mapping workshop on universal reporting parameters for the speech of individuals with cleft palate...
TRANSCRIPT
Mapping
Workshop on Universal Reporting Parameters for the Speech
of Individuals with Cleft Palate
Washington DC, April/May 2004
Tara Whitehill, Hong Kong
Proposed Three-Stage Plan
• Evaluating: The usual speech evaluation by the local speech specialist in cleft palate.
• Mapping: Conversion from the local speech evaluation to the universal reporting measures.
• Reporting: Universal parameters that describe the cardinal features of speech in individuals with cleft lip/palate and related craniofacial anomalies regardless of the individual’s primary language.
Already covered
• Evaluation (range of protocols and procedures; key issues)
• Parameters (proposal; key issues)
• Focus: mapping
Fanny LAW Sze Kuen
International Parameters for Cleft Palate Speech: A Pilot Study
Honours dissertation
Bachelor of Science (Speech & Hearing Science)
University of Hong Kong
May 2004
Supervisor: Dr. T. L. Whitehill
DRAFT - in progress
• 6 sets of client forms from 5 centres
• 5 clients each set = 30 client forms
• Original forms only (no recordings, no additional data such as transcriptions, unless provided on original form)
• Two groups of “mappers”– 5 “experts” - members of WG who supplied
forms
– 5 students - little/no previous experience cleft
• Used “current” version of parameters and guidelines
• Students given session to familiarize/
clarify terminology• Mapped from client forms onto parameters
form• Identifying information masked (client, examiner,
centre)
• Provided feedback on ease/difficulty of mapping each parameter– Either for client form or by set (centre)
• Weighted scores - not included in this study
“Outcome measures”
• Agreement– Within expert group– Within student group– Across all mappers
• Feedback (ease/difficulty)
Why use students?
• In future, unlikely mappers will be naïve
• However, “expert” group was restricted to individuals who had developed and debated the parameters
Agreement
• A big problem was “missing data”• OK if client form stated MD (or WNL or NAD),
but what if didn’t? - assume OK? tick MD?
• Calculated
(a) absolute agreement (Condition 1)
(b) disregarding “missing data” (Condition 2)
Trial Results amongexperts
Results amongstudents
Agreement and explanation
1 1 1 1 1 1 1 1 1 1 Agreed among experts, students and allmappers in both Condition 1 and 2.
2 1 1 1 1 2 2 2 2 2 Agreed among experts, and students in bothCondition 1 and 2
Not agreed among all mappers in bothconditions
3 1 MD 2 1 1 2 1 1 1 Not agreed among any groups in eitherconditions
4 1 MD 1 MD 1 1 1 1 1 Agreed among students in Condition 1 Agreed among experts in Condition 2 Agreed among all mappers in Condition 2
5 MD MD NA 2 MD 2 2 2 MD Trial for experts was deleted as there would be
no agreement when the MD and NA wereexcluded.
Agreed among students in Condition 2 Agreed among all mappers in Condition 2
Key: MD = missing data, NA = no response, 1 = WNL/Absent, 2 = Present
Table 1. Examples to illustrate how agreement was defined
Primary Parameters
Hypernasality
Agreement• Experts Condition 1 - 60%
Condition 2 - 63%• Students Condition 1 - 73%
Condition 2 - 73%• All Condition 1 - 57%
Condition 2 - 60%
Not much missing data
Primary Parameters
HypernasalityFeedback/Issues to consider
• Some difficulties when no. scale points differed (e.g., client form - 7 pt. scale; parameters form - 4 pt. scale)
• In such cases, mappers reported difficulty, and agreement decreased
• Implication: explicit instructions?• Another difficulty: “mild to moderate” or “moderate to
severe” (students selected less severe and experts more severe!)
• “Set B” - definition of “mild” = “evident but acceptable”. One expert rater mapped onto WNL.
