Sciences sociales et humaines
A study of the effect of speech rate on the articulation
of labiovelar stop consonants in fongbe: an EPG
and aerodynamic approach
Bertin C. YEHOUENOU
LPL Aix-en Provence
Introduction
The three phases are:
- phase i = the shutting phase, leading to
Notes on Fongbe and studies
- phase ii = the closure phase, followed by
on labiovelar stop consonants
- phase iii = the opening phase.
Attempts have been made to describe sorne of the uncommon
We can see, in the Fongbe consonant chart, that plain oral stop
sounds that occur in the languages of the world. One such sound
consonants occur in the language. Those have been clearly des-
is a stop consonant which is referred to with terrns as varied
cribed as using the same airstream mechanism, that is the egres-
as: double articulation, doubly articulated stop, labiovelar stop,
sive pulmonic airstream mechanism. They then display much
double articulated plosive, coarticulated stop, labiovelar arti-
of the pattern described in plain oral stops in the languages
culation, labial-velar double articulation, labial-velar stop.
of the world. The sounds concerned are: 1 p, b, t, d, cf, c, j,
Indeed, the sound in question is a stop consonant, an oral
k, g l, that is nine plain oral stops in number.
stop consonant. The most common of its type is made on an
As to Ikpl and Igbl which also appear in the stop invcntory
egressive pulmonic airstream mechanism. Il can be voice-
of Fongbe, little is known about the mechanism of their pro-
less as weil as voiced.
duction apart From tentative, uncommitted descriptions (Akoha,
Il is the purpose of the present study to investigate the nature
1991 ; Capo, 1993). We can only note that Fongbe has both
of the labial-velar stops in Fongbe. Fongbe is the language
voiceless and voiced double articulated stops, which at first
of the Fon people in South Benin (West Africa). Il is one of
seem to be in contrast with voiceless bilabial Ipl and voiccd
five sections into which the Gbé continuum has been broken
bilabiallbl respectively and also in contrast with voiceless velar
down. Gbé is cla~sified under Kwa languages which, in turn,
stop !kI and voiced velar stop Igl respecti vely.
belong to the Niger-Kordofanian family.
General surveys of African languages often list a series of traits
This will be done in the light of previous research and expe-
which are said to be characteristics of sub-sahara Africa as a
riment that have been undertaken by sorne investigators (such
linguistic area. These enumerations tend to include, among
as Sarah Garnes and Peter Ladefoged) and whose object has
other things, phonological features such as tone, click sounds,
been to make statements on both the articulatory and acoustic
implosives, labial-velar stops and the prevalence of open syl-
characteristics of labial-velar sotps in languages like Yoruba
lables. Of the characteristics many are in fact found in other
.and Ibibio. As far as plain oral stops are concerned, they are
world areas. For instance, a preference for open syllables or
produced in the following way : the soft palate is raised so that
tone. A few, however, are confined to Africa such as click
the nasal cavity is closed. There is an articulatory cl os ure ,
sounds which are not ofwide distribution since they are found
the location of which depends upon the active and the pas-
only in South Africa. Others are common in Africa and are
sive articulators involved, which in turn effect the quality of
found in other places. An example is labial-velar stops which
the oral stop concerned. Il can be labial, dental, alveolar, retro-
are the sounds under discussion here. Phoneticians often com-
flex, palatal, velar, uvular, glottal. Therefore the airstream is
pare labial-velar stops with plain stops with a view to working
completely obstructed. Pressure then builds up behind the clo-
out any significant similarities or differences ; Welmers, quoted
sure and an oral stop is forrned. When the articulators come
by Greenberg, states that :
apart, the airstream is released in a burst which is not always
audible. This processus is best explained by the three-phase
"There is an interesting geographical distribution of doubly
distinction which is found in phone tic the ory literature.
articulated stops in Africa. They occur primarily in languages
Abercrombie (1967: 147) for example, speaks of the three phases'
grouped in a strip from the Atlantic into the Central African
in the production of oral stops and says that this distinction
Republic across the West African bulge and somewhatfarther
is important in the function of stops in connected speech.
east." (Greenberg, 1983:5).
30 1
Rev. CAMES - Série B, vol. 00, 1998
Sciences sociales et humaines
Ian Maddieson (1984) for his part, observes that stops occur in
on labial and velar or post-velar gestures to complement infor-
the inventory of all known languages and have appropria-
mation obtained using EPG. So, this is very appropriate when
tely been regarded as the optimal consonant (Jakobson and
investigating coarticulatory effects in a VCV environment.
Halles, 1956:42). But he goes on to observe that labial-velar
Previous studies reporting on gestural timing in terms of EPG
stops are only found in twenty languages and represent 6.3 per
patterns and lor air pressure include Ladefoged (1962) who
cent of the languages investigated and mostly in African lan-
accounts for the behaviour of labiovelar stops in sorne African
guages such as Ibibio and Yoruba.
languages. D. Silverman and Jongho Jun (1994) investigate
stop consonant clusters using aerodynamic techniques only.
