Foreperiod and stuttering severity effects on acoustic laryngeal reaction time

/OURNAL

Of FLUENCY

DISORDERS

8 (1983)

183-205

Foreperiod and Stuttering Severity Effects on Acoustic Laryngeal Reaction Time Ben C. Watson University

of Connecticut,

Storrs

Peter j. Alfonso University

of Connecticut,

Storfs,

and Haskins Laboratories,

New Haven, Connecticut

An earlier paper presented a model of the laryngeal reaction time (LRT) paradigm that included several factors that appeared to affect LRT values. The present study assesses the effects of two of these factors: foreperiod former was assessed by the use of 13 foreperiod by classifying

experimental

factors significantly

subjects

affected LRT

and stuttering

durations.

as either

mild

LRT

values approached normal

stutterers’

LRT

foreperiods. vibration

values Results

initiation

are discussed

underlying

in

deficits underlying

stutterers’

of differential

values

posturing

severe at all and/or

delayed LRT values. We caution

alone are insufficient

to specify fully the nature of the

deficits.

A number

of

Starkweather

experiments et al.,

(most

1976;

notably

Cross

and

Adams

Luper,

that stutterers

as a group

are significantly

initiating

phonation

in response

to reaction

reaction time paradigm

that allowed

response, we unexpectedly

experiments significant

(Watson

Hayden,

Cross

slower

1976;

et al.,

1979)

than normals

signals.

Using

in

a simple

subjects 1 to 3 set to prepare for a failed to replicate the results of these

and Alfonso,

group difference

and

1979;

reported

known

mild stutterers’

increased, whereas

greater than normal

terms

Both

these factors demon-

Specifically,

values as foreperiod

remained significantly

that acoustic measurements

or severe stutterers.

values. More importantly,

strated a composite effect on group LRT differences.

severity. The

The latter was assessed

1982).

in laryngeal

That is, we failed reaction

time

to find a

(LRT) between

stutterers and nonstutterers, a difference we will refer to as the LRT effect. However, auditory vowel

we did find significant and visual

reaction

and phase-initial

vowel

within-group

signal conditions response

LRT differences and

conditions.

between The

between isolated

latter results

Address correspondence to Peter J. Alfonso, Department of Communication Sciences, Box U-85, University of Connecticut, Storrs, CT 06228. “Elsevier Science Publishing Co., Inc., 1983 52 Vanderbilt Ave., New York, NY 10017

0094-730X/83/$03.00

Ben C. Watson and Peter ). Alfonso

184

suggested to us that our

LRT

measurements

were indeed sufficiently

sensitive to detect an LRT effect if one existed. Other recent investigations have also failed to demonstrate a significant and adult subjects (Cullinan ner, 1981; Venkatagiri, by our original

and Springer,

1981, 1982).

experiment,

We are interested significant

as other recent experiments

those factors that form the basis for

between stutterers

end, we have conducted experiments paradigm developed

that

LRT effect.

in isolating

LRT differences

effect in both children Murphy and Baumgart-

The study reported here is motivated

as well

failed to demonstrate a significant

LRT

1980;

in our original

and their controls.

To this

based on the model of the LRT experiment.

The

model

includes

factors related to the perception of the reaction signal, production of the response, subjects

and factors specifically that influence

model “reaction sponse

type,”

significant function

LRT

signal

LRT

values.

For example,

modality,”

a production differences

of reaction signal

condition

related to characteristics

we included

a perceptual component,

component,

based on our

for both nonstutterers modality

(isolated vs phrase-initial

of stuttering

(visual

in the

and “re-

findings

and stutterers

vs auditory)

of as a

and response

vowel). The first purpose of the study

reported here is to investigate further the effects of two other factors on stutterers’

LRT

values as well

as on the LRT

included in the model as foreperiod

effect. These

and stuttering

severity.

that our failure to find a significant

LRT effect in our original

was related to our use of relatively

long foreperiods

moderate severity rating of our experimental

factors are We argued experiment

and to the mild-to-

group.

The foreperiod factor is included in the “perceptual component” of the model,

although

production

interval. In our experiments,

events

may also

occur during

this

foreperiod is defined as the interval between

the presentation of the warning cue and presentation of the phonate cue. Sufficiently known

long foreperiods provide the subject with time to prepare for a

response (Niemi

and Naatanen, 1981).

Preparatory activity that

may occur during the foreperiod includes perception of the warning cue, formulation

and transmission

of appropriate motor commands to posture

the speech mechanism for the required response, and movements of the various

components

prephonatory

of the speech mechanism

posture.

occurs is a function

The

to achieve the required

extent of preparatory

of foreperiod

duration. Thus,

activity

that actually

short foreperiods

may

ForeperiodlSeverity

restrict

Effects on Acoustic LRT

preparatory

perhaps,

activity

formulation

foreperiods

185

to perception

and transmission

may permit

mands and posturing

formulation

of the warning of

motor

commands;

and transmission

of the speech mechanism

cue and, long

of motor com-

prior to presentation of

the phonate cue. The

notion of a foreperiod

supported by Izdebski’s

effect on nonstutterers’

LRT

values is

(1980) observation of a U-shaped function when

LRT values are plotted across a range of increasing foreperiods. That is, he found that LRT values decrease to a minimum

as foreperiod increases to

about 1500 msec and then increase as foreperiod increases beyond 1500 msec. These results suggest that LRT values occurring at foreperiods

less

than 1500 msec may reflect the subject’s inability

to complete preparat-

ory activity.

msec may reflect the

increasing

subject’s inattention

LRT

values beyond 1500

to the task or failure to maintain the prephonatory

posture. We have argued previously particularly

dependent

hypothesized

on

that stutterers’

foreperiod

that when certain stutterers

posture the speech mechanism,

LRT

duration.

values may be

Specifically,

are given sufficient

We concluded that the long foreperiods

original

(1 to 3 set) provided stutterers

used in our

with ample time to

achieve the appropriate posture prior to the initiation contributed to our finding of a nonsignificant

of phonation and

LRT effect.

