t3k) /taka/ tanka and (2) (, ) before dental
or bilabialized prepalatal stopstrictive taxophones of /t, z/ (q, Q) and (, ):
(3bQi) /bazai/ banzai (2) /etjoo/ en>oo (~i2iP) /nizi/ nin-
jin and, naturally, () before the frequent stop(semi){con}strictive variant, (w,
), of /ka/ (k) ka\ (t3w, -3) /taka/ tanka.
In addition, we have (3) (, , , ) before nasals, /m, n, / (m n g) and again
() before r /R/ (), : (M3m) /Mmee/ unmei (3ni) /anai/ annai (~i-
2~i2k) /ninikM/ ninniku (2i, -3i) /aai, a-/ angai and (3'2k[]3i)
/aRakMsi/ anrakui
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 // and /i{i},
a{a}/): (3k'pi, 3w-, 3-, -'phi) /kapai/ kanpai ('t3k 'th-) /tako/ tako (3ip-
'p:M) /kippM/ kippu (3't -'th) /ata wata (k3p) /kepoo/ kenpoo
(word-initial /p/ only occurs in loanwords and onomatopeic forms). In emphatic
speech, there is more aspiration.
e greatest oddity regards /t/ (t) which, by assimilation, before /i, j/ (i, j) is
realized as a bilabialized prepalatal stopstrictive () (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 dierent from ()): (3i'i) /titi >i>i
(2t) /tjato/ >anto In addition and even more strangely /t/ (t), before
/M/ (M) is realized as a dental stopstrictive (q) (transliterated _): (3qM'Ni) ((-i))
/tMi/ _ugi
Let us observe, once and for all, that the palatal realizations of /k, g, /, (,
, N), are instead postpalatal, or retracted palatal, and can be represented better
with special symbols, ((, 8, )). ey may safely be used (although, more often,
(, , N) are used), after stating that they are realized in the rear part of their artic-
ulatory space, we dene postpalatal, which in any case remains distinct from the
prevelar articulation, (, , ), typical of most languages before front vowels and
(j), as in English (('hI;i, '>jl)) {Am. Engl. (('<jl))} (('s)) /'kIi, '<Egjl, 'sI-
I/ key regular singing.
us, we have: ((3ip'p:M, 3mi'8i, 3qM'i)) /kippM, migi, tMi/ kippu migi _ugi
or (3ip'p:M, 3mi'i, 3qM'Ni), provided (Ni) remains dierent from (~i) (therefore,
/ni/ (~i) should not be rendered as if it were (Ni), as we nd 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) /sMbeRM/ suberu (bi2ki) /baikai/ baikai (jM3bi) /zjMbi/ 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 /zj/ (j, Bj)
and zu /zM/ (QM, zM): (di3BiP -3iP) /daizi/ daijin (3jM'zM -'QM) /zjMzM
juzu Its normal distribution, non-emphatic and non-slow, is (B[j] |[j], [j]).
Also /g/ (g) g poses no problems, apart from a complementary (or alternative)
distribution with // () [g which sometimes is transliterated as as already said):
('g3g3k) /gogakM/ gogaku
Word-initial vowels, both at the beginning or in the middle of phrases and sen-
tences, are generally preceded by (), 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-nal vowels, mostly with a suspensive intoneme, can be followed by ([])
( with or without {an audible} release) as an alternative pronunciation instead of
a possible more normal partial devoicing of the last vocoid: ('[]3k|, '[]3k[]|)
/oke/ oke (3[]'sM|, 3[]'sM[]|) /asM 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 /j/ (j) (tran-
sliterated as V although /j/ (j) does not disappear at all): (3j'iP) /sjasi/ a-
in
e corresponding voiced sound, /z/ z is (z) (slow careful: (Q)) and (|Q,
Q). 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 //, it is re-
alized as a stopstrictive: (3k'z) /kaze/ kaze (slow careful: (3k'Q)), (3QM'P)
/zMa/ zuan (s3Q) /sezo/ senzo
In addition, we nd /z/, before i /i/ (i) (transliterated as j instead of z] and be-
fore yV /j/ (j) (transliterated as jV although /j/ (j) remains), which is realized as
prepalatal rounded (again, with reduced rounding) (Bi, Bj) (slow careful:
(i, j)) and (|i, |j i, j): ('FM3Bi) /hMzi/ fuji (3i'tP) /zite/
jiten (k2i) /kazi/ kanji
Approximants
12.2.4.1. Japanese has three approximants. e rst, /j/ (j) y is (voiced) palatal:
(3s2j'n3m, -2m, 3s2j'n3m) /sajonaRa, -Ra, sajoonaRa/ sayonara (-yoo-) (3j2i'i,
3j:'i) /jasiki yaiki (3mi'2k) /mijako/ miyako (j2MM) /jookjMM/ yoo-
kyuu (jMM3i) /sjMMsi/ uui
As can be seen from the examples, it remains unchanged in word-initial posi-
tion /j/ (j), whereas it is realized as a phonetic zero, (`), when preceded by
i /i/: /ij/ = (i). In non-slow speech, the same is possible for /ej/ = (), or ();
thus we will mark () (by using a palatal semi-approximant symbol): (3h') /he-
ja heya.
After the voiceless stop phonemes /pj, tj, kj/, /j/ is devoiced, (): (p, , );
but it remains (j) after other consonants (even if voiceless, /s, h/ (, ) s h] and
without being absorbed by /t, s, z, h/: (, j, Bj, j, j) (in spite of translitera-
tions such as >, , j). Sequences such as *yi, *ye do not occur.
Trills
12.2.5. Japanese has one phoneme of the trill type, which is similar to Spanish
r /R/ (R), as in interpretar /inteRpRe'taR/ (in&teRpRe'taR). A realization like this could
be sucient 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 rst taxophone of /R/ r is (m), lateral ap (or lateralized ap 10.13 in
NPT/HPh, besides 1.14.2-3 in this handbook), which occurs after vowels (even
within sentences): (3k'2m) /kaaRa/ kawara (3'mi) /eRi eri (the same phone al-
so occurs in American English before (), ('bm) /'bE/ better] A non-lateralized
ap is possible too, ([): (3k'2[) /kaaRa/ kawara (3'[i) /eRi eri (which is the
main American phone, in all contexts without (), ('b[i) /'bEi/ 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, 1.15.3),
which occurs after // (even within sentences) or after pauses: (b2i) /beRi/ ben-
ri (3M'i2Bi, -2i) /RMizi/ ruiji (32i'i, -:'i) /Rekisi/ rekii (3i) /RoRi/ ronri
For the sake of simplicity, we could say that the dierence between the two reali-
zations consists in dierent degrees of lateralization. In fact, (m) is less lateralized,
as lateralization is an added, or secondary, component (let us say 1); whereas, for
() lateralization is prevailing, or primary (let us say 2).
