DAF_DualTask_Thesis - Ideals - University of Illinois at Urbana

advertisement
DAF and Dual-Task Effects
The Effects of Delayed Auditory Feedback and Dual-Task on Speech Rate in Normally Fluent
Speakers
Rebecca Breanne Tibbs
Undergraduate Thesis
Dr. Torrey Loucks
University of Illinois at Urbana-Champaign
May 2011
1
DAF and Dual-Task Effects
2
Table of Contents
Title Page………………………………………………………………………………………
1
Table of Contents ……………………………………………………………………………....
2
Acknowledgements ……………………………………………………………………………
4
Abstract……………………………………………………………………………………………5
Introduction …………………………………………………………………………………….
6
Speech Rate……………………………………………………………………………
6
Dual-Task………………………………………………………………………………
7
Delayed Auditory Feedback…………………………………………………………… 10
Hypothesis……………………………………………………………………………………… 14
Performance Correlations………………………………………………………………. 15
Methods ………………………………………………………………………………………… 15
Participants……………………………………………………………………………….15
Tasks……………………………………………………………………………………. 16
Analysis…………………………………………………………………………………. 17
Disfluencies and Speech Errors………………………………………………………….17
Speech Rate …………………………………………………………………………….. 18
DAF and Dual-Task Effects
3
Results………………………………………………………………………………………….. 20
Descriptive Data………………………………………………………………………… 20
Analysis of Variance……………………………………………………………………. 20
Correlation Analysis……………………………………………………………………. 21
Discussion ……………………………………………………………………………………….23
Comparison of DAF and NAF …………………………………………………………..24
Support for Dual-Task Theories ………………………………………………………. .28
Application of Findings …………………………………………………………………29
Conclusion……………………………………………………………………………….30
References……………………………………………………………………………………… 31
DAF and Dual-Task Effects
ACKNOWLEDGEMENTS
This thesis would not have been possible without the guidance, experience, and patience of my
lab supervisor and professor, Dr. Torrey Loucks. Additionally, I would like to extend a special
thank you to HeeCheong Chon whose teaching and encouragement gave me the confidence to
complete this work.
I am grateful to the Speech and Hearing Science Department and the University of Illinois at
Urbana-Champaign for the opportunity to gain knowledge from the best researchers in the field.
Rebecca Breanne Tibbs
4
DAF and Dual-Task Effects
5
ABSTRACT
Speech rate is an important element in studying the planning and production of speech; it is often
used in the examination of stutterers and patients with Parkinson’s disease. Several factors can
have an effect on speech rate; people speaking under delayed auditory feedback (DAF) are
known to have a decreased rate of speech, likewise, simultaneously performing two tasks may
also elicit a delay in speech planning and production times. The goal of this study is to compare
the effects of a linguistic task and DAF in people who are normally fluent and their interactions
on speech rate. Previous studies have not made clear whether a dual-task influences speech rate
to the same extent as DAF. It is also unknown how a dual-task presented concurrently with DAF
will affect speech rate. Sixty-three normally fluent speakers performed reading tasks under four
conditions: NAF, DAF, dual-task NAF, and dual-task DAF. Speech rate analysis was performed
for each task and correlations were configured. It was determined that speech rate became
progressively slower in each task as follows – NAF >DT NAF > DAF > DT DAF –which
indicates that DAF affects speech rate to a greater extent that a dual-task alone. Sex correlations
revealed that females showed a significantly slower speech rate under DAF than males.
Otherwise, males and females performed similarly on each of the other tasks. The results will be
discusses further in the context of previous work on DAF, speech rate, and dual-task research.
DAF and Dual-Task Effects
6
Introduction
Speech Rate Analysis
Speech rate serves as a vital element in the study of speech production and speech
planning. There is not a universal technique for calculating speech rate, it can be measured in
words, syllables, or phones over a time period of a minute or second (Sawyer, Chon, Ambrose,
2008), although linguistic unit/second is the most commonly used rate format. Certain studies
have even used more than one measure of speech rate (Kelly, 1994; Kelly & Contour, 1992).
Some studies use the articulation rate of an entire utterance even if disfluencies are present
(disfluencies included), while in other studies only the fluent speech rate is calculated after
removing disfluencies, (Sawyer, Chon, & Ambrose, 2008). Rate of speech production has been
determined with stopwatches, video time codes, audio signals, or acoustic measurements
(Sawyer, Chon, & Ambrose, 2008).
Speech rate has been used to investigate disorders such as Parkinson’s disease and
stuttering. Rate measures have been included in diagnostic estimation of stuttering severity
(Logan & Contour, 1995). The unit that is measured has made a difference in the findings and
interpretations of studies; for example, measuring an entire disfluent utterance in phones per
minute will most likely yield different results than an utterance where the disfluencies have been
removed and measured in words or syllables per second (Sawyer, Chon, & Ambrose, 2008).
Hall, Amir, & Yairi’s (1999) longitudinal study of children with persistent stuttering reveals this
issue; the study found no significant differences between three groups of children over time
when examining perceptually fluent utterances measured in syllables per second, yet
measurements in phones per second revealed significantly lower speech rates in stuttering
children.
DAF and Dual-Task Effects
7
Speech rate has also been used to study a more internalized process, including the rate of
speech planning. Longer utterances involve longer planning periods in adults as shown by a
slower speech rate (Amster & Starkweather, 1987; Peters & Hulstijn, 1987). This longer
planning time could also negatively affect the fluency of an utterance (Sawyer, Chon, &
Ambrose, 2008).
The goal of this study is to compare the effects of a linguistic task and delayed auditory
feedback (DAF) on speech rate in people who are normally fluent and their interactions on
speech rate.
Dual-Task
The dual-task paradigm is best explained as any two tasks performed simultaneously in
which the attention of the performer could become divided across both tasks. Divided attention
may result in poorer efficiency on one or both tasks, or minimal changes. On many cases, there
are performance decrements that relate to numerous factors both associated with the task and
with the participant. A number of theorists have attempted to explain this paradigm and its
effects in a language context.