Greenberg (1970) says that here is an area outside of Africa
As to Marchal and Meynadier (1995), they investigated coar-
in which these sounds are found and that is the Kate ono group
ticulation in Ikll sequences in French using a multisensor
ofnon-Austronesian (lndo-Pacific) languages in north-eastern
approach involving EPG and oral airf1ow. Demolin (1992) exa-
New Guinea and sorne Autranesian languages of Melanesia
mines labiovelar stops found in sorne central African languages
have the sounds. In this respect, tms remark by Greenberg seems
and he draws on aerodynamic parametres of oral airf10w and
important which says that :
pharyngeal pressure as well as video images showing the
"Moreover for many languages we still have only word lists.
lips and lower jaw movement to account for the characteris-
In the absence of systematic phonology the non-occurrence
tics of labiovelar stops in three central African languages.
ofthese sounds cannot with certainty be deducedfrom their
His conclusions as drawn from the pharyngeal pressure traces,
absence in a specifie list, only their existence from their occur-
are that in the articulation of these complex sounds, there is
rence." (Greenber, 1983:5).
a front - back movement of the tongue body accompanied some-
times by a lowering of the larynx on the one hand and a spe-
Hypothesis
cific lower jaw movement different from the one involved
One trend in the phonetic literature is to regard and treat the
in simple velar or bilabial stops. Finally Connell examines
structure of labiovelar stop consonnant as behaving like clus-
the status of the sole voiceless labiovelar stop Ikpl found in
ters. In this respect, attempts have been made to explain the
Ibibio, a Nigerian language and he provides acoustic, aero-
patterns ofthe gestures involed in the production ofthese com-
dynamic, laryngographic as well as EPG accounts, stating that
plex articulations.
the labial release is achieved subsequent to the velar one.
One such attempt is the use of the EPG technique correlated
In the present study, by means of EPG and aerodynammic tech-
with aerodynamic technique to account for the spatio-temporal
niques, we investigate the effect of speech rate on the spatio-
organization of articulatory gestures during the production
temporal organization of the labiovelar stop consonants /kpl
of labiovelar stop consonants.
and Igbl in Fongbé. To ourknoweledge, no previous study has
reported on this factor. To account for the degree ofthis effect,
It is well accepted nowadays that the EPG technique is used to
we consider as relevant: il timing of the tongue body and of
obtain quantitative and qualitative data on patterns of lin-
the labial closure and release together with the articulatory
gual contacts with the hard palate during continuous speech
phase durations as indicated by onset, c10sure and release
(Hardcastle et al. 1989). The technique indicates the pre-
corresponding to the volume or rarefaction of air pressure
sence of contact while not providing any direct information on
and airf10w relative to plain velar consonants, iil magnitude of
the nature of the contact of the tongue with the palate.
the velar constriction in both Ikpl and Igb/, that is in terms
Because of the limitation of the EPG technique, one has to
of the number of activated electrodes as compared with simple
resort to an aerodynamic technique to acquire information
velar counterparts.
Material and Method
low or high tone and carries the sentence stress. Only data
for five repetitions were segmented.
Subject, recording and data acquisition
The multisensor, multichannel system developped in Reading
(Hardcastle et al. 1989 ; Gibbon et al., 1993) was used to acquire
The sole subject was an adult male who is a native speaker
EPG, intraoral pressure, pharyngeal pressure, oral airf10w as
of the Ouidah dialect of Fongbé. Eight Fongbé real words were
well as acoustic data. The sound signal was sampled at 20 khz
embedded in a carrier sentence. They were repeated ten times
using a Sennheiser MK 40 P48 microphone with simultaneous
at normal speech rate and at fast speech rate yielding 160 tokens.
recording on a Sony DAT DTC - 1000ES. For this study we
The words were : gi, gbi, ki,ikplils, gè, gbè, kè kpè. Each word
onl)' processed the data supplied by the audio, EPG, intraoral
was preceded by vowel Iii or lei so that the vocalic symme-
and pharyngeal pressure channels. A signal editor named
tric context obtained is of the ICiC/ or ICeCI pattern where
Phonedit was used to simultaneously and synchronously dis-
C is either a velar consonant or a labiovelar consonant. Syllables
play the various articulatory and aerodynamic parameters
containing the velar or labiovelar consonant have either a
and to carry out both segmentation and label1ing.
Rev. CAMES - Série B, vol. 00, 1998
1
31
Sciences sociales et humaines
Segmentation and labelling
Consequently, Ik/, in normal speech rate, has a me an dura-
tion of 165 ms and Ikpl has a me an duration of 172 ms.
Because of the nature of the data, segmentation was done in
This is significant of the faet that both Ikl and/kpl are pro-
the following way: segmentation of Ik/ and Igl was based on
grammed simultaneously and that, as Sara Garnes says, Ikpl
the EPG frames and information on the labial release part of
articulation constitutes a single unit of timing.
Ikpl and Igbl was drawn from the variations in both the intraoral
and pharyngeal pressure as weil from the oral airflow trace. So
ln fast speech rate, Ik/ has a mean duration of 53 ms and Ikpl
the following phases were identified and labelled : the onset of
has 64 ms with a standard deviation of 14.5. This clearly shows
the closure, the complete closure, the release of the closure.
the effect ofhypoarticulation which yields reduction and causes
A script was specially designed which permitted us to observe
the overall duration of the consonant to be shorter than in
and quantify the activated electrodes in each phase corres-
the case of normal speech. Tables 1 and II show the duration
ponding to the completing of the velar gesture.
of consonants and figures 1, 2, 3 and 4 show speech signaIs
and speetrograms for ligil, ligbil, li kil, /ikpil for both normal
The EPG data was segmented using the following labels:
and fast rates of speech.
