Variability in Word Recognition by Adults with Cochlear Implants: The Role of Language Knowledge Aaron C. Moberly, M.D. CI2015 Washington, D.C.
Disclosures ASA and ASHFoundation Speech Science Research Grant Triological Society Career Development Award No conflicts of interest
Problem Substantial variability remains in speech recognition among CI users Much of this variability is not understood CIs deliver highly degraded representations of speech, especially in the spectral domain
Outcome Variability Variability is enormous and frustrating Likely relates to abilities within four areas: AS: Auditory Sensitivity PO: Perceptual Organization LS: Language Skills CF: Cognitive Factors
Auditory Sensitivity
Perceptual Organization
Language Skills
Cognitive Factors
Outcome Variability First, recognize what causes it. Then, specifically address the sources of variability. 10
Problem Speech scientists have long known that speech recognition requires more than auditory sensitivity to acoustic cues Listeners must also know what cues should be paid perceptual attention in order to recover sound units (phonemic structure)
Problem Postlingually deafened adults with CIs should have developed refined perceptual attention strategies CI users might shift their perceptual attention to more available cues This shift may not support the most effective speech recognition
Methods (1) Perceptual attention: ba - wa contrast Normal-hearing listeners heavily weight a spectral cue Auditory sensitivity: Discrimination testing of spectral versus amplitude changes Speech recognition: CID word list (% correct) ba wa
Methods (1) 21 postlingually deafened adult CI users Ages 18 to 62 years Results compared with 15 adult NH listeners
Results (1) Wide variability in perceptual attention to spectral and amplitude cues among CI users
Word recognition versus attention to spectral cue 100 90 Word recognition % correct 80 70 60 50 40 30 20 standardized β =.77 10 0 0 1 2 3 4 5 6 7 8 9 10 FRT weighting factor Weighting factor Spectral cue
Results (1) Auditory sensitivity did not independently predict word recognition
Methods (2) Perceptual attention: cob - cop NH listeners: duration cue sa - sha NH listeners: static spectral cue Auditory sensitivity: Discrimination testing of spectral and duration cues Speech recognition: CID word list (% correct) cob sha cop sa
Methods (2) 30 adult postlingually deafened CI users Ages 52 to 88 years 20 adult NH listeners
Word recognition versus attention to spectral cue Weighting factor Dynamic spectral cue (cob-cop)
Results (2) Word recognition was better for those with better spectral discrimination (80% versus 56.4%), p <.001). Word recognition was not predicted by sensitivity to duration cue. Auditory sensitivity did not predict perceptual attention to those cues.
Interim Discussion CI users who are sensitive to and perceptually attend to dynamic spectral structure have the best speech recognition Weighting of spectral but not duration or amplitude cues predicts word recognition
Phonemic Awareness (3) Understanding speech requires storing speech using phonological codes. Hearing loss diminishes access to acoustic structure supporting these codes. We examined abilities to access and manipulate phonological representations ( phonemic awareness skills) for CI users. Phonemic awareness examined as predictor of variability in word recognition. 23
Participants 30 adult postlingually deafened CI users Ages 52 to 88 years 20 adult NH listeners 24
Methods All participants underwent cognitive screen to rule out dementia (MMSE) Word recognition in quiet (CID word list) 3 phonemic awareness tasks presented audiovisually: Initial Consonant Choice (ICC) Final Consonant Choice (FCC) Backwards Words (BW) 25