Kinsbourne and Hicks (1978) provided a neurophysiological explanation of the dual-task
paradigm which they refer to as the Functional Distance Hypothesis. The amount of interference
of dually performing tasks is inversely related to the proximity of brain regions activated in each
task. As a result, two tasks activating brain regions with great distance between one another will
result in relatively low interference. On the other hand, two tasks in brain regions of close
proximity will increase the neural interference and, therefore, decrease the performance
efficiency of one or both tasks (Dromey, 2008). Dromey tested this proposal by examining dualtask effects on speech, fluency, and manual motor tasks; his subjects performed activities with
DAF and Dual-Task Effects
8
the left and right hands in an attempt to maximize the interference that right hand activities may
have on the speech and language’s left hemisphere. The differences in task performance among
each task showed slight differences, however were not drastic or enough to fully support the
theory. The results of his study showed that Kinsbourne and Hick’s (1978) theory may have
underestimated the system’s complexity (Dromey, 2008). Though the functional distance
hypothesis provides a testable account of dual task interference, it does not account for the
complexity of dual-task influences.
Capacity theories offer a different explanation of dual-task effects. Leclercq (2002) states
that there are several processors in the brain which are associated with different types of stimuli;
if two tasks depend on the same processor then an interference will occur and lead to a decrease
in performance of one or both tasks. Likewise, there will be no decline in performance if the
tasks rely on different processors (Allport et al., 1972). This theory suggests that in a dual-task
“attentional resources” are shared in a “graded fashion” among both tasks (Dromey, 2008); thus,
at least one task must become impaired if there are insufficient resources.
A third account of dual-tasks was presented by Wickens (1984) and is known as the
Time-Sharing Model. This model “suggests a series of rapid and smooth transitions between two
tasks,” (Dromey, 2008). Wickens (1984) proposed that if, by preference, we focus more on one
of the tasks, we may be “attending” to that task “for a greater length of time before switching to
the other” task (Dromey, 2008). The task that receives less attention will suffer. However, this
theory along with the Capacity Theory fails to explain what the resource sharing or time-sharing
mechanisms are. Furthermore, observational experiments have shown limitations in the two
explanations (Leclercq, 2002). From a behavioral perspective, the Functional Distance
DAF and Dual-Task Effects
9
Hypothesis is useful because it predicts that more similar tasks will be affected by a dual-task
paradigm more than dis-similar tasks.
Similarly, the Central Bottleneck Model explains the information processing of dual-tasks
as an all-or-none model (Tombu & Jolicaeur, 2002). The effect occurs as the initial task gains
access to the bottleneck first and information processing must complete before task two may
begin. The initial delay of task two accounts for the declinations in performance that may occur
as the secondary task is completed.
Dual-tasks have been used to study speech production in fluent speakers and persons who
stutter. Studies of dual-tasks with fluent speakers have concluded that syntactic, phonological,
and articulatory systems remain unaffected by a secondary task (Power, 1985; Rummer, 1996).
This suggests that in normally fluent people it is the nature of the secondary language tasks that
determines if interference will occur, rather than their language ability. On the other hand,
persons who stutter appear to be uniquely “sensitive to interference from concurrent attentiondemanding cognitive processing particularly when phonological coding is involved,” (Bosshardt,
2004).
Normally fluent people, or people with presumably “healthy” speech systems, have an
increased reaction sense that seems to be able to reduce the amount of interference from a dualtask (Raichle et al., 1994; Jueptner et al., 1997; Beilock et al., 2002). A recent study directly
tested whether people who stutter are uniquely sensitive to dual task effects. Normally fluent
people were first found to have a faster rate of speech on a practiced syllable reading task
compared to people who stutter (Smits-Bandstra & De Nil, 2008). When the researchers
introduced a dual right-handed motor task to the same subjects, the normally fluent people
adjusted and compensated for the interference at a much faster rate with practice than people
DAF and Dual-Task Effects
10
who stutter (Smits-Bandstra & De Nil, 2008). Most importantly, these fluent speech systems of
the control subjects were initially interrupted by the introduction of a secondary task that slowed
their rate of speech. The dual-task slowed speech rate to a greater extent in the stuttering
subjects.
Dromey & Shim (2008) examined the speech, fluency, and motor tasks of twenty
normally speaking adults in a dual-task activity. They found that there was an overall decrease
in the articulatory parameters of lip displacement and peak velocity, with an increase in the
sound pressure level. Essentially, the dual-task resulted in slower and hypo-articulated speech
rate. It is likely that the decreased velocity of speech movements lent to a decreased speech rate
in this dual-task study of normally fluent people.
In light of these previous findings, a decrease in speech rate can be expected in normally
fluent people when performing a concurrent linguistic task. It is not clear however whether a
dual-task affects speech rate to the same extent as delayed auditory feedback (DAF). It is well
known that DAF slows speech rate in most speakers, especially when the delay is over 50 ms. It
is also unknown how a dual-task presented concurrently with DAF will affect speech rate.
Delayed Auditory Feedback
There are relatively few studies of dual-task effects on speech rate compared to numerous
reports showing that delayed auditory feedback (DAF) slows speech rate. DAF is a simulated
delay in the auditory feedback of speech created by holding the microphone signal within a
buffer for a specified period. DAF effects were first reported by B.S. Lee (1950) who classified
the phenomena as “startling” because many of his subjects were described as having a
“quavering slow speech of the type associated with cerebral palsy” and exhibited reddening of
the face indicating tension (Lee, 1950). Following Lee’s study, other research revealed that DAF
DAF and Dual-Task Effects
11
appeared to have opposing effect on persons who stutter by actually improving fluency (Nessel,
1958; Lotzmann, 1961; Chase et al., 1961; Bohr, 1963; Zerneri, 1966; Soderberg, 1969). Now,
certain devices are even sold commercially to assist in the treatment of stuttering (Borsel, 2007).
It is important to consider that DAF does not benefit all people who stutter and those who do
show improved fluency vary greatly in the degree benefit (Bloodstein, 1995; Ward, 2006).
One hypothesis suggests that stutterers become more fluent under DAF because of increased
focus on phonation which is indicated by a simultaneous increase in speech duration—or
decreased speech rate (Wingate, 1976). Wingate (1976) calls this a “slowing down” process that
alleviates many of the processing demands on the brain that may cause stuttering. Howell and
Archer (1984) explained the effects of DAF on speech rate and speech error by suggesting that
speakers use auditory feedback control mechanisms more effectively after the speech is no
longer under a “normal temporal relationship.” Speech disfluencies and slow durations are
accounted for by the fact that speakers are using a delayed feedback to monitor their articulatory
output (Black, 1951).