- ACE (or 1) which identifies the Approach to the Closure and
which indicates the approach of the articulatory target (the velar
occlusion). It is placed at the point where the first electrode or
Table I. Showing the mean duration (in milliseconds) of the
group of electrodes start coming together from each side at
simple velar and labiovelar consonant in lei context
various places on the palatogram in order to achieve the C!o-
Consonants
Normal
Fast
sure in the posterior rows.
Ik/ [kè] "hernia"
148
- SCE (or
53
2) which identifies the exact moment when the
stop closure is achieved. This label is placed on the first pala-
Ikpl [kpè] "charm"
146
65
tographic image where there is complete closure, which cor-
responds to the presence of a horizontal bar of activated
Igl [gè] "tocs"
125
50
electrodes in one or more posterior rows.
Igbl [gbè] "voice, speech"
110
48
- SRE (or 3) is the label we used which marks the stop release
of the closure as it is measured from the EPG data. It is placed
on the last EPG frame showing a complete closure.
As a matter of fact, measurements were taken from both the
Table II. Showing the mean duration (in milliseconds) of
oral airflow trace and intraoral and pharyngeal pressure trace
the simple velar and labiovelar consonants in Iii context
and correlated with those taken from the EPG data.
Consonants
Normal
Fast
Results and discussion
Ikl [kil "bamboo"
165
76
Timing of closlJre and release/Articulatory
Ikpl [ikpiI8] "place name" 172
75
phase duration
Igl [gi] "maize paste"
174
106
Measurements of the total duration of both simple Ikl and
Igbl [gbi] "to defy"
142
39
Igl and complex Ikpl and Igbl show a specific pattern for
each speech rate.
32
Rev. CAMES - Série B, vol. 00, 1998
1
Sciences sociales et humaines
5
-------------~-------------
•,
A
,\\
,\\
"
.~ (, fI il ., ,\\ 1\\ " 1\\ Il " .•
_:-
L
__ L
_
~
·5
- - - - - - - - - - - - -
l'HO
1300
13.50
1400
l"~O
1.500
:1
.\\
Hz
Am Jitudc....
~
lli'--
~------~.-------~--
..-----~------~------~-~
oouo 1: - .. - - - .. - - - - .. - .. f .......... - - -- - -.. - ; - - - .... -- - --- - .. f .. - - - - - -
.. ---~ -------.... ---.. ~ ----.. ------.... ~ ---
~
~ ~
~
~
~
~
~
4000 : - - - - - - - - - - - - -
- - - - - -
-
- - -
- -
-
-
-
-
-
- - - -
-
- - - - -
- - - - - - - - - - - - -
- - - - - - - - - - - - -
- - -
-~iiiii;~,~
t-.. ---_.... -.. ---; ---...... _-
-:ktrÎ:'.
'OdO ' ..
+ - .. - _ .. - - .. -
13.50
1400
Figure 1: showing spectrogram of /igi/ in normal rale of speech.
,
-
,
,
,
1
- - - - - -, - - - - - - - - - - - ,- - - - - - - - - - - .... - - - - - - - - - - ... - - - - - - - - - -
,
,
,
,
,
,
A A :~ ~ A A ~ b ~ .
"
1'" V' 1-;,
V ;li ~r ,,"' '''i'''"; il"
,
,
,
,
,
1
l
,
,
,
,
,
., ------:-----------:-----------~----------~----------
:
- - - - - - - - - -
- - - - - - - - - - -: - - - - - - - - - - -: - - -
,
,
•
1
1
•
•
1
,
,
,
,
,
1
1
,
,
,
2660
;;1680
2100
2120
;;1140
i
1
1
•
,
,
1
l
,
,
l
,
]OÛÛ
- - - - - -:- - - - - - - - - - -:- - - - - - - - - - -~ - - - - - - - - - -} - - - - - - - - - - ~ - - - - - - - - - - ~ - - - - - - - - - - -:- - - - - - - - - - -:- --
,
,
,
,
,
,
,
,
-
-
-
-
-
- , -
-
-
-
-
-
-
-
-
-
- , -
-
-
-
-
-
-
-
-
-
-
t-
-
-
-
-
-
-
-
-
-
-
..
-
-
-
-
-
-
-
-
-
-
..
-
-
-
-
-
-
_ _
-
-
"1
-
1
-
-
-
, -
_ _
600Û
,
.
,
,
,
,
,
(
5000
-4000
______!__ ~ ~ ~ ~ ~ ~ ~ ~ ~!~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ::~ ~ ~ ~ ~ ~ ~ ~ ~ :~ :~ ~ ~ ~ ::~ :
.. _~~ :~ ~ ~ ~ ~ ::~ __ .. :
:::~i~ ~ ~
_____, _:_.;~ ,-..:.f.:::;:"",:;,;j::~..··,;..j$(··..:~1:~"'@~;'Ü::;i~~ _ _ _ _ _ _ _ _ _ _
_
+
~~ __ ~ __.... .-
' .._j
_~.
:_.__
?QOQ
lOÛO
-.II'fI'I.:'11Il'.'' .i -----~-- -- - - - - -- -- - -- ...........'