Occasionally, lateral realizations can be heard as well, (l) (which is still alveo-
lar), and postalveolar ones (of various manners of articulation respectively: ap,
tap, stop, apped lateral, lateral: (, e, , , $). Obviously, these phones need not
be actively acquired it is sucient 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 (i, i, i) /ti, ki, gi/ >i, ki gi (i Bi, |i,
i i) /si zi hi/ i ji hi\ (~i2jMM) /nizjMM/ ninjuu (3k'Ni2mM) /kaiRM/
kagiru (3i') /tie/ >ie (3i'2mM) /kieRM/ kieru (3i'2k) /gikM/ giwaku (3i-
'm) /sim ima (3'Bi) /azi/ aji (3i'mi) /zimi jimi ('m3i) /mahi/ mahi
In addition, we nd: (~j, Nj) /nj, j/ ny, gy (, , j) /tj, kj, gj/ >, ky gy (j
Bj, |j, j j) /sj zj hj/ j hy\ (~jMM22k) /njMMakM/ nyuugaku (N-
2Nj) /aja/ angya (MM3) /tjMMtjo/ >uu>o (3MMMM3j) /kjMM-
kjMMsja/ kyuukyuua (jMM2~jMM) /gjMMnjMM/ gyuunyuu (3i~jMM3s)
/sinjMMsee/ innyuusei (kN2jMM) /kajMM/ kan'yuu (3id3j) /zidoos-
ja/ jidooa (3j'kM) /hjakM hyaku
Again, we must remember that, for /k, g, / + /i, j/, the actual articulation is
postpalatal ((, 8, )) (rather than fully palatal, (, , N)), and that (j) remains.
cess). So, we regularly have: (mi, |i, i) /Ri/ ri (mi, pi) /mi, pi/ mi, pi (bi, 6i)
/bi/ bi. us: (326i'2mM) /obieRM/ obieru (3mi'n2mi) /minoRi/ minori (3i's2k) /Ri-
sokM/ risoku (i3i) /siRi/ inri (2pi2q) /epitM/ enpi_u
In addition, /0j/ (0j) CyV, ChV\ (mj, |j, j) /Rj/ ry (mj, p) /mj, pj/ my py
(bj, 6j) /bj/ by: (3j) /eRjo/ enryo (jMM) /RjMM/ ryuu (3p'p:2k) /Roppja-
kM/ roppyaku (bj) /bjoo/ byoo (bM2mj2k) /bMmjakM/ bunmyaku
(3Qimj) /zaiRjoo/ 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: (2kMm36M) /kMRa-
bM/ club (3s[]tmi3[i]) /sMtoRaiki/ (workers') strike (3s[]tmi3k[]) /sM-
toRaikM/ (baseball) strike. 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: (paa3ti)
/paat-i/ party (3di'm3k[]3taa) /d-iRekMtaa/ director ([]3i) /tjezi/
change ([j]t3t:) /djetto/ jet ('[j]3F[]) /sjehM/ chef (3FM'i2mM2mM,
-3mM3mM, 'i-, 'Fi-, 'Fi-) /hMiRMmM, h[M]iRMmM/ lm (i3F3~ii) /sihMo-
nii/ symphony (3kq3n) /katMoone/ canzone.
Some examples clearly show that, besides placing some phones into new combi-
nations, certain sequences are slightly denipponized (as the possible dropping,
in these words but not in genuine ones, of (j) after prepalatal {rounded} articula-
tions), becoming slightly more international.
Structures
Taxophonics
12.3.1.1. e basic things have already been said. In fact, we have seen devoiced
vocoids ( 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 mora coincides with a light syllable, as the one
formed by a short vowel (/i, e, a, o, M/), or by //, or else by the rst element of a
geminate consonant (/-0-0/ (0-0:)).
A half-heavy syllable corresponds to a geminate vowel (/ii, ee, aa, oo, MM/) or a
diphthong, or to a short vowel + // (//) or + the rst 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) + // (//)
or + the rst element of a geminate consonant: /0[-0]/.
As we have seen from various previous examples, in Japanese, vowel length is dis-
tinctive, short and long or rather geminate (or doubled) vowels: (s) /so, so
so (one mora), (s) /soo, soo/ soo (two morae); ('t3i) /toki/ toki, (t3i) /too-
ki/ tooki. Even consonant length is distinctively present, as in (3i't) /kite/ (from ki-
ru /kiRM/ to wear) and /kite kite (from kuru /kMRM/ to come), both with two
morae; (3it't:) /kitte[]/ (stamp) and ('it3t:, 3it't:) /kitte kitte (from kiru /ki-
RM/ to cut), both with three morae: /ki-t-te/; (3g'k) /gaka/ gaka artist, (3gk-
'k:) /gakka/ gakka lesson.
From a phonetic point of view, a doubled consonant consists of two morae: the
rst one coincides with the rst 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 (3st't:) sotto softly (while a similar Italian word sotto /'sotto/ under
is ('sot:to) in an intoneme, or ('sotto) in a preintoneme). Let us observe well and
listen even more carefully to the dierence 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.
Moraic 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 //
(P-), not with VnV /n/ (-n), and the same goes for Vn'yV /j/ (P-j),
which is dierent from VnyV /nj/ (-~j)): (t3d) /tedoo/ tendoo (, a sim-
ilar Italian word, tendo /'tEndo/, or ('tEn:do) in an intoneme).
However, in Japanese, in addition to // (P-) and /n/ (-n), as in (P3i)
/ai/ an'i ('3~i) /ani/ ani, we can also have // (-) (which is the combina-
tion of // and //): (N2Ni) /ei/ engi and /n/ (-n) too (combination
of // and /n/): (2n) /ana/ anna (3~i) /ani/ anni ( Italian: /'anna, 'an-
ni/ ('an{:}na, 'an{:}ni) Anna, anni] Let us add this not useless example: (hP3j)
/ho-ja/ hon'ya
erefore, the structure of // (, , , N, , P) (one mora) is dierent from the
geminate one /0-0/ (two morae, or three including the rst vowel: /-0-0/).
However, there is no dierence for the counting of morae.
12. japanese 383
12.3.2.2. Only in the following table (which relates to 12.3), we will show a
tonemic () and a tonetic pattern (, where (3) indicates a low-pitched syllable,
with the vowel timbre of /o/; whereas (2) 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 () with typic-
ally oriental graphic complexities. We also show a phonotonetic pattern turned
into a more orthodox one (), originating from pattern (). To indicate any mo-
ra, here we use (), /o/, o $ (the last one to indicate katakana, in ):
12.3.2.3. In type- tonetic transcriptions, the rst instance (/oooo/) diers from
the second (/oooo), 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 [] 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 reect the rst ones in a more abstract way:
tonemic and tonetic (, ); even the katakana transliteration or transcription
() shows the same characteristic, but in a more abstract way in comparison with
real transcriptions (, , ).
We do not use transcriptions of the type /oooo/ (^), 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 nd both types and with katakana, and type with transcriptions.
As far as the indication or pitch variants is concerned, as we have done in
12.0.2, for hiragana ((3i'm2n, 3i'm3n 3i'm2&n) /hiRaana -na) and ka-
takana ((3k't3kn, 3k't2kn) /katakana/), the most important thing is to in-
dicate them (unless there are usage dierences). Our phonotonetic transcription
also shows their (wider or narrower) diusion and recommendability (which is to
be interpreted gradually from the rst onwards).
On the other hand, in a pronouncing dictionary, which would be worth publish-
ing (using a transliteration, followed by the ocial writing and, of course, by an
IPA transcription, certainly a phonemic one), preferences could and should be
shown, by indicating /hiRaana, hiRaana, hiRaana and /katakana, katakana/,
obviously in a shortened form: /hiRaana, -Ra- -na and /katakana, -ka-/. 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, .