Speech is typically delayed 25, 50, or 200 ms depending on the desired intervention
(Stuart et al., 2002). It is puzzling that not everyone is susceptible to DAF to the same degree
(Howell & Archer, 1984). Some factors have been identified that appear to enhance a person’s
susceptibility; these include personality differences (Korrowbrow, 1955; Rankin & Balfrey,
1966), fewer speech disfluencies (Butler & Galloway, 1957), or whether a person relies on
auditory or oral sensory feedback mechanisms (Yates, 1963). Other factors that influence the
fluency of a speaker are gender (Bachrach, 1964), age (Siegel, Fehst, Garber, & Pick, 1980), and
if the speaker is speaking his/her first language (McKay, 1970). The fluency of a person
speaking under natural feedback conditions is inversely related to the fluency of a person under
DAF and Dual-Task Effects
12
DAF conditions; this suggests that normally fluent speakers may use their auditory feedback
mechanisms less than people who stutter (Howell & Archer, 1984).
It is quite clear that DAF affects the motor system in speech output because at longer
delays it induces part-word disfluencies, part and whole-word repetitions, and an increase in
speech errors—all of which are not linguistic errors (Chong et al., ASHA presentation, Chicago,
2008). Stuart et al. (2002) noted that several studies have shown that DAF also provokes
changes in speech rate and reading time durations, lengthened voicing, a louder speaking rate,
and changes in the aerodynamics of speech (Black, 1951; Fukawa et al., 1988; Howell, 1990;
Longova et al., 1970; Lee, 1950, 1951; McKay, 1968; Siegel et al., 1982; Stager et al., 1997;
Stager and Ludlow, 1993). In normal speakers and at long delays (>100 ms), the effects of DAF
have been likened to simulated stuttering—it forces normally fluent people to “artificially
stutter” (Stuart et al., 2002).
It is interesting that at shorter delay times (<50 ms) the speech production system tends to
compensate for the delay and apparently show minimal evidence of a speech rate effect. Stuart
et al. (2002) found that speech rate was reduced when the feedback delays were above 25 ms;
from there the reduction in rate is directly proportional to the delay time. It is at delays of 200
ms or greater that most speakers (fluent or stutterers) show disfluencies and profoundly slower
speech (Stuart et al., 2002). Since syllables are typically 200 ms long, DAF effects suggest that
the syllable is the unit that controls production of speech (Howell & Archer, 1984). The DAF
phenomenon also suggests that the peripheral feedback mechanisms (in this case, auditory) can
affect motor control of speech (Stuart et al., 2002). But since there are clear differences in the
way normally fluent people react to DAF compared to stutterers, Stuart et al. (2002) deemed
DAF and Dual-Task Effects
13
DAF induced disfluencies as a “poor analog of stuttering,” although, we consider that DAF
induced changes in speech production have implications for stuttering.
As an intervention technique, DAF may “restore the sensitivity in the left hemisphere of
those who stutter” (Stuart et al., 2002). DAF may instead impose an increased cognitive load on
speech as subjects attempt to compensate for DAF induced interruptions in speech output. This
cognitive load interpretation differs from a speech motor interpretation or auditory feedback
interpretation. Interferences directly related to DAF may arise as a sort of distraction mechanism
or, because DAF is dividing attention between the cognitive intent of a message and attempts at
speech production DAF disrupts the normally seamless translation of linguistic message by
forcing the subject to pay attention both to the message and control of the oral and laryngeal
articulators. But this has not been proven. DAF effects may be equally well-explained as
interfering with the auditory-motor pathways.
Certain Evidence discussed previously suggests both dual-task conditions and DAF can
slow speech rate (Wingate, 1976). If the affect of DAF on speech rate in any given speaker is
correlated with speech rate effects caused by appropriate dual-task scenarios, then DAF effects
on rate could be interpreted as having a cognitive basis. If the effects of DAF on speech rate are
not correlated with dual-task effects, then there may be separate mechanisms underlying DAF
effects; which we consider to arise from interference in auditory-to-motor processing. It may also
turn out that a significant correlation between dual-task induced and DAF induced changes in
speech rate does not explain all the variance. Then when DAF and a dual-task are combined,
cognitive and auditory-to-motor processes may combine in some additive manner to slow speech
to a greater degree than when presented separately.
DAF and Dual-Task Effects
14
Hypothesis
Main Effects of Conditions on Speech Rate
Articulation Rate Analysis of Single vs. Dual Task
Single Task
Dual-Task
NAF Condition
Fastest Artic
Rate
2nd Fastest Artic
Rate
DAF Condition
3rd Fastest Artic
Rate
Slowest Artic
Rate
Single Task/NAF: Articulation rate will be fastest under natural auditory feedback and when an
individual is engaged in a single speaking task.
Dual Task/NAF: Articulation rate will be slower than the Single Task/NAF condition when the
individual’s attention is divided by a dual-task condition.
Single Task/DAF: Articulation rate will be slower under DAF than both of the NAF conditions.
Dual Task/DAF: Articulation rate will the slowest when under DAF and the individual’s
attention is divided by a dual-task condition.
Articulation Rate: Single Task/NAF > Dual Task/NAF > Single Task/DAF > Dual Task/DAF
DAF and Dual-Task Effects
15
Performance Correlations
If DAF effects on speech rate arise due to similar mechanisms as dual-task effects, then the
speech rate of an individual under DAF condition will be significantly and positively correlated
with performance in the dual-task condition.
a) Speech rate under the single-task/DAF will be significantly and positively correlated with
speech rate in the dual-task/NAF condition
b) Individuals who have a slower rate of speech under both the single task/DAF and dualtask/NAF conditions will show slower rates of speech under dual-task/DAF and vice
versa.
Methods
Participants
The participants consisted of 63 normally fluent speakers (30 males and 33 females). The
males ranged in age from 18; 6 to 29; 7 years; months (mean = 20; 10). The females ranged from
18; 6 to 22; 9 (mean = 20; 2). The criteria for selection of participants included: (a) monolingual
native English speaker, (b) between 18 to 30 years (to avoid decreased hearing sensitivity after
30 years as noted by Pearson et al., 1995), (c) no hearing, speech-language, psychiatric or
neurological disorders, (d) normal hearing sensitivity at 20 dB HL for 500, 1000, 2000, and 4000
Hz tested with a GSI 17 audiometer (Grason-Stadler Inc), and (e) right handed according to the
Edinburgh Handedness Inventory (Oldfield, 1971). The handedness inventory scores were a
mean of 92.41 (SD = 9.00) for males and 91.72 (SD = 12.58) for females.
