voice onset time in aphasia, apraxia of speech and...
TRANSCRIPT
clinical linguistics amp phonetics 2000 vol 14 no 2 131plusmn 150
Voice onset time in aphasia apraxia ofspeech and dysarthria a review
PA SCA L A UZOUsup2 CA NA N OEgrave ZSA NCA Ksup3 R ICHA RD J MOR R ISsect MA RY JA N sup3 FR A NCIS EUSTA CHEpara andDIDIER HA NNEQUIN sup3
sup2 FeAcirc deAcirc ration de Neurologie (Rouen F ) et UniteAcirc INSERM 320 (Caen F )sup3 FeAcirc deAcirc ration de Neurologie (Rouen F )sectDepartment of Communication Disorders Tallahassee Florida 32306-1200USAparaService de neurologie et UniteAcirc INSERM 320 (Caen F )
(Received 14 August 1998 accepted 7 June 1999)
Abstract
Voice onset time ( VOT ) is an objective temporal acoustic parameter dereg ned asthe time between the release of the oral constriction for plosive production andthe onset of vocal fold vibrations Many researchers consider VOT to be themost reliable acoustic cue for the distinction between voiced and voiceless stopsPrevious studies have explored the physiological and linguistic factors underlyingVOT production in normal speakers across several languages A major clinicalgoal of acoustic analysis in speech disorder is to establish a correlation betweenthe acoustic abnormalities and the phonetic perturbations VOT could thus beused as an acoustic parameter that indicates the phonetic contrast between voicedand voiceless stops This paper includes a critical review of the measurement ofVOT factors of VOT variability and the e ect of neurogenic communicationdisorders on VOT We review the VOT data from subjects who exhibit aphasiaapraxia of speech and dysarthria These studies reveal that VOT perturbationsin aphasia have been interpreted as phonemic or phonetic errors while VOTabnormalities in apraxia of speech and dysarthria grossly remacr ect loss of motorcontrol
Keywords voice onset time acoustic aphasia apraxia of speech dysarthria
Introduction
Voice onset time ( VOT) is an objective acoustic parameter remacr ecting motor speechcontrol It has been measured in normal speakers and those with speech disordersto test linguistic hypotheses or to measure the timing between the structures of the
Address correspondance to Dr P Auzou Service de Neurophysiologie Hopital charlesNicolle 1 Rue de Germont 76031 Rouen Cedex France
Clinical L inguistics amp PhoneticsISSN 0269-9206 printISSN 1464-5076 online Ntilde 2000 Taylor amp Francis Ltd
httpwwwtandfcoukjournalstf 02699206html
P Auzou et al132
larynx and the oral cavity VOT is dereg ned as the time between the release of theoral constriction for plosive production and the onset of vocal-fold vibration ( Liskerand Abramson 1964 1967 Zlatin 1974) This temporal parameter is the mostreliable acoustic cue for the distinction between voiced and voiceless stops Theseconsonants are characterized by the creation of a pressure pulse during the completeocclusion in the vocal tract
Production of word-initial voiceless stops involves a period of articulatory closureduring which the vocal folds are maintained in a relatively open position withoutglottal pulsing During the closure phase intraoral pressure increases rapidly untilit becomes equal to the subglottal pressure ( Hertegard and Gau n 1995) Theequalization of the two pressures creates an aerodynamic environment that inhibitsvocal fold vibration The vocal folds are also actively abducted to interrupt voicingTo end the closure interval the oral contact is released resulting in a rapid macr ow ofair out of the mouth and a rapid drop in the intraoral air pressure This airmacr owcreates a transient burst of acoustic energy Although the glottis remains open atthe moment of articulatory release the vocal folds then approximate to begin theglottal vibration of the following vowel
The duration of the abduction-adduction gesture of the vocal folds is a primarycue to the phonetic voicelessplusmn voiced distinction between the plosive cognates( Benguerel Hirose Sawashima and Ushijima 1978 Hirose and Gay 1972)Abduction of the vocal folds is mainly controlled by the posterior cricoarythenoidmuscle and their adduction is mainly controlled by the interarytenoid muscle(Cooper 1992) Hence the accurate production of stop consonants requires closecoordination between the larynx and the lips tongue and jaw It has been suggestedthat the glottal opening gesture maximum glottal opening and articulatory closurerelease may be coordinated in terms of a reg xed temporal targetrsquo ( Zebrowski Contureand Cudahy 1985) The timing control of related articulators to achieve a temporaltarget suggests that the control of speech movements occurs over aggregates ratherthan individual articulators ( LoEgrave fqvist and Lindblom 1994 Munhall LoEgrave fqvist andKelso 1994) These speech movements might be comprised of a set of motorprograms specireg ed for the individual phonemes that would be activated andsequenced into larger aggregates associated with syllables words and phrases toallow meaningful communication (Gracco and LoEgrave fqvist 1994)
In nearly all languages VOT plays a primary role in distinguishing betweenvoiced and voiceless plosive cognates ( Lisker and Abramson 1964) However acomplete acoustical description of plosive consonants must take into considerationother parameters such as the presence or absence of fricative noise upon consonantalrelease intensity of the noise and duration of the burst release or formant transitions(De Mori and Flammia 1993 Lisker and Abramson 1964 Stevens and Klatt 1974Sussman and Shore 1996) VOT denotes the coarticulatory timing control betweenlaryngeal and oral articulation Factors that can a ect the magnitude of VOT includesize of glottal opening transglottal pressure and vocal fold tension ( LoEgrave fqvist 1992Tyler and Watterson 1991)
The purpose of this paper is to present a review of current information on VOTThe reg rst section includes a review of the methodological considerations when makingVOT measurement Then a section on normative VOT data follows After that isa section on the linguistic and physiological variations that can alter VOT values innormal speech production The reg nal section includes a review of the informationon VOT production in subjects who have aphasia apraxia of speech and dysarthria
VOT and speech disorders 133
The measurement of VOT methodological considerations
The measurement of VOT is usually obtained from wideband spectrograms accordingto the procedure reg rst recommended by Lisker and Abramson ( 1964) That is VOTcorresponds to the interval between the onset of the energy burstrsquo representing therelease of an articulatory constriction and the reg rst of the regularly spaced verticalstriations representing the vocal fold vibration ( reg gure 1) The instant of release isdereg ned as the point where the spectrogram shows an abrupt spectral change repres-enting the transient noise burst When the vocal fold vibration precedes the releaseas in voiced stops the VOT is given a negative value and is called voicinglead ( Lisker and Abramson 1964 Macken and Baron 1980ab Zlatin andKoenigsknecht 1976) On the other hand when the release precedes the vocal foldvibration as in voiceless stops the VOT value is positive and called voicing lag
Lag time measurements are often taken from the burst to the onset of the reg rstformant ( F1) of the following vowel In some studies lag time measurements weretaken from the closure release at the beginning of the burst to the beginning of thesecond formant ( F2) of the following vowel (Gandour 1985 Klatt 1975) Thechoice of the second formant as a delimiter of VOT is particularly functional forvoiced stops because voicing frequently occurs during the lag time making it di cultto determine exactly where the reg rst formant of the vowel begins (Davis 1995)
Figure 1 Oscillogram of the syllable [ka] in a normal subject (top) and the correspondingwide-band spectrogram (bottom) V OT represents the duration of time between theopening of the lips (a on the oscillogram and arsquo on the spectrogram) and the regularstriations representing the vocal fold vibrations (b on the oscillogram and brsquo on thespectrogram)
P Auzou et al134
Generally some delay occurs between the onset of F2 and F1 However thisdi erence is three milliseconds or less for both short and long lag stops Thus usingF2 onset does not introduce an important measurement error
VOT measurements are usually done from isolated words or syllables or fromwords or syllables embedded in carrier phrases (eg Say ___ again Say ___ instead)Using a carrier phrase has two e ects on the subjectrsquo s speaking rate First the longerutterance causes the subject to time the utterance phonemes and syllables in a mannersimilar to conversational speech rather than the prolongation common to citationspeech Second the e ects of prepausal phoneme lengthening in utterance reg nal syl-lables is removed by embedding the target phoneme in the middle of the phrase
Digitized acoustic signals allow the experimenter to obtain time-synchronizedwideband spectrographic and oscillographic displays Rather than using the spectro-gram several authors proposed directly measuring VOT from the oscillogram( Keller 1990 Lane Wozniak and Perkell 1994 Tyler and Watterson 1991 Volaitisand Miller 1992) Using a display of the digitized speech signal these authorsmeasured VOT from the onset of the plosive release burst to the reg rst zero crossingat the onset of periodicity in the waveform However others report that the simultan-eous examination of oscillograms and spectrograms provides a greater accuracy ofmeasurement ( Bortolini Zmarich Fior and Bonifacio 1995 Davis 1995 PetrosinoColcord Kurcz and Yonker 1993) Whenever precise measurement is questionableon the spectrogram the oscillographic display can be used to substantiate onset ofeither the release burst or voicing for the vowel
A physiologic variation in measuring VOT was reported by Morris and Brown( 1987) They used the oscillographic displays from intraoral air pressure changesand a simultaneous signal from a contact microphone on the neck to assess VOTBrown Morris and Weiss ( 1993) reported that this system provided output compar-able to measures made with spectrograms and expanded digital oscillographicdisplays
In some utterances it is not possible to determine the location of the burst( reg gure 2) or the onset of regular striations on the spectrogram ( reg gure 3) Forexample the burst does not occur when a subject fails to achieve full closure in theproduction of stop consonants Similarly the vowel onset may be di cult to deter-mine (Sweeting and Baken 1982) Lisker and Abramson ( 1964) noted the risk ofincluding a short span of pulsation as brief as a single cycle of vocal fold movementthat would be too weak to be audible ( reg gure 4) They believed that such occasionalerrors make no signireg cant di erence in the use of VOT to categorize stop consonantsThe combined use of oscillograms and spectrograms may reduce the frequency ofsuch errors In addition a double burst may make it di cult to determine when therelease the lip or tongue closure occurs ( reg gure 5) In these cases measurements areusually taken from the start of the reg rst burst (Davis 1995) Finally few researchersreport the percentage of unmeasurable VOTs the data available indicate that lessthan 4 of the productions cannot be measured ( Itoh Sasanuma TatsumiMurakami Fukusako and Suzuki 1982 Sweeting and Baken 1982)
Several authors report that the reliability of VOT measurements is high ConnorLudlow and Schultz ( 1989) completed two analyses of the spectrograms producedby 12 normals and 12 dysarthric patients with Parkinsonrsquo s disease ( PD) The inter-measurement reliability was 45 (range 41plusmn 68) for the syllables produced bythe normal subjects and 44 (range 17plusmn 67) for the syllables produced by thePD subjects Caruso and Burton ( 1987) found mean intrajudge measurement errors
VOT and speech disorders 135
Figure 2 Same representation as in Fig1 for two syllables [pa] of a patient with spasticdysarthria No burst was identireg able
of 556msec and interjudge ones of 475msec Other authors have reported closeragreement in terms of percentage agreement within a selected range of durationdi erences such as 966 of reliability measures being within 2 msec (Sweeting andBaken 1982) 100 agreement within 3 msec (Davis 1995) and 975 agreementwithin 5 msec ( Zlatin 1974) Similarly Hoit Solomon and Hixon ( 1993) reported943 interjudge agreement and 951 intrajudge agreement
Norms of VOT production
VOT data have been reported for normal speakers of many languages These studiesindicate similar VOT categories across languages Lisker and Abramson ( 1964)reported mean VOT values for word-initial stop consonants data from normalsubjects who spoke 11 di erent languages However each language was representedby a small pool of subjects The greatest number of studies have been done usingEnglish speaking subjects The results from many of these studies are shown intable 1 The VOT values vary for each consonant and may be best represented as arange of durations Table 1 reveals that the range of VOT values produced for agiven consonant is comparable across studies and across di erent phonetic contextsExtensive cross-language studies by Lisker and Abramson ( 1964 1967) demonstratedthat three categories of stops emerge along the VOT continuum with reasonableagreement as to category boundaries across languages
( 1 ) `Voicing leadrsquo negative VOT values ranging from about Otilde 125 to Otilde 75 msecwith a median value of Otilde 100msec Italian or French voiced stops are ofthis type ( Bortolini et al 1995 Ryalls Provost and Arsenault 1995)
P Auzou et al136
Figure 3 Same representation as in Fig1 for the [ta] of a patient with spastic dysarthriaThe striation does not become regular enough to determine the beginning
( 2 ) Short voicing lagrsquo positive VOT values ranging from 0 to +25 msec witha median value of +10 msec English voiced stops and Italian voiceless stopsare of this type
( 3 ) `Long voicing lagrsquo large positive VOT values ranging from +60 to+100 msec with a median value of +75 msec English voiceless stops areof this type
However Lisker and Abramson ( 1964) reported that in English the phonemes[b] [d] and [g] used two of the VOT categories Their group of four speakersexhibited VOT values for [b] that ranged from either Otilde 130 to Otilde 20 msec (voicinglead ) or from 0 to +5 msec (short lag) The individual speakers did not tend toexhibit both VOT categories One speaker produced 95 of all the stops with voicinglead and a second one produced the remaining 5 The other two speakers producedalmost exclusively positive values or short lag ( 4142 for one speaker and 5658 forthe other) Because the voiced stops in English occur in two VOT categories whereasthe voiceless stops occur in one VOT category the VOT values of voiceless stopstend to be more normally distributed than their voiced counterparts ( Zlatin 1974)