12.3.2.4. A short mono-moraic syllable may have two pitch possibilities: /ne/
ne sound, tone (absence of akusento), /ne ne root (presence of akusento), but
tonetically they are both non-low ( said in a mid pitch: ('n); the same goes for
/ki/ ki spirit, /ki ki tree, ('i), and /ha/ ha leave, /ha ha tooth, ('h).
In the case of two morae, we can have (s) /soo/ soo monk, (s) /soo/ soo
villa, inn, which are monosyllables realized right as (s), with half-low pitch
but slightly falling or rising, respectively, since they combine mid and low, or low
and mid, pitch ( 12.4).
12. japanese 385
/, / =
(, P), (3, 3P)
/, / =
386 a handbook of pronunciation
Stress in Japanese
12.3.2.6. True problems begin with bisyllables, though. In fact, there are dier-
ences between (3'm) /ame/ ame candy, ('3m) /ame/ ame rain, and (3n'mM)
/neRM/ neru to sleep, ('n3mM) /neRM/ neru annel. Bisyllables of two morae,
that is with two light syllables, are stressed on the second syllable, unless akusento
follows the rst one, which is then stressed: (3k'k) /koko/ koko (3'Bi) /azi/ aji
(3M') /Me/ ue (3i'M) /iM/ iu (3'i) /oi/ oi and (3't) /oto oto (3'i) /asi ai
(3mM'm) /mMRa mura (3qM'Ni) /tMi _ugi (3i') /sio io (3i') /ie ie How-
ever, we have: ('d3m) /doRe/ dore ('3i) /aki/ aki ('qM3m) /tMma/ _uma .
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 dierences (which are pitch dierences), in
phonemic transcriptions, since sometimes they are the only actual dierences (but
12.4 must be carefully considered): (3'mi) /omoi/ omoi (3mi) /omoi/ omoi
(3k's) /kasoo/ kasoo (3ks) /kasoo/ kasoo (3i'mi) /iRai/ irai (2imi) /iRai/ irai
(3'i) /aoi/ aoi (3i) /aoi/ aoi (3i'i) /siai/ iai (3i'n) /kinoo/ kinoo (3j't)
/jotee/ yotei (2ii) /hiai/ higai (but ('i3Nj) /kijoo/ kigyoo] (2kMm) /kM-
Roo/ kuroo (2bm) /baRee/ baree
More: (3iki) /kikai/ kikai (3ikP) /sike/ iken (2ikP) /ike/ iken (3i'kP)
/ike/ iken (2gzP) /goze/ gozen (2QM6P) /zMbo/ zubon (k2Bi) /koozi/
kooji (k3Bi) /koozi/ kooji (N2Ni) /ei/ engi (i2d) /aida/ aida (hi3mM)
/haiRM/ hairu (3t) /kjooto/ Kyooto (ki2Ni) /kaii kaigi (di3k) /dai-
12. japanese 387
kM/ daiku (b3i) /beRi/ benri (mi2n) /mina/ minna (3miq'q:M) /miqqM
mit_u (3k'k:) /akka/ akka (but: ('k3k:) /akka/ akka ('3:j) /Ressja/ resa
where pitch prevails on other factors).
Generally, in four-mora bisyllables, stress falls on the rst syllables, unless it is a
light one ( with just one mora) or there is an akusento after the second one (or if
the rst is only half-heavy, checked by (==:)): (jMM26j) /zjMMbjoo/ juu-
byoo (jMM36j) /zjMMbjoo/ juubyoo (h2k) /hookoo/ hookoo (jMM-
3d) /zjMMdoo/ juudoo (k2Bj) /koozjoo/ koojoo (3kBj) /koozjoo/ koo-
joo (3tiFMM) /taihMM/ taifuu
More: (k2t) /ketoo/ kentoo (3kt) /ketoo/ kentoo (s2t) /se-
too/ sentoo (s3t) /setoo/ sentoo (3sPs) /sesee/ sensei (iP2P) /kie/
kin'en (jMM26iP) /jMMbi/ yuubin (3gk'k:) /gakkoo/ gakkoo (3t':MM) /tok-
kjMM/ tokkyuu (but: (b3:P) /bottja/ boc>an] (3'iiP) /ekii/ ekiin (3ip-
'p:i, 3i-) /sippai/ ippai (t2) /tookjoo/ Tookyoo
12.3.2.9. In ve-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: (3k&t2zM'k3mM) /katazMkeRM/ katazukeru (3i&n26ik3mi) /inabi-
kaRi/ inabikari (3&t2kn3k) /otokonoko/ otokonoko (3n&Bi2m3i) /nezi-
maasi/ nejimawai (3j&m2n63mi) /jamanoboRi/ yamanobori (3&kM2mi'm-
2n) /okMRimono/ okurimono (3&2kM's2m, 22k-) /okjakMsama/ okyaku-
sama (3&t3qMdi3sP) /otetMdaisa/ ote_udaisan (3&m2mi'k3BiP, -3iP) /ameRi-
ka zi/ amerikajin
More: (3s&t2kM'm2n) /setakMmono/ sentakumono (3i&i2~i'i2BjMM) /iti-
nitizjMM/ i>ini>ijuu (dN2i2s't36M) /dekisMtoobM/ denki-sutoobu
(MM22kMmj3mi) /tjMMokMRjooRi/ >uugokuryoori (k2djMM3t-
3k) /koodazjMMtakM/ koodan-juutaku (d2b's3qM36i, d2b 's-) /da-
boosetMbi/ danboo-se_ubi (2m'2mi&sP) /omaaRisa/ omawarisan (3~jMM-
&2kM2ikP) /njMMakMsike/ nyuugaku-iken (m2ii2i3qM, -i3q) /ma-
tiaisitM/ ma>iaii_u
12.3.2.10. In verbs in -ru generally stress is on the preceding syllable and akusen-
to immediately after it: (26'3mM) /oboeRM/ oboeru (k2'3mM) /kaaeRM/
kangaeru (2qM'm3mM) /atMmaRM/ a_umaru (i2m'63mM) /siRabeRM/ 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 reection (also
on the paramount importance that a real pronouncing dictionary would have {by
using IPA} for akusento stress, segments, devoicing, ): (3FMjM'3[i]3i) /hMjM-
esiki/ huyugeiki from (3FM'jM) /hMjM huyu ('k3[i]3i) /kesiki/ keiki
Other examples: (3jM&~jMM2kM'd3m3n) /jMnjMMkMdamono/ yunyuukuda-
mono from (3jM'~jMM) /jMnjMM/ yunyuu (3kM'd3m3n) /kMdamono/ kudamo-
no and more: (jMM2j2k'z3m) /jMMjakezoRa/ yuuyakezora from (jMM2j2k)
/jMMjake/ yuuyake ('s3m) /soRa/ sora (3s&t2k3m) /satookoRo/ satogokoro
from (3s't) /sato/ sato (3k'k3m) /kokoRo/ kokoro.