The studies referred to earlier that reported a significant (and used isolated vowels response

conditions

as the response, a task similar

in our original

warning cues in their experimental Cross and Luper, 1979;

experiment)

posture

experiments warning

Cross et al., 1979).

before the presentation

that report significant

cue may reflect stutterers’

did not incorporate

Consequently,

LRT

LRT

the response.

it cannot be

achieved the appropriate

of the phonate cue. Thus, effects but do not include

difficulty

with posturing

mechanism prior to phonation onset as well as difficulties initiating

LRT effect

to one of the

designs (Adams and Hayden, 1976;

determined if the stutterers in these experiments response

time to

they will demonstarte LRT values similar

to those of normals. experiment

we

a

the speech

associated with

It seems possible that certain stutterers’

delayed

values may be related to posturing, that is, prephonatory events (as

suggested

by Freeman

delayed LRT

values

and Ushijima,

1978),

may be more directly

while

other

stutterers’

related to initiation

of the

response or perhaps a combination of posturing and initiation activities.

If

186

Ben C. Watson

and Peter 1. Alfonso

this is the case, one may suspect that certain stutterers’ approach normal values as foreperiod ers’ LRT

values could remain

throughout in this

LRT

increases. However,

significantly

the entire range of foreperiods.

values will other stutter-

greater than normal values The first hypothesis

study states that there is a foreperiod

under test

effect on stutterers’

LRT

values. To test this notion, we extended the range of the foreperiods from 100 to 3000

msec. Specifically,

those stutterers

with

deficits

only

in

posturing the speech mechanism will demonstrate LRT values approaching normal values as foreperiod deficits in initiating

increases, whereas those stutterers

with

the response, or in both posturing and initiation,

will

demonstrate LRT values significantly

greater than normal values through-

out the range of short to long foreperiods. The second factor that may affect stutterers’ severity,

included

in the model

results of several studies (Hayden, 1975; Alfonso,

LRT values is stuttering

under “subject

characteristics.”

Lewis et al., 1979;

1982) suggest that mild stutterers

may exhibit

The

Watson and

LRT values more

similar to normals than would severe stutterers.

Additional support for this

notion

obtained

is found

experiment

in a comparison

rating of our experimental et al. (19811,

for

differences

yet the

results

groups. Finally,

are clearly

in the stuttering

hypothesis stutterers’

significantly

and nonstutterers

under test

procedures were very

different.

of similar We

duration,

suggest

that

ratings of the experimental

severity effect on timing

Specifically,

is found

she observed that severe

longer vocal and manual “execution”

time values than did nonstutterers mild stutterers

severity

support for a stuttering

displayed

Reich

of these studies may, in part, be attribut-

in data reported by Borden (1982). stutterers

However,

as moderate-to-severe,

Both included foreperiods

between the results

able to differences

classified

LRT effect. The experimental

in the two studies.

example,

subjects

in our original

The average severity

group was mild-to-moderate.

using stuttering

obtained a significant similar

of results

and in a study by Reich et al. (1981).

while none of the differences between reached significance.

is that there

is a stuttering

Thus,

the second

severity

effect on

LRT values. That is, we expect that a group of severe stutterers

will demonstrate greater LRT values than will a group of mild stutterers. The two hypotheses just described assess the independent effects of foreperiod

and stuttering

pared to nonstutterers.

severity

Of interest,

on stutterers’

LRT

values when com-

however, is the relationship

between

ForeperiodlSeverity

foreperiod

Effects on Acoustic

and stuttering

LRT

severity.

187

Consequently,

the second purpose of

this study was to assess the combined effect of foreperiod and stuttering severity

on stutterers’

certain

stutterers’

LRT

LRT

values. values

For example,

would

foreperiod increases, in that these stutterers’ primarily tively,

related to difficulty

we hypothesized

significantly

different

foreperiods,

implying

related to difficulty

initiating

and initiation

stutterers,

classified

“posture” mild

whereas

normals’

by severity,

primary

as

Alterna-

values would

throughout

the entire

delayed LRT

remain

range of

values may be of

could be characterized according to the hypothesis.

That is, is it the case that

is posturing

primary

posturing and response initiation? Specifically,

that

We wanted to ascertain if groups of

difficulty

severe stutterers’

LRT

the response or, perhaps, a combination

difficulties.

versus the “initiation”

stutterers’

values

posturing the speech mechanism.

that these stutterers’

posturing

normal

delayed LRT values may be

that other stutterers’ from

we hypothesized

approach

difficulty

the speech mechanism is some combination

of

The third hypothesis tested this notion.

we expected that mild stutterers’

normal values, whereas severe stutterers’

LRT

values will approach

LRT values will

remain signifi-

cantly greater than normal values, as foreperiod increases. In summary,

the first

purpose of this study was to determine

the

effects of two factors included in the model of the LRT paradigm (Watson and Alfonso, second

1982) on the LRT effect and on stutterers’

purpose

was

deficits-posturing

to test

the

notion

versus initiation-underlie

LRT values. The

that qualitatively

different

mild and severe stutterers’

delayed LRT values.

Subjects Subjects participating in this study included 10 adult stutterers and 5 adult nonstutterers. stutterers’

LRT

In order to test the effect of stuttering

values,

it was necessary

subjects on this dimension. separate

analyses

pathologist during

of

subjectively

Stutterers

severity.

to classify

the experimental

First,

a certified

speech-language speech and speech

Passage of the stuttering

second certified speech-language pathologist objectively speech samples using the Stuttering the Stuttering

Interview

on

were classified on the basis of three

rated the conversational

reading of the Rainbow

severity

A

rated the same

Severity Index (SSI) (Riley,

(SI) (Ryan, 1974).

subjects.

1972) and

Ben C. Watson and Peter 1. Alfonso

188

The

results

of the stuttering

indicate that the experimental groups: five severe stutterers

severity

analysis,

shown

subjects could be classified and five mild stutterers.

Weiss,

1965;

Izdebski,

1980),

1,

Since reaction time

values may be affected by subject sex and age (Birren 1955;

in Figure

as two distinct and Botwinick,

we matched the control

against the average age and sex ratio of the two stuttering

group

groups.