Thai is a language that uses all three voicing categories Lisker and Abramson
VOT and speech disorders 137
Figure 4 Oscillogram and spectrogram of a [ka] in a normal subject If using only theoscillogram the beginning of the periodic vibration may be doubtful (b) W hen usingcombined oscillogram and spectrogram the ambiguity disappears (correct beginningin brsquo)
( 1964) reported that their Thai speakers used lead short lag and long lag respectivelyfor voiced unaspirated voiceless and aspirated voiceless stops
Across languages the same consonant can demonstrate notable VOT di erencesin its phonetic expression For instance the mean VOT values for voiceless conson-ants are much lower in Thai or Korean than in English However common character-istics can be observed in the data across language In most languages the VOTvalues for voiced and voiceless stops are produced in discrete duration ranges thatcorrespond to the voicing categories That is the categories of VOT values areseparated by a range of times in which no production occurs This categorizationindicates active control for the voiced and voiceless cognates in the timing relation-ship between the articulatory release and the onset of vocal fold vibration Thearticulatory gestures for short voicing lag stops are easier than for the other twotypes of VOT Voicing lead requires more muscles gestures than those needed forshort voicing lag stops The production of initial stop consonants with voicing leadnecessitates some mechanisms external to the larynx to sustain an adequate transglot-tal pressure drop during the closure In contrast the stop consonants with longvoicing lag require more carefully controlled timing between the oral stop and
P Auzou et al138
Figure 5 Oscillogram and spectrogram of a [ka] in a normal subject Sometimes a doubleburst is observed We consider the reg rst to measure V OT
laryngeal closure For these consonants the adduction of the vocal folds requiresmore complex muscle activity ( Dixit 1989 Tyler and Watterson 1991 Vohr Garcia-Coll and Oh 1988 Westbury and Keating 1986) In producing the voicing contrasta speaker needs to coordinate the timing of velopharyngeal closure supraglottalarticulators closure vocal fold oscillation and supraglottal articulator release
Another cross language commonality is that VOT varies regularly with the placeof articulation in all languages The lag time values tend to increase as the positionof occlusion moves posteriorly within the oral cavity the mean VOT values for [p]are shorter than for [t] which in turn are shorter than for [k] ( reg gure 6) ( Baum andRyan 1993 Klatt 1975 Miller Green and Reeves 1986 Volaitis and Miller 1992)However the range of VOT values overlap extensively across places of articulationfor both voiced and voiceless consonants
Factors that can alter VOT production
Changes occur in VOT as a consequence of physiological di erences (such as agelung volume) pathological status ( hearing impairment depression) and di erentlinguistic tasks (speech task speech rate phoneme environment ) One such factorthat has been investigated is the age of the speaker Variations in VOT have beenreported across the life span Eguchi and Hirsh ( 1969) reported high VOT variability
VOT and speech disorders 139
Tab
le1
Nor
mat
ive
valu
esof
VO
Tin
Eng
lish
liter
atur
ein
mill
isec
onds
(mea
nsan
dst
anda
rdde
viat
ions
orra
nge
ifav
aila
ble)
S
ome
ofth
emar
eap
prox
imat
edfr
omregg
ures
L
iske
ran
dA
bram
son
(196
4)ga
vetw
ora
nges
ofva
lues
for
voic
edst
ops
corr
espo
ndin
gre
spec
tive
lyto
lead
and
shor
tla
g(s
eete
xt)
C
Con
sona
nt
V
Vow
el
Num
ber
ofA
utho
rssu
bjec
tsp
tk
bd
gC
onte
xt
Lis
ker
and
458
(20
120
)70
(30
105
)80
(50
135
)Otilde
101
(Otilde13
0Otilde
20)
Otilde10
2(Otilde
155
Otilde40
)Otilde
88(Otilde
150
Otilde60
)A
bram
son
1964
1(0
5)
5(0
25
)21
(03
5)
Kla
tt
1975
347
6570
1117
27Sa
yC
Vin
stea
dSw
eeti
ngan
d30
766
(Ocirc4
6)
139
(Ocirc1
4)
Ita
CV
Bak
en
1982
Har
dcas
tle
etal
4
653
8082
816
915
721
3Si
ngle
wor
d19
85M
orri
san
dB
row
n25
7065
100
1987
Car
uso
and
Bur
ton
862
5(2
54
)71
9(1
94
)74
8(7
48
)19
7(8
5)
214
(49
)35
2(8
4)
Say
CV
Cag
ain
1987
Lee
etal
19
8810
725
(26
9)
Tw
enty
-one
For
rest
etal
19
898
46(1
0)
12(3
)B
uyB
obby
aP
oppy
Bau
man
dR
yan
1085
9511
010
1530
Ple
ase
say
CV
C19
93B
row
net
al
1993
1257
652
64
168
Spea
kC
aag
ain
Pet
rosi
noet
al
1089
(20
)14
(24
)Sa
yC
atag
ain
1993
Dav
is
1995
1090
30
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
P Auzou et al132
larynx and the oral cavity VOT is dereg ned as the time between the release of theoral constriction for plosive production and the onset of vocal-fold vibration ( Liskerand Abramson 1964 1967 Zlatin 1974) This temporal parameter is the mostreliable acoustic cue for the distinction between voiced and voiceless stops Theseconsonants are characterized by the creation of a pressure pulse during the completeocclusion in the vocal tract
Production of word-initial voiceless stops involves a period of articulatory closureduring which the vocal folds are maintained in a relatively open position withoutglottal pulsing During the closure phase intraoral pressure increases rapidly untilit becomes equal to the subglottal pressure ( Hertegard and Gau n 1995) Theequalization of the two pressures creates an aerodynamic environment that inhibitsvocal fold vibration The vocal folds are also actively abducted to interrupt voicingTo end the closure interval the oral contact is released resulting in a rapid macr ow ofair out of the mouth and a rapid drop in the intraoral air pressure This airmacr owcreates a transient burst of acoustic energy Although the glottis remains open atthe moment of articulatory release the vocal folds then approximate to begin theglottal vibration of the following vowel
The duration of the abduction-adduction gesture of the vocal folds is a primarycue to the phonetic voicelessplusmn voiced distinction between the plosive cognates( Benguerel Hirose Sawashima and Ushijima 1978 Hirose and Gay 1972)Abduction of the vocal folds is mainly controlled by the posterior cricoarythenoidmuscle and their adduction is mainly controlled by the interarytenoid muscle(Cooper 1992) Hence the accurate production of stop consonants requires closecoordination between the larynx and the lips tongue and jaw It has been suggestedthat the glottal opening gesture maximum glottal opening and articulatory closurerelease may be coordinated in terms of a reg xed temporal targetrsquo ( Zebrowski Contureand Cudahy 1985) The timing control of related articulators to achieve a temporaltarget suggests that the control of speech movements occurs over aggregates ratherthan individual articulators ( LoEgrave fqvist and Lindblom 1994 Munhall LoEgrave fqvist andKelso 1994) These speech movements might be comprised of a set of motorprograms specireg ed for the individual phonemes that would be activated andsequenced into larger aggregates associated with syllables words and phrases toallow meaningful communication (Gracco and LoEgrave fqvist 1994)
In nearly all languages VOT plays a primary role in distinguishing betweenvoiced and voiceless plosive cognates ( Lisker and Abramson 1964) However acomplete acoustical description of plosive consonants must take into considerationother parameters such as the presence or absence of fricative noise upon consonantalrelease intensity of the noise and duration of the burst release or formant transitions(De Mori and Flammia 1993 Lisker and Abramson 1964 Stevens and Klatt 1974Sussman and Shore 1996) VOT denotes the coarticulatory timing control betweenlaryngeal and oral articulation Factors that can a ect the magnitude of VOT includesize of glottal opening transglottal pressure and vocal fold tension ( LoEgrave fqvist 1992Tyler and Watterson 1991)
The purpose of this paper is to present a review of current information on VOTThe reg rst section includes a review of the methodological considerations when makingVOT measurement Then a section on normative VOT data follows After that isa section on the linguistic and physiological variations that can alter VOT values innormal speech production The reg nal section includes a review of the informationon VOT production in subjects who have aphasia apraxia of speech and dysarthria
VOT and speech disorders 133
The measurement of VOT methodological considerations
The measurement of VOT is usually obtained from wideband spectrograms accordingto the procedure reg rst recommended by Lisker and Abramson ( 1964) That is VOTcorresponds to the interval between the onset of the energy burstrsquo representing therelease of an articulatory constriction and the reg rst of the regularly spaced verticalstriations representing the vocal fold vibration ( reg gure 1) The instant of release isdereg ned as the point where the spectrogram shows an abrupt spectral change repres-enting the transient noise burst When the vocal fold vibration precedes the releaseas in voiced stops the VOT is given a negative value and is called voicinglead ( Lisker and Abramson 1964 Macken and Baron 1980ab Zlatin andKoenigsknecht 1976) On the other hand when the release precedes the vocal foldvibration as in voiceless stops the VOT value is positive and called voicing lag
Lag time measurements are often taken from the burst to the onset of the reg rstformant ( F1) of the following vowel In some studies lag time measurements weretaken from the closure release at the beginning of the burst to the beginning of thesecond formant ( F2) of the following vowel (Gandour 1985 Klatt 1975) Thechoice of the second formant as a delimiter of VOT is particularly functional forvoiced stops because voicing frequently occurs during the lag time making it di cultto determine exactly where the reg rst formant of the vowel begins (Davis 1995)
Figure 1 Oscillogram of the syllable [ka] in a normal subject (top) and the correspondingwide-band spectrogram (bottom) V OT represents the duration of time between theopening of the lips (a on the oscillogram and arsquo on the spectrogram) and the regularstriations representing the vocal fold vibrations (b on the oscillogram and brsquo on thespectrogram)
P Auzou et al134
Generally some delay occurs between the onset of F2 and F1 However thisdi erence is three milliseconds or less for both short and long lag stops Thus usingF2 onset does not introduce an important measurement error
VOT measurements are usually done from isolated words or syllables or fromwords or syllables embedded in carrier phrases (eg Say ___ again Say ___ instead)Using a carrier phrase has two e ects on the subjectrsquo s speaking rate First the longerutterance causes the subject to time the utterance phonemes and syllables in a mannersimilar to conversational speech rather than the prolongation common to citationspeech Second the e ects of prepausal phoneme lengthening in utterance reg nal syl-lables is removed by embedding the target phoneme in the middle of the phrase
Digitized acoustic signals allow the experimenter to obtain time-synchronizedwideband spectrographic and oscillographic displays Rather than using the spectro-gram several authors proposed directly measuring VOT from the oscillogram( Keller 1990 Lane Wozniak and Perkell 1994 Tyler and Watterson 1991 Volaitisand Miller 1992) Using a display of the digitized speech signal these authorsmeasured VOT from the onset of the plosive release burst to the reg rst zero crossingat the onset of periodicity in the waveform However others report that the simultan-eous examination of oscillograms and spectrograms provides a greater accuracy ofmeasurement ( Bortolini Zmarich Fior and Bonifacio 1995 Davis 1995 PetrosinoColcord Kurcz and Yonker 1993) Whenever precise measurement is questionableon the spectrogram the oscillographic display can be used to substantiate onset ofeither the release burst or voicing for the vowel
A physiologic variation in measuring VOT was reported by Morris and Brown( 1987) They used the oscillographic displays from intraoral air pressure changesand a simultaneous signal from a contact microphone on the neck to assess VOTBrown Morris and Weiss ( 1993) reported that this system provided output compar-able to measures made with spectrograms and expanded digital oscillographicdisplays
In some utterances it is not possible to determine the location of the burst( reg gure 2) or the onset of regular striations on the spectrogram ( reg gure 3) Forexample the burst does not occur when a subject fails to achieve full closure in theproduction of stop consonants Similarly the vowel onset may be di cult to deter-mine (Sweeting and Baken 1982) Lisker and Abramson ( 1964) noted the risk ofincluding a short span of pulsation as brief as a single cycle of vocal fold movementthat would be too weak to be audible ( reg gure 4) They believed that such occasionalerrors make no signireg cant di erence in the use of VOT to categorize stop consonantsThe combined use of oscillograms and spectrograms may reduce the frequency ofsuch errors In addition a double burst may make it di cult to determine when therelease the lip or tongue closure occurs ( reg gure 5) In these cases measurements areusually taken from the start of the reg rst burst (Davis 1995) Finally few researchersreport the percentage of unmeasurable VOTs the data available indicate that lessthan 4 of the productions cannot be measured ( Itoh Sasanuma TatsumiMurakami Fukusako and Suzuki 1982 Sweeting and Baken 1982)
Several authors report that the reliability of VOT measurements is high ConnorLudlow and Schultz ( 1989) completed two analyses of the spectrograms producedby 12 normals and 12 dysarthric patients with Parkinsonrsquo s disease ( PD) The inter-measurement reliability was 45 (range 41plusmn 68) for the syllables produced bythe normal subjects and 44 (range 17plusmn 67) for the syllables produced by thePD subjects Caruso and Burton ( 1987) found mean intrajudge measurement errors
VOT and speech disorders 135
Figure 2 Same representation as in Fig1 for two syllables [pa] of a patient with spasticdysarthria No burst was identireg able
of 556msec and interjudge ones of 475msec Other authors have reported closeragreement in terms of percentage agreement within a selected range of durationdi erences such as 966 of reliability measures being within 2 msec (Sweeting andBaken 1982) 100 agreement within 3 msec (Davis 1995) and 975 agreementwithin 5 msec ( Zlatin 1974) Similarly Hoit Solomon and Hixon ( 1993) reported943 interjudge agreement and 951 intrajudge agreement
Norms of VOT production