Furthermore: (3~i'M3mM3m) /niMRMma/ niguruma from ('~i) /ni ni (3kM-
'mM2m) /kMRMma/ kuruma (3's3k3z) /asakaze/ asakase from ('3s) /asa/ asa
(3k'z) /kaze/ kaze (3iP2s[]&t2tk3ii) /isMtatokoohii/ insutantokoohi
from (3iP2s[]t3t) /isMtato/ insutanto (k3ii) /koohii/ koohi (gi2k'kM-
3BiP, -3iP) /gaikokMzi/ gaikokujin from (gi2k2k) /gaikokM/ gaikoku
(iP) /zi/ jin (gi2k&kM2Bit3m3k, -2i-) /gaikokMzitooRokM/ gaiko-
kujin-tooroku from (gi2k'kM3BiP, -3iP) /gaikokMzi/ gaikokujin (t2m2k)
/tooRokM/ tooroku
/jasasii ho/ yasaii hon (without modications), (ii 3Bi6i3i, 3i-) /ii zibiki/ ii
jibiki (from (3i'6i2i) /zibiki, with lowering of -biki] (3sM'zM3ii 3h) /sMzMii
heja suzuii heya (in (3h') /heja, -ya is lowered), (3sPsd3s, -s:, -s) /sesee
desM/ sensei desu (with stress reduction on ('d3s, -s:, -s) /desM/ and lowering of
de-]
More: (3t't2m 2'm2imi hP) /totemo omosiRoi ho/ totemo omoiroi
hon (from (3'm2imi), with raising of o- and lowering of hon] (3i'Bj2~i 2F-
&kM2z'qM2n 'm2di) /hizjooni hMkMzatMna modai/ hijoo ni fukuza_una
mondai (with raising of fu- and mon-] ('mt3t: 3jsMi hP) /motto jasMi ho/
motto yasui hon (with lowering of -sui and hon] (3'h2j 2g'zi2m3s, -s:, -s) /o-
hajoo gozaimasM/ ohayoo gozaimasu (with raising of go-]
Some others: (3'mi33t 3gzi3s, -s:, -s) /aRiatoo gozaimasM/ arigatoo gozai-
masu (with lowering of -zaima-\ everything is low, except -ri-] (d3m 3mi33t
3gzi3m3s, -s:, -s) /doomo aRiatoo gozaimasM/ doomo arigatoo gozaimasu
(with lowering of -ri- and -zaima-\ everything is low, except doo- which is half-low).
In addition: (M2i 'k2m '3i m3d| 3&mM2i'm3s, -s:, -s) (in the case of fast
speech with no breaks: (M2i 'k2m '3i m3d 3mM3im3s, -s:, -s); but, separate-
ly, word by word: (3M'i, 3k'm, '3i, 'm3d, 3&mMi'm3s, -s:, -s)} /Mti kaRa eki ma-
de aRMki masM/ u>i kara eki made arukimasu (please, note the stress change in the
rst rhythm groups, mainly due to /i/ devoicing).
Also: ('~i3Bi&k3m\ s3i m3d| 3i'mM3j3~i\ 3im3s, -s:, -s) possibly with
fewer breaks, in less slow speech: ('~i3Bik3m s3i m3d| 3i'mM3j3~i 3im-
3s, -s:, -s). On the other hand, in separate rhythm groups, we have: ('~i3Bi, 3k'm,
s3i, 'm3d| 3i'mM3j3~i, 3i'm3s, -s:, -s) (with modications on kara] /nizi
kaRa sazi made zimMsjoni imasM/ niji kara sanji made jimuo ni imasu ('j3Bi
3md3~i\ 3k'k2~i 2i't 2kM2dsi) (slower: ('j3Bi 3md3~i\ 3k'k2~i\ 2i't 2kM2d-
si); on the other hand, in separate rhythm groups, we have: ('j3Bi, 3md3~i, 3k-
'k~i, 3i't, 3kM2dsi), with stress modications on made ni in comparison with
made] /jozi madeni kokoni kite kMdasai/ yoji made ni koko ni kite kudasai
-be- may give a similar prominence eect. 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: (3i, 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: (k3ii) /koohii/ koohii (coee), (2h-
t3mM) /hoteRM/ hoteru (hotel).
Intonation
/ / ( ) (( )) /./ (13)
/ / ( ) (( )) /?/ (31)
/ / ( ) (( )) // (^)
/ / ( ) (( )) /,/ (2)
392 a handbook of pronunciation
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 (k) /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:
/./: (3&k2mi2msd3i3t13) /akaRimase desita./ Wakarimasen desita.
/?/: (3&k2mi'm3i3t3k31) /akaRimasitaka?/ Wakarimasita ka?
//: (3FM'jM dt3t: 3km3d^ 3i2i'm3i3t13) /hMjM datta keRedo ikimasita./ Fu-
yu datta keredo, ikimasita.
Typically, Japanese has a peculiar paraphonic setting, with lowered larynx @,
mostly for men.
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 dierent
normalized versions. We start with the (neutral) Japanese pronunciation of (neu-
tral British) English this is the rst 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 reections.
At the end, as always, there is the version which gives the English pronuncia-
tion of Japanese, by neutral British speakers, uent 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 rst.
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 0.12).
Japanese text
Phonotonemic transcription
Phonotonetic transcription
Vowels
13.1 Esperanto ( E-o] presents only the ve most normal and natural vowel
phonemes (which are the most widespread in the various languages of the world):
(i, , a, , u) /i, e, a, o, u/ ( 13.1). e ve E-o vowels are realized very much
like the ve (stressed) vowels of Spanish, as in: ('iR, 'tRs, 'maR, 'ds, 'tu) /'iR, 'tRes,
'maR, 'ds, 'tu/ ir tres mar dos t. In the most international pronunciation of Es-
peranto, e o always have the timbres (, ), both in stressed and unstressed sylla-
bles. Here are some examples: ('tRiki, 'b-l, 'ka;Ra, 'd;m, 'u;nu) /'trinki, 'eble, 'ka-
ra, 'domo, 'unu/ trinki eble kara domo unu to drink, perhaps, dear, house, one.
e ocial Esperanto diphthongs are of the (i, u) /i, u/ 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 ( 3.1.2-3, about Italian
diphthongs, and NPT/HPh: 1.16 2.10-1), as, for instance, /eo, io/ in (&ggRa-
'fi) /geogra'fio/ geografio geography. Given the nature of Esperanto orthogra-
phy, a more logical solution Vi, Vu 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): ('m;n, 'fndi) /'mo-
no, 'fendi/ mono fendi money, to split. ile /m/ never assimilates: (m'fa;z,
&mmkn'sRv) /em'fazo, memkon'servo/ emfazo memkonservo emphasis, self-
conservation, there are various possible assimilations for /n/ (M, ~, ), which ren-
der the pronunciation more uent and natural: (iM'f;Ra, 'ma~Gi, si~'j;R, 'SRa-
k) /in'fera, 'manGi, sin'joro, 'Sranko/ infera mangi sinjoro sranko infernal, to
eat, gentleman, cupboard (including (n, ~) ((, n)): ('sndi, 'ma~Gi) /'sendi,
'manGi/ sendi mangi to send, to eat).
However, between lexemes (including prexes) no assimilation takes place (on
the contrary, we also nd secondary stresses, even on syllables contiguous to a
stressed one): (&kun'm;ti, &kun'v;ni, &sn'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:
('hnts) /'hInts/ hints); then, there are (c) c and its voiced counterpart, (G) g (as in
English: ('fc, 'G;) /'fEc, 'GOE/ fetch joy).
ey always maintain the values we have seen, before any vowel or consonant:
('qnt, 'pa;q) /'qent, 'paqo/ cent paco hundred, peace, (c'va;l, f'li;ca) /ce'va-
lo, fe'lica/ cevalo felica horse, happy, (GaR'd;n, 'pa;G) /Gar'deno, 'paGo/ gar-
deno pago garden, page (of book). 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 zero]\ ('dz) /'edzo/ edzo husband.