Test Stimuli Figure 2 illustrates effect of foreperiod

one sequence of the stimuli

on LRT

variable interstimulus

interval

require

breathe normally

that subjects

Consequently,

used to assess the

values. Each sequence was separated by a (ISI) of 8 to 12 sec. ISIS of this duration between response

sequences.

subjects are not able to remain in a phonatory position

-

EXPERIMENTAL

SEX -

AQE

QROUP

SI

M

48;4

severe

severe

severe

severe

2

F

36;3

severe

severe

severe

severe

3

M

31;6

severe

severe

severe

severe

PUBJECT

1

4

881

SUBJECTIVE

F

30;o

severe

severe

severe

M

w 33;8

severe

severe

moderate

6

M

44:5

mild

mild

mild

7

M

41;4

mild

mild

mild

mild/ moderate mild

8

M

26;9

mild

ioderate

mild

mild

9

M

22;8

mild

mild

mild

mild

mild

mild

mild

mild

5 Mean A08

IA%" Apc

-

M -

z&s

11

M

48;7

normal

12

F

35;8

normal

13

M

26;lO

normal

M

26;0

normal

M -

%ifL

14 &

Figure 1:

severe I

normal

Results of the stuttering

A

severity

analysis.

severe

Foreperiodbeverity

Effects on Acoustic LRT

189

Phonation Interval

Reaction Signal onoe, onael

Variable I.S.I.

b .

Termination Signal

t

I

,

l

\

1

Variable Foreperiod S-12 Seconds

_

100-3000 msec.

_

2 Seconds

1 Second

-

.

Figure 2: One sequence of stimuli used to assess the effect of foreperiod on LRT. The reaction signal varied from 100 to 3000 msec. Reaction signal onset served as the warning cue, offset as the phonate cue.

between responses.

The reaction signal consisted of the synthetic

vowel

/a/. Onset of the reaction signal served as the warning cue, and the offset served as the phonate cue. Subjects

were instructed

to “get ready” to

phonate when and only when they heard the warning cue. Duration

of

the reaction signal varied from 100 to 500 msec in IOO-msec increments, 700 to 1500 msec in 200-ms 500-mesc

increments,

increments,

and from 2000 to 3000 msec in

a total of 13 foreperiods.

A “terminate

phonation”

signal was presented 2 set after the phonate cue. The terminate consisted

of the synthetic

vowel

replicated five times, randomized, Haskins

Laboratories

/i/. Each of the 13 sequences

signal was

and output onto audiotape using the

Plus Code Modulation

(PCM) system.

Procedures Stimulus

sequences were presented simultaneously

seated in a soundproof recorder.

Subjects’

to the subject,

booth, and to track one of a two-track

responses

tape

were recorded on track two of the tape

recorder. Subjects were instructed to phonate the vowel /a/ immediately at the offset

of the reaction

signal

and to continue

phonation

until

presentation of the terminate signal. All subjects were allowed 21 training sequences,

including

long and short foreperiods.

required fewer than the maximum the relatively

Although

most subjects

number of training sequences to learn

simple task, all subjects were exposed to training sequences

Ben C. Watson and Peter I. Alfonso

190

containing

long and short foreperiods.

sented in two 7-min

Response

sequences were pre-

tests separated by an optional

3- to 5-min

rest

interval. Fluency Criteria We followed

the same procedures used in our original

to ensure that only fluent responses were analyzed. instructed

to identify

any production

Second, the experimenter dysfluent. Finally,

experiment

subjects were

that they thought was dysfluent.

noted any production

that he thought was

No responses were omitted on the basis of the first two criteria. productions

were excluded from the data set if the waveform

showed certain irregulatirites ing,

First,

such

as isolated

phonation.

As

that may be related to nonaudible stutter-

pitch

a result

pulses

before the onset

of the third

excluded from the mild stutterers’

criterion,

three

of continuous responses

were

data set, one response was excluded

from the severe stutterers’ data set, and no responses were excluded from the nonstutterers’ stutterers,

data set. Thus,

322 LRT values were measured for mild

324 values were measured for severe stutterers,

and 325 values

were measured for nonstutterers. Measurements Data were analyzed with the aid of a computer waveform-editing system

at Haskins

Laboratories.

Temporal

resolution

analyzer is accurate to one-tenth of a millisecond

of the waveform

(Nye et al., 1975). LRT

values were defined as the interval between the offset of the phonate cue and the onset of the first regular pitch pulse of the voiced vowel /a/. Statistical Analyses All data were subjected to several multiple

correlation

(MCR) analyses (Cohen & Cohen, 1975) for the following the procedure

permits

analysis

of interaction

regression

reasons. First,

effects between interval

(foreperiod) and nominal (stuttering severity) level independent variables, a capability

not provided

procedures.

Second,

MCR

specific group comparisons.

by traditional analysis Finally,

multiple

permits

analysis

experimenter

of variance selection

of

MCR analysis allows for the evalua-

ForeperiodlSeverity

Effects on Acoustic LRT

tion of nonlinear

relationships,

between foreperiod

such as the hypothesized

and LRT.

ment was a subjects within

191

relationship

The statistical

design used in this experi-

groups (normal,

mild, severe) by condition

(foreperiod) repeated measures MCR. This

design requires separate MCR

analyses to determine (1) the significance of the between-subject ing severity)

main effect and (2) the within-subject

effect and interaction MCR

analysis

(stuttering

severity

was conducted

coded to permit separate comparisons analysis

and between mild

foreperiod)

to determine

stuttering severity factor. For this analysis, stutterers

x

main

effect. The first

the significance

of the

the subject group variable was between nonstutterers

and severe stutterers.

was conducted to determine

(stutter-

(foreperiod)

The

the signifiance

and mild

second MCR

of the foreperiod

factor and the interaction betweeen stuttering severity and foreperiod. For this analysis, the subejct group variable was, once again, coded to permit comparisons mild

between normals

and severe stutterers.

determine

the

foreperiod

magnitude

and LRT

of

MCR analysis

the

nonlinear

effect. Finally,

values at each foreperiod

parametric Randomization the criteria data (Siegel,

as well as between was conducted to

relationship

between

for each group in order to determine whether there

was an optimal foreperiod mean LRT

and mild stutterers

The third

comparisons

between group

were conducted using the non-

Test for Independent Samples, since several of

required by parametric analyses were not fulfilled

by these

1956).

RESULTS Figure 3 displays

a summary

Each data point

in this

‘The

of LRT

figure

values for the complete data set.’

represents

present study reports results obtained

from statistical

data set. In so doing, it is consistent

with

stutterers.

are sometimes

However,

minimum

LRT

procedures

two procedures

values

is that

responses occurring

LRT

prior

to group

values

(1978)

most LRT

faster

of the complete nonstutterers

used to eliminate

the maximum

comparisons.

significantly

analysis

studies comparing The than

rationale

for

the mean

and lzdebski

either

reflect

before the phonate cue, whereas values significantly

mean reflect the subjects’ inattention and Shipp

the average of all analyzed

with and

of these

anticipatory

slower than the

to the task. As an example of one procedure, lzdebski

(1980)

used statistical

tests to eliminate

only significant

outliers.

