Lecture Eleven Theories of disfluencies in speech

1
Lecture Eleven - Theories of disfluencies in
speech
Orienting Question Describe and evaluate
approaches for accounting for dysfluency?
2
Overview of dysfluencies the theories
consider Normal developmental
dysfluency Stuttering in its early and late
forms Telegraphic speech (loss of function words)
and loss of inflectional endings on words in
different disorders
3

• Observations
• Alternative account of how linked together
  (EXPLAN), hearing and
• synchronicity
• 2. Wing-Kristofferson and DAF. Steves Dow and
  Moruzzi results
• 3. I dont
• 4. We also thought there might be promise in this
  (we work with children).
• Work with Steve and Downing-Wilson etc a) are
  temperament
• differences but vary what are between studies, b)
  dont correlate with
• language. Havent looked at correlation with
  anxiety yet.
• General issue
• 5. Temperament, as well as anxiety, might be
  secondary effects of
• stuttering.
• 6. How good is startle response measure. Might
  merit extending.
• 7. Can the genes responsible be predicted on
  independent and a
• priori grounds?

4
Two directions from which we can approach
theory

• From the point of view how a constellation of
  biological, behavioural and environmental factors
  can lead to stuttering (psychological). Says
  what stuttering is and treatment implications are
  not the primary focus.
• By taking a stance on what stuttering is and how
  it can be treated. The success of these sorts of
  theories depends on whether they can locate and
  identify cases of stuttering and whether the
  treatments suggested by theory work.
• Distinction not absolute

5
I will start with a position A theory (theory
one) and then consider how similar facts have
been approached from position B (theory two).

• A definition of stuttering
• fluent speech with stuttering
• events that are interspersed.
• Stuttering is a language problem.

6
What are stutterings?

• Johnsons list (reorganised into the classes
  labelled stalling and non-stalling disfluencies)
• Stallings
• Pauses (filled and unfilled)
• Word repetitions (in, in the morning)
• Phrase repetitions (in the, in the morning)
• Non-stallings
• Prolongations (ssister)
• Part-word repetitions (s.s.sister)
• Word breaks (s-ister)
• Revisions (my uunc., my mothers brother)
• Abandonments like non-sequitors (will not say
  much about these)

7
What words are stuttered?

• In adults (Brown, 1945)
• Content words (nouns, verbs etc)
• Long words
• Words starting with consonants
• Words starting early in sentences
• (Wingate, 1988, 2002 adds stressed words)
• Many of these are correlated with content words
• In children
• Function words (pronouns, articles) Bloodstein
• Tend to be short, start with vowels and not
  stressed (3 of the remaining 4 characteristics)
  because of the confound between these factors and
  whether the word is content or function in type

8
This definition only partly meets the person who
stutters experience they feel that all their
speech is affected and even after recovery,
they still have this same feeling Social and
emotional concomitants of stuttering (including
temperament in children) The language
definition does not address the biological facts
behind stuttering That there is a large genetic
component That there are functional (possibly
even structural) differences between the brains
of people who stutter and those who do not
9
B. Practical questions Early diagnosis Changes
over development What treatments are effective
10
Theory one The multifactoral theory of
stuttering. (Smith, de Nil etc) Maintains that
many these factors affect stuttering

• Evidence for
• Actual research into multiple factors
• Smith, motor output and how affected by
  linguistic variables
• Limited range of factors that are examined
• SLI measure
• Evidence against
• Pay lip service to these factors, but do not come
  up with causal models. E.g. does anxiety directly
  influence whether words are stuttered or not or
  do people feel anxious because they stutter?

11
B. Theory two Demands and capacities
(Starkweather and others)

• Not research-evidence based
• A way of describing the factors (social,
  emotional, linguistic, pragmatic, biological ..)
  that affect an individual in therapy.
• A clinical tool, not so much evidence based.
• Yaruss OASES model does related things

12
Single factor/ single issue theories subsumed
under position A Note Researchers may focus
their investigations on a particular factor and
point out its relevance. They would not
necessarily rule out other factors. This can be
a result of the need for close experimental
control, in other cases it can be a result of
promoting one factor as paramount.
13
Theory three Stuttering emerges in child
development as the speakers grammatical system
starts to develop (Bernstein Ratner).