VOT data have been reported for normal speakers of many languages These studiesindicate similar VOT categories across languages Lisker and Abramson ( 1964)reported mean VOT values for word-initial stop consonants data from normalsubjects who spoke 11 di erent languages However each language was representedby a small pool of subjects The greatest number of studies have been done usingEnglish speaking subjects The results from many of these studies are shown intable 1 The VOT values vary for each consonant and may be best represented as arange of durations Table 1 reveals that the range of VOT values produced for agiven consonant is comparable across studies and across di erent phonetic contextsExtensive cross-language studies by Lisker and Abramson ( 1964 1967) demonstratedthat three categories of stops emerge along the VOT continuum with reasonableagreement as to category boundaries across languages
( 1 ) `Voicing leadrsquo negative VOT values ranging from about Otilde 125 to Otilde 75 msecwith a median value of Otilde 100msec Italian or French voiced stops are ofthis type ( Bortolini et al 1995 Ryalls Provost and Arsenault 1995)
P Auzou et al136
Figure 3 Same representation as in Fig1 for the [ta] of a patient with spastic dysarthriaThe striation does not become regular enough to determine the beginning
( 2 ) Short voicing lagrsquo positive VOT values ranging from 0 to +25 msec witha median value of +10 msec English voiced stops and Italian voiceless stopsare of this type
( 3 ) `Long voicing lagrsquo large positive VOT values ranging from +60 to+100 msec with a median value of +75 msec English voiceless stops areof this type
However Lisker and Abramson ( 1964) reported that in English the phonemes[b] [d] and [g] used two of the VOT categories Their group of four speakersexhibited VOT values for [b] that ranged from either Otilde 130 to Otilde 20 msec (voicinglead ) or from 0 to +5 msec (short lag) The individual speakers did not tend toexhibit both VOT categories One speaker produced 95 of all the stops with voicinglead and a second one produced the remaining 5 The other two speakers producedalmost exclusively positive values or short lag ( 4142 for one speaker and 5658 forthe other) Because the voiced stops in English occur in two VOT categories whereasthe voiceless stops occur in one VOT category the VOT values of voiceless stopstend to be more normally distributed than their voiced counterparts ( Zlatin 1974)
Thai is a language that uses all three voicing categories Lisker and Abramson
VOT and speech disorders 137
Figure 4 Oscillogram and spectrogram of a [ka] in a normal subject If using only theoscillogram the beginning of the periodic vibration may be doubtful (b) W hen usingcombined oscillogram and spectrogram the ambiguity disappears (correct beginningin brsquo)
( 1964) reported that their Thai speakers used lead short lag and long lag respectivelyfor voiced unaspirated voiceless and aspirated voiceless stops
Across languages the same consonant can demonstrate notable VOT di erencesin its phonetic expression For instance the mean VOT values for voiceless conson-ants are much lower in Thai or Korean than in English However common character-istics can be observed in the data across language In most languages the VOTvalues for voiced and voiceless stops are produced in discrete duration ranges thatcorrespond to the voicing categories That is the categories of VOT values areseparated by a range of times in which no production occurs This categorizationindicates active control for the voiced and voiceless cognates in the timing relation-ship between the articulatory release and the onset of vocal fold vibration Thearticulatory gestures for short voicing lag stops are easier than for the other twotypes of VOT Voicing lead requires more muscles gestures than those needed forshort voicing lag stops The production of initial stop consonants with voicing leadnecessitates some mechanisms external to the larynx to sustain an adequate transglot-tal pressure drop during the closure In contrast the stop consonants with longvoicing lag require more carefully controlled timing between the oral stop and
P Auzou et al138
Figure 5 Oscillogram and spectrogram of a [ka] in a normal subject Sometimes a doubleburst is observed We consider the reg rst to measure V OT
laryngeal closure For these consonants the adduction of the vocal folds requiresmore complex muscle activity ( Dixit 1989 Tyler and Watterson 1991 Vohr Garcia-Coll and Oh 1988 Westbury and Keating 1986) In producing the voicing contrasta speaker needs to coordinate the timing of velopharyngeal closure supraglottalarticulators closure vocal fold oscillation and supraglottal articulator release
Another cross language commonality is that VOT varies regularly with the placeof articulation in all languages The lag time values tend to increase as the positionof occlusion moves posteriorly within the oral cavity the mean VOT values for [p]are shorter than for [t] which in turn are shorter than for [k] ( reg gure 6) ( Baum andRyan 1993 Klatt 1975 Miller Green and Reeves 1986 Volaitis and Miller 1992)However the range of VOT values overlap extensively across places of articulationfor both voiced and voiceless consonants
Factors that can alter VOT production
Changes occur in VOT as a consequence of physiological di erences (such as agelung volume) pathological status ( hearing impairment depression) and di erentlinguistic tasks (speech task speech rate phoneme environment ) One such factorthat has been investigated is the age of the speaker Variations in VOT have beenreported across the life span Eguchi and Hirsh ( 1969) reported high VOT variability
VOT and speech disorders 139
Tab
le1
Nor
mat
ive
valu
esof
VO
Tin
Eng
lish
liter
atur
ein
mill
isec
onds
(mea
nsan
dst
anda
rdde
viat
ions
orra
nge
ifav
aila
ble)
S
ome
ofth
emar
eap
prox
imat
edfr
omregg
ures
L
iske
ran
dA
bram
son
(196
4)ga
vetw
ora
nges
ofva
lues
for
voic
edst
ops
corr
espo
ndin
gre
spec
tive
lyto
lead
and
shor
tla
g(s
eete
xt)
C
Con
sona
nt
V
Vow
el
Num
ber
ofA
utho
rssu
bjec
tsp
tk
bd
gC
onte
xt
Lis
ker
and
458
(20
120
)70
(30
105
)80
(50
135
)Otilde
101
(Otilde13
0Otilde
20)
Otilde10
2(Otilde
155
Otilde40
)Otilde
88(Otilde
150
Otilde60
)A
bram
son
1964
1(0
5)
5(0
25
)21
(03
5)
Kla
tt
1975
347
6570
1117
27Sa
yC
Vin
stea
dSw
eeti
ngan
d30
766
(Ocirc4
6)
139
(Ocirc1
4)
Ita
CV
Bak
en
1982
Har
dcas
tle
etal
4
653
8082
816
915
721
3Si
ngle
wor
d19
85M
orri
san
dB
row
n25
7065
100
1987
Car
uso
and
Bur
ton
862
5(2
54
)71
9(1
94
)74
8(7
48
)19
7(8
5)
214
(49
)35
2(8
4)
Say
CV
Cag
ain
1987
Lee
etal
19
8810
725
(26
9)
Tw
enty
-one
For
rest
etal
19
898
46(1
0)
12(3
)B
uyB
obby
aP
oppy
Bau
man
dR
yan
1085
9511
010
1530
Ple
ase
say
CV
C19
93B
row
net
al
1993
1257
652
64
168
Spea
kC
aag
ain
Pet
rosi
noet
al
1089
(20
)14
(24
)Sa
yC
atag
ain
1993
Dav
is
1995
1090
30
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders 133
The measurement of VOT methodological considerations
The measurement of VOT is usually obtained from wideband spectrograms accordingto the procedure reg rst recommended by Lisker and Abramson ( 1964) That is VOTcorresponds to the interval between the onset of the energy burstrsquo representing therelease of an articulatory constriction and the reg rst of the regularly spaced verticalstriations representing the vocal fold vibration ( reg gure 1) The instant of release isdereg ned as the point where the spectrogram shows an abrupt spectral change repres-enting the transient noise burst When the vocal fold vibration precedes the releaseas in voiced stops the VOT is given a negative value and is called voicinglead ( Lisker and Abramson 1964 Macken and Baron 1980ab Zlatin andKoenigsknecht 1976) On the other hand when the release precedes the vocal foldvibration as in voiceless stops the VOT value is positive and called voicing lag
Lag time measurements are often taken from the burst to the onset of the reg rstformant ( F1) of the following vowel In some studies lag time measurements weretaken from the closure release at the beginning of the burst to the beginning of thesecond formant ( F2) of the following vowel (Gandour 1985 Klatt 1975) Thechoice of the second formant as a delimiter of VOT is particularly functional forvoiced stops because voicing frequently occurs during the lag time making it di cultto determine exactly where the reg rst formant of the vowel begins (Davis 1995)
Figure 1 Oscillogram of the syllable [ka] in a normal subject (top) and the correspondingwide-band spectrogram (bottom) V OT represents the duration of time between theopening of the lips (a on the oscillogram and arsquo on the spectrogram) and the regularstriations representing the vocal fold vibrations (b on the oscillogram and brsquo on thespectrogram)
P Auzou et al134
Generally some delay occurs between the onset of F2 and F1 However thisdi erence is three milliseconds or less for both short and long lag stops Thus usingF2 onset does not introduce an important measurement error
VOT measurements are usually done from isolated words or syllables or fromwords or syllables embedded in carrier phrases (eg Say ___ again Say ___ instead)Using a carrier phrase has two e ects on the subjectrsquo s speaking rate First the longerutterance causes the subject to time the utterance phonemes and syllables in a mannersimilar to conversational speech rather than the prolongation common to citationspeech Second the e ects of prepausal phoneme lengthening in utterance reg nal syl-lables is removed by embedding the target phoneme in the middle of the phrase
Digitized acoustic signals allow the experimenter to obtain time-synchronizedwideband spectrographic and oscillographic displays Rather than using the spectro-gram several authors proposed directly measuring VOT from the oscillogram( Keller 1990 Lane Wozniak and Perkell 1994 Tyler and Watterson 1991 Volaitisand Miller 1992) Using a display of the digitized speech signal these authorsmeasured VOT from the onset of the plosive release burst to the reg rst zero crossingat the onset of periodicity in the waveform However others report that the simultan-eous examination of oscillograms and spectrograms provides a greater accuracy ofmeasurement ( Bortolini Zmarich Fior and Bonifacio 1995 Davis 1995 PetrosinoColcord Kurcz and Yonker 1993) Whenever precise measurement is questionableon the spectrogram the oscillographic display can be used to substantiate onset ofeither the release burst or voicing for the vowel
A physiologic variation in measuring VOT was reported by Morris and Brown( 1987) They used the oscillographic displays from intraoral air pressure changesand a simultaneous signal from a contact microphone on the neck to assess VOTBrown Morris and Weiss ( 1993) reported that this system provided output compar-able to measures made with spectrograms and expanded digital oscillographicdisplays
In some utterances it is not possible to determine the location of the burst( reg gure 2) or the onset of regular striations on the spectrogram ( reg gure 3) Forexample the burst does not occur when a subject fails to achieve full closure in theproduction of stop consonants Similarly the vowel onset may be di cult to deter-mine (Sweeting and Baken 1982) Lisker and Abramson ( 1964) noted the risk ofincluding a short span of pulsation as brief as a single cycle of vocal fold movementthat would be too weak to be audible ( reg gure 4) They believed that such occasionalerrors make no signireg cant di erence in the use of VOT to categorize stop consonantsThe combined use of oscillograms and spectrograms may reduce the frequency ofsuch errors In addition a double burst may make it di cult to determine when therelease the lip or tongue closure occurs ( reg gure 5) In these cases measurements areusually taken from the start of the reg rst burst (Davis 1995) Finally few researchersreport the percentage of unmeasurable VOTs the data available indicate that lessthan 4 of the productions cannot be measured ( Itoh Sasanuma TatsumiMurakami Fukusako and Suzuki 1982 Sweeting and Baken 1982)
Several authors report that the reliability of VOT measurements is high ConnorLudlow and Schultz ( 1989) completed two analyses of the spectrograms producedby 12 normals and 12 dysarthric patients with Parkinsonrsquo s disease ( PD) The inter-measurement reliability was 45 (range 41plusmn 68) for the syllables produced bythe normal subjects and 44 (range 17plusmn 67) for the syllables produced by thePD subjects Caruso and Burton ( 1987) found mean intrajudge measurement errors
VOT and speech disorders 135
Figure 2 Same representation as in Fig1 for two syllables [pa] of a patient with spasticdysarthria No burst was identireg able
of 556msec and interjudge ones of 475msec Other authors have reported closeragreement in terms of percentage agreement within a selected range of durationdi erences such as 966 of reliability measures being within 2 msec (Sweeting andBaken 1982) 100 agreement within 3 msec (Davis 1995) and 975 agreementwithin 5 msec ( Zlatin 1974) Similarly Hoit Solomon and Hixon ( 1993) reported943 interjudge agreement and 951 intrajudge agreement
Norms of VOT production
VOT data have been reported for normal speakers of many languages These studiesindicate similar VOT categories across languages Lisker and Abramson ( 1964)reported mean VOT values for word-initial stop consonants data from normalsubjects who spoke 11 di erent languages However each language was representedby a small pool of subjects The greatest number of studies have been done usingEnglish speaking subjects The results from many of these studies are shown intable 1 The VOT values vary for each consonant and may be best represented as arange of durations Table 1 reveals that the range of VOT values produced for agiven consonant is comparable across studies and across di erent phonetic contextsExtensive cross-language studies by Lisker and Abramson ( 1964 1967) demonstratedthat three categories of stops emerge along the VOT continuum with reasonableagreement as to category boundaries across languages
( 1 ) `Voicing leadrsquo negative VOT values ranging from about Otilde 125 to Otilde 75 msecwith a median value of Otilde 100msec Italian or French voiced stops are ofthis type ( Bortolini et al 1995 Ryalls Provost and Arsenault 1995)
P Auzou et al136
Figure 3 Same representation as in Fig1 for the [ta] of a patient with spastic dysarthriaThe striation does not become regular enough to determine the beginning
( 2 ) Short voicing lagrsquo positive VOT values ranging from 0 to +25 msec witha median value of +10 msec English voiced stops and Italian voiceless stopsare of this type
( 3 ) `Long voicing lagrsquo large positive VOT values ranging from +60 to+100 msec with a median value of +75 msec English voiceless stops areof this type
However Lisker and Abramson ( 1964) reported that in English the phonemes[b] [d] and [g] used two of the VOT categories Their group of four speakersexhibited VOT values for [b] that ranged from either Otilde 130 to Otilde 20 msec (voicinglead ) or from 0 to +5 msec (short lag) The individual speakers did not tend toexhibit both VOT categories One speaker produced 95 of all the stops with voicinglead and a second one produced the remaining 5 The other two speakers producedalmost exclusively positive values or short lag ( 4142 for one speaker and 5658 forthe other) Because the voiced stops in English occur in two VOT categories whereasthe voiceless stops occur in one VOT category the VOT values of voiceless stopstend to be more normally distributed than their voiced counterparts ( Zlatin 1974)