' .
' "
.•
1000
- _- - - -~:... - - - -.."':.~....: - - -:- -- - - - - - - - -:- - - - - - - - - - - - - - - - - - - :..:: _.=~ :;::~:::;:;,;.;
~ _:~:~.:::.-·-::-·-·-:::.--:.:··:;::···~·.;;,-·-·'=";::;;,·:.::··w-:::.:.~· _. _ ';: __ ,'-: __
2t.60
2660
2100
2120
:1740
Figure 1 (bis) : showing spectrogram of /igi/ in fast rate of speech.
Rev. CAMES - Série B, vol. 00, 1998
33
Sciences sociales et humaines
~f~ nn__nf -_-_-- -;-
n__un _----n,' _-_n_ni
u
-
u
_
- -
- -
o -Ov I~ .. Ô.,. '\\;,.'''.::: ·:,",~·l..,.rl.Y\\. \\: "'..
:
'
lm,""
••~ "'ç ~v 'l.,. "v "v '\\:
. ,
. .
. .
"
'r---- -- ---T- -- ---- --- --- T----- ------- -1- ------ ---- ---1 -- ---- ----- --- [--- ---- -------r--~
1400
14jO
DOO
IjjO
•
1
,
,
,
_
J
_ _
~
~
~
Figure 2 : showing spectrogram of /igbi/ in normal rate of speech
~~l:Z:~~~~lal!Yflj i limil'PT• •I~r~iml&kiDi
f.lle
Idit
.bat~-Ijng
.Çon'lpulll'lg
l.oc,(~
~ind,:,~,,'
!:1~I!=.
II 1i
,!
Il
, ,
,
,
- - - - - - _ ... - - - - - - - _ .... - - - - - - - --
,
,
~
--L-
---.J
~~---~---~'--
3060
3080
3100
31:lû
3140
,1
..d
tl
~;:Th1J'E~Tti1:'\\;n::::'jrG''':':;:rEg~::jT~~'''li@T.'N,~~:&'~_'''M~l&''tMm~8_
mil"!"}
3158.9 ms
1
Hz
1 Amplitude 1 "j ...:::J
lb
,
1
"
1
1
- -' - - - - - - - - - ~ - - - - - - - - -: - - - - - - - - - ~ - - - - - - - - -: - - - - - - - - - ~ - - - - - - - - -: - - - - - - - - - ~ - - - - - - - - -; - - - - - - - - -
,
,
1
l
,
,
,
,
- -,- - - - - - - - - .. - - - - - - - - -,- - - - - - - - - .. - - - - - - - - -,- - - - - - - - - ... - - - - - - - - -,- - - - - - - - - ... - - - - - - - - -,- - - - - - - --
,
,
,
,
l
,
,
,
,
,
,
,
,
,
,
,
-
- . - -
-
-
-
-
-
-
-
T
-
-
~
-
-
-
-
-
- , - -
-
-
-
-
-
-
-
T
-
-
-
-
-
-
-
-
- , - -
-
-
-
-
-
.- -
T
-
-
-
-
-
-
-
-
- , - -
-
-
-
-
-
-
-
T
-
-
-
-
-
-
-
-
- , - -
-
-
-
-
-
- -
,
,
,
,
.
,
.
.
- -' - - - - - - - - - ~ - - - - - - - - -: - - - - - - - - - ~ - - - - - - - - -: - - - - - - - - - ~ - - - - - - - - -: - - - - - - - - - ~ - - - - - - - - -: - - - - - - - - -
,
.
,
,
,
,
,
17000
,'H
6000
1.
.:'iOOO
1«
4~)l..'O
3ûût'l
1
JOOO
- - - -è""XT' , -- - - - - : ~..~ ~ ~ ~ ~ ~ ~ ~:~ ~ ~ ~ ~ ~ ~ ~~;I;.~.i:.!!!!~~.ê=
1000
3lJO
3140
Figure 2 (bis): showing spectrogram of /igbi/ in fast rate of speech
34 1
Rev. CAMES - Série B, vol. 00, 1998
Sciences sociales et humaines
-
-
-
r
-
-
-
-
-
-
-
-
-
-
-
-
-
., -
-
-
-
-
-
-, -
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
.,
,
,
,,: ,." ,', ,~
,~ i\\L~.!\\I\\J\\ < ' :
:
...'
ft, il ,h ,~ :,1 fi ~ f. ,~ ,~.'
•
'cr ", " \\, " \\U V' ,'. J. '1 '
,. • "':
.
\\/ Il ~ ~: ~ v ; ,} c,'- ,
1
l
,
1
1
1
1
.J
•
L
.J
'
L
.J
. j
l
,
,
l
,
1
1
1
l
"
l
'
"
1
1
.
"
1150
1)00
1250
1300
Figure 3 : showing spectrogram of /iki/ in normal rate of speech
Figure 3 (bis): showing spectrogram of /iki/ in fast rate of speech
Rev. CAMES - Série B, vol. 00, 1998
35
Sciences sociales et humaines
1
l
,
,
l
,
,
,
,
-
_ _
-
_ _
-
-
.., -
-
-
-
-
-
-
-
-
T
-
-
-
-
-
-
-
-
- , -
-
-
-
... - - - ~ - - ~ - - ,- - - - - - - - - , - - - - - - - - - ~ - - - - - - - - .., - - - - - - - - - ... - -
,
,
.