• Evidence for
• Stuttering starts at about the age speakers start
  to use content words.
• Limited amount of evidence that syntactic factors
  influence stuttering
• Evidence against
• Many studies have failed to find support for
  syntactic problems being associated with
  stuttering (see Nippold reviews).
• Children with syntactic deficit (e.g. SLI) dont
  stutter
• Is only concerned with how the problem starts,
  not how it can be treated

14

• Theory four The covert Repair
• Hypothesis (Kolk and Postma, 1997 -review)
• Stuttering is a result of a slow phonological
• system.
• b. This makes speakers prone to make errors.
• c. They detect these errors internally (covertly)
  
• through the perceptual system
• d. Stuttering is a reflection of the processes by
  
• which these errors are corrected
• NB a could be right whether or not b-d are

15
Theory four The covert Repair Hypothesis (Kolk
and Postma, 1997)

• Evidence for
• Yaruss and Conture there is a correlation
  between phonological errors and stuttering in
  children who stutter
• Isolated piece of support
• Evidence against
• Does not address changes in stuttering over
  development
• Talk on EXPLAN criticizes CRH interpretation and
  offers an alternative account

16
Theory five Stuttering is a subtle motor control
deficit (van Lieshout, Alfonso, Max, ..)

• Evidence for
• Studies like those of Smith show subtle motor
  problems.
• Motor changes (speech rate) affect fluency
• Van Lieshout suggests changes in speech affected
  over development due to motor maturation (does
  not specify how)
• Evidence against
• Motor problems are subtle
• People with motor problems (e.g. dysarthria) do
  not necessarily stutter
• EXPLAN argues that the reason stuttering is
  difficult to see when linguistic or motor factors
  alone are investigated is because a model is
  needed of their joint action and it is the way
  this process fails that leads to stuttering (an
  account of developmental changes is also
  included)
• Smith takes a similar view

17
Theory six The importance of genetic factors
(Yairi and Ambrose) Not so much a theory, really
a strong finding that these workers, among
others, have established and pushed
18
Twin Models

• A variance decomposition (A, C, E) can be applied
  to liability, where the correlations in
  liability are determined by path model
• This leads to an estimate of the heritability of
  the liability

19
ACE Liability Model
1
1/.5
E
C
A
A
C
E
L
L
1
1
Unaf
Unaf
Twin 1
Twin 2
20
Example of model fitting for one year only (here
for liability to stuttering age 2).
21
Age and sex differences
22
Persistent/Recovered (age seven)
23
The genetic transmission process is not
completely understood Chromosonal analyses are
starting to be conducted unlikely that a
single gene is involved
24
Theory seven The importance of CNS factors
(Ingham and others)

• Evidence for
• Ingham et al. argue that not a structural
  problem. May be functional.
• Sommer et al. tract connecting Brocas and
  Wernickes areas (sensory-motor?) on left is
  poorer in people who stutter
• Our findings bilateral

25
Delayed gt normal auditory feedback
R
L
-48
48
6
Significantly more activity during DAF than NAF
in superior temporal gyrus bilaterally, right
anterior insula, frontal operculum and cerebellum
26
Diffusion Tensor Imaging in PDS

• We obtained diffusion tensor data in 9 PDS and 11
  controls
• Diffusion of water easiest along the long axis of
  fibres compared with across it
• From measuring diffusion of water in each voxel
  we can infer the predominant direction of fibres
• Fractional anisotropy (FA) is a measure of
  coherence of fibre orientation
• FA is low if fibre orientation is disrupted or
  perhaps reflects poor myelination of those fibres

27
Analysis of DTI data

• 9 PDS and 11 controls
• 2.5mm3 voxels 2 averages 60 directions (1.5T
  Sonata)
• FA maps calculated