Thai is a language that uses all three voicing categories Lisker and Abramson
VOT and speech disorders 137
Figure 4 Oscillogram and spectrogram of a [ka] in a normal subject If using only theoscillogram the beginning of the periodic vibration may be doubtful (b) W hen usingcombined oscillogram and spectrogram the ambiguity disappears (correct beginningin brsquo)
( 1964) reported that their Thai speakers used lead short lag and long lag respectivelyfor voiced unaspirated voiceless and aspirated voiceless stops
Across languages the same consonant can demonstrate notable VOT di erencesin its phonetic expression For instance the mean VOT values for voiceless conson-ants are much lower in Thai or Korean than in English However common character-istics can be observed in the data across language In most languages the VOTvalues for voiced and voiceless stops are produced in discrete duration ranges thatcorrespond to the voicing categories That is the categories of VOT values areseparated by a range of times in which no production occurs This categorizationindicates active control for the voiced and voiceless cognates in the timing relation-ship between the articulatory release and the onset of vocal fold vibration Thearticulatory gestures for short voicing lag stops are easier than for the other twotypes of VOT Voicing lead requires more muscles gestures than those needed forshort voicing lag stops The production of initial stop consonants with voicing leadnecessitates some mechanisms external to the larynx to sustain an adequate transglot-tal pressure drop during the closure In contrast the stop consonants with longvoicing lag require more carefully controlled timing between the oral stop and
P Auzou et al138
Figure 5 Oscillogram and spectrogram of a [ka] in a normal subject Sometimes a doubleburst is observed We consider the reg rst to measure V OT
laryngeal closure For these consonants the adduction of the vocal folds requiresmore complex muscle activity ( Dixit 1989 Tyler and Watterson 1991 Vohr Garcia-Coll and Oh 1988 Westbury and Keating 1986) In producing the voicing contrasta speaker needs to coordinate the timing of velopharyngeal closure supraglottalarticulators closure vocal fold oscillation and supraglottal articulator release
Another cross language commonality is that VOT varies regularly with the placeof articulation in all languages The lag time values tend to increase as the positionof occlusion moves posteriorly within the oral cavity the mean VOT values for [p]are shorter than for [t] which in turn are shorter than for [k] ( reg gure 6) ( Baum andRyan 1993 Klatt 1975 Miller Green and Reeves 1986 Volaitis and Miller 1992)However the range of VOT values overlap extensively across places of articulationfor both voiced and voiceless consonants
Factors that can alter VOT production
Changes occur in VOT as a consequence of physiological di erences (such as agelung volume) pathological status ( hearing impairment depression) and di erentlinguistic tasks (speech task speech rate phoneme environment ) One such factorthat has been investigated is the age of the speaker Variations in VOT have beenreported across the life span Eguchi and Hirsh ( 1969) reported high VOT variability
VOT and speech disorders 139
Tab
le1
Nor
mat
ive
valu
esof
VO
Tin
Eng
lish
liter
atur
ein
mill
isec
onds
(mea
nsan
dst
anda
rdde
viat
ions
orra
nge
ifav
aila
ble)
S
ome
ofth
emar
eap
prox
imat
edfr
omregg
ures
L
iske
ran
dA
bram
son
(196
4)ga
vetw
ora
nges
ofva
lues
for
voic
edst
ops
corr
espo
ndin
gre
spec
tive
lyto
lead
and
shor
tla
g(s
eete
xt)
C
Con
sona
nt
V
Vow
el
Num
ber
ofA
utho
rssu
bjec
tsp
tk
bd
gC
onte
xt
Lis
ker
and
458
(20
120
)70
(30
105
)80
(50
135
)Otilde
101
(Otilde13
0Otilde
20)
Otilde10
2(Otilde
155
Otilde40
)Otilde
88(Otilde
150
Otilde60
)A
bram
son
1964
1(0
5)
5(0
25
)21
(03
5)
Kla
tt
1975
347
6570
1117
27Sa
yC
Vin
stea
dSw
eeti
ngan
d30
766
(Ocirc4
6)
139
(Ocirc1
4)
Ita
CV
Bak
en
1982
Har
dcas
tle
etal
4
653
8082
816
915
721
3Si
ngle
wor
d19
85M
orri
san
dB
row
n25
7065
100
1987
Car
uso
and
Bur
ton
862
5(2
54
)71
9(1
94
)74
8(7
48
)19
7(8
5)
214
(49
)35
2(8
4)
Say
CV
Cag
ain
1987
Lee
etal
19
8810
725
(26
9)
Tw
enty
-one
For
rest
etal
19
898
46(1
0)
12(3
)B
uyB
obby
aP
oppy
Bau
man
dR
yan
1085
9511
010
1530
Ple
ase
say
CV
C19
93B
row
net
al
1993
1257
652
64
168
Spea
kC
aag
ain
Pet
rosi
noet
al
1089
(20
)14
(24
)Sa
yC
atag
ain
1993
Dav
is
1995
1090
30
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
P Auzou et al134
Generally some delay occurs between the onset of F2 and F1 However thisdi erence is three milliseconds or less for both short and long lag stops Thus usingF2 onset does not introduce an important measurement error
VOT measurements are usually done from isolated words or syllables or fromwords or syllables embedded in carrier phrases (eg Say ___ again Say ___ instead)Using a carrier phrase has two e ects on the subjectrsquo s speaking rate First the longerutterance causes the subject to time the utterance phonemes and syllables in a mannersimilar to conversational speech rather than the prolongation common to citationspeech Second the e ects of prepausal phoneme lengthening in utterance reg nal syl-lables is removed by embedding the target phoneme in the middle of the phrase
Digitized acoustic signals allow the experimenter to obtain time-synchronizedwideband spectrographic and oscillographic displays Rather than using the spectro-gram several authors proposed directly measuring VOT from the oscillogram( Keller 1990 Lane Wozniak and Perkell 1994 Tyler and Watterson 1991 Volaitisand Miller 1992) Using a display of the digitized speech signal these authorsmeasured VOT from the onset of the plosive release burst to the reg rst zero crossingat the onset of periodicity in the waveform However others report that the simultan-eous examination of oscillograms and spectrograms provides a greater accuracy ofmeasurement ( Bortolini Zmarich Fior and Bonifacio 1995 Davis 1995 PetrosinoColcord Kurcz and Yonker 1993) Whenever precise measurement is questionableon the spectrogram the oscillographic display can be used to substantiate onset ofeither the release burst or voicing for the vowel
A physiologic variation in measuring VOT was reported by Morris and Brown( 1987) They used the oscillographic displays from intraoral air pressure changesand a simultaneous signal from a contact microphone on the neck to assess VOTBrown Morris and Weiss ( 1993) reported that this system provided output compar-able to measures made with spectrograms and expanded digital oscillographicdisplays
In some utterances it is not possible to determine the location of the burst( reg gure 2) or the onset of regular striations on the spectrogram ( reg gure 3) Forexample the burst does not occur when a subject fails to achieve full closure in theproduction of stop consonants Similarly the vowel onset may be di cult to deter-mine (Sweeting and Baken 1982) Lisker and Abramson ( 1964) noted the risk ofincluding a short span of pulsation as brief as a single cycle of vocal fold movementthat would be too weak to be audible ( reg gure 4) They believed that such occasionalerrors make no signireg cant di erence in the use of VOT to categorize stop consonantsThe combined use of oscillograms and spectrograms may reduce the frequency ofsuch errors In addition a double burst may make it di cult to determine when therelease the lip or tongue closure occurs ( reg gure 5) In these cases measurements areusually taken from the start of the reg rst burst (Davis 1995) Finally few researchersreport the percentage of unmeasurable VOTs the data available indicate that lessthan 4 of the productions cannot be measured ( Itoh Sasanuma TatsumiMurakami Fukusako and Suzuki 1982 Sweeting and Baken 1982)
Several authors report that the reliability of VOT measurements is high ConnorLudlow and Schultz ( 1989) completed two analyses of the spectrograms producedby 12 normals and 12 dysarthric patients with Parkinsonrsquo s disease ( PD) The inter-measurement reliability was 45 (range 41plusmn 68) for the syllables produced bythe normal subjects and 44 (range 17plusmn 67) for the syllables produced by thePD subjects Caruso and Burton ( 1987) found mean intrajudge measurement errors
VOT and speech disorders 135
Figure 2 Same representation as in Fig1 for two syllables [pa] of a patient with spasticdysarthria No burst was identireg able
of 556msec and interjudge ones of 475msec Other authors have reported closeragreement in terms of percentage agreement within a selected range of durationdi erences such as 966 of reliability measures being within 2 msec (Sweeting andBaken 1982) 100 agreement within 3 msec (Davis 1995) and 975 agreementwithin 5 msec ( Zlatin 1974) Similarly Hoit Solomon and Hixon ( 1993) reported943 interjudge agreement and 951 intrajudge agreement
Norms of VOT production
VOT data have been reported for normal speakers of many languages These studiesindicate similar VOT categories across languages Lisker and Abramson ( 1964)reported mean VOT values for word-initial stop consonants data from normalsubjects who spoke 11 di erent languages However each language was representedby a small pool of subjects The greatest number of studies have been done usingEnglish speaking subjects The results from many of these studies are shown intable 1 The VOT values vary for each consonant and may be best represented as arange of durations Table 1 reveals that the range of VOT values produced for agiven consonant is comparable across studies and across di erent phonetic contextsExtensive cross-language studies by Lisker and Abramson ( 1964 1967) demonstratedthat three categories of stops emerge along the VOT continuum with reasonableagreement as to category boundaries across languages
( 1 ) `Voicing leadrsquo negative VOT values ranging from about Otilde 125 to Otilde 75 msecwith a median value of Otilde 100msec Italian or French voiced stops are ofthis type ( Bortolini et al 1995 Ryalls Provost and Arsenault 1995)
P Auzou et al136
Figure 3 Same representation as in Fig1 for the [ta] of a patient with spastic dysarthriaThe striation does not become regular enough to determine the beginning
( 2 ) Short voicing lagrsquo positive VOT values ranging from 0 to +25 msec witha median value of +10 msec English voiced stops and Italian voiceless stopsare of this type
( 3 ) `Long voicing lagrsquo large positive VOT values ranging from +60 to+100 msec with a median value of +75 msec English voiceless stops areof this type
However Lisker and Abramson ( 1964) reported that in English the phonemes[b] [d] and [g] used two of the VOT categories Their group of four speakersexhibited VOT values for [b] that ranged from either Otilde 130 to Otilde 20 msec (voicinglead ) or from 0 to +5 msec (short lag) The individual speakers did not tend toexhibit both VOT categories One speaker produced 95 of all the stops with voicinglead and a second one produced the remaining 5 The other two speakers producedalmost exclusively positive values or short lag ( 4142 for one speaker and 5658 forthe other) Because the voiced stops in English occur in two VOT categories whereasthe voiceless stops occur in one VOT category the VOT values of voiceless stopstend to be more normally distributed than their voiced counterparts ( Zlatin 1974)
Thai is a language that uses all three voicing categories Lisker and Abramson
VOT and speech disorders 137
Figure 4 Oscillogram and spectrogram of a [ka] in a normal subject If using only theoscillogram the beginning of the periodic vibration may be doubtful (b) W hen usingcombined oscillogram and spectrogram the ambiguity disappears (correct beginningin brsquo)
( 1964) reported that their Thai speakers used lead short lag and long lag respectivelyfor voiced unaspirated voiceless and aspirated voiceless stops
Across languages the same consonant can demonstrate notable VOT di erencesin its phonetic expression For instance the mean VOT values for voiceless conson-ants are much lower in Thai or Korean than in English However common character-istics can be observed in the data across language In most languages the VOTvalues for voiced and voiceless stops are produced in discrete duration ranges thatcorrespond to the voicing categories That is the categories of VOT values areseparated by a range of times in which no production occurs This categorizationindicates active control for the voiced and voiceless cognates in the timing relation-ship between the articulatory release and the onset of vocal fold vibration Thearticulatory gestures for short voicing lag stops are easier than for the other twotypes of VOT Voicing lead requires more muscles gestures than those needed forshort voicing lag stops The production of initial stop consonants with voicing leadnecessitates some mechanisms external to the larynx to sustain an adequate transglot-tal pressure drop during the closure In contrast the stop consonants with longvoicing lag require more carefully controlled timing between the oral stop and
P Auzou et al138
Figure 5 Oscillogram and spectrogram of a [ka] in a normal subject Sometimes a doubleburst is observed We consider the reg rst to measure V OT
laryngeal closure For these consonants the adduction of the vocal folds requiresmore complex muscle activity ( Dixit 1989 Tyler and Watterson 1991 Vohr Garcia-Coll and Oh 1988 Westbury and Keating 1986) In producing the voicing contrasta speaker needs to coordinate the timing of velopharyngeal closure supraglottalarticulators closure vocal fold oscillation and supraglottal articulator release
Another cross language commonality is that VOT varies regularly with the placeof articulation in all languages The lag time values tend to increase as the positionof occlusion moves posteriorly within the oral cavity the mean VOT values for [p]are shorter than for [t] which in turn are shorter than for [k] ( reg gure 6) ( Baum andRyan 1993 Klatt 1975 Miller Green and Reeves 1986 Volaitis and Miller 1992)However the range of VOT values overlap extensively across places of articulationfor both voiced and voiceless consonants
Factors that can alter VOT production
Changes occur in VOT as a consequence of physiological di erences (such as agelung volume) pathological status ( hearing impairment depression) and di erentlinguistic tasks (speech task speech rate phoneme environment ) One such factorthat has been investigated is the age of the speaker Variations in VOT have beenreported across the life span Eguchi and Hirsh ( 1969) reported high VOT variability
VOT and speech disorders 139
Tab
le1
Nor
mat
ive
valu
esof
VO
Tin
Eng
lish
liter
atur
ein
mill
isec
onds
(mea
nsan
dst
anda
rdde
viat
ions
orra
nge
ifav
aila
ble)
S
ome
ofth
emar
eap
prox
imat
edfr
omregg
ures
L
iske
ran
dA
bram
son
(196
4)ga
vetw
ora