,
.
,
,
.
,
.
.
"
'
1
l
"
.S
__ - _ - - _...J -
-
-
-
-
-
-
-
-
~ - - - - - - - - -:- - - - - - - - - ... - - - - - - - - -,- - - - - - - - - : - - - - - - - - -:- - - - - - - - - --: - - - - - - - - _ ... - -
.
,
L_-~--~----_:-"'-'----------':::-'::-'-------:-:13:':>-=-O-----:-14::0:::0-~------;174>';:O:-------;1~------u'5-0-----~lO~O-O~1 =<
,
l
,
,
,
--~--------~---------~---------:---------~--
.
,
-
-
-
- --. - - - - -
- - - .... - - - - - - - - -,- - -
î - -
- - r - - - - - -
Figure 4
showing spectrogram of /ikpi/ in normal rate of speech
Figure 4 (bis)
showing spectrogram of /ikpi/ in fast rate of speech
36
Rev. CAMES - Série B, vol. 00, 1998
Sciences sociales et humaines
As shown for Ibibio and Mangbetu by Connel and Demolin
by the velar contact on the EPG frames and the period of
respectively, our data also provide evidence for the existence
closure for the entire stop consonant as associated with the
of asynchrony in the clos ure and release of the articulation
speech signal. There is a substantial difference between the
of Fongbe /kpl and Igb/. This is indicated by acoustic data which
velar closure duration (62 ms) and the total duration of the stop
shows that F2 transition associated with the release of the labio-
consonant (130 ms). When we consider aerometric data, the
velar stop is negative and then is a typicallabial transition.
velar part of Ikpl or Igbl shows a closure duration of 62 ms
In fact, the consonant release occurs at a frequency lower
(measured from Igbel as we take into account thë duration
than the adjacent vowel frequency. It is thus clear that the labial
of positive pharyngeal pressure) and the volume of oral air flow
release is achieved after the velar one has occurred. Further
whereas the total duration of the stop consonant is estimated
evidence of the asymmetry of the two gestures is provided
at 174 ms as measured from the speech signal. Further evi-
by the combination ofEPG and aerometric techniques. This is
dence is seen in the decrease in the intraoral pressure as shown
seen when we consider the period of velar closure as shown
in figures 5 and 6.
~4
"'Cl
1300
IBO
HOO
lBO
1500
1550
11SO
1300
1350
1400
14jO
1"(110
1.'50
..
Figure 5 : showing speech signal, pharyngeal pressure for Igbl in normal l'ale of speech in /il context
Figure 5 (bis) : showing speech signal, pharyngeal pressure for Igbl in fast rate of speech in IiI context
Rev. CAMES - Série B, vol. DO, 1998
1 37
Sciences sociales et hummnes
1400
15':'0
1600
~~~---=:J
1300
1350
HOû
145.)
IS00
=--=0
__ : =
1600
=li1MJvow
1600
Figure 6 : showing speech signal, pharyngeal pressure for Ikpl in normal rate of speech in Iii context
i
1
1
~:::'::n
1
1·
1
,EPG
.
~l
2.00') ~1
,1Ilo........rj 1l'Mi=c.IlW,,",·
I;mMaii>J..~16
IliYPhonWt
~Pi>inlS_p'o·ikpn4<,,,j fi' 110ê
Figure 6 (bis) : showing speech signal, pharyngeal pressure for ikpl in fast rate aï speech in Iii comex[
38 1
Rev. CAMES - Série B, vol. 00, 1998
Sciences sociales et humaines
Magnitude of velar constriction
Comparing simple velar and Iki and Igl with the simple velar
part of /kpl and Igbl respectively, we saw little difference as to
The use of the EPG technique has been decisive in accoun-
the patterning of the articulatory gesture. In both cases, we seg- .
ting for the lingual contact with the palate during the articu-
mented and labelled the three phases (onset, complete clo-
lation of our test consonants. Although the artificial palate only
sure (corresponding to maximum number of activated elec-
covers the pre-velar region of the roof of the mou th, we were
trodes) and offset phase as specified in 2b.
able to observe the following :
In both consonantal contexts, complete closure was materia-
In the context of high front vowel Iii and mid high front
liezed by the horizontal bar of activated electrodes. There is
vowel lei, contact with the pre-velum and the back of the tongue
also presence of lateral contact which is less consistent for /k/.
was clearly visible. This was consistent with aIl velar and labio-
Figures 7 illustrate complete c10sure (maximum c1osure) normal
velar consonants at normal rate of utterance.
rate of speech.
Figure 7 : showing EPG frame illustrating maximum closure for /ikpi/in normal rate of speech.
-liïEl
lBülJO Hl"
~ F'.3mt Sh<':tP PlO - tl:~;1.lffa;~ __ l ''-6' "-15~54
Figure 7 (bis) : showing EPG frame illustrating maximum c10sure for /ikpi/in fast rate of SPl:CC:l.