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, 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 ('js, 'wn:) /'jEs, 'wIn/ yes win\ ('js, 'v;j, si~'j;R,
'wst) /'jes, 'vojo, sin'joro, 'westo/ jes vojo sinjoro esto yes, road, gentleman,
west.
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, 'Raiti, 'nau, 'laudi)
/'tui, 'raitoi, 'nau, 'laudi/ tuj rajtoj na ladi immediately, rights, nine, praise.
Apart from this last case of greater phonetic naturalness, i and u always need to be
distinguished from j and before vowels: (mi';l) /mi'elo/ mielo honey, ('mj;l)
/'mjelo/ mjelo spinal cord, (fi'ask, &mani';R) /fi'asko, mani'ero/ asko maniero
asco, manner.
Let us also consider the following cases, for which (as will be seen in 13.2.5.2)
j is forced to have two dierent values owing to the stress-assignment rule: ('plid)
/'pleido/ plejdo plaid, but (pl'ja;do) /ple'jado/ Plejado Pleiad.
In lexical derivation, when is followed by a vowel, we have two possibilities,
/w, u/: ('na;wa, 'naua) /'nawa, 'naua/) naa ninth; the same goes for (bal'da;wa,
-aua) /bal'dawa, -aua/) baldaa prompt, from ('baldau) /'baldau/) balda soon.
400 a handbook of pronunciation
13.2.5.3. In addition, even for stressing, let us consider cases (that we present
here in advance) such as: ('js, 'jam) /'jes, 'jam/ jes jam yes, already and ('is, 'iam)
/'ies, 'iam/ ies iam someone's, once, and (s'i;fi, 'sil, ba'lau, 'baldau) /so'ifi, 'soi-
lo ba'lau, 'baldau/ soi sojlo balau! balda to be thirsty, threshold, sweep!,
soon. 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 rst ocial Esperanto publication by Zamenhof appeared in 1887. e Inter-
national Phonetic Association was founded in 1886 and the rst 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;n, 'kRp, 'tR;) /'rano, 'korpo, 'tre/ rano korpo
tre frog, body, very (much). Any other pronunciation of /r/ (although frequent,
because of the dierent national languages used by Esperanto speakers) is not neu-
tral.
Laterals
13.2.7. Lastly, we nd a pure alveolar lateral, (l), with no particular nuances (so
it is dierent from the English {non-prevocalic} l sound as in ('w:{}) well(s)]\
('la;n, mul't;ga) /'lano, mul'tega/ lano multega wool, very very. Not even a
slight palatalization ((), perhaps before /i/) would be acceptable: ('li;gi) /'ligi/ li-
gi to bind 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, 'faci) /'alta, 'falci/ alta falci high, mow.
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 dialect 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-nal) 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 healthy, (the letter) A. In checked stressed syllables (either in word-
-internal or word-nal position), Esperanto vocoids are always short: ('paRt) /'par-
to/ parto part. 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{}, 'w;I 'wIT, 'wID/-D) /'wEI{z}, 'wEId 'wEIt,
'wEId/ way(s) wade wait waider]\ ('uR) /'euro/ ero Euro, ('ti) /'tie/ tie
there.
Consonant clusters
13.3.2. Some Esperantists strive to follow to the letter 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 each letter must be pronounced clearly separated from the neigh-
boring ones, his sole aim was to make people avoid overly marked ethnic pronun-
ciations. Surely, he had no clear intention to lead them to reect 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 dierent current ethnic pronunciations (even by ex-
pert and uent Esperantists): (u'R;p) Eropo *('pP, j'>p) ('laudi) la-
di *('l_i, 'l;DIi, 'lAdi, 'laodi, 'Oudi) ('pa~-j) panjo *('pa;No, 'phni) ('lg)
longe *('l, 'lO, 'lAI) ('vRt) vorto *('vOto, 'v;TU, 'f<to, 'vtP) ('sta;Ri) stari
*('Sta:i, 'sT>i, 'stORi) (la'tmp) la tempo *('tEm:b, l'thEmp{h}P) (in'tnsa)
intensa *(in'dEn:Qa, 't:sa, in'tns) (la'pa;q) la paco *(la'Faq:qP, la'baqqso, l-
'pto) (mi'G;jas) mi gojas *(miG'GOj:jasse, mi'Z:jasse) (b'l;q) ebleco *(eb-
'b]q:q, ji'b;q) (la'ku;b) la kubo *(la'guubbo, la'h:BP) (li'b;lo) libelo *(ib-
'bE;, li'B;lo) ('kRu;c) kruco *('kru:So, k'Rju;Cx) ('pa;Si) pasi *('paS:Si, 'paai) (mi-
'p;tas) mi petas *(mi'bE;Edas:, mi'phe:t{h}as) ('js) yes *(', 'jE).
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 dierences 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 // to a following consonant ( with (mp, Mf,
~c, N, k) ) 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: ('kmbi, km'pR;ni) /'kombi,
kom'preni/ kombi kompreni to comb, to understand, (kM'v;ni, kM'f;si)
/kon'veni, kon'fesi/ konveni konfesi to be suitable, to confess, (kn'du;ki, 'kn-
tRau) /kon'duki, 'kontrau/ konduki kontra to drive, against, (k~'cRt) /kon-
'certo/ koncerto concerto, (k'gR;s, k'ka;va) /kon'greso, kon'kava/ kongreso
konkava congress, concave.
But (&kunm'tb-la, &kun&pRduk'ta;d, &kun&bata'lant) /kunme'tebla, kunpro-
duk'tado, kunbata'lanto/ kunmetebla kunproduktado, kunbatalanto combinable,
coproduction, comrade in arms; in fact, they belong to dierent lexemes, because
they are compounds.
13. esperanto 403
Syllabication
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 I'm your brother's friend would
certainly not be *('mi; 'stas 'la; a'mi;k 'd; 'via 'fRa;t), 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;k
&dvia'fRa;t).
As can be seen from previous transcriptions, even in compounds, we nd dier-
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
renement of using instead the allomorphs with 'st-: (&mistas&ci'ti, &listsf'li;ca)
/mistasci'tie, listosfe'lica/ Mi 'stas ci tie Li 'stos felica I'm here, he'll be happy
(either only when pronouncing or when writing as well by providing a consider-
able visual-mnemonic help). Even the phrase (ti'stas) /tio'estas/ tio estas that is,
when it is not emphasized, surely gains by becoming (&tistas) /'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? To be, or not to be?.
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 (&aSD'Th-
Dm, --), which is quite normal for I should have told them while on the con-
trary something like *(5a; S&hv'Thd&m, --) would be quite unusual in-
13. esperanto 405
13.3.6. Returning to our rst example, we will have the following realization:
(&mistasla'mi;k d&via'fRa;t), which is possible to write as: Mi 'stas l'amiko de via
frato. e elision of la is ocial, 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 she saw him, ('daknal&vi) /'dankonalvi/ dankon al vi
thank you, ('ti&ci) /'tioci/ tio ci this, ('cu;&n21) /'cune?/ cu ne? isn't it?, aren't
you?, doesn't he?, haven't they?