Procedures
The participants were seated in a sound-treated booth and fitted with headphones (Sennheiser
HD280 professional with 8Hz to 25,000Hz frequency response) and a microphone (Shure
WL185 Cardioid Lavalier Microphone) that was connected to a preamplifier (Shure MX1BP).
The microphone was placed 10 centimeters away from the individual’s mouth. Audio and video
DAF and Dual-Task Effects
16
digital recordings were obtained for each participant during the reading tasks (SONY digital
camera (model DCR-VX2000) burned onto a DVD with Panasonic DVD Video Recorder (model
DMR-T2020), and audio recordings obtained with a HHB CDR830 BurnIT CD Recorder).
In the DAF condition, auditory feedback was delivered through the headphones with the
DAF/FAF Assistant software program (version 1.1, http://www.artefactsoft.com/) with a latency
of 250 milliseconds (Fairbanks, 1955; Goldiamond, 1965; Van Borsel, Sunaert, & Engelen,
2005). (The 250 ms latency was confirmed using an oscilloscope that showed the 220
millisecond delay time in the program involved an extra 30 millisecond-processing latency). In
conjunction with several previous studies, the subject’s own voice was also amplified before
delivering it back through the headphones (e.g., MacKay, 1970; 95dB SPL; Siegel et al., 1980:
95dB SPL; Van Borsel, Sunaert, & Engelen, 2005: 96dB SPL). The intensity level was calibrated
with a Brüel & Kjær sound level meter (type 2235). The participants were asked to confirm that
the intensity of their vocal feedback (equivalent to 95dB SPL at 1000Hz) was “loud but O.K.”
According to Cox, Alexander, Taylor, & Gray (1997), this loudness is ranked as 6 on a 7-point
scale of Loudness Categories, which defines level 1 as “very soft” and level 7 as “uncomfortably
loud.” Cox et al. (1997) previously reported that a level 6 ranges from 70—100dB HL for the
speech of normal hearing people. In our study, loudness was only reported as uncomfortable by
two participants, in which cases the intensity was reduced until the level 6 contour was reached.
Sound levels then remained constant throughout the data collection session. The intensity levels
were implemented in the aNAF conditions.
Tasks
For the single-tasks, each participant read two passages—one under DAF and the other
aNAF conditions. The passages contained 123 and 125 words and were balanced for phonemes
DAF and Dual-Task Effects
17
and set at an appropriate reading level for teen ages and adults (SSI-3, Riley, 1994). For dualtasks, each participant was asked to read two passages—one under DAF and one NAF
conditions—which contained 129 and 134 words, balanced for phonemes and slightly below an
appropriate reading level for teens and adults (SSI-3, Riley, 1994). While reading each passage,
participants were instructed to push a button every time they encountered the /s/ phoneme. The
button pushes were recorded to verify the task performance using a Windaq data acquisition
system (Dataq Instruments). Each time the button is pushed the system sends a TTL pulse to the
instrumentation, which then records acoustic input on a separate channel. Channel 1 recorded
speech acoustics and Channel 2 recorded button presses.
Analysis
The collected speech samples recorded under DAF and aNAF conditions were
transcribed using the Systematic Analysis of Language Transcripts (SALT) Program (Miller &
Chapman, 1996). The entire speech sample was used for each reading task.
Disfluencies and speech errors
Each transcript was analyzed for disfluencies and speech errors. Following Ambrose and
Yairi (1999), nine types of disfluencies were divided into two categories: stuttering-like
disfluencies (SLD) and other disfluencies (OD). The SLD class refers to repetitions (singlesyllable word and part-word repetitions) and disrhythmic phonation (including blocks and
prolongations). Part-word echo (PWE) and wavering voice (WV) were also included in the
repetition and disrhythmic phonation categories; however, these are not typically present in
natural speech conditions, as reported by Lee (1950a, 1950b, 1951) and Fairbanks and Guttman
(1958) in early DAF studies. The OD class refers to interjections, revision/abandoned utterances
DAF and Dual-Task Effects
18
and multisyllable/phrase repetitions (see Ambrose & Yairi, 1999, p. 899). Each transcript was
coded with the identity of each disfluency and the number of SLD and OD per 100 syllables
were calculated for all the reading tasks. Speech articulatory errors (SE), omissions,
substitutions, and additions were identified and the number per 100 syllables was calculated.
Speech rate
The articulation rate was calculated as the number of perceptually fluent content syllables
divided by the duration of fluent speech in seconds for each utterance based on the acoustic
record. For the calculations, the durations of (1) disfluencies and (2) pauses after disfluencies or
pauses greater than 250 ms were excluded from the sample (Andrews et al., 1982; Hall, Amir, &
Yairi, 1999; Miller, Grosjean, & Lomanto, 1984). In order to control for the influence of the
number and duration of pauses on articulation rate, pauses after any SLD or OD were also
excluded (Adams & Ramig, 1980; Chon, Sawyer, & Ambrose, 2007; Zellner, 1994).
Utterances were captured for articulation rate analysis using the GoldWave software
program (version 5.23, Craig, 2008) and converted to wave file format at a sampling rate of 44
kHz. The Praat software program (version 5.0.13, Boersma & Weenink, 2008) was then used to
measure: 1) the duration of each utterance, 2) the duration of each pause between utterances to
determine if it exceeded 250 ms, and, 3) the duration of each disfluency along with the following
pause. Each of these durations was subtracted from the total duration of the utterance. Speech
errors were not excluded from the rate analysis if they did not disrupt the fluent flow of speech
because they carry linguistic meaning, with the exception of omissions. Using the vertical cursor
of the Praat software, durations and onset/offset points were verified by comparing the time
wave form and corresponding spectrogram and confirmed with the audio signal.
DAF and Dual-Task Effects
19
The onset point was determined by the initial acoustic energy shown on the time
waveform and spectrogram and referencing it perceptually with the audio signal. The offset
point was determined by the final acoustic energy shown on the time waveform and spectrogram
and also perceptually confirmed with the audio signal. Pauses within utterances were measured
from offset of acoustic energy to the following onset of acoustic energy, and the duration was
recorded.
In accordance with Throneburg and Yairi (2001) and Stuart and colleagues (2008), the
duration of SLD’s, PWE’s, WV, and OD’s were measured as follows:

Interjections and revisions/ abandoned utterances: onset of sound to offset of disfluency.

Multisyllable/phrase, single-syllable word and part-word repetitions and PWE: entire
disfluency, including repetitions and pauses between phonations.

Prolongations and WV: onset to offset point of disfluency.