As an example of the second procedure, Reich et al. (1981) omitted the fastest and

slowest

responses of each subject before group comparisons.

will discuss the effects of various data-reduction

In a forthcoming

procedures on the LRT

effect.

paper, we

Ben C. Watson and Peter 1. Alfonso

192

550

Nonstutterers

Mom 279.23

SD. 27.03

Mild Stutterers

340.35

46.26

Severe Stutterers

390.19

Qroup _ 500

f

1V 3 1:i I !i

460

48.00 T

T

7 6 400 :: I .S

..

‘.+.....”

. ..’ ..a.- ,

1;

5 350

J_L____-l ’ I I I

44

3 8 300

‘t

l

T

250

200

Fcrqmrbd Figure 3:

Acoustic

LRT

In Msoc.

values in group means and standard deviation disper-

sions for the 13 foreperiod

conditions.

Each data point represents the individual

subject averages pooled across the five subjects in each group.

responses

values

per subject

are expressed

dispersions Also

shown

pooled

in

across the five subjects in each group. LRT

group

means

and two

standard

for the three subject groups and 13 foreperiod are group means and standard deviations

the 13 foreperiod

conditions.

LRT

deviation conditions.

collapsed across

values for nonstutterers

are shown as

closed circles, for mild stutterers as open circles, and for severe stutterers as open triangles.

Note that this figure demonstrates that LRT

function of subject group and foreperiod.

varies

as a

ForeperiodlSeverity

Effects on Acoustic LRT

193

TABLE 1 Summary

of Main and Interaction

Effects F

Main Effects

2,12

f3.88a*b,c

Stuttering severity Foreperiod Interaction

ti

3.1 ahc

12,144

Effect

Stuttering severity by foreperiod

0.52

24.144

T.99 (2,12) = 6.93. bF.99 (12,144) = 2.31. ‘F.95 (24,144) = 1.59.

The

stuttering

first

two

severity

hypotheses

in this

effects on LRT.

main effects as well as the stuttering effect are summarized

in Table

study

The results severity

1. This

predicted foreperiod

and

of MCR a.nalyses of these by foreperiod

table shows

interaction

that both, the

stuttering severity and foreperiod factors are significant @I < 0.01). Partial regression MCR

are presented

coefficients in Table

obtained from the between-subjects

2. Coefficients

for both the nonstutterer

versus mild stutterer and mild versus severe stutterer group comparisons were significant LRT

fp
These

values were significantly

results

different

indicate that the three groups’ when collapsed

across the 13

foreperiod conditions. Table 3 shows

results

describing the relationship group. Second-order

of analyses of the power of the polynomial between foreperiod and LRT for each subject

polynomials

were found for the nonstutterers

TABLE 2 Partial

Regression

Coefficients

for Stuttering

Comparison Nonstutterers Mild stutterers

Severity

vs mild stutterers vs severe stutterers

V.99 (1,12)= 9.33.

Factor F

df

-57.36

14.1ga

1,12

-53.59

1 2.3ga

1,12

8

and

Ben C. Watson and Peter 1. Alfonso

194

TABLE 3 Summary

of Power

Polynomial

Analysis

Power Term Nonstutterers

Mild stutterers

Severe stutterers

of Foreperiod F

Inc R Square

df

Linear

0.27

4.17

1,ll

Quadratic

0.30

08.10”

1,lO

Cubic

0.06

1.45

I,9 1,ll

Linear

0.17

2.27

Quadratic

0.33

11.61b

1,lO

Cubic

0.16

4.64

1,9

Linear

0.11

1.39

Quadratic

0.36

13.33‘

Cubic

0.26

"F.95 (1,ll) =

4.84.

bF.95 (1,lO) =

4.96.

1,ll 1,lO

8.66d

1,9

‘F.99 (1,lo) = 10.04. dF.95 (1,9)=

mild

5.12.

stutterers.

minimum

That

is,

LRT

values

ship between foreperiod

and LRT

following

analysis

subjects,

a second-order

decrease to a

was also reported by lzdebski

(1980)

of a reduced data set. He found, using only normal relationship

However,

our

foreperiod

for severe stutterers

shows

for these subjects

and then increase as foreperiod increases. A nonlinear relation-

data indicate

that a third-order

between

foreperiod

that the relationship is different.

polynomial

and LRT.

between

LRT

For these subjects,

also becomes significant

and

Table

3

and ap-

proaches the second-order term in best describing the shape of the curve. This

implies

that LRT

values for severe stutterers

minimum,

then increase to a maximum,

foreperiod

increases.

These

results

tend to decrease to a

and then decrease again as

emphasize

the difference

severe stutterers versus mild stutterers and nonstutterers.

between

For example, the

data shown in Figure 3 for mild stutterers

and nonstutterers

show single

maximum

a single inflection

point in the

curve.

and minimum

A curve

deviations, Analysis tively.

of predicted

LRT

values,

representing

least-squared

was obtained by solving regression equations for each group.

of predicted

nonstutterers

values, yielding

curves

and mild stutterers

For severe stutterers,

indicates

that the inflection

occur at 2000

points

for

and 1500 msec, respec-

there is less difference

between maximum

ForeperiodlSeverity

and minimum and 2500

Effects on Acoustic LRT

LRT values

and the curve

of 2000

lzdebski

For the severe stutterers,

(1980).

foreperiod

for mild

stutterers

msec. Thus, minimum membership. maximum

for the

LRT values

severe

is, both the stuttering

severity

between

and severe

foreperiods.

stutterers

nonstutterers’

hypothesis

stated

stutterers’

of foreperiod.

Our original

between

nonstutterers

and 3-set

foreperiods.