It is worthwhile comparing the following cases: (Si'vi;dis 'lin) /Si'vidis 'lin/ si vidis
lin she saw him, ('dakn al'vi;) /'dankon al'vi/ dankon al vi thank you!, (&ti-
'ci;) /tio'ci/ tio ci this here, (&cu'n;21) /cu'ne?/ cu ne? isn't that so?.
e case of ju (mal)pli, des (mal)pli is very interesting for stressing: ('ju; &pli-
vipa'R;las2 2 'ds &malpli&mikm'pR;nas23) /'ju plivipa'rolas 'des malplimikom'pre-
nas./ ju pli vi parolas, des malpli mi komprenas the more you speak, the less I un-
derstand, ('ju; &pli'mult2 2 'ds &pli'b;n23) /'ju pli'multe 'des pli'bone/ ju pli multe
des pli bone the more, the better.
Intonation
13.3.7. 13.3 shows the preintonemes and intonemes of neutral E-o (free from
any inuence from ethnic languages), so we will see only three fundamental ex-
amples:
/./: (&ili'ft kn'sultas &l{a}sp'RantaM &vRt'a;Rn23) /ili'ofte kon'sultas l{a}espe-
13.3. Esperanto preintonemes and intonemes.
'rantan vort'aron./ Ili ofte konsultas la esperantan vortaron ey often consult the
Esperanto dictionary.
/?/: (&cu&ili'ft kn'sultas &l{a}sp'RantaM &vRt'a;Rn21) /cuili'ofte kon'sultas
l{a}espe'rantan vort'aron./ Cu ili ofte konsultas la esperantan vortaron? Do they of-
ten consult the Esperanto dictionary?.
//: (&ili'ft kn'sultas &l{a}sp'RantaM &vRt'a;Rn2 2| &sd'tu;t &sn'fRukt23) /ili'ofte
kon'sultas l{a}espe'rantan vort'aron| sed'tute sen'frukte./ Ili ofte konsultas la esper-
antan vortaron, sed tute senfrukte ey often consult the Esperanto dictionary,
but that's of no use.
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: , (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 alter the very essence of E-o itself (whereas this is
normal for natural languages)! us some evolution and structural simplication
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 nd 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 diculties (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 dicult 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 26, (&sp'Rant &dudk'ss), 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 simplied, by removing the source of some main problems,
when we consider the dierent 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 sucient 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 eect the for-
mer substitution: (x) = (k), as for kaoso kemio jakto chaos, chemistry, yacht; al-
so the change (Z) = (G) has already been recorded: jazo = gazo jazz. 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 chorus, since we already have horo hour and koro heart,
the new form koruso has been added in order to nally replace it. For holero chol-
era, since kolero anger is already there, the solutions can be: holero or kolro
(26). In a series of oppositions like justa just, fair and gusta right, exact, we
may insert juro law, juro = guro vow, jako jacket, jako = gako jack as well.
As far as jeto jet, geto jetty, jeto throw are concerned, if this last one becomes
('G;t) yeto (26), the second example can safely become (G't) yeteo (26),
approaching the related form (m'l) moleo mole.
A really concrete case is ajo thing (and the corresponding sux aj with an
analogous meaning, which is very widely used) and ago age. Here, so as not to
risk overlapping with the exclamation aj! ow!, ouch!, from which in E-o it is cer-
tainly possible to derive ajo a cry of pain, 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;Z) into ('a;G) [ajo =
ago = ayo 26), and ('a;G) into ('aiG) [ago = ajgo = ajyo 26) or probably, bet-
ter still, into ('iG) [ago = ejgo = ejyo 26).
any consideration about the perversion of the language, since they are not at all
natural, nor necessary).
Obviously, as seen above, would become w (w), ('na;wa, 'naua) [naa = nawa
26), ninth; however g would become y (G), without causing much ado, given
that even in Spanish, in many contexts and variants, for y we nd a very similar
pronunciation, ('pa;G) [pago = payo 26) page (of book). In addition, c would
become q (c), and here again we nd similar pronunciations in Chinese and Al-
banian, (c'va;l) [cevalo = qevalo 26) horse. (On the other hand, does Espe-
ranto not want to be the second language for all 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;S) [so = xo 26) sh.
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;l) celo aim. For in-
stance, is not h a real nothing for most Romance-language speakers: ('ha;vi) havi
to have (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
normalized accents, mostly as useful introductory examples of the phonetic meth-
od of natural phonetics. As always, rst 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 antaeniris 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, nne, 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'nRda 'vnt2 &kaila'su;n2| dis'pu;tis23 'caR2 ciu &lla'du;2 a'sRtis23
&sti&pli'fRta2 &ll[a]a'lia23| n&tium'mnt2 2| &ili&k'vi;dis23 &vja'Gantn23 &kiuan&tawn-
'i;Ris23 vl'vit n&siaman't;l23|| la'du &dispu'tanti2 'd;2 d'qi;dis23 &kkn&sid'Ra;ts2
&pli'fRta2 2|| &tiulla'du;2| &kiusuk'q;ss2 &igila&vja'Gantn2 2| &fR'pRni &siamman't;-
ln23||
la'nRda 'vnt2 2| &k'blvis &tR'fRt23| 'sd2 'ju; &pliGi'bl;vis2 2| 'ds2 &plila&vja'Gan-
t2 &mall'zigis &siamman't;ln23| &kaifiM'fi;n2| la&kmpa'tinda 'vnt2| &dvisR'zig-
ni23|| 'tiam2| la'sun mn'tRi;Gis2 &nlaci';l23| &kaimal'lg 'pst2 2| la&vja'Gant2 al-
&kiu[]stis'vaRm2 2 &fR'pR;nis23 &laman't;ln23|| 'til2| la'nRda 'vnt2 &dvis-ag'ns-
ki23 &kla'su;n2 &stas&pli'fRta23 &lla'du;23||
&cuvi'Sa;tis21 la&hist'Rin2| &cuniRi'p;tu&Gin21|||)
British pronunciation
American pronunciation
Italian pronunciation
13.5.7. ('i;am2 la'nOrda 'vEn:to2 &kaila'su:n2| dis'pu:tis23 'ca:r2 'ciu ella'du2 a'sEr:tis23
&stipli[f]'fOr:ta2 &llaa'li;a23| en&tiumo5men:t12| &iliek'vi:diz23 &voja'Gan:to23 kiu&antawe-
'ni:Riz23 vol'vi:te en&siaman'tE:lo23|| la'du[d] dispu'tantoi2 'dO[d]2 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. ('im2 l'nd 'vntP2 &il'snP2| _isp+is3 3 [T]'S:2 [T]'Si ll'd2 sE-
+is3 3 &s+ipi'ft2 ≪li3 3| 3n&+im5mntP1 1| &ilivi_is3 3 &v,Zntn3 3 &int-
eniis3 3 vl'vite 3n&simntElP3 3|| l'd &_isp'tntPi2 'dP2 d[t]si_is3 3 &ekn&side-
'tPs2 &pi5ft1 1|| &+ill'd2| &isk'sEsPs2 &iilv,5Zntn1 1| f'peni &simntE-
ln3 3||
l'nd 5vntP1 1| k'blvis tfte3 3| 'sEd2 ', &pi[D]Zi5blvis1 1| 'dEs2 &pil&v,'Zn-
tP2 &mllO'ziis &simntEln3 3| &ifin'fine2| l&kp+ind 'vntP2| &devisezigni3 3||
'+im2| l'snP mn'ti[D]Zis2 &3nl[T]SelP3 3| &iml'lNe 5pste1 1| l&v,ZntP2 -
&i'Es+is 5vme1 1 fpenis3 3 &lmntEln3 3|| '+il2| l'nd 'vntP2 &devi-sgns-
i3 3 &el'snP2 &sts&pift3 3 &ll'd3 3||