nges
ofva
lues
for
voic
edst
ops
corr
espo
ndin
gre
spec
tive
lyto
lead
and
shor
tla
g(s
eete
xt)
C
Con
sona
nt
V
Vow
el
Num
ber
ofA
utho
rssu
bjec
tsp
tk
bd
gC
onte
xt
Lis
ker
and
458
(20
120
)70
(30
105
)80
(50
135
)Otilde
101
(Otilde13
0Otilde
20)
Otilde10
2(Otilde
155
Otilde40
)Otilde
88(Otilde
150
Otilde60
)A
bram
son
1964
1(0
5)
5(0
25
)21
(03
5)
Kla
tt
1975
347
6570
1117
27Sa
yC
Vin
stea
dSw
eeti
ngan
d30
766
(Ocirc4
6)
139
(Ocirc1
4)
Ita
CV
Bak
en
1982
Har
dcas
tle
etal
4
653
8082
816
915
721
3Si
ngle
wor
d19
85M
orri
san
dB
row
n25
7065
100
1987
Car
uso
and
Bur
ton
862
5(2
54
)71
9(1
94
)74
8(7
48
)19
7(8
5)
214
(49
)35
2(8
4)
Say
CV
Cag
ain
1987
Lee
etal
19
8810
725
(26
9)
Tw
enty
-one
For
rest
etal
19
898
46(1
0)
12(3
)B
uyB
obby
aP
oppy
Bau
man
dR
yan
1085
9511
010
1530
Ple
ase
say
CV
C19
93B
row
net
al
1993
1257
652
64
168
Spea
kC
aag
ain
Pet
rosi
noet
al
1089
(20
)14
(24
)Sa
yC
atag
ain
1993
Dav
is
1995
1090
30
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders 135
Figure 2 Same representation as in Fig1 for two syllables [pa] of a patient with spasticdysarthria No burst was identireg able
of 556msec and interjudge ones of 475msec Other authors have reported closeragreement in terms of percentage agreement within a selected range of durationdi erences such as 966 of reliability measures being within 2 msec (Sweeting andBaken 1982) 100 agreement within 3 msec (Davis 1995) and 975 agreementwithin 5 msec ( Zlatin 1974) Similarly Hoit Solomon and Hixon ( 1993) reported943 interjudge agreement and 951 intrajudge agreement
Norms of VOT production
VOT data have been reported for normal speakers of many languages These studiesindicate similar VOT categories across languages Lisker and Abramson ( 1964)reported mean VOT values for word-initial stop consonants data from normalsubjects who spoke 11 di erent languages However each language was representedby a small pool of subjects The greatest number of studies have been done usingEnglish speaking subjects The results from many of these studies are shown intable 1 The VOT values vary for each consonant and may be best represented as arange of durations Table 1 reveals that the range of VOT values produced for agiven consonant is comparable across studies and across di erent phonetic contextsExtensive cross-language studies by Lisker and Abramson ( 1964 1967) demonstratedthat three categories of stops emerge along the VOT continuum with reasonableagreement as to category boundaries across languages
( 1 ) `Voicing leadrsquo negative VOT values ranging from about Otilde 125 to Otilde 75 msecwith a median value of Otilde 100msec Italian or French voiced stops are ofthis type ( Bortolini et al 1995 Ryalls Provost and Arsenault 1995)
P Auzou et al136
Figure 3 Same representation as in Fig1 for the [ta] of a patient with spastic dysarthriaThe striation does not become regular enough to determine the beginning
( 2 ) Short voicing lagrsquo positive VOT values ranging from 0 to +25 msec witha median value of +10 msec English voiced stops and Italian voiceless stopsare of this type
( 3 ) `Long voicing lagrsquo large positive VOT values ranging from +60 to+100 msec with a median value of +75 msec English voiceless stops areof this type
However Lisker and Abramson ( 1964) reported that in English the phonemes[b] [d] and [g] used two of the VOT categories Their group of four speakersexhibited VOT values for [b] that ranged from either Otilde 130 to Otilde 20 msec (voicinglead ) or from 0 to +5 msec (short lag) The individual speakers did not tend toexhibit both VOT categories One speaker produced 95 of all the stops with voicinglead and a second one produced the remaining 5 The other two speakers producedalmost exclusively positive values or short lag ( 4142 for one speaker and 5658 forthe other) Because the voiced stops in English occur in two VOT categories whereasthe voiceless stops occur in one VOT category the VOT values of voiceless stopstend to be more normally distributed than their voiced counterparts ( Zlatin 1974)
Thai is a language that uses all three voicing categories Lisker and Abramson
VOT and speech disorders 137
Figure 4 Oscillogram and spectrogram of a [ka] in a normal subject If using only theoscillogram the beginning of the periodic vibration may be doubtful (b) W hen usingcombined oscillogram and spectrogram the ambiguity disappears (correct beginningin brsquo)
( 1964) reported that their Thai speakers used lead short lag and long lag respectivelyfor voiced unaspirated voiceless and aspirated voiceless stops
Across languages the same consonant can demonstrate notable VOT di erencesin its phonetic expression For instance the mean VOT values for voiceless conson-ants are much lower in Thai or Korean than in English However common character-istics can be observed in the data across language In most languages the VOTvalues for voiced and voiceless stops are produced in discrete duration ranges thatcorrespond to the voicing categories That is the categories of VOT values areseparated by a range of times in which no production occurs This categorizationindicates active control for the voiced and voiceless cognates in the timing relation-ship between the articulatory release and the onset of vocal fold vibration Thearticulatory gestures for short voicing lag stops are easier than for the other twotypes of VOT Voicing lead requires more muscles gestures than those needed forshort voicing lag stops The production of initial stop consonants with voicing leadnecessitates some mechanisms external to the larynx to sustain an adequate transglot-tal pressure drop during the closure In contrast the stop consonants with longvoicing lag require more carefully controlled timing between the oral stop and
P Auzou et al138
Figure 5 Oscillogram and spectrogram of a [ka] in a normal subject Sometimes a doubleburst is observed We consider the reg rst to measure V OT
laryngeal closure For these consonants the adduction of the vocal folds requiresmore complex muscle activity ( Dixit 1989 Tyler and Watterson 1991 Vohr Garcia-Coll and Oh 1988 Westbury and Keating 1986) In producing the voicing contrasta speaker needs to coordinate the timing of velopharyngeal closure supraglottalarticulators closure vocal fold oscillation and supraglottal articulator release
Another cross language commonality is that VOT varies regularly with the placeof articulation in all languages The lag time values tend to increase as the positionof occlusion moves posteriorly within the oral cavity the mean VOT values for [p]are shorter than for [t] which in turn are shorter than for [k] ( reg gure 6) ( Baum andRyan 1993 Klatt 1975 Miller Green and Reeves 1986 Volaitis and Miller 1992)However the range of VOT values overlap extensively across places of articulationfor both voiced and voiceless consonants
Factors that can alter VOT production
Changes occur in VOT as a consequence of physiological di erences (such as agelung volume) pathological status ( hearing impairment depression) and di erentlinguistic tasks (speech task speech rate phoneme environment ) One such factorthat has been investigated is the age of the speaker Variations in VOT have beenreported across the life span Eguchi and Hirsh ( 1969) reported high VOT variability
VOT and speech disorders 139
Tab
le1
Nor
mat
ive
valu
esof
VO
Tin
Eng
lish
liter
atur
ein
mill
isec
onds
(mea
nsan
dst
anda
rdde
viat
ions
orra
nge
ifav
aila
ble)
S
ome
ofth
emar
eap
prox
imat
edfr
omregg
ures
L
iske
ran
dA
bram
son
(196
4)ga
vetw
ora
nges
ofva
lues
for
voic
edst
ops
corr
espo
ndin
gre
spec
tive
lyto
lead
and
shor
tla
g(s
eete
xt)
C
Con
sona
nt
V
Vow
el
Num
ber
ofA
utho
rssu
bjec
tsp
tk
bd
gC
onte
xt
Lis
ker
and
458
(20
120
)70
(30
105
)80
(50
135
)Otilde
101
(Otilde13
0Otilde
20)
Otilde10
2(Otilde
155
Otilde40
)Otilde
88(Otilde
150
Otilde60
)A
bram
son
1964
1(0
5)
5(0
25
)21
(03
5)
Kla
tt
1975
347
6570
1117
27Sa
yC
Vin
stea
dSw
eeti
ngan
d30
766
(Ocirc4
6)
139
(Ocirc1
4)
Ita
CV
Bak
en
1982
Har
dcas
tle
etal
4
653
8082
816
915
721
3Si
ngle
wor
d19
85M
orri
san
dB
row
n25
7065
100
1987
Car
uso
and
Bur
ton
862
5(2
54
)71
9(1
94
)74
8(7
48
)19
7(8
5)
214
(49
)35
2(8
4)
Say
CV
Cag
ain
1987
Lee
etal
19
8810
725
(26
9)
Tw
enty
-one
For
rest
etal
19
898
46(1
0)
12(3
)B
uyB
obby
aP
oppy
Bau
man
dR
yan
1085
9511
010
1530
Ple
ase
say
CV
C19
93B
row
net
al
1993
1257
652
64
168
Spea
kC
aag
ain
Pet
rosi
noet
al
1089
(20
)14
(24
)Sa
yC
atag
ain
1993
Dav
is
1995
1090
30
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
P Auzou et al136
Figure 3 Same representation as in Fig1 for the [ta] of a patient with spastic dysarthriaThe striation does not become regular enough to determine the beginning
( 2 ) Short voicing lagrsquo positive VOT values ranging from 0 to +25 msec witha median value of +10 msec English voiced stops and Italian voiceless stopsare of this type
( 3 ) `Long voicing lagrsquo large positive VOT values ranging from +60 to+100 msec with a median value of +75 msec English voiceless stops areof this type
However Lisker and Abramson ( 1964) reported that in English the phonemes[b] [d] and [g] used two of the VOT categories Their group of four speakersexhibited VOT values for [b] that ranged from either Otilde 130 to Otilde 20 msec (voicinglead ) or from 0 to +5 msec (short lag) The individual speakers did not tend toexhibit both VOT categories One speaker produced 95 of all the stops with voicinglead and a second one produced the remaining 5 The other two speakers producedalmost exclusively positive values or short lag ( 4142 for one speaker and 5658 forthe other) Because the voiced stops in English occur in two VOT categories whereasthe voiceless stops occur in one VOT category the VOT values of voiceless stopstend to be more normally distributed than their voiced counterparts ( Zlatin 1974)
Thai is a language that uses all three voicing categories Lisker and Abramson
VOT and speech disorders 137
Figure 4 Oscillogram and spectrogram of a [ka] in a normal subject If using only theoscillogram the beginning of the periodic vibration may be doubtful (b) W hen usingcombined oscillogram and spectrogram the ambiguity disappears (correct beginningin brsquo)
( 1964) reported that their Thai speakers used lead short lag and long lag respectivelyfor voiced unaspirated voiceless and aspirated voiceless stops
Across languages the same consonant can demonstrate notable VOT di erencesin its phonetic expression For instance the mean VOT values for voiceless conson-ants are much lower in Thai or Korean than in English However common character-istics can be observed in the data across language In most languages the VOTvalues for voiced and voiceless stops are produced in discrete duration ranges thatcorrespond to the voicing categories That is the categories of VOT values areseparated by a range of times in which no production occurs This categorizationindicates active control for the voiced and voiceless cognates in the timing relation-ship between the articulatory release and the onset of vocal fold vibration Thearticulatory gestures for short voicing lag stops are easier than for the other twotypes of VOT Voicing lead requires more muscles gestures than those needed forshort voicing lag stops The production of initial stop consonants with voicing leadnecessitates some mechanisms external to the larynx to sustain an adequate transglot-tal pressure drop during the closure In contrast the stop consonants with longvoicing lag require more carefully controlled timing between the oral stop and
P Auzou et al138
Figure 5 Oscillogram and spectrogram of a [ka] in a normal subject Sometimes a doubleburst is observed We consider the reg rst to measure V OT
laryngeal closure For these consonants the adduction of the vocal folds requiresmore complex muscle activity ( Dixit 1989 Tyler and Watterson 1991 Vohr Garcia-Coll and Oh 1988 Westbury and Keating 1986) In producing the voicing contrasta speaker needs to coordinate the timing of velopharyngeal closure supraglottalarticulators closure vocal fold oscillation and supraglottal articulator release
Another cross language commonality is that VOT varies regularly with the placeof articulation in all languages The lag time values tend to increase as the positionof occlusion moves posteriorly within the oral cavity the mean VOT values for [p]are shorter than for [t] which in turn are shorter than for [k] ( reg gure 6) ( Baum andRyan 1993 Klatt 1975 Miller Green and Reeves 1986 Volaitis and Miller 1992)However the range of VOT values overlap extensively across places of articulationfor both voiced and voiceless consonants
Factors that can alter VOT production
Changes occur in VOT as a consequence of physiological di erences (such as agelung volume) pathological status ( hearing impairment depression) and di erentlinguistic tasks (speech task speech rate phoneme environment ) One such factorthat has been investigated is the age of the speaker Variations in VOT have beenreported across the life span Eguchi and Hirsh ( 1969) reported high VOT variability
VOT and speech disorders 139
Tab
le1
Nor
mat
ive
valu
esof
VO
Tin
Eng
lish
liter
atur
ein
mill
isec
onds
(mea
nsan
dst
anda
rdde
viat
ions
orra
nge
ifav
aila
ble)
S
ome
ofth
emar
eap
prox
imat
edfr
omregg
ures
L
iske
ran
dA
bram
son
(196
4)ga
vetw
ora
nges
ofva
lues
for
voic
edst
ops
corr
espo
ndin
gre
spec
tive
lyto
lead
and
shor
tla
g(s
eete
xt)
C
Con
sona
nt
V
Vow
el
Num
ber
ofA
utho
rssu
bjec
tsp
tk
bd
gC
onte
xt
Lis
ker
and
458
(20
120
)70
(30
105
)80
(50
135
)Otilde
101
(Otilde13
0Otilde
20)
Otilde10
2(Otilde
155
Otilde40
)Otilde
88(Otilde
150
Otilde60
)A
bram
son
1964
1(0
5)
5(0
25
)21
(03
5)
Kla
tt
1975
347
6570
1117
27Sa
yC
Vin
stea
dSw
eeti
ngan
d30
766
(Ocirc4
6)
139
(Ocirc1
4)
Ita
CV
Bak
en
1982
Har
dcas
tle
etal
4
653
8082
816
915
721
3Si
ngle
wor
d19
85M
orri
san
dB
row
n25
7065
100
1987
Car
uso
and
Bur
ton
862
5(2
54
)71
9(1
94
)74
8(7
48
)19
7(8
5)
214
(49
)35
2(8
4)
Say
CV
Cag
ain
1987
Lee
etal
19
8810
725
(26
9)
Tw
enty
-one
For
rest
etal
19
898
46(1
0)
12(3
)B
uyB
obby
aP
oppy
Bau
man
dR
yan
1085
9511
010
1530
Ple
ase
say
CV
C19
93B
row
net
al
1993
1257
652
64
168
Spea
kC
aag
ain
Pet
rosi
noet
al
1089
(20
)14
(24
)Sa
yC
atag
ain
1993
Dav
is
1995
1090
30
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders 137
Figure 4 Oscillogram and spectrogram of a [ka] in a normal subject If using only theoscillogram the beginning of the periodic vibration may be doubtful (b) W hen usingcombined oscillogram and spectrogram the ambiguity disappears (correct beginningin brsquo)
( 1964) reported that their Thai speakers used lead short lag and long lag respectivelyfor voiced unaspirated voiceless and aspirated voiceless stops