Blocks: if identifiable from the video recording and the audio signal, onset of the
tense/abnormal hesitation to the offset of the disfluent word. In unidentifiable onset
points, the onset of the word to the onset of the following phonation—to include the
following pause—to insure accurate measurement.
Once the durations were calculated and the number of fluent syllables was determined,
articulation rate was calculated for each utterance. The average was then found for all utterances
within the reading sample for each subject.
DAF and Dual-Task Effects
20
Results
Descriptive Data
The means and standard deviations of each sex group (male and female) are shown in
Figure 1 for each experimental condition. It is clear that that speech rate was slowest in the
combined DT-DAF condition. Both the DAF and DT-NAF conditions were slower than the
baseline NAF condition with the DAF condition being descriptively slower than the DT-NAF
condition.
Analysis of Variance
The 2*4 factorial ANOVA comparing feedback by sex indicated a significant feedback
effect (F(3,174)=187.6, p<.001), but a main effect for gender was not detected. However, a
feedback by sex interaction (F(3,174)=2.9, p<.05) was observed. Post-hoc comparisons with
Bonferroni correction indicated that speech rate in each feedback condition was significantly
different from NAF, which indicates speech rate was significantly affected by each manipulation.
The changes in speech rate showed the following pattern - NAF > DT NAF > DAF > DT DAF –
that is apparent in Figure 1. Statistically, speech rate was slowest in the combined condition and
DAF affected speech rate more than the dual task under NAF. Inspection of the feedback by sex
interaction indicated the female group showed a significantly slower speech rate under DAF
compared to males (Figure 1). Otherwise, both groups had similar speech rates in each condition.
Correlation Analysis
Correlations between each of the 4 conditions were determined with the pearson r
method. The pattern of correlations for the full group is shown in Table 1 with asterisks
indicating significance (p<.05, Bonferroni correction). Correlations were also determined on a
gender basis to identify whether different associations may be present for the male and female
DAF and Dual-Task Effects
21
speakers (Tables 2 & 3; p<.05, Bonferroni correction). Figure 2 shows a scatter plot of the
significant association between NAF and DAF for the male and female speakers. The slope of
the association is indicated by a linear regression trend line calculated using the minimal least
squares approach (Males – 0.766; Females – 0.265). A comparison of these specific correlation
coefficients indicated the association between NAF and DAF is not equal for males and females
(Z= 2.67, p=0.008), which supports the interpretation that the correlation is higher in adult males.
A similar approach was taken to test the equality of correlations that were significant for one sex
but not the other sex.
Figure 1
Speech Rate in Males and Females
7.00
6.00
Syllables / s
5.00
4.00
male
female
3.00
2.00
1.00
0.00
NAF
DAF
DT NAF
Feedback Condition
DT DAF
DAF and Dual-Task Effects
Figure 2
NAF vs DAF
5.5000
5.0000
4.5000
DAF
male
4.0000
female
3.5000
Linear (male)
3.0000
2.5000
2.0000
3.0000
4.0000
5.0000
6.0000
7.0000
NAF
Table 1 Males and Females
Conditions
NAF
DAF
DT NAF
NAF
1.000
DAF
.537*
1.000
DT NAF
.481*
.355*
1.000
DT DAF
.450*
.806*
.500*
DT DAF
1.000
22
DAF and Dual-Task Effects
23
Table 2 Males
Conditions
NAF
DAF
DT NAF
NAF
1.000
DAF
.265
1.000
DT NAF
.527*
.289
1.000
DT DAF
.289
.753*
.574*
DT DAF
1.000
Table 3 Females
Conditions
NAF
DAF
DT NAF
NAF
1.000
DAF
.766*
1.000
DT NAF
.456
.471
1.000
DT DAF
.605*
.856*
.463
DT DAF
1.000
Discussion
The results of this study provide evidence that DAF has a greater effect on speech rate
than dual-task which is one indication that DAF effects on rate are not likely simply a dual-task
effect. DAF acts on language processing mechanisms in the brain for the motor planning and
production of speech. Since subjects showed a slower speech rate under DAF than under DTNAF, we expect that the speaker is not simply being distracted as if DAF is a secondary task. In
natural speaking conditions, auditory feedback of speech production plays a relatively subtle role
in the central speech processing of typical speakers because of its slow return-time and because
speech is a highly practiced skill.
DAF and Dual-Task Effects
24
The correlation analysis provides further evidence for difference between mechanisms of
dual-task and DAF. When examining both males and females together, the correlation of
performance on a dual-task compared to performance under DAF is modest at r=0.36 (p < 0.05).
Thus, susceptibility to decreased speech rate under DAF is only partially related to speech rate
under a dual-task condition. Interestingly, males have a higher correlation of r=0.47 (p < 0.05),
which indicates performance under dual-task conditions is associated with performance under
DAF. The female correlations between DAF and dual-task alone are not statistically
significant—speech rate in these two conditions is weakly associated for females. These modest
correlations indicate DAF and dual-task are only weakly associated so different factors need to
be considered in accounting for how these manipulations influence speech rate.
Other evidence suggests the motor production of speech is being affected by DAF to a
greater extent than the planning of speech production. This is made clear by the types of
disfluencies induced by DAF. These disfluencies include part-word disfluencies, part and wholeword repetitions, and an increase in speech errors—all of which differ from linguistic
disfluencies such revisions and long pauses (Chong et al., ASHA presentation, Chicago, 2008).
Because DAF elicits disfluencies related to motor production and the speech rate effects are
weakly correlated with effects of dual-task on speech rate, we consider different processes lead
to slowing of speech rate by these experimental manipulations
Comparison of DAF and NAF
The effects of DAF and NAF on speech rate of the speakers are differed with regards to
several factors. When participants performed the reading task under NAF with no concurrent
task, speech rate was consistently around the 5-6 syllables per second range which serves as a
baseline for comparisons with the other conditions. There appeared to be no significant
DAF and Dual-Task Effects
25
difference in rate between sexes; thus, males and females performed similarly under natural
reading conditions. Because there was not an initial difference between sexes, it can be assumed
that any differences in other conditions are the result of experimental manipulations.