Rev. CAMES - Série B, vol. 00, 1998
1
39
Sciences sociales et humaines
These patterns indicate that there exists sorne similarity bet-
One may then suggest that, because of the relative asym-
ween simple ve1ar lkJ and complex consonant /kpl and is a fur-
metry of velar and labial gestures in the articulation of labio-
ther piece of evidence that both consonants are programmed
velar stop consonants in Fongbe, the gestural overlap model
similarly and constitute a single unit of timing consequen.tly
can help to describe and account for the sequence of ges-
a single articulary gesture. This result is in accordance with
tures and consequently the timing of the phases involved. This
Connell who observed, for Ibibio, that there is a similairy
view is supported by D. Byrd (1992) although this author
between these two consonants with regard to their dynamics.
remarks that labiovelar stops are distinguished from to sequences
As a matter of fact, a different pattern is obtained in the case
by the principle that single complex segments have duration
of fast speech rate. As was predicted, the EPG technique is
comparable to that of simple segments of the same phonetic
appropriate in that it helps to see that the total duration of com-
c1ass as was shown in Timing ofclosure and release/Al1iculatory
plete cIosure (maximum cIosure) is significantly shorter than
pase duration.
in the case of normal speech and is averaged at 50 ms. Table III
With the EPG data, we were able to see that the Fongbe velar
shows the du ration of cIosing phase for normal rate of speech
and labiovelar consonants Ikpl and Igbl are subject to coarti-
and for fast rate of speech. Besides, a look at EPG frames reveals
culation and we do not notice any backward movement of
the presence of a gap in the posterior row as a central electrode
the dorso-velar contact during the phase of complete c10-
is not activated while contact persists in the lateral part of
sure. This may be interpreted as an evidence that Fongbe simple
the palatogram.
velar Ik! and Igl on the one hand, and the velar part of Fongbe
Table III : showing the mean duration (in milliseconds) of
labiovelar Ikpl and Igbl are articulated in the velar region as
maximum cIosure as measured from EPG data consonant in Iii
indicated by our data. The gap might be the evidence of incom-
context
piete velar the due to aerodynamic constraints as weil as to the
temporal organization of the articulatory gestures.
Consonants
Normal
Fast
lkJ [kil "bamboo"
145
45
Conclusion
/kpl [ikpilè] "place name"
102
44
The significance of the present study rests entirely both on
Igl [gi]. "maize paste"
145
79
the methodology used and the results achieved. It enabled
Igbl [gbi] "to defy"
51
32
us to test and confirm the theoretical hypothesis as to the
principle of phonetic gesture economy which seems to be
particularly applicable when we compare normal rate of speech
As it appears, our data reveal changes in the magnitude of
with fast rate of speech. The EPG technique, when it is cor-
lingual - palatal contact as weil as in the duration of contact.
related with aerodynamic as weil as acoustic data, made it cIear
Thi~ means that the rate of utterance contributes to the spa-
that, as far as the articulation of labiovelar stop consonants
tial and temporal reduction of /kpl and 19b/. Consequently, tem-
is concerned, the temporal organization of articulatory gestures
poral displays (contact profiles) corresponding to /kpl an Igbl
is dissimilar and thus specifie to each of the spl:laking situa-
as compared to Ik/ and Igl respectively show the number of
tions investigated. Evidence is then given for the existence
electrodes contacted in the back region of a sequence of frames
of context specific constraints in terms of the effect of speech
over time. The duration of lingual palatal contact in the back
rate on complex consonant articulation. In other word~, our
region for the environ ment investigated is estimated in terms
data provided evidence concerning the velar articulatory ges-
of an articulatory index which corresponds to the sum of
activited electrodes relative to the total number of electrodes.
ture and we were able to account for the relative timing of
For /kp/, it is evaluated at 38 electrodes in the case of normal
the labial and velar gestures. Despite the tendency for incom-
rate of utterance and 31 electrodes in the case of fast rate (Iii
piete velar clos ure as may be inferred from the gap noticed
context) ; whereas Igbl shows the following : 39 electrodes for
in the posterior row of the palatogram, it is c1ear that the
normal rate of utterance and 31 electrodes for fast rate.
velar place of articulation is maintained. This is in line with
Carré and Mrayati (1990) whose production model supports
As shown by Hardcastle (1985) investigating phonetic and syn-
the view that in fast speech rate, phonetic gestures may be
tactic constraints on lingual coarticulation in /kil sequence, our
incompletely achieved although the targeted place of articu-
data seem to reveal that the rate of utterance exerts sorne
lation does not change. It appears that, at least for our sub-
influence on the degree of coarticulation.
ject, speech production cornes down to a process whereby a
Besides, it may be inferred that there is a specifie gcsture orga-
speaker adapts his articulation to speaking situation. Howew"
nization at work here which is dependent on the rate of vtte
wc must he eautious and not Jose sight of the fm:t th:!t our c0'1'u"
;ance and which results in gcstural overlap duc to the ;:hangc
'S~lS i"èad, \\'}c lh~?n nccd tC1 in\\'c~t~gat(' spcnltancous l:,pCC'Cr. -,,~_'i"-'
in the duration of the closure. This brings us back to our initial
more subjects in order to see if the observations made with our
statement as regards the tendency of sorne phoneticians to treat
subject are idiosyncratic or constitute a feature of the labio-
labiovelars as a cIuster of consonants.
velar stop consonants in Fongbe. 0
40 1
Rev. CAMES - Série B. vol. 00. 1998
Sciences sociales et humaines
Bibliography
FOLEY W. A. (1986) The Papuan Languages of New Guinea. Cambridge
:CUP.