Across languages the same consonant can demonstrate notable VOT di erencesin its phonetic expression For instance the mean VOT values for voiceless conson-ants are much lower in Thai or Korean than in English However common character-istics can be observed in the data across language In most languages the VOTvalues for voiced and voiceless stops are produced in discrete duration ranges thatcorrespond to the voicing categories That is the categories of VOT values areseparated by a range of times in which no production occurs This categorizationindicates active control for the voiced and voiceless cognates in the timing relation-ship between the articulatory release and the onset of vocal fold vibration Thearticulatory gestures for short voicing lag stops are easier than for the other twotypes of VOT Voicing lead requires more muscles gestures than those needed forshort voicing lag stops The production of initial stop consonants with voicing leadnecessitates some mechanisms external to the larynx to sustain an adequate transglot-tal pressure drop during the closure In contrast the stop consonants with longvoicing lag require more carefully controlled timing between the oral stop and
P Auzou et al138
Figure 5 Oscillogram and spectrogram of a [ka] in a normal subject Sometimes a doubleburst is observed We consider the reg rst to measure V OT
laryngeal closure For these consonants the adduction of the vocal folds requiresmore complex muscle activity ( Dixit 1989 Tyler and Watterson 1991 Vohr Garcia-Coll and Oh 1988 Westbury and Keating 1986) In producing the voicing contrasta speaker needs to coordinate the timing of velopharyngeal closure supraglottalarticulators closure vocal fold oscillation and supraglottal articulator release
Another cross language commonality is that VOT varies regularly with the placeof articulation in all languages The lag time values tend to increase as the positionof occlusion moves posteriorly within the oral cavity the mean VOT values for [p]are shorter than for [t] which in turn are shorter than for [k] ( reg gure 6) ( Baum andRyan 1993 Klatt 1975 Miller Green and Reeves 1986 Volaitis and Miller 1992)However the range of VOT values overlap extensively across places of articulationfor both voiced and voiceless consonants
Factors that can alter VOT production
Changes occur in VOT as a consequence of physiological di erences (such as agelung volume) pathological status ( hearing impairment depression) and di erentlinguistic tasks (speech task speech rate phoneme environment ) One such factorthat has been investigated is the age of the speaker Variations in VOT have beenreported across the life span Eguchi and Hirsh ( 1969) reported high VOT variability
VOT and speech disorders 139
Tab
le1
Nor
mat
ive
valu
esof
VO
Tin
Eng
lish
liter
atur
ein
mill
isec
onds
(mea
nsan
dst
anda
rdde
viat
ions
orra
nge
ifav
aila
ble)
S
ome
ofth
emar
eap
prox
imat
edfr
omregg
ures
L
iske
ran
dA
bram
son
(196
4)ga
vetw
ora
nges
ofva
lues
for
voic
edst
ops
corr
espo
ndin
gre
spec
tive
lyto
lead
and
shor
tla
g(s
eete
xt)
C
Con
sona
nt
V
Vow
el
Num
ber
ofA
utho
rssu
bjec
tsp
tk
bd
gC
onte
xt
Lis
ker
and
458
(20
120
)70
(30
105
)80
(50
135
)Otilde
101
(Otilde13
0Otilde
20)
Otilde10
2(Otilde
155
Otilde40
)Otilde
88(Otilde
150
Otilde60
)A
bram
son
1964
1(0
5)
5(0
25
)21
(03
5)
Kla
tt
1975
347
6570
1117
27Sa
yC
Vin
stea
dSw
eeti
ngan
d30
766
(Ocirc4
6)
139
(Ocirc1
4)
Ita
CV
Bak
en
1982
Har
dcas
tle
etal
4
653
8082
816
915
721
3Si
ngle
wor
d19
85M
orri
san
dB
row
n25
7065
100
1987
Car
uso
and
Bur
ton
862
5(2
54
)71
9(1
94
)74
8(7
48
)19
7(8
5)
214
(49
)35
2(8
4)
Say
CV
Cag
ain
1987
Lee
etal
19
8810
725
(26
9)
Tw
enty
-one
For
rest
etal
19
898
46(1
0)
12(3
)B
uyB
obby
aP
oppy
Bau
man
dR
yan
1085
9511
010
1530
Ple
ase
say
CV
C19
93B
row
net
al
1993
1257
652
64
168
Spea
kC
aag
ain
Pet
rosi
noet
al
1089
(20
)14
(24
)Sa
yC
atag
ain
1993
Dav
is
1995
1090
30
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
P Auzou et al138
Figure 5 Oscillogram and spectrogram of a [ka] in a normal subject Sometimes a doubleburst is observed We consider the reg rst to measure V OT
laryngeal closure For these consonants the adduction of the vocal folds requiresmore complex muscle activity ( Dixit 1989 Tyler and Watterson 1991 Vohr Garcia-Coll and Oh 1988 Westbury and Keating 1986) In producing the voicing contrasta speaker needs to coordinate the timing of velopharyngeal closure supraglottalarticulators closure vocal fold oscillation and supraglottal articulator release
Another cross language commonality is that VOT varies regularly with the placeof articulation in all languages The lag time values tend to increase as the positionof occlusion moves posteriorly within the oral cavity the mean VOT values for [p]are shorter than for [t] which in turn are shorter than for [k] ( reg gure 6) ( Baum andRyan 1993 Klatt 1975 Miller Green and Reeves 1986 Volaitis and Miller 1992)However the range of VOT values overlap extensively across places of articulationfor both voiced and voiceless consonants
Factors that can alter VOT production
Changes occur in VOT as a consequence of physiological di erences (such as agelung volume) pathological status ( hearing impairment depression) and di erentlinguistic tasks (speech task speech rate phoneme environment ) One such factorthat has been investigated is the age of the speaker Variations in VOT have beenreported across the life span Eguchi and Hirsh ( 1969) reported high VOT variability
VOT and speech disorders 139
Tab
le1
Nor
mat
ive
valu
esof
VO
Tin
Eng
lish
liter
atur
ein
mill
isec
onds
(mea
nsan
dst
anda
rdde
viat
ions
orra
nge
ifav
aila
ble)
S
ome
ofth
emar
eap
prox
imat
edfr
omregg
ures
L
iske
ran
dA
bram
son
(196
4)ga
vetw
ora
nges
ofva
lues
for
voic
edst
ops
corr
espo
ndin
gre
spec
tive
lyto
lead
and
shor
tla
g(s
eete
xt)
C
Con
sona
nt
V
Vow
el
Num
ber
ofA
utho
rssu
bjec
tsp
tk
bd
gC
onte
xt
Lis
ker
and
458
(20
120
)70
(30
105
)80
(50
135
)Otilde
101
(Otilde13
0Otilde
20)
Otilde10
2(Otilde
155
Otilde40
)Otilde
88(Otilde
150
Otilde60
)A
bram
son
1964
1(0
5)
5(0
25
)21
(03
5)
Kla
tt
1975
347
6570
1117
27Sa
yC
Vin
stea
dSw
eeti
ngan
d30
766
(Ocirc4
6)
139
(Ocirc1
4)
Ita
CV
Bak
en
1982
Har
dcas
tle
etal
4
653
8082
816
915
721
3Si
ngle
wor
d19
85M
orri
san
dB
row
n25
7065
100
1987
Car
uso
and
Bur
ton
862
5(2
54
)71
9(1
94
)74
8(7
48
)19
7(8
5)
214
(49
)35
2(8
4)
Say
CV
Cag
ain
1987
Lee
etal
19
8810
725
(26
9)
Tw
enty
-one
For
rest
etal
19
898
46(1
0)
12(3
)B
uyB
obby
aP
oppy
Bau
man
dR
yan
1085
9511
010
1530
Ple
ase
say
CV
C19
93B
row
net
al
1993
1257
652
64
168
Spea
kC
aag
ain
Pet
rosi
noet
al
1089
(20
)14
(24
)Sa
yC
atag
ain
1993
Dav
is
1995
1090
30
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders 139
Tab
le1
Nor
mat
ive
valu
esof
VO
Tin
Eng
lish
liter
atur
ein
mill
isec
onds
(mea
nsan
dst
anda
rdde
viat
ions
orra
nge
ifav
aila
ble)
S
ome
ofth
emar
eap
prox
imat
edfr
omregg
ures
L
iske
ran
dA
bram
son
(196
4)ga
vetw
ora
nges
ofva
lues
for
voic
edst
ops
corr
espo
ndin
gre
spec
tive
lyto
lead
and
shor
tla
g(s
eete
xt)
C
Con
sona
nt
V
Vow
el
Num
ber
ofA
utho
rssu
bjec
tsp
tk
bd
gC
onte
xt
Lis
ker
and
458
(20
120
)70
(30
105
)80
(50
135
)Otilde
101
(Otilde13
0Otilde
20)
Otilde10
2(Otilde
155
Otilde40
)Otilde
88(Otilde
150
Otilde60
)A
bram
son
1964
1(0
5)
5(0
25
)21
(03
5)
Kla
tt
1975
347
6570
1117
27Sa
yC
Vin
stea
dSw
eeti
ngan
d30
766
(Ocirc4
6)
139
(Ocirc1
4)
Ita
CV
Bak
en
1982
Har
dcas
tle
etal
4
653
8082
816
915
721
3Si
ngle
wor
d19
85M
orri
san
dB
row
n25
7065
100
1987
Car
uso
and
Bur
ton
862
5(2
54
)71
9(1
94
)74
8(7
48
)19
7(8
5)
214
(49
)35
2(8
4)
Say
CV
Cag
ain
1987
Lee
etal
19
8810
725
(26
9)
Tw
enty
-one
For
rest
etal
19
898
46(1
0)
12(3
)B
uyB
obby
aP
oppy
Bau
man
dR
yan
1085
9511
010
1530
Ple
ase
say
CV
C19
93B
row
net
al
1993
1257
652
64
168
Spea
kC
aag
ain
Pet
rosi
noet
al
1089
(20
)14
(24
)Sa
yC
atag
ain
1993
Dav
is
1995
1090
30
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
P Auzou et al140
Figure 6 Mean and standard deviations for each voiceless stops by 5 normal subjects TheV OT values increase from [pa] to [ta] and from [ta] to [ka]
in preadolescent children speakers of American English These authors speculatedthat the extremely precise temporal coordination involved in VOT requires a fullymature nervous system In contrast Ryalls and Larouche ( 1992) found comparableVOT values between children and adult speakers of QueAcirc bec French However thedi ering results from these two studies may remacr ect a di erence in the VOT categoriesused for phoneme contrast within the language
Three steps in the developmental sequence for the acquisition of the voicingcontrast have been described The stops reg rst appear around six months of age withVOT values randomly distributed from voicing lead to long voicing lag At one yearof age the stop productions separate into the appropriate phonemic categories( Tyler and Watterson 1991) Then the children modify their productions towardadult VOT ranges for voiced (short lag in English) and voiceless stop consonants( long lag in English) Most English-speaking children develop the voicing contrastby approximatively 30 months of age ( Kewley-Port and Preston 1974 Macken andBarton 1980a) Surprisingly a comparable study of Spanish-learning childrenshowed that they acquire the Spanish voicing contrast after the age of 4 years( Macken and Barton 1980b) The developmental pattern among young childrenindicates that short voicing lag stops are easier to produce successfully than the twoother categories No consistent gender based di erence in VOT has been found atany age level (Sweeting and Baken 1982 Zlatin 1974)
As stated above for children greater VOT variability has been reported for olderadults The mean VOT values for labial (Sweeting and Baken 1982) and velarplosives ( Petrosino et al 1993) are similar in young and aged male subjects thereforethe older subjects produced VOTs that remained within the appropriate VOT cat-egories However for the voiceless plosives the older subjects produced shortermean VOTs for [p] among the male subjects (Sweeting and Baken 1982) and for[p] and [t] among the female subjects ( Morris and Brown 1987) Thus the oldersubjects exhibited reduced VOT di erences between the voiced and voiceless stopsFinally the older male and female subjects exhibited greater VOT variability thandid the younger subjects ( Morris and Brown 1994 Sweeting and Bacon 1982)
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders 141
In normal speech production VOT values vary with the speaking rate VOTdecreases as the rate increases indicating that the overall rate changes are imple-mented at the segmental level ( Baum and Ryan 1993 Diehl Souther and Convis1980 JaEgrave ncke 1994 Miller 1981 Miller et al 1986) A slower speaking rate resultsnot only in longer VOT but also in a wider range of VOT values ( Miller and Volaitis1989 Volaitis and Miller 1992) This e ect is particularly prominent in English forvoiceless stop consonants while values for voiced stops remain relatively stableThus the di erence between VOTs of voiced and voiceless consonants is reducedwhen the speaking rate increases The changes in VOT values are also more markedon velar consonants than for apical or labial consonants Thus consonants withlonger VOT values velar and voiceless stops permit greater decreases when speakingrate increases ( Baum and Ryan 1993) However normal subjects maintain bimodalVOT distributions for all places of articulation clearly separating the voiced andvoiceless consonants ( Baum and Ryan 1993) Moreover the modireg cations of VOTsinduced by changes in speaking rate are more marked for aspirated than for unaspir-ated stops ( Miller et al 1986 Pind 1995 1996)
Measurements have been made of VOT in English plosives followed by vowelsin word initial medial and reg nal positions using both stressed and unstressed syl-lables Few studies report the VOT values in consonant clusters Klatt ( 1975) showedthat VOTs were longer before sonorant consonants than before vowels They werealso longer before the high vowels [I] and [u] than before mid- or low vowels [ae]and [a] In some way the voicing seems to be more di cult to initiate when thelaryngeal frequency is high The high vowels are produced with greater longitudinalvocal fold tension due to a greater contraction in the cricothyroid muscle ( LoEgrave fqvistBaer McGarr and Story 1989) The increased vocal fold tension has been associatedwith longer VOT values ( LoEgrave fqvist 1992)
The physiologic interaction of the larynx and the respiratory system can a ectVOT values Hence the VOT of [p] is longer at high lung volume and shorter atlow lung volume in normal subjects ( Hoit et al 1993) For example at high lungvolumes the diaphragm usually macr attens and pulls the trachea and larynx caudallyexerting a force that tends to abduct the vocal folds ( Zenker 1964) This linkbetween VOT and lung volume must be kept in mind when studying VOT in subjectsexhibiting aphasia apraxia of speech or dysarthria The abnormally long or shortVOT values produced by subjects with neurogenic speech problems are interpretedto remacr ect impaired coordination between the larynx and the upper airways In factsome disorders such as PD are also associated with the use of abnormal lung volumeranges during speech production ( Murdoch Chenery Bowler and Ingram 1989Solomon and Hixon 1993)
Changes in the VOTs produced by a group of mountain climbers who wererecorded while they were ascending Mt Everest have been interpreted similarly( Lieberman Protopapas and Kanki 1995) These authors found that the meaninterval that di erentiated the VOTs of their voiced and voiceless stop consonantsdecreased at increasing altitudes from Base Camp to Camp Three They concludedthat smaller lung volumes were available to the climbers in the thinner air at higheraltitudes Thus their reg ndings agree with those of Hoit et al ( 1993)
As opposed to the high altitude di erences in healthy subjects obstructivepulmonary disease does not appear to a ect VOT No di erences were foundbetween healthy and asthmatic subjects for the VOT of word initial [t] ( LeeChamberlain Loudon and Stemple 1988) Additionally Lee et al reported that in
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
P Auzou et al142
their healthy subjects VOT was signireg cantly reduced in the initial [t] of 21rsquo whencounting loudly in comparison to counting at normal intensity
Finally VOT patterns have been studied in people exhibiting depression andhearing impairment Depressed patients have signireg cantly shorter VOT values com-pared to normals ( Flint Black Campbell-Taylor Gailey and Levinton 1992) Thisreg nding is of interest when considering neurological disorders in which depression isencountered for example Parkinsonrsquo s disease Two studies report that speakers borndeaf or deafened before learning English present a weaker distinction between voicedand voiceless consonants ( Lane et al 1994 Monsen 1976) The subjects in thesestudies did not use longer VOT durations for the voiceless consonants For Frenchspeaking children Ryalls and Larouche ( 1992) found no di erence between hearing-impaired children and age-matched subjects for the six stop consonants As beforenoted this cross language di erence in results may remacr ect a di erence in the categor-ies of VOT used for making phoneme contrasts within the languages