Although the purpose of this report is not to examine the effects of dual-task on
disfluencies, the results are considered important for providing a context in which to understand
changes in speech rate. The number of stuttering-like disfluencies (SLDs) under natural reading
conditions remained near zero, and equal among males and females. Although there was great
variation among subjects, speech errors (SEs) appeared more frequently in reading tasks, which
is typically found. Other disfluencies (ODs), which are classified as interjections, whole-word or
phrase repetitions and phrase revisions appeared most frequently, with a higher percentage
among the male subjects. It is important to consider that both stuttering-like and other
disfluencies occur naturally in the speech of normal speakers so when examining the effects of
other stimuli baseline performance in more typical or unperturbed speaking conditions must be
documented.
Disfluency data for the DAF task showed significant increase relative to the NAF and
dual-task NAF conditions. SLDs increased from near zero to an average of 7.7 SLDs per 100
syllables, SEs increased from .34 to 1.37 errors per 100 syllables; ODs, on the other hand,
decreased from 0.91 to 0.57 per 100 syllables. Disfluency data for the combined DAF/dual-task
condition revealed that SLDs was 10.35 per 100 syllables, which is an increase of 3 disfluencies
from the DAF reading task and an increase of almost10 disfluencies from the NAF dual-task
activity. Therefore, DAF increases stuttering-like disfluencies to a much greater extent than dualtask. Although disfluencies are removed from the calculation of speech rate, the high frequency
could be a contributor to the significant decrease in speech rate. The number of SEs and ODs
DAF and Dual-Task Effects
26
did not change significantly from the other tasks, which indicates that naturally occurring speech
disfluencies were not affected by DAF or the dual-task. The differences in SLD and speech error
rate provide one indication that the DAF and dual-task conditions do not have parallel effects on
speech production. There may also be individual differences in susceptibility to disfluencies
versus susceptibility to speech rate changes. By monitoring both disfluencies and rate changes in
a large group of speakers such individual differences can be identified.
In terms of effects on speech rate, s spoke significantly slower under the dual-task
condition as rate fell to between 4-5 syllables per second in both males and females. Because
speech rate reduced 1-2 syllables it can be inferred that performing a dual-task slows a person’s
speaking rate significantly and consistently. The 0.48 Pearson correlations between speech rate
in the NAF and dual-task condition (p < 0.05) revealed that performance under NAF is
associated with performance under dual-task regardless of sex. However, the same correlation
comparisons were not significant between males and females; again, there is no significant sex
different in performance when the groups were split into males and females. Both sexes are
affected to the same degree when a dual-task is introduced into an NAF reading task. Disfluency
data for the addition of a dual-task to natural reading conditions reveals only minimal differences
from NAF. These differences are not statistically significant and more likely due to random
chance; therefore, dual-task has no effect on the number of disfluencies on the speech of
normally fluent speakers.
Under DAF alone, subjects showed a significant decrease in speech rate compared to
both the NAF and NAF dual-task conditions. The average speaking rate was only 3-4 syllables
per second, which proves that DAF slows speech rate to a greater extent than a dual-task. It was
in this task that a significant difference between males and females was detected. The average
DAF and Dual-Task Effects
27
speech rate of females was just below 4 syllables per second, while the average for males was
just above 4 syllables per second. This difference was not expected because males are typically
expected to have a slower speech rate than females.
A Pearson correlation of 0.54 (p < 0.05) was found between speech rate in the NAF task
and speech rate in the DAF task regardless of sex. This significant correlation suggests that
persons who exhibit the fastest speaking rates under NAF will have the fastest speaking rates
under DAF; likewise, the subjects with slow speech rates under NAF will have slower speech
rates under DAF. In other words, the delay of speech rate affects all speakers, but naturally fast
speakers will still be faster under DAF than naturally slow speakers. A lower Pearson
correlation of 0.355 (p < 0.05) was found between rate on the dual-task NAF manipulation and
rate under DAF. Therefore there is some relation between speech rates across the experimental
manipulations but other factors must be considered to understand the findings. This correlation
appears to be more affected by sex differences than the other correlations because males had a
correlation of 0.471 (p < 0.05) between NAF and DAF rate, while for females it was 0.289 and
non-significant. Rate under NAF in females is therefore less strongly associated with rate under
DAF. This sex difference in associations between speech rate relationships of rate that is
revealed with DAF is a novel phenomenon that does not occur in dual-task activities.
For the combined DAF plus dual-task condition, the mean speech rate exhibited a further
decrease to around 3.5 syllables per second, which is the slowest rate of all four conditions.
Males had a slightly higher average rate than females but the difference was not significant. The
Pearson correlation of 0.806 (p < 0.05) showed that performance on the DAF task is strongly
associated with performance on the dual-task DAF performance. Performance on NAF and dualtask NAF are not as strongly associated, yet both were significantly associated with correlations
DAF and Dual-Task Effects
28
of 0.450 and .500 respectively. Similar correlations hold true for the performance of the male
participants (what are the correlations); however, it appeared that for females performance on the
initial baseline NAF reading task was little indication of the effects of dual-task DAF (correlation
of only 0.289).
Support for dual-task theories
Albeit less than DAF, the results of the study conclude that engaging in a dual-task does
indeed create deficits in the speech rate performance of the speaker; to better apply these
confirmations we must look back to the dual-task theories previously discussed.
The speech rate delays that appeared when subjects performed two linguistic tasks
simultaneously (i.e. reading and phoneme monitoring) could arise from the concepts that
Kinsbourne and Hicks (1978) proposed in their Functional Distance Hypothesis. Theoretically
both tasks required the participants to utilize the language left hemisphere of their brain, so it is
possible that interference occurred and leading to slower speech rate. Although, as Dromey
(2008) points out, the complexity of most speech experiments are difficult to attribute solely to a
simple explanation of postulated differences between brain areas.
However, the Central Bottleneck Model (Tombu & Jolicaeur, 2002) may account for the
results of this study to a greater extent. Performance of the dual-task activity requires the brain to
process and deliver the speech output mechanism as well the phonemic output of a specific
phoneme. The Central Bottleneck Model (Tombu & Jolicaeur, 2002) explains that these
processes occur in an all-or-none fashion; in other words, both tasks cannot be squeezed through
the bottleneck—or cognitively processed—simultaneously. One task must reach the neck
slightly before the other, causing a delay in the processing of the second task.
In this case, the
DAF and Dual-Task Effects
29
/s/ phoneme monitoring task could be reaching the bottleneck slightly before the speech output
begins. Thus, speech rate becomes delayed because it reaches the cognitive processors slightly
after the recognition of phonemes.
Understanding these dual-task concepts lends to the appreciation of the importance of
attention in speech. In simultaneously performing two tasks that rely on the similar processing
functions (i.e. language processors), deficits will occur because of the inability for both tasks to
be processed at the same time.