ABRY C. and PERRIER P. (1995) Le contrôle des mouvements audibles
GARNES S. (1975) An acoustic analysis of double articulations in Ibibio.
et visibles dans la parole. Actes Ecole d'Eté, Juillet 1995, Marseille.
Ohio State University Working Papers in Linguistics Proceedings of the
AKOHA B. (1991) Syntaxe et Lexicologie du Fongbé. Thèse de doctorat
Conference on African Linguistics 20, 44-55.
d'Etat Paris Sorbonne
GIBBON F., HARDCASTLE W.J. and NICOLAIDIS K (1993)
ANDERSON S.R. (1976) On the description of multiply articulated stops.
Temporal and spatial aspects of lingual coarticulation in /kil sequences: a
Journal of Phonetic 4, 17-27.
cross linguistic investigation. Language and Speech 36 (2, 3), 261-277.
BYRD D. (1992) Perception of assimilation in consonant c1usters :
GREENBERG J.H. (1983) Sorne areal characteristics of African lan-
a gestural mode!. Phonetica 49 : 1-24
guages. In Dihoff, Ivan R. (ed.) CUITent Approaches to African Linguistics
1 : 3-22.
BYRD, DANI (1995) Using regions and indices in EPG data reduction.
Journal of Speech and Hearing Research. Vo!. 38, August 1995,
HARDCASLTE W. J. JONES W., KNIGHT C. TRUDGEON A. and
p.821-827.
CALDER G. F. (1989) New developments in electropalatography : astate
of the art report. Clinicallinguistics and Phonetics 3 : 1-38.
BYRD, DANI (1996) Influences on articulatory timing in consonant
sequences Journal of Phonetics 24,209 - 244
HARDCASTLE W.J. (176) Physiology of Speech Production. London:
Academy Press.
BYRD, DANI (1992) Perception of assimilation in consonant clusters :
a gestural mode!. Phonetica 49 : 1 - 24.
HARDCASTLE W.J. and ROACH P.J. (1979) An instrumental investi-
gation of coarticulation in stop consonant sequences. Hollien, H. and
BYRD, DANI (1995) Using regions and indices in EPG data reduction.
Hollien, P. (eds.). CUITent Issues in the Phonetic Sciences, l, Amsterdam:
Journal of Speech and Hearing Research. Vol. 38, August 1995
John Benjamin, p. 531-540.
p.821-827.
HARDCASTLE W.J. (1972) The use of lectropalatography in Phonetic
BYRD, DANI (1996) Influences on articulatory timing in consonant
Research : Phonetica 25, 197-215.
sequences Journal of Phonetics 24, 209-244.
HARDCASTLE W.J. (1985) Sorne phonetic and syntactic constraints on
BYRD, DANI (1992) Perception of assimilation in consonant c1usters :
lingual coarticulation during /kil sequences. Speech Communicaiton 4,
a gestural mode!. Phonetica 49, 1-24.
247-263.
CAPO H.B.C. (1980) Un regroupement des parlers Gbe. Africana
HARDCASTLE W.J., GIBBON F. and NICOLAIDIS K. (1991) EPG
Marburgensia 1 : 3-23.
data reduction methods and their implications for studies of lingual coarti-
CAPO H.B.C. (1992) Let us joke over it : Nigeria is a Tower of Bade!.
culation. Journal of Phonetics 19 (3), 251-266.
Ilorin: Ilorin Press and Garome : Labo GBE Int.
HARDCASTLE W.J. VAXELAlRE B. GIBBON F., HOOLE P. and
CARRÉ R. and MRAYATI M. (1990) Articulatory-acoustic-phonetic
NGUYEN N. (1996) Tongue kinematics in /kil c1usters and singleton lkJ :
relations and modelling, regions and modes. Hardcastle, W.J. and Marchal,
a combined EMAlEPG study. Actes Congrès d'Information et de phoné-
A. (eds) Speech production and speech modelling, Vo!. 55, NATO ASI
tique d'Australie, p. 555-558.
series. Dordrecht, Boston and London: Kluwer, pp. 211-240.
HENTON Co, LADEFOGED P. and MADDIESON 1. (1992) Stops in
CATFORD J.C. (1977) Fundamental Problems in Phonetic. Edinburgh :
the World's Languages. Phonetica 49, 65-101.
University of Edinburgh Press.
JAKOBSON R., FANT G. and HALLE M. (1972) Preliminaries to spee-
CONNELL B. (1989) Instrumental evidence for photicallly graduai sound
ch analysis : the distinctive features and their cOITelates. Cambridge: MIT.
change. Work in Progress (Univesity of Edinburgh) 22 : 58-55.
LADEFOGED P. (1964) A phonetic Study of West African Languages:
CONNELL B. (1994) The structure of labial-velar stops. Journal of
an Auditory-Instrumental Survey. Cambridge: C.U.P.
Phonetics, 22 : 441 - 476
LADEFOGED P. and MADDIESON 1. (1996) The Sounds of the
CONNELL B. (1987) Temporal aspects of labiovelar stops. Work in
World's Languages. Oxford: Blackwell.
Progress (Department of Linguistics, University of Edinburgh) 20, 53-60.
MADDIESON 1. (1984) Patterns of sounds. Cambridge: C.U.P.