Aphasia
VOT has been measured in aphasic patients to help di erentiate their phonemic andphonetic error patterns ( Baum and Ryan 1993 Blumstein Baker and Goodglass1977 Itoh et al 1982 Ryalls et al 1995) One way that the VOT production ofthe aphasic patients matched normal speakers was that both macr uent and nonmacr uentaphasics maintained the normal pattern of longer VOT values for velars in compar-ison to labial and alveolar stops ( Baum and Ryan 1993) However several studieshave found VOT di erences between phonemic and phonetic errors A phonemicerror phoneme substitution is remacr ected by a VOT value lying within the normalrange of the corresponding cognate In contrast a phonetic error phoneme distortionis remacr ected by a VOT value lying between the normal ranges of VOT for the voicedand voiceless categories Blumstein ( 1980) provided the following example of thetwo error types In English the normal VOT range of the voiced bilabial [b] is fromOtilde 105 to +15 msec and the VOT range of its cognate [p] is from +35 to +105 msecFor a [b] a VOT greater than +35 msec falls in the VOT range of [p] and wouldbe a phonemic error whereas a value between +15 and +35 msec would be aphonetic error
Using these criteria Blumstein ( 1980) compared the VOT productions fromsubjects exhibiting three types of aphasia She reported that the Brocarsquos aphasicsmade the greatest number of errors they made errors during 40 of their VOTproductions For these speakers 65 of the errors were phonetic The conductionand the Wernickersquo s aphasics made errors during 29 and 8 of their VOT produc-tions respectively Even among the macr uent aphasics who exhibited lower error ratesat least 50 of the errors were phonetic in nature Several studies indicate thatBrocarsquos aphasics present a continuum of VOT values with overlap between theshort and long lag bilabial phonemes ( Baum and Ryan 1993 Blumstein et al1977 Blumstein 1980 Gandour and Dardarananda 1984 Gandour PonglorpisitKhunadorn Dechongkit Boongird and Boonklam 1992) Other studies show thatdespite producing VOTs that occurred between the ranges for [b] and [p] Wernickersquo saphasics maintained a bimodal VOT distribution The preserved separation of theVOTs for the cognate pairs indicates that the Wernickersquo s aphasics better maintainedthis phonetic distinction ( Baum and Ryan 1993 Blumstein et al 1977 Itoh et al1982 Shewan Leeper and Booth 1984 Tuller 1984) Hence the results for macr uent
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders 143
aphasics are interpreted as dereg cits of phonological planning rather than an articulat-ory impairment On the contrary the phonetic error patterns of the nonmacr uentaphasics correspond to dereg cits in temporal motor programming Thus Blumstein( 1980) concluded that all of her aphasic subjects exhibited dereg cits in control of thetiming between their laryngeal and supralaryngeal articulatory adjustmentsHowever the macr uent aphasic subjects ( both Wernickersquo s and conduction aphasics)exhibited timing control problems that were mild enough so that the voiced andvoiceless productions were still distributed into relatively discrete VOT ranges
In contrast a comparable study using French aphasic patients failed to reg ndmore phonetic errors among Brocarsquos aphasics than among macr uent aphasics ( Ryallset al 1995) In addition Buckingham and Yule ( 1987) and Gandour et al ( 1992)suggested that some of the phonemic substitutions produced by macr uent aphasics couldremacr ect subtle motor disturbances rather than impairments in phonological planning
Apraxia of speech
Few researchers have studied VOT in patients with apraxia of speech ( FreemanSands and Harris 1978 Hardcastle Barry and Clark 1985 Itoh et al 1982) Apraxia of speech has been dereg ned as a disorder of motor programming for speechwith substitutions that tend to be variable and unpredictable (Darley Brown andSwenson 1975) Instrumental analyses have shown a high proportion of phoneticdistortions in the speech of these patients particularly in areas such as timing andcoordination ( Itoh and Sasunuma 1984 Kent and Rosenbek 1983) which mightlead to some troubles in VOT production Previous studies using reg ber-optic andcomputer-controlled X-ray microbeam systems clearly indicate a defective temporalorganization of articulatory movements in one apraxic patient ( Itoh SasanumaHirose Yoshioka and Ushijima 1980 Itoh Sasanuma and Ushijima 1979)
As was reported for nonmacr uent aphasics some authors report that apraxic subjectsexhibit a considerable overlap in VOT distribution between long lag and short lagcognates ( Freeman et al 1978 Itoh et al 1982) In contrast Hardcastle et al( 1985) found a clear separation of long lag and short lag stops at the alveolar andvelar places of articulation but a slight overlap at the bilabial place in one apraxicpatient For this patient both the bilabial and alveolar long lag stops showedconsiderable variability in VOT values The authors interpreted this general patternas a problem in sensorimotor programming of speech with errors characterizedprimarily as selection sequencing and temporal integration of target articulationsSpatial and temporal variability in achieving the articulation goals for this apraxicspeaker seemed more marked than the low level of spatial and temporal variabilityfrequently observed among dysarthric patients
Dysarthria
Dysarthria is a term referring to a group of speech disorders resulting from damageto neural mechanisms that regulate speech movements The classireg cation of thedysarthrias according to the physiological approach of Darley et al ( 1975) associatesspastic dysarthria with lesions of the upper motor neurons macr accid dysarthria withlesions of the lower motor neurons ataxic dysarthria with cerebellar disease hypo-kinetic dysarthria with extrapyramidal disorders such as Parkinsonrsquo s disease and
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
P Auzou et al144
hyperkinetic dysarthria with extrapyramidal disorders such as choreoathetosis ordystonia When several of these systems are a ected the dysarthria is called mixed
Several studies have included VOT measurements of the speech produced bysubjects exhibiting di erent types of dysarthria The speech samples used haveincluded both oral reading and spontaneous speech tasks Since Brown and Docherty( 1995) found no consistent e ect of sampling task (reading or spontaneous speech)on VOT of dysarthric patients the results from these studies should be comparable
Morris ( 1989) reported the VOT values for the three voiceless stops [p] [t] and[k]produced during repeated syllable tasks He analysed audio tapes from 20 patientsrepresenting four types of dysarthria reg ve spastics reg ve macr accids reg ve ataxics and reg vehypokinetics however he did not include a control group The results from thisstudy are summarized in reg gure 7 He found that the mean VOT values increased asthe position of occlusion moves posteriorly with much overlap of values among thethree consonants When comparing the mean VOT values across dysarthria typeshe found that the VOTs produced by the spastic dysarthrics for [t] were signireg cantlyshorter than those produced by the macr accid and ataxic dysarthrics SimilarlyHardcastle et al ( 1985) reported that spastic patients produce shorter VOTs thannormally speaking control subjects
Increased variability in the VOTs produced by dysarthric subjects has beenreported in several studies In his comparison of subjects exhibiting di erent dys-arthrias Morris ( 1989) reported that the patients with macr accid and ataxic dysarthriaexhibited signireg cantly greater VOT variability than those with spastic and hypo-kinetic dysarthria In the macr accid dysarthria group the interspeaker VOT variabilitywas great whereas intraspeaker VOT variations were similar to those produced byspastic and hypokinetic dysarthrics In contrast the speakers with ataxic dysarthriaexhibited not only interspeaker but also intraspeaker VOT variability Other studieshave conreg rmed the reg nding of greater VOT variability among ataxic patients( Ackermann and Hertrich 1993 Keller Vigneux and Laframboise 1991) SimilarVOT variability was also found among patients with Huntingtonrsquo s disease a hyper-kinetic neurogenic disorder patients ( Hertrich and Ackermann 1994) and stutteringadults even during non stuttering periods (JaEgrave ncke 1994) In their study of eight
Figure 7 Mean and standard deviations of each voiceless stops according to the type ofdysarthria type (adaptated from Morris 1989)
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders 145
patients with amyotrophic lateral sclerosis ( ALS) and eight age-matched controlsCaruso and Burton ( 1987) reported no signireg cant di erences among the mean VOTsof the six stop consonants However the variability of the VOT was greater in theALS group for all of these consonants except [p]
Results concerning the VOTs produced by subjects who have PD vary amongthe studies Forrest Weismer and Turner ( 1989) compared PD patients with age-matched normal subjects They measured the VOT for the bilabial stops ( reg ve [b]and one [p]) in sentences such as rsquoBuild a big buildingrsquo or rsquoBuy Bobby a poppyrsquo The PD speakers produced longer VOTs than control speakers for all reg ve [b]productions but only the utterance initial [b] productions were signireg cantly longerThe authors interpreted this as representing movement initiation problems at thelaryngeal level No signireg cant di erence occurred for the [p] VOTs in rsquopoppyrsquo eventhough these VOTs tended to be longer than those produced by the normal speakersThis lack of signireg cant di erence for the VOTs of long lag stops between normaland PD speakers was conreg rmed in two studies of isolated and repeated CV syllables(Cohen Laframboise Labelle and Bouchard 1993 Connor et al 1989) In a studyof French speaking subjects the VOTs of [p] [t] and [k] produced during a repeatedsyllable task by 18 PD patients were found to be signireg cantly longer than thoseproduced by 12 normally speaking control subjects (OEgrave zsancak Auzou JanLeAcirc onardon Gaillard and Hannequin 1997) In contrast to the reports that subjectswith PD use longer VOTs Flint et al ( 1992) found signireg cantly decreased VOTsin patients with PD versus normally speaking controls for initial [p] [t] and [k]consonants This reg nding is unique among the VOT studies of subjects with PD
Since the French subjects with PD exhibited signireg cantly longer VOTs for [p][t] and [k] than the normally speaking control subjects and the VOT di erences forthe English speakers with PD were signireg cantly longer only for [b] the cross languageresults from PD patients appear to be contradictory However the contradiction ispresent only if the plosives are considered in terms of the voicedplusmn voiceless contrastThe results can be more easily interpreted in term of the short voicing lag-longvoicing lag contrast ( table 2 ) In French the voiceless stops use the short voicinglag and the voiced stops use the voicing lead Thus the French voiceless stops areproduced with similar VOTs as the English voiced stops We can then summarizethe data from the previously reported studies as follows in subjects with PD theshort lag durations are longer than normal and the long lag durations are normalor shorter The reduction of the timing interval between these two VOT categoriesmay lead to an overlap between the two stop consonant cognates and the perceptionof phonetic errors as mentioned by Lieberman Kako Friedman TajchmanFeldman and Jiminez ( 1992) As Tyler and Watterson ( 1991) speculated it appearsthat the excess muscular tension exhibited by PD speakers inhibits appropriately
Table 2 Modireg cations of V OT mean values in hypokinetic dysarthria
Authors Short lag Long lag
Forrest et al 1989 increase NConnor et al 1989 NFlint et al 1992 decreaseLieberman et al 1992 overlappingCohen et al 1993 NOEgrave zsancak et al 1997 increase
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
P Auzou et al146
timed approximation and initiation of vibration leading to longer short lag dura-tions To some extent the VOT modireg cations in PD might also result from areduction in the coordination among the di erent structures in the speech productionmechanism
Conclusion
Instrumental analysis provides quantitative and objective data on a wide range ofspeech parameters and greatly enhances the scope of auditory-based perceptualjudgments of speech Our review suggests that VOT is a reliable temporal parameterwhen appropriate methodological precautions are followed such as using time-synchronized wideband spectrographic and oscillographic displays
VOT remacr ects the temporal control between the larynx and the lips tongue andjaw It can be used to dereg ne the phonetic voicedplusmn voiceless contrast and has been afunctional tool for exploring the acoustic aspects of some speech disorders Whenthe subjects have had neurologically based communication disorders researchershave varied their use of VOT depending on the whether the subjects exhibitedaphasia or dysarthria When the subjects exhibited aphasia the authors usuallyreported the distribution of VOT values along a continuum rather than the meanvalues within a voicing category When so used these data indicated the clinicaldistinction between phonetic errors corresponding to a dereg cit in temporal motorplanning and phonemic errors remacr ecting a dereg cit of phonological planningHowever some authors suggest that part of the phonemic substitutions producedby macr uent aphasics could also remacr ect subtle motor disturbances rather than impair-ments in phonological planning
When the subjects exhibited dysarthria the goal was to evaluate articulatoryimpairments and VOT productions have mainly been interpreted by using meanvalues or standard deviations When VOT abnormalities are present they mainlycorrespond to a shortening of long lag durations or a lengthening of short lagdurations These alterations are usually interpreted as a loss of timing control butother variables such as the size of the glottal opening the transglottal pressure andthe vocal fold tension can also a ect the VOT values For example the VOT valuesmay be inmacr uenced by the respiratory impairment typically found among people withPD In addition the variability of VOT productions has been shown to be animportant factor to consider in speakers with several types of dysarthria Even whennormal mean values occur increased variability has been found in ataxic andhyperkinetic dysarthric subjects
Our analysis of the literature leads us to use descriptive statistical measures ofvariability to complement the mean VOT values such as standard deviations or thecoe cient of variation In addition using graphs to represent the distributions ofthe VOT di erences between the voiceless and voiced stop cognates often clarireg esthe data patterns Previous studies also show the importance of measuring VOT atdi erent speaking rates The combined use of these di erent methodological toolsthat may allow the distinction of specireg c diagnostic patterns among the di erentdysarthria types