• Tract-based spatial statistics
• Nonlinear registration
• Skeletonization of white matter
• Local search for highest FA in tract
• Statistics calculated on the skeleton with
  permutation testing

28
Reduced White Matter Integrity Underlying Motor
Cortex
R
L
1
22
Red average white matter for all subjects
Green skeleton region in which data are
compared
Blue significantly higher FA in Controls than
in PWS
29
(No Transcript)
30
Reduced White Matter Integrity in Arcuate
Fasciculus
-44
-50
Red average white matter
Green skeleton
Watkins et al., in prep
Blue significantly higher FA in Controls than
in PWS
31
Reduced white matter integrity in the arcuate
fasciculus
32
Summary of DTI findings

• Replication of previous DTI study in PDS of left
  sensorimotor representation of the articulators
• But
• Our data shows bilaterally reduced FA, consistent
  with a developmental disorder with no apparent
  plasticity to aid reorganisation of function
• Do recovered developmental stutterers show the
  same structural differences or are these
  unilateral?
• Also - we saw reduced FA in arcuate fasciculus
  bilaterally, consistent with a disruption of
  motor-to-sensory and sensory-to-motor loops in PDS

33
Theory seven The importance of CNS factors
(Ingham and others)
Observations Does this suggest less reliant on
sensory feedback? Phrenology
34
Theories that address how the disorder should be
treated.
Demands and capacities (theory two) can be seen
how to tailor needs to problems of a particular
client.
35

• Theory eight (Alm)
• The brain has two parallel systems the medial
  and the lateral system
• and stuttering is related to the medial system
  which includes the basal
• ganglia and other structures in the mid brain
  involved in automatization
• of behavior. Main problem is impaired go signal
  for syllables.
• 2. Under AAF conditions, speech timing is shifted
  to the lateral system
• (includes cerebellum).
• 3. Likes sub-typing.
• 4. Stuttering related to a sensitive or
  reactive temperament that
• increases neuromuscular reactivity (measures as
  magnitude of the
• acoustical startle response).
• 5. Found higher startle response, and higher
  level of anxiety
• (acknowledges that this may be a secondary
  effect).
• 6. Notes a slight negative relation startle
  reactivity and anxiety
• 7. Genetic effects might be due to a high number
  of dopamine D2
• receptors in the motor part of the basal ganglia.

36
Theory nine The learning-theory based Lidcombe
therapy

• Reinforce fluent speech
• Time out disfluent speech
• Evidence for
• Extensive funding and randomized control trials.
  Probably the most investigated therapy in the
  world
• Evidence against
• Dutch study shows no better than other techniques
• What is fluent and what is disfluent see EXPLAN
  talk

37
Theory ten Mirror neuron account of the effects
of altered auditory feedback
Altered auditory feedback improves speech
control of speakers who stutter
38
Theory ten Mirror neuron account of the effects
of altered auditory feedback
Doesnt work because corrects a hearing
deficit as in a monitoring account (e.g. covert
repair hypothesis) Mirror neurons motor neurons
that fire when watch a behaviour. Suggested that
involved in sensory motor integration (in the
appropriate part of brain to do this)
39
Theory ten Mirror neuron account of the effects
of altered auditory feedback

• Evidence for
• Allows an interpretation of feedback (not really
  evidence)
• Tasks that used are distant from the neurons
  themselves
• Evidence against
• A flashing light (which could not be used for
  sensory-motor integration) has similar effect

40
Theory eleven Parental interaction

• Evidence for
• Researchers at the Michael Palin centre for
  stammering children locate problems specific to
  father or mother (not published to date)
• Dialog data beginning to be examined
• Evidence against
• Paucity of data at present

41
Conclusions

• Single factor account unlikely to operate
• Warning if promote a multifactor theory, do so in
  a way it can be tested
• Practical (e.g. clinical) accounts should not
  ignore known fluency enhancing effects
• Personal belief is that linguistic and motor
  factor interaction is crucial (subject of next
  talk