Hence,

significant

differences

between

only at foreperiods

post-hoc

zation

Test for Independent

shown

below

and

difference

the

symbols

between

severe stutterers’ all

13 foreperiods

values

foreperiods,

revealed

that

M line

mild

nor-stutterers’

group

S refer

connecting means.

fp < 0.05). greater

at I-, 2-, to find

stutterers’

LRT

we expected

and severe stutterers’

These hypotheses

were

by using the Randomi-

Results of these comparisons to

mild

groups

Results

On the other less than

and

severe

of this

no

analysis

stutterers, significant reveal

greater than nonstutterers’ hand, 1100

are

N refers to nonstutter-

indicates

than nonstutterers’

is, at foreperiods

as a

nonsignificant

stutterers

and

group comparisons

and

and

between

study, we expected

nonstutterers’

LRT values are significantly

are significantly

across the 13

LRT values

and mild-moderate

Samples.

group

as between

LRT values.

the abscissa in Figure 3. The symbol A solid

significant

as well

severity)

in the present

to

coefficients

was a difference

experiment

between

the

effect on nonstutterers’

short and long foreperiods.

tested by conducting

respectively.

regression

less than 1100 msec. Conversely,

differences

at both

by

stutterers

of this study. That

are collapsed

there

(grouped

differences

to find significant

that

effect on the

To summarize,

main effect reflects

LRT values

1300

factor were shown

partial

has a greater

at a

of group

and mild

LRT values.

and mild stutterers

when

foreperiod

and

LRT values

of nonstutterers

LRT values than on severe stutterers’

The third

occurred

has a greater

factor and foreperiod severity

by

the fastest LRT values

but seems to be around

In addition,

900

occurred reported

fastest LRT values

the first two hypotheses

nonstutterers

Finally,

mild stutterers’

ers,

points,

the results

at which

stutterers’

significantly.

that the stuttering

differences

with

is less clear,

thus far support

affect LRT values

function

consistent

it appears that foreperiod

and minimum

results reported

values

inflection

LRT values also seem to vary as a function

Finally,

it does

revealed

msec,

of 500 msec. The foreperiod

occurred

mild

has two

msec. Note also that the fastest LRT for nonstutterers

at a foreperiod

than

195

mild

at only msec.

stutterers’

that at LRT

5 of the first 7

However,

we

Ben C. Watson and Peter /. Alfonso

196

unexpectedly

found significant

and mild stutterers msec. Thus,

LRT

results

differences

as a function

of foreperiod

differences between nonstutterers’

and

LRT values, but only partially support our hypothesized

between nonstutterers’

general, these results

and mild

demonstrate

proach those of nonstutterers stutterers’ LRTs

between nonstutterers

equal to and greater than 1100

of group comparisons

clearly support our hypothesized severe stutterers’

differences

at 4 of 6 foreperiods

stutterers’

LRT

that mild stutterers’

LRT

as foreperiod

remain significantly

increases,

values.

In

values ap-

whereas

greater than nonstutterers’

severe

throughout

the entire range of foreperiods. DISCUSSION The first

interesting

finding

of the present study is that of a significant

stuttering severity factor. This finding is consistent with reaction time data for complex Using

vocal and manual responses

the same stuttering

constant foreperiod differences

subjects

reported by Borden (1982).

used in the present study

and a

equal to 1 set, she also observed significant

group

between nonstutterers

and severe stutterers

of perceptually fluent counting and finger-tapping between nonstutterers reach significance. stuttering

and mild

Thus,

stutterers

for the execution

responses. Differences

for the same tasks failed to

results of this and the Borden study indicate that

severity affects group reaction time differences for both simple

and complex vocal responses as well as for manual responses.

Further-

more, these results suggest that the delayed LRT values demonstrated by the experimental

subjects may represent an underlying

motor control in stutterers

as a group. Finally,

these results suggest that

the magnitude of the delay, and correspondingly deficit,

is reflected in the stuttering

measurements

alone do not permit

cesses occurring discussion, motor-control

severity analysis

deficit in general

the magnitude of the

rating. Of course, acoustic of the motor-control

pro-

before the onset of the acoustic response. Later in this

we will

suggest

procedures

that may allow

analysis

of

processes during posturing and response onset.

Perhaps the most interesting

and important finding

the composite effect of the stuttering

severity

of this study is

and foreperiod factors on

the significance of group LRT differences between stutterers terers. Specifically,

and nonstut-

we observed that mild stutterers’ LRT values approach

Foreperiodheverity

normal

Effects on Acoustic LRT

values

as foreperiod

values are significantly of foreperiods. our original

increases,

whereas

severe stutterers’

greater than normal values throughout

LRT

the range

These results are in general agreement with the findings of

LRT experiment.

LRT differences stutterers

197

That study failed to show significant

between nonstutterers

for foreperiods

equal to 1, 2, and 3 sec. Although

study reports nonsignificant this range, it should

group

and a group of mild to moderate the present

differences at only two of six foreperiods

be pointed out that the results

in

of the present study

reflect fewer subjects per group, fewer responses per subject, and the use of nonparametric statistics. supports

our original

stutterers’

With these differences aside, the present study

experiment

and nonstutterers’

LRT

differences between nonstutterers’ Throughout

in that the differences

between mild

values are significantly

less than the

and severe stutterers’

LRT values.

this article, we have noted that long foreperiods

permit

subjects to complete activity required to posture the speech mechanism for the voiced response. LRT

values

approach normal

severe stutterers’ contribute mild

LRT

the finding

values as foreperiod

LRT

values

for

the two

with regard to the comparisons

stutterers,

our results

primary

generally

difficulty

vibration,

since their LRT

severe stutterers

Results

Following

this

example,

better understand stutterers’

delayed

How-

initiating

between nonstutterers

and

suggest that these stutterers deficits.

have been studied with (Botwinick

respect to their

and Thompson,

1966).

we have chosen

to study

the posture

of the reaction time

response

in an attempt to

the qualitative LRT

and

that mild

have some difficulty

of foreperiod

components

components

of stutterers.

values do not become identical with those of

as a function

and motor

groups

the hypothesis

of the comparisons

Reaction time responses

initiation

whereas

the speech mechanism.

may have both posturing and vibration-initiation premotor

increases,

between nonstutterers

support

is posturing

ever, it is also likely that our mild stutterers the nonstutterers.

that mild stutterers’

values do not, suggests that different deficits may

to delayed

Specifically, stutterers’

Consequently,

values.

differences These

in the deficits

components

and

underlying

are schematically

represented in Figure 4. The posture component is represented by a series of processes formulation

related to perception of the warning and transmission

speech mechanism,

of neuromotor

and/or phonate cue,

commands to posture the

posturing of the speech mechanism for the required

Figure 4:

Onset

Rk&t&

,

-

F&&E

Articulatory (V.T.Shape)

Laryngeal eaistancel

components

Posture Time

ExaZ~ion POSTURE cornrnands

mw3on and

n Trans -

Commands

-

b

Frmwlatiotl Respiratory (Psgl

Postural Processes

Processes

Newomotol r

Posture and initiation

lritaraction of Modality Y-l ana hltanaity Parcepts

Derceptual hocesses

POSTURE COMPONENT

of the laryngeal

MlJSCleS

reaction

time response.