CONNELL BRUCE (1991a) Accounting for the reflexes of labiovelar
MADDIESON 1. (1993) "Investigating Ewe articulations with electroma-
stops. Rossi, M. (ed.) Proceedings of the 12th ICPhS, Aix-en-Provence,
gnetic articulatography". UCLA Working Papers in Phonetics 85·22-53.
Vol. 3.110-113.
MADDIESON 1.' and LADEFOGED P. (1989) "Multiply articulated seg-
CONNELL BRUCE (1991b) Phonetic aspects of the Lower Cross
ments and the feature hierarchy", UCLA Working Papers in Phonetics 72,
Languages and their implications for sound chane. Ph.D Dissertation,
116-138.
Edinburgh.
MARCHAL A. and MEYNADIER Y. (1995) Contribution à l'analyse de
DEMOLlN D. (1991) Les consonnes labiovêlaires du Mangbetu.
la coarticulation linguale dans la séquence consonantique /kil en français.
Pholia 6,85-105.
Travaux de l'Institut de Phonétique d'Aix 16 : 61·74.
DEMOLIN DIDIER et TESTON BERNARD (1996) Aspects aérodyna-
MARCHAL A., NGUYEN N. and HARDCASTLE W.J. (1995)
miques et articulatoires des occlusives labiovélaires. Actes XXIe J.E.P.,
Multitiered phonetic approach to speech labelling. Speech Communication 5
p. 11-114.
: 149-157.
DOGIL G. (1988) On the acoustic structure of multiply articulated stop
MARCHAL ALAIN (1988) Coproduction : evidence from EPG data,
consonants (labiovelars). Wiener Linguistische Gazette 42-43, 3-55.
Speech Communication 7; 287-295.
DUEZ DANIÈLE (1992) Second formant locus-nucleus patterns:
OHALA J.J. (1979) Universals of labial velars and de Saussure's chess ana-
an investigation of spontaneaous French speech. Speech Communication,
logy. Proceedings of the 9th International Congress of Phonetic Sciences.
II : 417-427. North Holland.
Vol. 2, Copenhagen: Institute ofPhonetics, 41-47.
DUEZ D. (1995) On spontaneous French speech: aspects of the reduction
OHALA J.J. and LORENTZ J. (1977) The story of [w] : an exercice in the
and contextual assimilation of intervocalic voiced stops. Journal of
phonetic explanation for sound patterns. Berkeley Linguistic Society,
Phonetics, 23 : 407-427.
Proceedings of the Annual Meeting, 3, 577-599.
ELUGBE B. O. (1989) Comparative Phonology of the Edoid Languages.
PAINTE\\.c. (1978) Implosives, inherent pitch, tongenesis, and laryngeal
Port Harcourt: University of Port Harcourt Press.
mechanisnis. Journal of Phonetics 6, 249-274.
Rev. CAMES - Série B, vol. 00, 1998
1
41
Sciences sociales et humaines
PARKER FRANK (1979) Acoustic cues and consonant cluslers. Hollien,
SILVERMAN DANIEL AND JONGHO Jun (1994) Aerodynamic evi-
H. and Hollien, P. (eds.) Curent Issues in the Phonetic Sciences, 2,
dence for articulatory overlap in Korean.
Amsterdam: John Benjamin, p. 1095-1104.
SUSSMAN M. HARVEY (1991) The representation of stop consonants in
three dimensional acoustic space; Phonetica, Vol. 48, N°l, October 1991 ;
PITERMAN M" CROCEA S, and SCHOENTGEN J. (1996) Influence
p.18-31.
de la vitesse d'élocution et de l'accent sur les cibles vocaliques aux niveaux
acoustiques et quasi articulatoires. Actes des Journées d'Etudes sur la
WELMERS W. (1973) African Language Structures. Berkeley:
University of California Press.
Parole 96 : 95-98.
ZIMA P. (1985) Labiovelar stop in the Djougou Dendi dialect of Songhay.
PONELIS F. (1974) On the dynamics of velarization and labialization :
Actas Universitatis Carolinae-Philologica 3, Phonetica Pragensia,
sorne Bantu evidence. Studies in African Lingu\\stic 5( 1), 27 -58.
VII, 97-103.
PUECH G. (1989) L'opposition implosive /mi-voisées en Bekwel. Paper
ZSI~A E.C. and BYRD D. (1990) Acoustic evidence for gestural overlap
presented at the J9th Colloquium on African Languages and Linguistics.
in consonant sequences. Journal of the Acoustical Society of America 88 :
Leiden, The Nelherlands.
S82.
-Thetimingofbothlingualandlabialgesturesintheproductionoflabiovelarstopconsonants
(.)
ca is investigated in this paper. Evidence is provided as to the effect of speech rate (normal
J..
-enandfast)onthecompleteorpartialrealizationofthegestures.Articulatoryasweilas
.c aerodynamic eues show the presence of specifie constraints in each case. Consequently
ct EPG readings for /ikpi/ and /iki/ or /ege/ and /egbe/ sliow more activated electrodes in
the posterior row in normal speech rate than in fast speech rate as both intraoral and
pharyngeal pressures and oral airflow readings are used to monitor the bilabial closure
and release relative to the velar gesture.
42 1
Rev. CAMES - Série B, vol. 00, 1998