Primary Parameters
Hyponasality
Agreement• Experts Condition 1 - 53%
Condition 2 - 96%• Students Condition 1 - 87%
Condition 2 - 100%• All Condition 1 - 50%
Condition 2 - 97%
Excellent, when missing data is accounted for
Primary Parameters
HyponasalityFeedback/Issues to consider
• Missing data
• Otherwise, excellent
• (binary scale)
Primary Parameters
Audible Nasal Emissionwith/without Nasal Turbulence
Level One (absent, present, MD)
• Experts Condition 1 - 63%
Condition 2 - 87%
• Students Condition 1 - 77%
Condition 2 - 80%
• All Condition 1 - 57%
Condition 2 - 70%
Level Two (subcategories)
• Experts Condition 1 - 23%
Condition 2 - 60%
• Students Condition 1 - 30%
Condition 2 - 38%
• All Condition 1 - 17%
Condition 2 - 27%
Primary Parameters
Audible Nasal Emissionwith/without Nasal Turbulence
Feedback/Issues to consider• Low agreement for subcategories• “phoneme specific” - e.g. “audible for plosive and /s/”• Confusion about pervasiveness e.g. “mild but
consistent”, “phoneme specific and frequent”• Results of mirror test (if not audible in speech)• (NE w/wout nasal turbulence - not raised here)
Primary Parameters
Weak Oral Pressures
Agreement• Experts Condition 1 - 17%
Condition 2 - 75%• Students Condition 1 - 13%
Condition 2 - 62%• All Condition 1 - 0%
Condition 2 - 46%
Primary Parameters
Weak Oral Pressures
Feedback/Issues to consider• All 9 mappers found this parameter difficult to map• Primarily because of lack of data on client forms - no
form had specific category for this• Mappers needed to derive the information from other
information; particularly difficult for students• Also, students unclear about definition
Primary Parameters
Substitution ErrorsLevel One (absent, present, MD)
• Experts Condition 1 - 57%
Condition 2 - 73%
• Students Condition 1 - 73%
Condition 2 - 85%
• All Condition 1 - 57%
Condition 2 - 70%
Level Two (subcategories)
• Experts Condition 1 - 10%
Condition 2 - 30%
• Students Condition 1 - 10%
Condition 2 - 15%
• All Condition 1 - 10%
Condition 2 - 20%
• Agreement not bad for level one, but very poor for subcategories; missing data was not the main problem
Primary Parameters
Substitution Errors
Feedback/Issues to consider• All forms included level one (present/absent/MD)• Subcategories: most forms had patterns, but few
included frequency information • Some patterns which expert listeners considered
‘related to cleft palate’ not included here (e.g. nasal substitution/realization’, ‘double articulation’)
• Confusion between this parameter and ‘Developmental Delay or Other Articulation/Phonological Errors’ (even for experts) e.g. omission, backing /t/ -> [k].
Secondary Parameters
Errors related to dentition, occlusion, palatal vault configuration
Agreement• Experts Condition 1 - 13%
Condition 2 - 69%• Students Condition 1 - 30%
Condition 2 - 59%• All Condition 1 - 7%
Condition 2 - 53%
Better when missing data taken into consideration, but still pretty poor
Secondary Parameters
Errors related to dentition, occlusion, palatal vault configuration
Feedback/Issues to consider• No specific category for this on client forms;
information had to be extracted from other information
• Some forms (Sets B, E, F) included categories such as ‘lateralization’, ‘palatalization’, ‘dentalization’ - but still not clear whether these due to organic factors or “Other Articulation/Phonological Errors” - even for experts
• Errors related to hearing problem? fistula?
Secondary Parameters
Developmental Delay or Other Articulation/Phonological Errors
Agreement• Experts Condition 1 - 16%
Condition 2 - 72%• Students Condition 1 - 30%
Condition 2 - 50%• All Condition 1 - 10%
Condition 2 - 43%
Again, better when missing data taken into account, but still pretty poor
Secondary Parameters
Developmental Delay or Other Articulation/Phonological Errors
Feedback/Issues to consider• Mappers needed to make judgement regarding
whether errors were related to cleft or not • Confusion with both “Substitution Errors” and “Errors
related to dentition..”• Client forms rarely specified cause or explanation of
error patterns - mappers needed to judge• Special cases e.g. “no oral consonants”
Secondary Parameters
Voice/Laryngeal Disorder
Agreement• Experts Condition 1 - 56%
Condition 2 - 90%• Students Condition 1 - 63%
Condition 2 - 97%• All Condition 1 - 60%
Condition 2 - 96%
Agreement was good, especially when missing data was taken into account
Secondary Parameters
Voice/Laryngeal Disorder
Feedback/Issues to consider• Similar to hyponasality - mappers were familiar with
term• When data was provided on client form, it was clear• Binary choice; no subcategories
Global RatingSeverity/Intelligibility
• Not decided at time of this study
• Only 2 of the 6 sets included such data
Conclusions
• Reliability lower among students than experts
• Experts were those who developed form
• Who will be mapping in the future?