close

Вход

Забыли?

вход по аккаунту

?

PowerPoint-presentation

код для вставкиСкачать
SPEECH SAMPLES
AND THEIR INFLUENCE ON
SPEECH OUTCOME
Gunilla Henningsson
SWEDEN
How can we get comparable data sets?
Two important variables to consider:
п‚·
nature of the speech sample
п‚·
type of speech data
TYPE OF SPEECH SAMPLE
a variable influencing
SPEECH OUTCOME
Clinical Experiences: assessment
of speech samples based on
• syllable repetition : /papapa/ /pipipi/
• spontaneous speech
• will result in different conclusions
about the quality of the patientВґs
speech, i.e. on speech outcome
Speech sampling context(s):
1)
2)
3)
4)
5)
6)
7)
automatic speech (counting, days
of week, months of year, etc.)
memorized material (nursery
rhymes, prayers, etc.)
syllable repetition
single words/articulation test
sentence repetition
oral reading of sentences
spontaneous connected speech
Regardless of which speech
samples we are using there is in
fact the individual target speech
sounds on which we base our
descriptions
Examples from the battery of
The Universal Reporting
Parameters shows the same thing:
(1) hypernasality: high vowels, some
vowels, glides, liquids
(2) hyponasality: what else could we
listen to but nasals?
(3) atypical backing of dental/alveolar
target to velar
individual target sounds in
isolation are not to be
recommended as speech
samples.
THUS:
the individual target speech
sounds must occur in
a phonetic context
Two types of speech samples
recommended for reporting:
• single (short) words – as the
main speech sample
• single short sentences – as a
supplementary type of speech
sample
The main reason : Speech outcome
is influenced not only by:
• the individual target speech
sounds
but also by
• the phonetic context of the target
sound
Common specific requirements for the
structure of speech material
in terms of single words and
short sentences:
Birgit Hutters, phonetician,University of
Copenhagen, Denmark
Recommended by the
Eurocran Speech Group
EUROCRAN: European
Collaboration on Craniofacial
Anomalies
Webside: link to the Eurocran
speech project : “Good Practice
Archive for Speech”
www.eurocran.org
Each language needs to work
out
a word list
and
a sentence list
With adaptation to the specific
language:
The specific requirements when
the TARGET SOUND occurs in
WORDS:
Single consonant
inventory should include:
• all pressure consonants
• all or some non-pressure
oral consonants
• one or more nasal
consonants (+hyponasality)
Vowel inventory should
include:
• all or some high vowels
• all or some low vowels
• some non-high/low vowels
As to phonetic context:
• the number of contextual
sounds should be limited
and the following should be
avoided:
п‚· pressure consonants within
the word other than the target
sound: ex
gate, geese, pick
п‚· nasal consonants:
ten, can, pin
vowels of different height (in case
of more vowels in the word):
daddy , baby, potato
For 'target' consonants in
clusters, due to loading of the vpmechanism on target cons :
include following types:
• clusters with non-maximum loading (i.e.
oral non-pressure consonants: (sleep,
flag)
• clusters with maximum loading (i.e.
nasal consonants: snow, snail, small)
• other types of clusters
As to the position of the target
consonants in the word, the
consonant should occur in:
• 'strong position'
• other positions - if relevant for the
consonant/language in question
'Strong position' : where the
target sound is:
• most distinctly articulated
• most easily recognizable
• minimally influenced by the
context
Next requirement with regard to the
number of representations of each
target consonant:
• single target consonants:
• three times in 'strong position',
• twice in other positions
• in clusters: twice
Other requirements for the
structure of the word list:
• the target consonants - and
subsequently the words - should be
randomly ordered
• high vowels should occur in
approximately 10 of the words
following a target consonant in
'strong position'
Ex from English word list, 3 year old
children (Scandcleft project, Eurocran)
target word
target sound
• pea
• doll
• peep
pdp-
(ideal ex)
• kick
k-
(target sound - twice - OK)
• gate
g-
(velar + dental - not recommended)
(ideal ex)
(target sound – twice - OK)
SHORT SENTENCES:
• The following target consonants
should occur in 'strong' and
possibly other positions
• all single pressure consonants
• nasal consonant (+ hyponasality)
• others as required
With regard to vowels, the
following is suggested
vowel height should be taken
into consideration when deciding
on the phonetic content of the
sentences
A few words of cross-linguistic
comparisons
• Cleft Palate Craniofacial J: Sept 2004
• Speech Outcome Following Treatment
in Cross-Linguistic Cleft Palate Studies:
Methodological Implications.
Birgit Hutters, Gunilla Henningsson
The impact of the cleft condition will
differ across languages depending
on how vulnerable a language is in
terms of:
• prevalence of high pressure
consonants
• prevalence of high vowels
Two additional requirements for the
target sounds intended for
comparison - they should :
• be of similar phonetic content
• occur in similar phonetic context
We have two controlled speech
samples we all should use:
single short words
short sentences
What about repeated
syllables such as /pipipi/ ?
No standardization for:
• the SLP´s rep behavior
• numbers of syllables used
• pronounciation of the vowel
• some syllable chains may be real
words in some languages
What about automatic
speech and memorized
material ?
What about automatic speech
cont.
• Should NOT be used within
the battery of The Universal
Reporting Parameters
What about spontaneous
speech ?
• impossible to control from a
universal aspect
• impossible to describe the base for
the assessment
• the length of the speech will vary
• the phonetic context will vary
• the speech tempo will vary
However:
• spontaneous speech could still be
assessed but in terms of intelligibility or with other rating scales.
• but once again: spontaneous
speech should be rated and
reported separated from single
short words and short sentences
Assessment and rating of
HYPERNASALITY
• from high vowels within short
words - or
• from low vowels if this is the
purpose of our assessment
• keep assessment and rating of high
and low vowels separated
Hypernasality cont.
• Recommendation: assessment of
10 short words with high vowels
• The longer the speech samples
with a mixture of vowels - the more
difficult to listen to the individual
vowels and rating of hypernasality
For the clinical speech assesment
of an individual patient , the SLP is
free to use whatever speech
samples he/she likes, but they will
not be usable within the battery of
The Universal Reporting
Parameters
The point is:
in order to be able to report on
speech outcome in a
meaningful way - the requirements for speech samples
should be considered in the
clinical assessment
Each cleft palate team within each
language needs to establish:
• a word list as much as possible in
accordance to the recommended
requirements
• a sentence list as much as
possible in accordance to the
recommended requirements
How to elicit single word
speech material?
Eurocran recommendations (1)
• parents must not interfere with the
childВґs talk
• naming of the word - first choice
• semantic prompting -next choice
• final choice : repetition of the word
• After each naming the SLP should
repeat the target word
Further recommendations (2)
• the speech samples: tape
recordings of high standard
• high quality equipment (see
Eurocran recommendations)
• the target sound realisation:
transcriptions from the recordings
should be used for assessing cleft
palate speech characteristics:
Further recommendations (3)
This means that:
• each target speech sound
realisation could include more
than ONE speech characteristic
(ex dental backing to velar +
nasal emission)
Further recommendations (4)
• Words and sentences should
include an established number of
target speech sounds
• then: type of speech
characteristics – AND number of
correct and incorrect responses
could be summed up and
transferred into any scoring system
Further recommendations (5)
• CSAG (Great Britain)
• the ScandCleft Project (longitudinal, multicenter study:10 european
cleft palate centres, 5 languages
• Swedish Articulation and Nasality
Test: SVANTE (publ 2005)
Документ
Категория
Презентации
Просмотров
4
Размер файла
890 Кб
Теги
1/--страниц
Пожаловаться на содержимое документа