Lecture 5 - Supra threshold Speech Recognition I Flashcards
What are the purposes of speech recognition testing?
1) Assess precision of speech transmission system
2) Access precision of individual’s speech reception
- Assumes an ideal transmission system
- Clinical implications: diagnostic purposes, estimating functional deficits of hearing loss
3) Both purposes incorporate understanding of speech
- Using supra threshold presentation to determine how well an individual performs
What are the requirements of a speech recognition test?
1) Material should be representative of the language
- Concept of validity: performance should reflect the person’s ability to understand every day speech
- Use material that represents the language
- Test materials developed to reflect the sounds in average speech
2) Test results should be reliable (repeatable)
3) Test should be easy to score
4) Materials should be relatively difficult to obtain a measure of the person’s speech intelligibility problem
- Nonsense Syllables: too abstract, too difficult for some people; measures absolute recognizability rather than intelligibility
- Monosyllabic words: not abstract, not too much redundancy, a measure of intelligibility
- Multisyllabic words: scores are influenced by knowledge of the language (lexicon)
- Sentences: knowledge of syntax, semantics, and linguistic affects performance
What is the history of monosyllabic word tests (Harvard PAL)?
1st attempt to develop a test of speech recognition
- Called Harvard PAL PB-50s
Purpose for test development: assess communication
- System or individual
First requirement: wanted a valid test
- Phonetic balancing scheme: individual phonemes occur as frequently on the list as they do in the language
- Based on published data of phonetic composition of 100,000 words
Second Requirement: monosyllabic in structure
- Measure individual’s ability to discriminate among fine elemetns of speech (multisyllabic words - too easy, nonsense syllables - too difficult)
Third requirement: lists of equal difficulty
- Concept of list equivalence
- Range of difficulty on each list
Fourth requirement: words should be in common usage
Describe the Harvard PAL PB50 monosyllabic word test.
- 20 lists, 50 words/list
- Never recorded at Harbard
- Later recording (very difficult test)
- Most widely used as MLV
- Carrier phrase: “Say the word….”
- Peak a word in carrier phrase at 0 VU
Describe the CID W-22 PB
Criteria:
- Monosyllables
- Highly familiar words
- Phonetic balance: based on printed and spoken English
Developed 4 lists, 50 words each
- Recordings with Hirsh as speaker
- List 1 was easier than Lists 2, 3, 4
- High test-retest reliability
Describe a psychometric functions.
- X axis is presentation level in dB SPL
- Y axis is the percentage of correct score
- Linear segment: steady or fixed increase in speech recognition score
- We’re interested in what level the listeners score 100%
- We’re interested in the linear segment
- PB Max: where the person reaches their maximum score on a speech recognition test
- Increases 5% per dB (steep rise)
How do the psychometric functions of different speech tests differ?
CID W-22
- Steeper slope
PAL PB 50s
- Has a gradual (shallow) slope (3-4% per dB)
What is the NU-6?
Northwestern University Test No. 6 (NU6)
- Tillman & Carhart, 1966
Goal for test development: list equivalence
Scheme of phonemic balancing
- Based on frequency of occurrence of phonemes within their set of words, not on English
Developed 4 lists; 50 words/list
- Recordings made (male & female talkers)
- Highly common monosyllabic words
- Looked at the frquency occurrence of each phoneme as it appears on the original list of words
- New recordings on CD made by VA with female talker
What is the psychometric function of the NU-6?
Normal hearing and conductive hearing losses
- Slope: 6% dB, 100% score at 24 dB SL re: SRT
People with sensorineural hearing loss
- Slope: 3.5%; plateau at 40 dB SL re: SRT
- Mean plateau was 92% correct
- Good inter-list equivalence
What are the effects of MLV vs. recorded materials on performance on PB wordlists?
1) MLV vs. Recorded Materials
Advantages of recorded materials
- Easier to administer (words peaking correctly)
- Standardized - can compare results between clincs
- More reliable
Advantages of MLV
- More flexible presentation
- Faster
Disadvantages of MLV - talker differences
- ASHA recommendations - recorded preferred
- Research findings - inconsistent
What is the performance-intensity function (PIF)?
- Only way to be sure that you’re measuring PB-max
- Limitations: takes time
- Compromise: estimate level needed for PB-max
What are the effects of presentation level on performance on PB word lists?
Presentation Level
Goal: often want to measure best performance
Procedures that are not recommended:
- MCL does not equal the level needed for PB-max
- Fixed presentation (sensation) level above SRT
Procedures that are recommended:
- Fixed SPL (95 dB SPL)
- Just below LDL
- Adaptive procedure to estimate level
What are the effects of word familiarity on performance on PB word lists?
Word Familiarity
- Individual’s lack of familiarity will affect some
What are the effects of half vs. whole lists on performance on the PB word lists?
Half vs. whole lists
Advantages of half lists
- Shorter test time
- May be sufficient for diagnostic purposes
- High intra-list reliability (1/2 list to whole list)
Disadvantages of half lists
- Sacrifice phonetic balance (lose face validity)
- Sacrifice inter-list equivalence
- Lower reliability than full lists
Decision based on purpose of testing
- Estimating everyday performance –> full list
- Diagnostic purposes –> half lists may be OK
What is the binomial probability theorem and how does it relate to clinical testing?
- The theorem shows that the standard deviation is inversely proportional to the square root of the number of test items (N)
Implications
- As N increases, standard deviation decreases
(results are more reliable with longer lists)
- As percent correct score varies:
Scores near 0 and 100% are more reliable
Scores near 50% are least reliable
Implications for clinical testing:
- Use longer lists when possible
- Use tables to indicate standard deviation of a given score to determine if a second score is significantly different from the first score