COMPONENT

Initiation Processes

Aal (4Gbttal Raaiatarcr?)

#WI

Contraction of Raspkatory

INITIATION

+Muscle RT. 2 -Movement R.T. Y Acoustic R.T.

Transmission and Execution of INITIATION Commands

Neuromotor Processes

. Reaction Signal Offset

Formulation of HlTlATlON Commands

Neuromotor Processes

PhzFe

Foreperiodbeverity

response,

Effects on Acoustic LRT

and formulation

of neuromotor

response. The formulation occur

simultaneously

of neuromotor with

the

by processes commands

gestures. The initiation

the response.

The

(3) acoustic output.

which

foreperiod

duration

of

component is represented

consequences

of neuromotor

of executing

(2) articulator

Figure 4 demonstrates permits

the may

and transmission

and execution

commands are: (1)muscular adjustments, finally,

to initiate

Postural processes are also taken to

related to the transmission for

commands

commands for initiation

formulation

neuromotor commands for posturing. include prephonatory

199

completion

these

movement, and,

the special case in

of all postural

activity

prior to the presentation of the phonate cue. The interval required for perceptual processing of the warning and phonate cue will (Elliot,

1968;

conflicting

vary as a function

Murray,

1970;

stutterers

and nonstutterers

visual stimuli

Conversely,

not conclusive

whether

modality

experimenters

stimulus

time

stimulus

1970),

1982)

modality and intensity

may reflect

as a variable with

is determined

variables

that of

are presented at

to the subject and the experimental

of perceptual processes. interval

the failure

levels. In addition, cognitive and affec-

as well

may interact

it is

effect.

manner and, more importantly,

differences

equal intensity

Thus,

the LRT

to ensure that visual and auditory stimuli

(Murray,

stimulus-related

influences

has shown that stimulus

reaction

psychophysically

processing

(1982) failed to find significant

modality

tive factors, such as instructions

duration

but not for

differences for auditory or visual stimuli.

Kohfeld (1971)

Lehtonen,

is

modality on the LRT

have been reported for auditory

parameters interact in a complex

setting

There

group reaction time differences between

Watson and Alfonso

LRT

However,

and intensity

1982).

by McFarlane and Prins (1978) and McFarlane and Shipley

between-group

cross

modality

and Alfonso,

evidence regarding the effect of stimulus

effect. For example, significant

(1981).

of stimulus

Watson

stimulus

Thus,

foreperiod parameters

the duration

(Niemi

and

to alter the

of the perceptual The

effects of

may be reduced by maintaining

constant

modality and intensity

by several variables.

parameters for all subjects. Although

it is

not possible to measure the duration of perceptual processes in humans directly,

Wall

physiological trical

activity

et al. (1953)

provide an estimate of this interval based on

data obtained from anesthetized in pyramidal

tract neurons

animals.

Recording elec-

in the motor

cortex,

they

Ben C. Watson and Peter 1. Alfonso

200

observed a latency of approximately visual stimulus the contribution

of perceptual processing

may be relatively duration

small. To summarize,

of perceptual

stimulus

intensity interval

(i.e., formulation peripheral values

processes

and modality

may be held relatively The

30 msec between the onset of a

and the onset of neural activity.

These data suggest that

activity to overall

LRT

values

it is not possible to measure the

directly.

However,

by controlling

parameters, the duration of this interval

constant across subjects.

required for the completion and transmission

musculature)

in the simple

of neuromotor

processes

of appropriate neural commands to the

may also contribute

minimally

to overall

reaction time paradigm. Estimates

LRT

of formulation

latencies are not available for human subjects. However, the transmission velocity

of

neural

approximately

impulses

along

the recurrent

56 meters/second in nonstutterers

laryngeal

nerve

is

(Flisberg and Lindholm,

1970). This value, in addition to a residual latency of 1.5 to 2.5 msec due to synaptic junctions fibers

(Basmajian,

and the decreasing diameter of peripheral

1970),

latency in nonstutterers

yields

an estimated

of approximately

maximum

3.0 msec. Thus

although the duration of perceptual and neuromotor

nerve

transmission it appears that

processing compo-

nents in the LRT paradigm cannot be directly measured, it is likely that the contribution

of both these processes to group LRT differences is relatively

insignificant. Posturing the speech mechanism for the onset of an isolated, voiced vowel requires articulatory

muscular

systems.

the optimization

adjustments

in the respiratory,

In the respiratory

of thoracic

system, these adjustments

muscle

tension.

Optimal

levels, in turn, facilitate rapid generation of sufficient for phonation

initiation

muscular adjustments phonation. laryngeal

(Baken et al., 1979). adjustments postures

result

appropriate

these systems will occur simultaneously. nature of the posturing activity within the qualitative versus

interaction

in respiratory

voiceless

result in

muscle

tension

subglottal pressure

In the laryngeal

system,

in achievement

of supra-

for the required

response

(e.g., the isolated vowel /a/). We assume that posturing

differences

and

modify vocal-fold tension and position to facilitate

Articulatory vocal-tract

laryngeal,

vowels.

Furthermore,

activity within

it is likely that the

any system is, in part, a function of

between systems.

For example, there may be

and laryngeal coupling for the onset of voiced In addition,

articulatory

postures

may affect

Foreperiodheverity

Effects on Acoustic

LRT

201

laryngeal posturing (i.e., constricted open vocal-tract configurations). In the aerodynamic domain, respiratory respect to lung volume.

