PSYC 2207 Speech Lecture 3 part one Orienting Question

1
PSYC 2207 (Speech) Lecture 3 (part one)
Orienting Question
Outline and evaluate Levelts monitoring theory
of speech production
2
Levelts perceptual loop theory as an account of
speech production that relies on perceptual
processes

• Diagram shows
• LHS conceptualizer, formulator, articulation
• RHS audition, speech comprehension
• Note
• Speech comprehension is the speech perception
  system
• The flow of information between the processes
• The conceptualizer has a monitor that receives
  input from the conceptualizer and the speech
  comprehension system

3
Levelts way of modelling these phenomena
4
What the links between the perception and the
production systems allow the model to achieve 1.
All speakers make errors on some occasions and
they can realize and do something about this 2.
The perception system is sitting there and, if it
can detect these errors, it could signal them. 3.
No extra mechanism is needed (perception does
what it always does, Postma, 2000) 4. Perception
supplies an indication of what was actually
produced and the conceptualizer has the original
intention 5. The monitor compares the two. If
they are the same, speech has been produced
correctly, if not (in Levelts account), speech
is re-initiated 6. Note for now (well see the
importance later) that there are two routes into
perception and to the monitor a. Before sound is
output (the internal loop) b. When speech is
actually produced (the external loop)
5
Evidence for the perceptual loops 1. The external
loop a. Speech errors and repairs Overt
errors Turn left at the, no, turn right at the
crossroads Left reparandum Right
alteration The first at the overshoot Pause
point of interruption No editing phrase Second
turn retrace Monitoring account - heard the
error and corrected it Repairs at syntactic,
lexical, phonemic levels Abandonment
6
Covert errors Turn, turn right at the
crossroads There may have been the same error as
in the earlier example that has been detected
over the internal loop and speech intercepted
before it is output Note 1. That later we will
return to consider events like these as features
in different forms of disfluent speech
(childrens stuttered and aphasic speech 2.
Levelt does not consider these to any extent but
later researchers do.
7
So for now we consider the external loop Other
evidence for the external loop - Altered
auditory feedback - if the sound is altered
experimentally, the perceptual system will detect
this, signal an error and the speaker will make
a correction. As none was required, the speaker
introduces rather than removes an error by the
alteration he or she makes (well consider
auditory feedback more later).
8
Problems for the monitoring account

• The relationship between speech errors and
  repairs
• speakers make errors that they do not repair
• these errors seem to have different
  characteristics to those that are repaired (seem
  to depend on different processes)
• errors are not always repaired (young children)
• Conclusion errors that are repaired are not
  representative of those that arise in production
  and repairing seems to be an optional process

9
Problems for the monitoring account

• Can speakers recover details about the sound of
  their voice from the sound that is output?
• Bone conducted sound
• is not like speech
• is as loud as direct speech

Key reference for Levelts theory, repairs and
their relationship to stutterings Levelt, W.
(1989). Speaking From intention to
articulation MIT Press
10
PSYC 2207 (Speech) Lecture 3 (part two)
Orienting Question
Do speakers who stutter make more speech errors
than fluent speakers or do they just have
problems timing their speech? Discuss your answer
with respect to theories of speech control
that either implicate errors or timing problems
11
Classic description of stuttering and
similarities between early childhood nonfluencies
and stuttering

• What is stuttering?
• Episodes of fluency interspersed with
  dysfluencies
• Based on assessment of speech
• Stuttering event counts are made

12
Stuttered dysfluencies Demos on
http//www.speech.psychol.ucl.ac.uk/index.html 1.
Word repetitions (e.g Katy, Katy) 2.
Interjections (e.g terms like erm but also
phonologically inappropriate sounds like
syllable-initial ng 3. Part-word repetitions
(e.g k..k..Katy) 4. Prolongations (e.g.
mmMother) 5. Other dysfluencies. Mainly broken
words (e.g di..nosaur) but also used as a
catch-all category for all kinds of
dysfluencies 6. Phrase repetition (e.g I Want, I
want) 7. Phrase revisions (e.g. my uunc,my
mothers brother) 8. Idea abandonment (like non
sequitors) None of these are exclusive to
stutterers
13
Who suffers from stuttering?