&[T]SviS+is1 1 l&ist'in2| &[T]S&niipet1 1 '[D]Zin2|||)
German pronunciation
Spanish pronunciation
13.5.10. ('ian2 la'nRa 'Bnto2 &kaila'u;no2| di'pu;ti3 3 'CaR2 'Ciu e{l}la'u2 a'Rti3 3
&ti&pli'fRta2 &{l}la'lia3 3| en&tiumo'mnto1 1| &ilie'Bi;i3 3 &Bo,a',anton3 3 &kiuantawe'ni;-
13. esperanto 413
Brazilian pronunciation
Lusitanian pronunciation
13.5.12. ('i4n2 ]'nOR 'vnt[u]2 &kai]'su;nu2| dSpu;t?S3 3 'Sa2 'Siu e{}]'u2 sEtNs3 3
&StN&p]i'fOta2 &o{}]a]i3 3| en&tiumu'mntu2 2| &i]ikvi;S3 3 &voZAntn3 3 &kiuAntjni;-
RS3 3 vo'vit e&si4m4nte;]3 3|| ]'u &Spu'tAntji2 'O2 si;S3 3 kNk&si'Ra;tuS2 &p]i-
'fOt2 2|| &tiue{}]'u2| &kiusuk'sE;sus2 'ii ]&voZ4nt3 3| fu'pRE;ni &si4m4nte;]3 3||
]'nOR 'vntu2 2| ek'b]OvS &tRfOt3 3| 'sE2 'u &p]iZ'b]O;vS2 2| 'dES2 &p]i]&voZAntu2
&mA{}]u'zi{S} &si4m4nte;]3 3| &kaifi'fi;n[]2| ]&kp'tind 'vnt[u]2| &dev{Z}zi-
ni3 3|| 'ti4| ]'sunu mn'tRi;ZS2 &en]Sje;]u3 3| &kaimA{}']9g 'pOSt[]2 2| ]&voZAnt[u]2
A&kiu'EStZ vaRm2 2 fupRE;nS3 3 &]m4nte;]3 3|| 'ti2| ]'nOR 'vnt[u]2 &viz-
nOSki3 3 &ki]'su;nu2 &StS&p]i fOt3 3 &e{}]u3 3||
&Suvi'Sa;tS21 &]aiStu'Ri2| Su&nii'pe;tu21 'Zi2|||)
414 a handbook of pronunciation
Russian pronunciation
Arabic pronunciation
Hindi pronunciation
Chinese pronunciation
Japanese pronunciation
Utilizable bibliography
Prelude ( 1)
ocial IPA, in transcriptions which are always prosodically decient, whereas in two pages
{arranged in three: 558-60} it tries to give accurate descriptions, which only reveal the
method's heaviness and complexity, with results that cannot be suggested and languages
which are unrecognizable even to native speakers, when one tries to reproduce them as rep-
resented; oIPA).
malmberg, b. (1974) Manuel de phontique gnrale Paris: Picard (IPA).
o'connor, j. d. (1973) Phonetics Harmondsworth: Penguin (IPA).
schubiger, m. (1977) Einfhrung 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 la phontique Paris:
puf (completed by the vinyl record of bouquiaux cloarec-heiss thomas; expanded
IPA).
trubeckoj, n. s. (1939) Grundzge der Phonologie Gttingen: Vandenhoeck Ruprecht
(1969) Principles of Phonology Berkeley: Univ. of California Press (non-IPA and sometimes
mixed, so that some symbols can have {very} dierent 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 ( 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*. Mnchen: 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 Clis: 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: cruttenden, a. <Gimson's Pronunciation of English (!); 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. (1960, 1918) An Outline of English Phonetics Cambridge: Heer (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 { 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 clever
editorial sta} had monstrosities like blaster with /s/, that is /s/!)
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, 1924) American Pronunciation George Wahr: Ann Arbor (American
English; meanwhile, pronunciation has changed, but it is still worthwhile seeing; there is
also an augmented edition by others, but with no real additions, 1994; IPA).
knott, t. a. (1953) A Pronouncing Dictionary of American English Springeld, .:
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 created in Britain and the usa 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) Springeld, Mass.: Merriam-Webster
(American pronunciation; interesting for its frequent phonetic variants, although it shows
absurd secondary stresses, 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 dierences; it shows cases of marked
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 dierences; it shows cas-
es of marked 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 italianoinglese}. Bologna: Zanichelli
(British pronunciation with only major American dierences; its transcriptions have been
revised by L. Canepari and A. Venturi but reduced by the editorial sta, however it pres-
ents the innovation of /E/ instead of /e/ and very many reduced forms, 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 identical forms, in addition to those with slightly dierent 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 dierences {failing
then to give the fundamental immediacy}. Besides, it uses dierent criteria and symbols for
the two accents { British and American}, leading the reader to think there are dierences
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 a /'EI, /; besides, it mixes up neutral pronunciations and others which are not (yet)
neutral, for ex. dune with British variant identical to June, /dZu:n/, or latter shown, in the
American pronunciation, exactly like ladder, /'ldr/, not even as a variant; mixed 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 Jones it provides a reliable survey, especially for Brit-
ish English; for American English, /:/ has been removed while it was present in the rst e-
dition {1990}, although it could be more useful than /r/, which is still there, whereas /o:r/
has become /oUr/, which is more economical, but far less appropriate; IPA).
Italian ( 3)
bonazzi, i. (1989) Dizione e qualit espressiva della voce Torino: Centro Maier (with 6 very
short audiocassettes; traditional pronunciation; hypodiacritical spelling).
camilli, a. fiorelli, p. (1965) Pronuncia e graa dell'italiano. renze: Sansoni (simplied
).
canepari, l. (1970) tljn prnnsexn in Le Matre Phontique, 133:6-8 (as all MPh con-
tributions, fully transcribed in IPA].
(1986) Italiano standard e pronunce regionali Padova: (with 2 enclosed audiocassettes,
the second one is about regional pronunciation; almost canIPA]
(2000 {amended modied 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 modied 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 koins; 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) Crso di pronnzia italiana Padova: Radar (with 14 vinyl records; tradition-
al pronunciation; hypodiacritical spelling]
migliorini, b. tagliavini, c. fiorelli, p. (1981) Dizionario d'Ortograa 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
muljaCi, 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 description 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 new tran-
slation from English, for instance with il vento del nord the North Wind instead of la tra-
montana. An answer was given that four Italian scholars had approved it for publication.
e only recording used for the transcription is partially better (as one can immediately
hear), being somehow less possessed than the transcription provided in the paper.