For example,

shown that a lung volume

posturing also occurs with

lzdebski

of approximately

and Shipp (1978)

50%

have

vital capacity yields

faster LRT values than do prephonatory lung volumes of 25 and 75% vital capacity. In addition,

Hoshiko

found that nonstutterers

(1965)

usually

initiate phonation from about 50% vital capacity. Thus, this value appears to represent

an optimal

lung volume

for the initiation

of vocal-fold

vibration. It is also true that LRT the initiation initiation

component.

neuromotor

values are affected by processes included in These

include transmission

commands,

movement of speech structures,

muscle

and execution

contraction,

and, finally,

of

coordinated

generation of the resultant

acoustic output. Reaction time measurements of the latter three processes can be obtained and are illustrated Lastly, would

we should

emphasize

delay initiation

vibration

or abnormally

may be prolonged

postured (i.e., simultaneous

in abnormally

in stutterers

the latency of

if the vocal folds

are

1978).

adduction and abduction, cf.

Hyperpostured

vocal folds would

likely

high levels of glottal resistance and, therefore, the

need for higher levels of subglottal

pressure,

whereas abnormally

tured vocal folds would prevent the accumulation pressure

deficits

For example,

that is, postured with excessive tension and adduction,

Freeman and Ushijima, result

that posturing

of the response.

onset for stutterers

“hyperpostured,”

in Figure 4.

to initiate vibration.

Finally,

markedly

of sufficient constricted

pos-

subglottal articulatory

postures increase supraglottal pressures and, thus, may prolong vibration onset latencies. The point we wish to make is that the delayed reaction time values in these instances would reflect postural rather than initiation deficits. We assume that the contribution study

to between-group

modality

and intensity

subjects.

In addition,

process stutterers’

to the

LRT

LRT

values

differences

of perceptual processes was

insignificant

since

parameters were held relatively it is likely effect was

constant for all

that the contribution insignificant.

approach those

The

in this stimulus

of neuromotor

finding

of nonstutterers

that

mild

as foreperiod

increases suggests that the primary difficulty

for this group of stutterers

related to posturing

However,

the speech mechanism.

is

since LRT values

202

Ben C. Watson

and Peter 1. Alfonso

for mild stutterers did not become identical with those of nonstutterers, is also

possible

initiating

that these

vibration

stutterers

as well. Conversely,

the finding that severe stutterers’

LRT values fail to approach those of nonstutterers

suggests that severe stutterers

it

have some degree of difficulty as foreperiod increases

may have difficulty

in both posturing the

speech mechanism and initiating vocal-fold vibration. is that the underlying

deficit may be qualitatively

and severe stutterers.

Unfortunately,

What is important,

different between mild

LRT measures obtained from acous-

tic analysis alone do not permit precise specification of the loci of deficits in phonation onset activity in these stutterers. that mild stutterers

For example, it is possible

have the same type of deficits as do severe stutterers

but to a lesser degree. Thus,

we feel that we have made the most of

acoustic measures of LRT. That is, we need to investigate those activities that occur before the onset of voicing. The advantage of obtaining simultaneous

measures in the acoustic,

movement, and EMG domains is discussed by Baer and Alfonso (in press). They suggest that simultaneous in LRT

experiments

measures may be particularly

because they provide information

prior to onset of the acoustic signal corresponding For example, processes

the combined

may be inferred

duration

from

regarding activity

to vocal-fold vibration.

of perceptual

EMG

signals

informative

and neuromotor

recorded from

intrinsic

laryngeal muscles. That is, the latency between the offset of the warning signal and the onset of the EMG signal in the laryngeal muscles may yield an estimate of the time required to complete perceptual and neuromotor processes. In addition, EMG measures may be useful in documenting the latency

of onset,

synergy,

and amount

of muscular

activity

during

prephonatory posturing of the speech system as well as during generation of subglottal chest-wall

pressure

by the respiratory

and vocal-fold

and transillumation information

movements,

instrumentation,

system.

Direct observation

respectively,

regarding the amount and coordination

laryngeal posturing activity as well as the interaction posturing and respiratory-system

of

via Respitrace (Cohn et al., 1977) may also

provide

of respiratory

and

between laryngeal

activity during the generation of subglot-

tal pressure. In conclusion, our original

the results of the present study support the results of

experiment

effect. Furthermore,

by demonstrating

the present results

a significant

stuttering

severity

support the notion that mild and

ForeperiodlSeverity

Effects on Acoustic

LRT

203

severe stutterers’ prolonged LRT values may reflect differential posturing and/or vibration

initiation.

tic analyses

alone will

contributing

to stutterers’

experiments

not specifically delayed

incorporating

movement,

reveal the nature of deficits

LRT values.

simultaneous

and EMG domains.

in initiating

Only through

A portion annual

phonation

of the data

convention

Los Angeles,

tion,

ported

California,

by NINCDS

gratefully

experimental assistance

awarded

the assistance

subjects.

We

a/so

in the preparation

described.

was first presented

798 7. This

November

at the Associa-

research

to Haskins

of Arlyne

wish

LRT

acoustic,

Speech-Language-Hearing

grant NSl3870

acknowledge

the

stutterers’ often reported

in this paper

American

in

the use of simultaneous

be systematically

reported

of the

We plan future

measures

measures can the nature of deficits underlying difficulty

deficits in

We recognize, however, that acous-

was

sup-

Laboratories.

We

Russo in classifying

the

to thank

Thomas

Baer

for

his

of this manuscript.

REFERENCES Adams,

M.R.,

and Hayden,

P. The ability

initiate and terminate Journal Baer, T.,

of Speech

and Hearing

and Alfonso,

PJ.

acoustic

measures

Science,

Vol. II, Normal

Gallagher, Baken,

On

S.A.,

1976,19,

and Weissman,

J. Muscles Alive.

to auditory stimuli.

290-296.

neuromuscular, In: Current

Speech. K.L.

and Hearing

Baltimore:

Williams

speakers to

of an isolated vowel. movement,

and

issues in Language

(D. Beasley, C. Prutting, T.

Eds.). San Diego: College Hill

of Speech

Birren, I.E., and Botwinick,

stutterers.

Research,

and Disordered

and normal

production

of speech articulation.

phonation.lournal

Borden, G.J. Initiation

of stutterers

during

simultaneous

and R.C. Daniloff,

R.J., Cavallo,

Basmajian,

phonation

Chest wall Research,

Press, in press.

movements

prior to

1979, 22, 862-872.

& Wilkins,

1970.

J. Age differences in finger, jaw, and foot reaction time

lournal

of Gerontology,

1955, 10, 429-432.

versus execution time during manual and oral counting by

Haskins

Laboratories

Status

Report

on Speech

Research,

1982,

SR-70, in press. Botwinick,

J., and Thompson,

time. journal

Cohen, J. and Cohen, P. Applied Behavioral

L.W.

of Experimental Sciences.