• Occurs in all cultures
• Estimate of incidence - about 10 some time in
  life
• Demographics
• mainly males
• disorder of childhood (about 90 recover
  spontaneously
• middle class children seem particularly prone

14
Stuttering

• Very intransigent to recovery if not recovered by
  teens
• Characteristics change developmentally - does
  change that happens lead to persistence?
• Stuttering events occur in all fluent speakers
• Not really errors - we call them fluency failures
• An intermittent problem - even in very severe
  stutterers
• NB although intermittent not random - linguistic
  factors

15
Johnsons categories divided into stalling and
advancing - STALLING
1. Word repetitions (e.g Katy, Katy) 2. Phrase
repetition (e.g I Want, I want) 3. Now extend to
filled / unfilled pauses Interjections (e.g
terms like erm but also phonologically
inappropriate sounds like syllable-initial
ng 4. Idea abandonment (like non sequitors)
16
Johnsons categories divided into stalling and
advancing - ADVANCING
5. Part-word repetitions (e.g k..k..Katy) 6.
Prolongations (e.g. mmMother) 7. Other
dysfluencies. Mainly broken words (e.g
di..nosaur) 8. Phrase revisions should be mainly
characterised by sticking on a content word where
sticking is indicated by the word being
incomplete as in the my uunc,my mothers
brother example. Complete function (in the, if
he) and complete content word abandonments (my
uncle, my mother..) should be rare as should
abandonment of part of a function word (and hhh,
and she) because all of these may arise from
selection errors
17
Health service issues

• Diagnosis
• is this being done correctly?
• If not, maybe children who are not really
  stuttering (NNF) are being treated

18
Health service issues

• Treatment outcome
• How can outcome be measured in children in the
  face of such a high rate of spontaneous recovery

If we knew what the difference between child and
adult stuttering was, we could develop treatments
for different age groups that have different
goals in mind - children - how to get them to
avoid taking a path that leads to
stuttering adults - how to reverse the patterns
that are characteristics of such speakers of this
age and - how to get those patterns to stick
19
Where does stuttering occur in speech?

• Browns adult factors (and differences in
  children)
• Adults stutter on content words, children on
  function words
• Adults stutter on consonants, children on vowels
• Problems occur in early positions in sentences
• Problems experienced on long words

20
Diagnosis and the related problem of separation
of normal nonfluency from stuttering Some rules
of thumb Conture - emergence of prolongations
is a sign stuttering is worsening Yairi Hall -
word repetitions are a characteristic of NNF,
part-word repetitions are a sign of
persistence Van Riper - when there is
acceleration across a series of repetitions, its
a danger sign Sampling problem - are there
children who show these signs at an early age but
are not detected because they are disguised by
the high number of children who recover
spontaneously? Could spontaneous recovery be
incorrect diagnosis?
21
What do we know about how stuttering changes over
development?

• Browns adult factors (and differences in
  children)
• Adults stutter on content words, children on
  function words
• Dysfluency type changes as children persist
• Conture
• change from repetitions to prolongations
• Yairi
• change from word to part-word repetitions

22
Comparison of what is known about adults with
what is known about children Want to emphasise
factor 1 - adults content words, children
function words. Bloodstein and Gantwerk (1967)
and Bloodstein and Grossman (1981) reported that
children who stutter are more likely to be
dysfluent on function words than on content
words. Children who stutter are dysfluent on the
simpler function words Certain explanations do
not apply at the onset of the disorder e.g.
Browns (1945) semantic explanation of all his
linguistic factors Wingate (1988) - all of
Browns factors can be explained on the basis
that stressed words are stuttered Semantic
difficulty cannot be the underlying cause as
children have difficulties on the semantically
simpler function words Wingates (1988)
explanation also ruled out - function words are
rarely stressed in these speakers
23
What would Clark and Clarks ideas mean when
applied to stuttered speech? Early speech
nonfluencies like I I I split. NB a function
plus a content word and the dysfluency is on the
function word (delaying saying split) Older
stutterers would be advancing I sssplit. NB
change in locuss and type of stuttering. Could
this lead to persistence? Phonological words and
stuttering locuss Function word position -
Au-Yeung et al (1998) Svartik study on fluent
speakers Which of the third person pronouns will
be stuttered/spoken fluently? He hit him and I
cried Exchange slides
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
English
28
(No Transcript)
29
Why do speakers advance to the content word when
they are not ready? Shown that the speech of
children who do not grow out of their stutter
show a change in the pattern of their speech.
They stop repeating function words and carry on
and try and say the content word before they are
ready. Using the earlier example, they might say
I ssssplit it. We have in this example a
prolongation of the s sound but might also
have a repetition of the first part of the word -
I s.s.s.split it. The change in the pattern
of stutterings is to part of the word rather
than the whole word and onto the more complex
class of word - the content word that was
mentioned above.
30
Is advancing a bad thing? Yes Follow up
groups Division into recovered and
non-recovered Improved Not improved Health
service provision implications Identification of
at-risk cases May be just better
diagnosis Supports a new proposal for treatment