Methods 3 phonemic awareness tasks presented audio-visually: Initial Consonant Choice (ICC) Final Consonant Choice (FCC) Backwards Words (BW) ACCESS 26
Methods 3 phonemic awareness tasks presented audio-visually: Initial Consonant Choice (ICC) Final Consonant Choice (FCC) Backwards Words (BW) PROCESSING 27
Results (3) Groups NH CI N Mean (SD) N Mean (SD) df t p Test Scores Word recognition (% correct) 19 97.1 (2.5) 28 66.5 (18.7) 45 8.76 <.001 ICC (% correct) 18 97.9 (2.7) 28 84.9 (19.7) 44 3.98 <.001 FCC (% correct) 19 86.8 (7.4) 27 64.4 (26.2) 44 3.60.001 BW (% correct) 19 66.6 (21.1) 28 58.0 (25.0) 45 1.16.253 28
29 Results
30 Results
Word recognition (percent correct) 100 90 80 70 60 50 40 30 20 10 0 β =.20 Results 0 10 20 30 40 50 60 70 80 90 100 BW (percent correct) 31
Discussion The ability to access phonemic structure was significantly degraded for postlingually deafened adult CI users compared with those with normal hearing The ability to process phonemic structure remained relatively intact The ability to access phonemic structure predicted 25 to 35% of the variance in word recognition for CI users. 32
Conclusions Future work to improve attention to spectral structure through CI device signal delivery, mapping strategies, auditory training Restoring phonemic awareness through intensive training Study semantic/syntactic/grammatical abilities to explain outcome variability 33
Outcome Variability First, recognize what causes it. Then, specifically address the sources of variability. 34
Outcome Variability First, recognize what causes it. Then, specifically address the sources of variability. 35
Acknowledgements Susan Nittrouer, Ph.D. Joanna Lowenstein, Ph.D. Amanda Caldwell-Tarr, Ph.D. Jamie Kuess Jessica Apsley Lauren Boyce Emily Hehl Jennifer Martin Demarcus Williams
Subject Age Gender Implant Processor Hearing aid? Age of Implant Side Tested Cause Hearing Loss 1 37 M Freedom Freedom Yes 30 Left Progressive as adult 2 40 M Nucleus 24 CP810 No 32 Left Menieres 3 30 F Nucleus 24 Freedom No 21 Right Congenital 4 31 M Freedom CP810 Bilateral CI 23 Left Progressive as child 5 29 M Freedom CP810 No 25 Right Progressive as child 6 29 M Advanced Bionics AB Harmony No 22 Right Ototoxicity as infant 7 32 F Nucleus 22 Freedom No 9.5 Left Meningitis and ototoxicity 8 29 F CI512 CP810 No 27 Right Meningitis and progressive as child 9 34 F Nucleus 24 CP810 No 25 Left Congenital progressive 10 18 F Cochlear Unknown Yes 16 Right Progressive as child 11 37 M Freedom Freedom Bilateral CI 33 Right Meningitis as adult 12 54 M Nucleus 24 CP810 Bilateral CI 48 Right Congenital progressive 13 47 F Nucleus 24 Freedom No 37 Left Congenital progressive 14 60 F Freedom Freedom Bilateral CI 54 Right Progressive as adult 15 46 M Nucleus 24 Freedom Bilateral CI 38 Left Progressive as adult 16 62 M CI512 CP810 No 60 Left Menieres 17 52 F Freedom Freedom Yes 48 Right Progressive as child 18 40 F Freedom Freedom No 33 Left Congenital progressive 19 57 F CI512 CP810 Bilateral CI 54 Right Congenital progressive 20 62 F Freedom CP810 Yes 62 Right Progressive as adult 21 62 F Freedom CP810 Yes 56 Right Progressive as adult
38
Groups NH CI Mean (SD) Mean (SD) t p N 19 28 Demographics Age (years) 63.7 (8.1) 66.7 (10.2) 1.07 0.29 SES (score) 32.8 (18.7) 25.3 (11.3) 1.71 0.09 Test Scores Reading (standard score) 104.1 (11.8) 99.6 (10.7) 1.30 0.20 Expressive vocabulary (standard score) 102.9 (19.6) 96.0 (15.3) 1.30 0.20 Cognitive MMSE (T score) 51.1 (8.2) 47.3 (8.5) 1.54 0.13 39