French ( 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 italianofrancese}. Bologna: Zanichelli (its
transcriptions have been revised by L. Canepari but reduced by the editorial sta, how-
ever it presents the innovation of neutral /, / instead of the old and outdated /, /; 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 franais).
capelle, g. (1968) Le rythme et l'intonation de la phrase franaise Paris: Didier (a vinyl record
and booklet, with no notation).
carton, f. (1979) Introduction la phontique du franais Paris: Bordas (almost IPA).
coveney, a. (2001) e Sounds of Contemporary French Exeter: Elm Bank Publications (not
too reliable for things dierent from French, reported from controversial written sources;
IPA).
gleissner, h. (1964) Franzsische Aussprache Leipzig: Verlag Enzyklopdie (a vinyl record and
booklet; IPA]
kaneman-pougatch, m. pedoya-guimbretire, e. (1989) Plaisir des sons Paris: Alliance
Franaise/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[) Phontique franaise Malm: Lromedel (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 franaise Paris/Gembloux:
Duculot (previous editions had an enclosed vinyl record and a less blank-page format; in-
conveniently separates common and proper names into two sections; IPA]
(1996) Orthographe et prononciation en franais 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 ( 5)
Spanish ( 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 fullled
canepari, l. (in prog.) Spanish Pronunciation*. Mnchen: 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 espaol).
dalbor, j. b. (1980) Spanish Pronunciation Fort Worth: Holt, Rinehart and nston (Amer-
ican with Iberian indications; numerical intonation with pitch levels; non-IPA]
Gran Diccionario de la lengua espaola (1985) Madrid: (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) Descripcin fono-tontica de las variedades regionales del espaol de Amrica
y Espaa Univ. of Venice, graduation thesis guided by L. Canepari (Iberian and American,
utilizable bibliography 419
Portuguese ( 7)
canepari, l. (in prog.) Portuguese Pronunciation*. Mnchen: 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 pronncias do portugus).
cristfaro silva, t. (2001) Fontica e fonologia do portugus So Paulo: Contexto (Brazilian;
with a ; IPA]
cuesta, p. vzquez da luz, m. a. mendes (1971) Gramtica portuguesa Madrid: Gredos
(Lusitanian with Brazilian indications; non-IPA]
Dicionrio da lngua portuguesa (1981) Porto: Porto Editora (Lusitanian; does not transcribe,
but indicates with italic letters in brackets occurrences of /'e, 'o/ [ ] cases of unstressed
/E, a, O/ [ , ] which is one of the uses of our /, A, / and consonant clusters includ-
ing x).
Dicionrio da lngua portuguesa contempornea (2001), Lisboa: Editorial Verbo (Lusitanian;
too much proudly presented as the publishing enterprise of the beginning of the 21st cen-
tury and of the 3rd millennium, which began before the French Revolution and nally led
to this two-volume dictionary; it does indicate metaphony, however, giving only plurals
with dierent 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 /, g, $, R, ], l/, instead of /, g, S, K, , L/, and even /, , , Z/ instead of /E, ,
O, Z/; semi-IPA).
grtner, e. thomaz jayme, j. (1984) Portugiesische Aussprache Leipzig: Verlag Enzyklopdie
(Brazilian; a vinyl record and booklet; almost IPA]
giangola, j. p. (2001) e Pronunciation of Brazilian Portuguese Mnchen: Lincom Europa
(without intonation; hybrid IPA with (y, w) for (j,i w, u))
irmen, f. cortes kollert, a. m. (1995) Taschenwrterbuch _ PortugiesischDeutsch {Deutsch
Portugiesisch} Berlin: Langenscheidt (Lusitanian; indicates metaphonetic variations, but
not systematically; presents /e/ + /S, Z, N, L, j, i/, for (), but (Ai), for (3I, I) and for (Ai,
4i); IPA]
mea, g. (2003) Dicionrio de PortugusItaliano Porto: Porto Editora (Lusitanian; in the pres-
ent reprint 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 (l), but uses (!) in place of (); in addition, it gives (j, w) for (i, u),
and () for (), instead of //; 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 dicionrio {italianoportugus} portugusitaliano (1993) So Paulo: Melhoramentos
(Paulista {pron. of So Paulo}; indicates some metaphonetic variations; IPA]
Pocket {EnglishPortuguese} PortugueseEnglish 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; cana old-style IPA).
viana, a. r. gonalves (1903) Portugais. Phontique et phonologie _ Morphologie _ Textes Leip-
zig: Teubner (Lusitanian; still the most reliable work, in spite of several inevitable misprints;
archaic IPA]
(1912) Vocabulrio Ortogrco e remissivo da lngua 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 ocial spelling, whereas that of the intermediate time was denite-
ly more careful}).
Russian ( 8)
Unfortunately, no IPA Russian pronouncing dictionary exists yet, with full transcriptions of the
dierent words (also avoiding archaic-Soviet morphonological or orthographical symbols,
mainly based on the Cyrillic alphabet).
Arabic ( 9)
Unfortunately, no Arabic pronouncing dictionary exists yet, with full IPA transcriptions of the
dierent words.
Hindi ( 10)
Unfortunately, no [IPA] Hindi pronouncing dictionary exists yet
chao, y. r. (1948) Mandarin Primer Cambridge (.): Harvard Univ. Press (with vinyl rec-
ords; non-IPA, but with tone-letters {now ocialized but trivialized by the reform 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).
kratochvl, p. (1968) e Chinese Language Today London: Hutchinson (Chinese/Karlgren-
ized IPA: with q for (M), that is with a vocoid still considered as if it were the intense con-
toid (); on the other hand, with the vowel Q for the true intense contoid ()]
norman, j. (1988) Chinese Cambridge: C. Univ. Press [Sino-like IPA: with , for (M, )]
Practical Chinese Reader (1985) Beijing: e Commercial Press (with audiocassettes; Chi-
nese/Karlgrenized IPA: with q for (M), that is with a vocoid still considered as if it were
the intense contoid (); on the other hand, it has the vowel Q for the true intense contoid
()]
qin, z. (1980) On Chinese Phonetics Beijing: e Commercial Press [Chinese/Karlgrenized
IPA: with q for (M), that is with a vocoid still consideres as if it were the intense contoid
(); on the other hand, it has the vowel Q for the true intense contoid ()]
shan pao li, j. (1975) Introduction to the Chinese Language Taipei: Ke Ji Publishing Co. [mod-
erately Sino-like IPA]
speSnev, n. a. (1980) Fonet$ka k$ta=ckogo rvka. Len$ngpad: *datelctbo Len$ngpadc-
kogo Yn$bepc$teta (a very hybrid IPA]
Unfortunately, in the Journal of the International Phonetic Association itself Standrad Chi-
nese (Beijing) has appeared {2003, 109-112}, which is semi-Karlgrenized IPA, with both (M,
) rendered as (), but described as () and (), including many more transcriptional queer
inaccuracies.
utilizable bibliography 423
Japanese ( 12)
Esperanto ( 13)
canepari, l. (1998) Non c' proprio niente da dire sulla pronuncia e la graa dell'esperanto? in
La linguistica, le lingue pianicate e l'Esperanto a special issue of L'Esperanto 44-55
[canIPA]
cresswell, j. hartley, j. sullivan, l. h. (1987) Teach Yourself Esperanto Sevenoaks:
Hodder Stoughton (with an audiocassette of Anglo-Esperanto; spelling).
kalocsay, k. waringhien, g. (1980[) Plena analiza gramatiko de esperanto Rotterdam:
(spelling).
wells, j. c. (1978) Lingvistikaj aspektoj de esperanto Rotterdam: / (spelling).
(s. a.) Esperanto Pronunciation Disc London: (a vinyl record and leaet; spelling).
zamenhof, l. l. (1962) Lingvaj respondoj Marmande: (spelling).