Application of Findings
DAF and dual-task effects on speech rate reveal that distinct mechanisms may affect
fluency that could have real-world application to the field of fluency. The effects caused by
DAF and dual-task differ in their nature. The affects of DAF could be partially related to the
interference in sensorimotor control in the auditory domain; the effects of dual-task, however,
are attributed to divided attention. Thus, the differences among the two variables give way for
the examination of the roles of audiomotor integration and attention in speech planning and
production.
As seen in the participants, both audiomotor integration and attention are contributors to
the fluency of speech. Upon deprivation of each component (i.e. audiomotor in DAF; attention
in dual-task), subjects exhibited significant speech rate delays. Under DAF, and the disruption
of the auditory feedback system, subjects exhibited a substantial amount of stuttering-like
disfluencies that were not observed under NAF or dual-task. For this reason, it can be assumed
that auditory integration is a key component in models of fluency.
Likewise, DAF slows speech rate to a greater extent than dual-task activities. The greater
effect suggests that audiomotor integration is essential for fluency. That is not to say that
DAF and Dual-Task Effects
30
attention is irrelevant; as the subjects portray, divided attention appears to be a significant
contributor to speech rate. Of equal importance, the combined DT-DAF condition elicited vital
information in the interpretation of fluency modalities. Although the results show a delayed
speech rate under both DAF and dual-task conditions in solidarity, when the tasks combine
together the effects are additively worsened (i.e. speech rate falls to an all-time low rate). This
occurrence confirms the contribution of both sensory and attention resources in the fluent
production of speech.
The differences in the contributions may lie within the domain of suprasegmental and
segmental aspects of speech rate. Attention plays a larger role in the control of suprasegmental
aspects of speech rate, while the audiomotor integration becomes important for both
suprasegmental and segmental speech domains. In other words, attention plays a role in the rate
of speech and audiomotor integration is important for both speech production and rate. Further
research could be conducted to understand the disfluency effects greater and to further the
application in the treatment and understanding of stuttering patients.
Conclusion
In conclusion, DAF and dual-task both contribute to the fluency of speech; DAF has
greater control over both the production and rate modalities, while dual-task divides attention to
affect only the rate modality of speech. DAF affects are a result of the disturbance in audiomotor
integration that results in disfluencies and slower speech rates.
DAF and Dual-Task Effects
31
References
Abbs, J.H., and Cole, K.J. (1982). Consideration of bulbar and suprabulbar afferent influences
upon speech motor coordination and programming. In Grillner, S., Lindblom, B.,
Lubker, J., and Persson, A. (Eds.), Speech Motor Control (pp159-186). Oxford:
Pergamon.
Allport, D.A., Antonis B., & Reynolds, P. (1972). On the division of attention: A disproof of the
single channel hypothesis. Quarterly Journal of Experimental Psychology, 24, 225235.
Amster, B.J., & Starkweather, C.W. (1987). Articulatory rate, stuttering, and speech motor
control. In: Peters HFM, Hulstijn, W. (Eds.). Speech motor dynamics and stuttering
(pp. 317-328) New York: Springer-Verlag.
Bachrach, D.L. (1964). Sex differences in reactions to delayed auditory feedback. Perceptual
and Motor Skills, 19, 81-82.
Beilock, S.L., Carr, T.H., MacMahon, C., & Starkes, J.L. (2002). When paying attention
becomes counterproductive: Impact of divided versus skill-focused attention on novice
and experienced performance of sensorimotor skills. Journal of Experimental
Psychology: Applied, 8, 6-16.
Black, J. (1951). The effect of side-tone delay upon vocal rate and intensity. Journal of Speech
and Hearing Disorders, 16, 50-56.
Bloodstein, O. (1995). A handbook on stuttering. London: Chapman & Hall.
Bohr, J. W. F. (1963). The effects of electronic and external control methods on stuttering: a
review of some research.J. South African Logopedic Soc., 10, 4-13.
DAF and Dual-Task Effects
32
Borden, G.J. (1979). An interpretation of research on feedback interruption in speech. Brain
Language, 7, 307-319.
Borden, G.J., and Harris K.S. (1984). Speech Science Primer. Baltimore: Williams and Wilkins.
Bosshardt, H.-G. (2006, July). Cognitive processing load as a determinant of stuttering:
Summary of a research programme. Clinical Linguistics & Phonetics, 20(5), 371-385.
Butler, R.A., and Galloway, F.T. (1957). Factoral analysis of the delayed speech feedback
phenomenon. Journal of the Acoustical Society of America, 29, 632-635.
Chase, R. A.; Sutton, S.; First, D.; Zubin, J. (1961). A developmental study of changes in
behavior under delayed auditory feedback. J. Genetic Psychol., 99, 101-112.
Cherry, E., and Sayers, B. (1956). Experiments upon total inhibition of stammering by external
control and some clinical results. J. Psychosomatic Research., 1, 233-246.
Dromey, C., & Shim, E. (2008, October). The effects of divided attention on speech motor,
verbal fluency, and manual motor task performance. Journal of Speech, Language, and
Hearing Research, 51(5), 1171-1182.
Fukawa, T., Yoshioka, H., Ozawa, E., and Yoshida, S. (1988). Difference of susceptibility to
delayed auditory feedback between stutterers and non-stutterers. Journal of Speech and
Hearing Research, 31, 475-479.
Hall, K., Amir, O., & Yairi, E. (1999, December). A longitudinal investigation of speaking rate
in preschool children who stutter. Journal of Speech, Language, and Hearing
Research, 42(6), 1367-1377.
Howell, P. (1990). Changes in voice level caused by several forms of altered auditory feedback
in fluent speakers and stutterers. Language Speech, 33, 325-338.
DAF and Dual-Task Effects
33
Howell, P., & Archer, A. (1984, September). Susceptibility to the effects of delayed auditory
feedback. Perception & Psychophysics, 36(3), 296-302.
Jueptner, M., Stephan, K.E., Frith, C.D., Brooks, D.J., Frackowiak, R.S.J., & Passingham, R.E.
(1997). Anatomy of motor learning, I: frontal cortex and attention to action. Journal of
Neurophysiology, 77, 1313-1324.
Kelly, E., & Contour, G. (1992). Speaking rates, response time latencies, and interrupting
behaviors of young stutterers, nonstutterers, and their mothers. Journal of Speech and
Hearing Research, 35, 1256-1267.