Premotor and motor components of reaction Psychology,

Multiple

Hillsdale,

1966, 71, 9- 15.

Regression/Correlation

Analysis

N.J.: Lawrence Erlbaum Associates,

for the 1975.

and Peter 1. Alfonso

Ben C. Watson

204

Cohn, M.A., Watson, for

International Cross,

H., Weisshaut,

Non-invasive

Symposium

D.E., and Luper,

children

R., Stott, F., and Sackner, M.A. A Transducer

Monitoring

on Ambulatory

H.L.

and adults. journal

Fluency Cullinan,

W.,

of Fluency

R.

Simple

23, 344visual

Psychonomical Flisberg,

K.,

of adult

1977.

59-77.

Effects of stimulus

stutterers

U.K.,

and nonstuttering

1979,4,

Disorders,

at the 2nd

Middlesex,

ear presentation on

and nonstutterers.

journal

of

45 -58. M.T.

and nonstuttering

1980,

search,

time 1979,4,

and ‘Springer,

stuttering Elliot,

reaction

Disorders,

Monitoring.

Voice reaction time of stuttering

Cross, D.E., Shadden, B.B., and Luper, H.L. the voice

Paper presented

of Respiration.

Voice

initiation

children.

journal

and termination of Speech and

times

in

Hearing

Re-

360.

and simple 1968,

Science,

and Lindholm,

T.

auditory

reaction

time:

A comparison.

10, 335-336.

Electrical

stimulation

of the human

laryngeal nerve during thyroid operation. Acta Otolaryngologica,

recurrent 1970, 263,

63-67. Freeman,

F.J.,

journal Hayden,

and Ushijima,

of Speech

T.

Laryngeal

and Hearing

Research,

P. The effects of masking

Hoshiko,

M.S.

Lung

Izdebski,

for

doctoral

stuttering.

and nonstutterers

dissertation,

and Hearing Simple

of phonation.

journal

interval on phonation initiation

and Hearing

Research,

Purdue

of Applied

1978,

21, 638-651.

reaction time as a function of stimulus

J., Ingham, R.].,

of Experimental

and Gervens,

intensity

Psychology,

and nonstutterers.

of the American

in decibels

1971,88,251-257.

A. The effect of stimulus

practice on voice reaction in stutterers the annual convention

reaction times.

1980, 23, 485-489.

reaction times for phonatory initiation. /ournal

Research,

of light and sound.lournal Lewis,

Unpublished initiation

K., and Shipp, T. Minimal D.L.

during

20, 480-482.

of Speech

of Speech Kohfeld,

volume

1965,

K. Effects of prestimulus

Journal

activity

21, 538-562.

1975.

Physiology, Izdebski,

1978,

and pacing on stutterers’

voice and speech onset time. University,

muscle

conditions

and

Paper presented at

Speech-Language-Hearing

Associa-

tion, Atlanta, Georgia, 1979. McFarlane,

S.C., and Prins,

D. Neural response time of stutterers and nonstutterers

in selected oral motor tasks. journal

of Speech

and Hearing

1978,

Research,

21,768-778. McFarlane,

S.C., and Shipley,

nonstutterers Speech Murphy,

under

and Hearing

M.,

K.G. Latency of vocalization

conditions Disorders,

and Baumgartner,

stuttering and nonstuttering

of auditory 1981,

cueing. journal

of

24, 307-312.

J.M. Voice children.

onset for stutterers and

and visual

initiation

lournal

and termination

of Fluency

Disorders,

time

in

1981,

6,

257-264. Murray,

H.G.

Stimulus

theory model. /ournal

intensity

and reaction time:

of Experimental

Psychology,

Evaluation

of a decision-

1970, 84, 383-391.

ForeperiodlSeverity

Niemi,

Effects on Acoustic

P., and Naatanen, R. Foreperiod 1981,

Bulletin, Niemi,

P., and Lehtonen, P.W.,

Reiss,

Montlick,

E. Foreperiod

L.J.,

1975,

Cooper,

SR-44,

A., Till,

Haskins

1981,24,

95-

Therapy

Nonparametric

McGraw-Hill, Starkweather,

Status

Mermelstein,

Report

P.,

and

and manipulation on Speech

of

Research,

and manual reaction times of

of Speech and Hearing

H.S.

nonstutterers.

instrument

for children

1972,37,

Research,

and adults.

lournal

of

314-322.

for Stuttering

in Children

and Adults.

Springfield,

1974. statistics

Hirschman,

Stutterers 1976,

Research,

Venkatagiri,

A reap-

for

the

Behavioral

Sciences.

York:

New

1956.

C.W.,

onset:

Venkatagiri,

R.M.,

H. Laryngeal

Research,

III.: Charles C Thomas,

tion

McGuire,

adults. /our&

severity

and Hearing

S.

intensity:

192-196.

Ryan, B. Programmed Siegel,

stimulus

107.

J., and Goldsmith,

G. A stuttering

Speech

F.S.

Laboratories

stuttering and nonstuttering Riley,

and visual

50, 73-82,

1982,

T. A digital pattern playback for the analysis

speech signals. Reich,

and simple reaction time. Psychological

89, 133-162.

praisal. Acta Psycho/o&a, Nye,

205

LRT

vs.

P., and Tannenbaum, nonstutterers.

journal

R.S. Latency of vocalizaof Speech

and

Hearing

19, 481-492.

Reaction time for voiced and whispered journal

of fluency

Disorders,

/a/ in stutterers

H.S. Reaction time for/s/ and lzl in stutterers and nonstutterers:

of discoordination

hypothesis.

/ournal

and

1981, 6, 265-271.

of Communication

Disorders,

A test 1982,

15, 55-62. Wall,

P.D.,

Remond,

action of visual raphy Watson,

and Clinical

A.G.,

and Dobson,

Neurophysiology,

B.C., and Alfonso, A.D.

journal

1953,

P.J. A comparison

stutterers and nonstutterers.journa/ Weiss,

R.L.

Studies

on the mechanism

afferents on motor cortex excitability. 5, 385-393. of LRT

of Fluency

and VOT

Disorders,

values between

1982, 7,219-241.

The locus of reaction time change with set, motivation, of Gerontology,

1965,

20, 60-64.

of the

Electroencephalog-

and age.