VOT could also be used as an indirect mean to study vocal tract motor controlAs speech remacr ects axial control the measure of VOT could be used to compare axialwith limb motor control Such comparisons would be of great interest for instancein PD because these two types of motor control are not inmacr uenced in the same
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders 147
manner by dopaminergic drugs VOT could be then used as an objective parameterto evaluate the e ect of new drugs on parkinsonian dysarthria in comparison tolimb tremor
Cross language studies using groups of subjects matched for diagnosis andseverity of communication disorder could help to clarify the VOT variations observedbetween languages Studies of this type will help researchers to better characterizethe common acoustic abnormalities underlying disordered speech production Forthis purpose we believe it is relevant to consider the consonants in terms of leadrsquo short lagrsquo and long lagrsquo rather than voiced and voiceless as this system betterdescribes the VOT categories used across di erent languages
Finally current physiological theories about VOT production indicate inter-actions between it and the physiologic adjustments that occur at the di erent levelsof the speech production mechanism Further studies are needed to establish dereg nit-ive normative data for these interactions In addition further studies should providedescriptions of the relationships between the clinical di erences in speech productionencountered as a result of motor disorders at the respiratory phonatory and articu-latory levels and their acoustic consequences
References
Ackermann H and Hertrich I 1993 Dysarthria in Friedreichrsquos ataxia Clinical L inguisticsand Phonetics 7 75plusmn 91
Baum S R and Ryan L 1993 Rate of speech in aphasia voice onset time Brain andL anguage 44 431plusmn 445
Benguerel A P Hirose H Sawashima M and Ushijima T 1978 Laryngeal control inFrench stop production A reg berscopic acoustic and electromyographic study FoliaPhoniatrica 30 175plusmn 198
Blumstein S E 1980 Production dereg cits in aphasia a voice-onset time analysis Brain andL anguage 9 153plusmn 170
Blumstein S E Baker E and Goodglass H 1977 Phonological factors in auditorycomprehension in aphasia Neuropsychologia 15 19plusmn 30
Bortolini U Zmarich C Fior R and Bonifacio S 1995 Word-initial voicing in theproduction of stops in normal and preterm Italian infants International Journal ofPediatric Otorhinolaryngology 31 191plusmn 206
Brown A and Docherty G J 1995 Phonetic variation in dysarthric speech as a functionof sampling task European Journal of Disorders of Communication 30 17plusmn 35
Brown W S Morris R J and Weiss R 1993 Comparative methods for measurementof VOT Journal of Phonetics 21 329plusmn 336
Buckingham H W and Yule G 1987 Phonemic false evaluation Theorical and clinicalaspects Clinical L inguistics and Phonetics 1 113plusmn 125
Caruso A J and Burton E K 1987 Temporal acoustic measures of dysarthria associatedwith amyotrophic lateral sclerosis Journal of Speech and Hearing Research 30 80plusmn 87
Cohen H Laframboise M Labelle A and Bouchard S 1993 Speech timing dereg citsin Parkinsonrsquos disease Journal of Clinical and Experimental Neuropsychology 15102plusmn 103
Connor N P Ludlow C L and Schulz G M 1989 Stop consonant production inisolated and repeated syllables in Parkinsonrsquos disease Neuropsychologia 27 829plusmn 838
Cooper M H 1992 Anatomy of the larynx In A Blitzer M F Mitchell C T SasakiS Fahn and K S Harris ( Eds) Neurologic Disorders of the L arynx (New YorkThieme Medical Publishers) pp 3plusmn 11
Darley F L Brown J R and Swenson W N 1975 Language changes after neurosurgeryfor parkinsonism Brain and L anguage 2 65plusmn 69
Davis K 1995 Phonetic and phonological contrasts in the acquisition of voicing Voice onsettime production in Hindi and English Journal of Child L anguage 22 275plusmn 305
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
P Auzou et al148
De Mori R and Flammia G 1993 Speaker-independantconsonant classireg cation in continu-ous speech with distinctive features and neural networks Journal of the AcousticalSociety of America 94 3091plusmn 3103
Diehl R Souther A and Convis C 1980 Conditions on rate normalization in speechperception Perception and Psychophysics 27 435plusmn 443
Dixit R P 1989 Glottal gestures in hindi plosives Journal of Phonetics 17 213plusmn 237Eguchi S and Hirsh I J 1969 Development of speech sounds in children Acta
Oto-laryngologica Suppl 257 1 plusmn 51Flint A J Black S E Campbell-Taylor I Gailey G F and Levinton C 1992
Acoustic analysis in the di erentiation of Parkinsonrsquos disease and major depressionJournal of Psycholinguistic Research 21 383plusmn 389
Forrest K Weismer G and Turner G S 1989 Kinematic acoustic and perceptualanalyses of connected speech produced by Parkinsonian and normal geriatric adultsJournal of the Acoustical Society of America 85 2608plusmn 2622
Freeman F Sands E S and Harris K 1978 Temporal coordination of phonation andarticulation in a case of verbal apraxia A voice onset time study Brain L anguage6 106plusmn 111
Gandour J 1985 A voiced onset time analysis of word-initial stops in Thai L inguistics ofthe T ibeto-Burman Area 8 68-80
Gandour J and DardaranandaR 1984 Voice onset time in aphasia Thai II ProductionBrain and L anguage 23 177plusmn 205
Gandour J Ponglorpisit S Khunadorn F Dechongkit S Boongird P andBoonklam R 1992 Stop voicing in Thai after unilateral brain damage Aphasiology6 535plusmn 547
Gracco V L and Lofqvist A 1994 Speech motor coordination and control Evidencefrom lip jaw and laryngeal movements Journal of Neuroscience 14 6585plusmn 6597
Hardcastle W J Barry R A M and Clark C J 1985 Articulatory and voicingcharacteristics of adult dysarthric and verbal dyspraxic speakers an instrumentalstudy British Journal of Disorders of Communication 20 249plusmn 270
Hertegard S and Gauffin J 1995 Glottal area and vibratory patterns studied withsimultaneous stroboscopy macr ow glottography and electroglottography Journal ofSpeech and Hearing Research 38 85plusmn 100
Hertrich I and Ackermann H 1994 Acoustic analysis of speech timing in Huntingtonrsquosdisease Brain and L anguage 47 182plusmn 196
Hirose H and Gay T 1972 The activity of the intrinsic laryngeal muscles in voicing controlAn electromyographic study Phonetica 25 140plusmn 164
Hoit J D Solomon N P and Hixon T J 1993 E ect of lung volume on voice onsettime ( VOT ) Journal of Speech and Hearing Research 36 516plusmn 521
Itoh M and Sasanuma S 1984 Articulatory movements in apraxia of speech In J CRosenbek M R McNeil and A E Aronson ( Eds) Apraxia of Speech San DiegoCollege Hill
Itoh M Sasanuma S and Ushijima T 1979 Velar movements during speech in a patientwith apraxia of speech Brain and L anguage 7 227plusmn 239
Itoh M Sasanuma S Hirose H Yoshioka H and Ushijima T 1980 Abnormalarticulatory dynamics in a patient with apraxia of speech X-ray microbeam observa-tion Brain and L anguage 11 66plusmn 75
Itoh M Samanuma S Tatsumi I Murakami S Fukusako Y and Suzuki T 1982Voice onset time characteristics in apraxia of speech Brain and L anguage 17 193plusmn 210
Jancke L 1994 Variability and duration of voice onset time and phonation in stutteringand non stuttering adults Journal of Fluency Disorders 19 21plusmn 37
Keller E 1990 Analyse phoneAcirc tique de la Dysarthrie In J L Nespoulous and M Leclercq( Eds) L inguistique et Neuropsycholinguistique Tendances actuelles (Paris SocieAcirc teAcirc deNeuropsychologie de Langue FrancEuml aise) pp 98plusmn 115
Keller E Vigneux P and Laframboise M 1991 Acoustic analysis of neurologicallyimpaired speech British Journal of Disorders of Communication 26 75plusmn 94
Kent R D and Rosenbek J C 1983 Acoustic patterns of apraxia of speech Journal ofSpeech and Hearing Research 26 231plusmn 249
Kent R D Weismer G Kent J F and Rosenbek J C 1989 Toward phonetic intelligibil-ity testing in dysarthria Journal of Speech and Hearing Disorders 54 482plusmn 499
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders 149
Kewley-Port D and Preston M S 1974 Early apical stop production A voice onsettime analysis Journal of Phonetics 2 195plusmn 210
Klatt D H 1975 Voice Onset Time frication and aspiration in word-initial consonantclusters Journal of Speech and Hearing Research 18 686plusmn 706
Lane H Wozniak J and Perkell J 1994 Changes in voice-onset time in speaker withcochlear implants Journal of the Acoustical Society of America 96 56plusmn 60
Lee L Chamberlain L G Loudon R G and Stemple J C 1988 Speech segmentdurations produced by healthy and asthmatic subjects Journal of Speech and HearingDisorders 53 186plusmn 193
Lieberman P Kako E Friedman J Tajchman G Feldman L S and Jiminez E B1992 Speech production syntax comprehension and cognitive dereg cits in Parkinsonrsquosdisease Brain and L anguage 43 169plusmn 189
Lieberman P Protopapas A and Kanki B G 1995 Speech production and cognitivedereg cits on Mt Everest Aviation Space and Environmental Medicine 66 857plusmn 864
Lisker L and Abramson A S 1964 A cross-language study of voicing in initial stopsAcoustical measurements Word 20 384plusmn 422
Lisker L and Abramson A 1967 Some e ects of context on voice onset time in englishstops L anguage and Speech 10 1 plusmn 28
Lofqvist A 1992 Aerodynamic measurements of vocal function In A BlitzerM F Mitchell C T Sasaki S Fahn and K S Harris ( Eds) Neurologic Disordersof the L arynx ( New York Thieme Medical Publishers) pp 98plusmn 107
Lofqvist A and Lindblom B 1994 Speech motor control Current Opinion in Neurobiology4 823plusmn 826
Lofqvist A Baer T McGarr N S and Story R S 1989 The cricothyroid muscle invoicing control Journal of the Acoustical Society of America 85 1314plusmn 1321
Macken M A and Barton D 1980a The acquisition of the voicing contrast in Englisha study of voice onset time in word-initial stop consonants Journal of Child L anguage7 41plusmn 74
Macken M A and Barton D 1980b The acquisition of the voicing contrast in Spanisha phonetic and phonological study of in word-initial stop consonants Journal of ChildL anguage 7 433plusmn 458
Miller J L 1981 Some e ects of speaking rate on phonetic perception Phonetica 38159plusmn 180
Miller J L and Volaitis L E 1989 E ect of speaking rate on the perceptual structureof a phonetic category Perception and Psychophysics 46 505plusmn 512
Miller J L Green K P and Reeves A 1986 Speaking rate and segments A look atthe relation between speech production and speech perception for the voicing contrastPhonetica 43 106plusmn 115
Monsen R 1976 The production of English stop consonants in the speech of deaf childrenJournal of Phonetics 4 29plusmn 42
Morris R J 1989 VOT and dysarthria a descriptive study Journal of CommunicationDisorders 22 23plusmn 33
Morris R J and Brown W 1987 Age-related voice measures among adult women Journalof Voice 1 38plusmn 43
Morris R J and Brown W S 1994 Age-related di erences in speech variability amongwomen Journal of Communication Disorders 27 49plusmn 64
Munhall K G Lofqvist A and Kelso J A S 1994 Lip-larynx coordination in speechE ects of mechanical perturbations to the lower lip Journal of the Acoustical Societyof America 3605plusmn 3616
Murdoch B E Chenery H J Bowler S and Ingram J C 1989 Respiratory functionin Parkinsonrsquos subjects exhibiting a perceptible speech dereg cit a kinematic andspirometric analysis Journal of Speech and Hearing Disease 54 610plusmn 626
Ozsancak C Auzou P Jan M LeAacute onardon S Gaillard M J and Hannequin D1997 Pseudobulbar and Parkinsonian Dysarthria an acoustical analysis Journal ofNeurology 244 S53
Petrosino L Colcord R D Kurcz K B and Yonker R J 1993 Voice onset time ofvelar stop productions in aged speakers Journal Perceptual and Motor Skills 7683plusmn 88
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111
VOT and speech disorders150
Pind J 1995 Speaking rate voice-onset time and quantity The search for higher-orderinvariants for two Icelandic speech cues Perception and Psychophysics 57 291plusmn 304
Pind J 1996 Rate-dependent perception of aspiration and pre-aspiration in IcelandicQuarterly Journal of Experimental Psychology 49A 745plusmn 764
Ryalls J and Larouche A 1992 Acoustic integrity of speech production in children withmoderate and severe hearing impairment Journal of Speech and Hearing Research35 88plusmn 95
Ryalls J Provost H and Arsenault N 1995 Voice onset time production in French-speaking aphasics Journal of Communication Disorders 28 205plusmn 215
Shewan C Leeper H and Booth J 1984 An analysis of voice onset time ( VOT ) inaphasic and normal subjects In J Rosenbek M McNeil and A Aronson ( Eds)Apraxia of speech (San Diego College Hill Press) pp 197plusmn 220
Solomon N P and Hixon T J 1993 Speech breathing in Parkinsonrsquos disease Journal ofSpeech and Hearing Research 36 294plusmn 310
Stevens K N and Klatt D H 1974 Role of formant transitions in the voicedplusmn voicelessdistinction for stops Journal of the Acoustical Society of America 55 653plusmn 659
Sussman H and Shore J 1996 Locus equations as phonetic descriptors of consonantalplace of articulation Perception and Psychophysics 58 936plusmn 946
Sweeting P M and Baken R J 1982 Voice onset time in a normal-aged populationJournal of Speech and Hearing Research 25 129plusmn 134
Tuller B 1984 On categorizing aphasic speech errors Neuropsychologia 22 547plusmn 557Tyler A A and Watterson T L 1991 VOT as an indirect measure of laryngeal function
Seminars in Speech L anguage 12 131plusmn 141Vohr B R Garcia-Coll C T and Oh W 1988 Language developmentof low birthweight
infants at two years Developmental Medicine and Child Neurology 30 608plusmn 615Volaitis L E and Miller J L 1992 Phonetic prototypes Inmacr uence of place of articulation
and speaking rate on the internal structure of voicing categories Journal of theAcoustical Society of America 92 723plusmn 735
Westbury J R and Keating P A 1986 On the naturalness of stop consonant voicingJournal of L inguistics 22 145plusmn 166
Zebrowski P M Conture E G and Cudahy E A 1985 Acoustic analysis of youngstutterersrsquo macr uency Preliminary observations Journal of Fluency Disorders 10173plusmn 192
Zenker W 1964 Question regarding the function of external laryngeal muscles In D Brewer( Ed ) Research potentials in Voice physiology (New York State University of NewYork) pp 20plusmn 40
Zlatin M A 1974 Voicing contrast Perceptual and productive voice onset time character-istics of adults Journal of the Acoustical Society of America 56 981plusmn 984
Zlatin M A and Koenigsknecht R A 1976 Development of the voicing contrast acomparison of voice onset time in stop perception and production Journal of Speechand Hearing Research 19 93plusmn 111