Kelly, E. (1994). Speech rates and turn-taking behaviors of children who stutter and their fathers.
Journal of Speech and Hearing Research, 37, 1284-1294.
Kinsbourne, M., & Hicks, R.E. (1978). Functional cerebral space: A model for overflow transfer
and interference effects in human performance: A tutorial review. In J. Requin (Ed.),
Attention and performance VII (pp. 345-362). Hillsdale, NJ: Erlbaum.
Korrowbrow, N. (1955). Reaction to stress: A reflection of personality trait organization.
Journal of Abnormal and Social Psychology, 51, 464-468.
Langova, J., Moravek, M., Novak, A., and Petrik, M. (1970). Experimental interference with
auditory feedback. Folia Phoniatr, 22, 191-196.
Leclercq, M. (2002). Theoretical aspects of the main components and functions of attention. In
M. Leclercq & P. Zimmermann (Eds.), Applied neuropsychology of attention: Theory,
diagnosis and rehabilitation (pp. 3-55). New York: Psychology Press.
Lee, B.S. (1950). Some effects of side-tone delay. Journal of the Acoustical Society of America,
22(5), 639-640.
DAF and Dual-Task Effects
34
Logan, K. & Contour, E. (1995). Length, grammatical complexity, and rate differences in
stuttered and fluent conversational utterances of children who stutter. Journal of
Fluency Disorders, 20, 35-61.
Lotzmann, G. (1961). On the use of varied delay times in stammerers. Folia Phoniatr., Basel, 13,
276-310.
MacKay, D.G. (1970). Why does language familiarity influence stuttering under delayed
auditory feedback? Perceptual and Motor Skills, 30, 655-669.
Näätänen, R. and Picton, T. (1987). The N1 wave of human electric and magnetic response to
sound: a review and an analysis of the component structure. Psychophysiology, 24,
375-425.
Nessel, E. (1958) Die Verzögerte Sprachrückkopplung (Lee effect) bei Stottern. Folia Phoniatr.,
Basel, 10, 87-89.
Peters, H. & Hulstijn, W. (1987). Programming and initiation of speech utterances in stuttering.
In: Peters HFM, Hulstijn, W. (Eds.). Speech motor dynamics and stuttering. New York:
Springer-Verlag.
Power, M.J. (1985). Sentence production and working memory. The Quarterly Journal of
Experimental Psychology, 37A, 367-385.
Raichle, M.E., Fiez, J.A., Videen, T.O., Macleod, A.M.K., Pardo, J.V., Fox, P.T., & Peterson,
S.E. (1994). Practice-related changes in human brain functional anatomy during
nonmotor learning. Cerebellar Cortex, 4, 8-26.
Rankin, R.J., and Balfrey, W.R. (1996). Impact of delayed auditory on the IPAT 8—Parallel
Form Anxiety Scales. Psychological Reports, 18, 583-586.
DAF and Dual-Task Effects
35
Rummer, R. (1996). Kognitive Beanspruchung beim Sprechen [Cognitive load during speech].
Weinheim: Psychologie Verlags Union.
Salmelin, R., Schnitzler, A., Schmitz, F., Jäncke, L., Witte, O.W., and Freund, H.J. (1998).
Functional organization of the auditory cortex is different in stutterers and fluent
speakers. NeuroReport, 9, 2225-2229.
Sawyer, J., Chon, H., & Ambrose, N. (2008, September). Influences of rate, length, and
complexity on speech disfluency in a single-speech sample in preschool children who
stutter. Journal of Fluency Disorders, 33(3), 220-240.
Siegel, G.M., Fehst, C.A., Garber, S.R., and Pick, H.L. (1980). Delayed auditory feedback with
children. Journal of Speech and Hearing Research, 23, 802-813.
Siegel, G.M., Schork, E.J., Jr., Pick, H.L., Jr., and Garber, S.R. (1982). Parameters of auditory
feedback. Journal of Speech and Hearing Research, 25, 473-475.
Smits-Bandstra, S., & De Nil, L. (2009, January). Speech skill learning of persons who stutter
and fluent speakers under single and dual task conditions. Clinical Linguistics and
Phonetics, 23(1), 38-57.
Soderberg, G. A. (1959). A study of the effects of delayed auditory side-tone on four aspects of
stutterer's speech during oral reading and spontaneous speech. Disseration (Doctoral)
- University Microfilms, The Ohio State University, Ohio.
Stager, S.V., and Ludlow, C.L. (1993). Speech production changes under fluency-evoking
conditions in non-stuttering speakers. Journal of Speech and Hearing Research, 36,
245-253.
DAF and Dual-Task Effects
36
Stager, S.V., Denman, D.W., and Ludlow, C.L. (1997). Modifications in aerodynamic variables
by persons who stutter under fluency-evoking conditions. Journal of Speech and
Hearing Research, 40, 832-847.
Stuart, A., Kalinowski, J., Rastatter, M., & Lynch, K. (2002, May). Effect of delayed auditory
feedback on normal speakers at two speech rates. The Journal of the Acoustical Society
of America, 111(5pt.1), 2237-2241.
Tombu, M., & Jolicaeur, P. (2002). All-or-none bottleneck versus capacity sharing accounts of
the psychological refractory period phenomenon. Psychological Research, 66, 274-286.
Van Borsel, J., Sierens, S., & Pereira, M. (2007). Using delayed auditory feedback in the
treatment of stuttering: Evidence to consider. Pro-Fono, 19(3), 323-332.
Van Riper, C. (1982). The Nature of Stuttering, 2nd ed. Englewood Cliffs, NJ: Prentice Hall.
Ward, D. (2006). Stuttering and cluttering: frameworks for understanding and treatment. New
York: Psychology Press.
Wickens, C.D. (1984). Processing resources in attention. In R. Parasuraman & D. F. Davies
(Eds.), Varieties of attention (pp.63-102). Orlando, FL: Academic Press.
Wingate, M.E. (1976). Stuttering: Theory and Treatment. New York: Irvington.
Yates, A.J. (1963). Delayed auditory feedback. Psychological Bulletin, 60, 213-232.
Zanini, S., Clarici, A., Fabbro, F., and Bava, A. (1999). Speaking speed effects on delayed
auditory feedback disruption of speech fluency. Perceptual Motor Skills, 89, 10951109.
Zerneri, L. (1966). Attempts to use delayed speech feedback in stuttering therapy. J. Français
d'oto-rhino-laryngol. chirurg. maxillo-facial, 15(4), 415-418.
Download