Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Temporal jitter mimics effects of aging on word identification and word recall in noise Brown, Sasha 2000

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2000-0636.pdf [ 8.23MB ]
Metadata
JSON: 831-1.0089735.json
JSON-LD: 831-1.0089735-ld.json
RDF/XML (Pretty): 831-1.0089735-rdf.xml
RDF/JSON: 831-1.0089735-rdf.json
Turtle: 831-1.0089735-turtle.txt
N-Triples: 831-1.0089735-rdf-ntriples.txt
Original Record: 831-1.0089735-source.json
Full Text
831-1.0089735-fulltext.txt
Citation
831-1.0089735.ris

Full Text

T E M P O R A L JITTER M I M I C S E F F E C T S OF A G I N G O N W O R D IDENTIFICATION A N D W O R D R E C A L L I N NOISE  by  SASHA BROWN  B . A . Simon Fraser University, 1997  A THESIS S U B M I T T E D I N P A R T I A L F U L F I L M E N T OF THE REQUIREMENTS FORTHE D E G R E E OF  M A S T E R OF SCIENCE in  T H E F A C U L T Y OF G R A D U A T E STUDIES (School o f Audiology and Speech Sciences)  W e accept this thesis as conforming to the required standard  UNIVERSITY OF BRITISH C O L U M B I A August, 2000 © Sasha Brown, 2000  In  presenting  degree freely  at  this  the  thesis  in  partial  fulfilment  of  University  of  British  Columbia,  I agree  available for  copying  of  department publication  this or of  reference  thesis by  this  for  his thesis  and  scholarly  or for  her  Department The University of British C o l u m b i a Vancouver, Canada  I further  purposes  gain  requirements that  agree  may  representatives.  financial  permission.  DE-6 (2/88)  study.  the  It  shall not  be is  that  the  Library  permission  granted  by  understood be  for  allowed  an  advanced  shall for  the that  without  make  it  extensive  head  of  my  copying  or  my  written  ABSTRACT  It is w e l l documented that older adults, even those with normal pure-tone thresholds, tend to have more difficulty perceiving speech and comprehending language in noise compared to younger adults (e.g., C H A B A , 1988). There is an increasing body o f literature suggesting that as people age, changes to the auditory system occur that decrease ability to process auditory temporal information, which is crucial to the analysis of the speech signal. Recently, the hypothesis has been put forward that the ageing auditory system becomes less able to code periodicity cues or synchrony, due to a disruption to the phase-locking capabilities o f the auditory neurons (Pass, 1998; PichoraFuller, Schneider, Pass & Brown, 2000). It has also been proposed that degradation caused by perceptual difficulties cascades into higher-level cognitive-processing problems. From an informationprocessing perspective, like that which guides and informs the studies in this thesis, a compromised signal requires more cognitive resources to help recover the signal, thereby leaving fewer resources for other processes, such as memory, which are necessary to language comprehension. The current experiments further investigated these hypotheses by creating an . asynchrony, like that believed to occur in the aged auditory system, and applying it to the high- and low-context S P I N - R sentences, which were presented concurrently with babble at different signal-tp-noise ratios (S/N). These jittered signals were then presented to young listeners with normal hearing to determine i f the asynchrony resulted in wordidentification and word-recall performance similar to that o f older adults who had  participated in an earlier study (Pichora-Fuller, Schneider, & Daneman, 1995). It was found that when presented with this jittered speech, the word identification performance o f young listeners hearing jittered speech almost perfectly matched the performance o f older listeners when sentence context was low; however, the young failed to demonstrate the extent o f benefit from high context that had been found for o l d listeners. Such differences in context effects were not apparent when recall was measured. It is suggested that the externally applied asynchrony resembles the asynchronous neural firings thought to characterize the auditory systems o f older adults, and that memory difficulties on auditory tasks may result from an interaction between age-related perceptual and cognitive decrements.  . ' '  iv  '  . .  T A B L E OF CONTENTS  ABSTRACT...,......:,.......  •  •.:  '.  :  ii  T A B L E OF CONTENTS  iv  LIST OF T A B L E S . . . , . . . . . ; . . . : . . . . . : LIST OF FIGURES.,  ...xiii :  :  xv  ACKNOWLEDGEMENTS  xviii  1. L I T E R A T U R E R E V I E W  1  1.1  Introduction;...;  1.2  Older Adults and Speech Perception  1.3  Speech Perception iri Poor Environments. 1.2.1  1.3  .',  1 3  •  4  Speech Perception with Compressed Speech...'.  5  Possible Contributors to Age-related Declines i n Speech . Comprehension  '.  6  1.3.1  Perceptual factors  6  1.3.2  Cognitive Factors  8  1.3.2.1 Rate o f Processing  8  1.3.2.2 M e m o r y C a p a c i t y . . . . . . . .  9  1.3.3  A h Interaction Between Perceptual and Cognitive Factors  10  1.3.3.1 Correlational Evidence o f an Interaction between Perception and Cognition  ...10  1.3.3.2 Possible Relationships Between Perceptual and Cognitive Declines with A g i n g . . . . .  12  v  1.3.3.2.1 Sensory Deprivation Hypothesis 1.3.3.2.2 Cognitive Declines  '. •;  13  1.3.3.2.3 Common Cause Hypothesis:.  .13  1.3.3.2.4 Perceptual Degradation.....  14  1.3.3.3 A Synthesis o f Perceptual and Cognitive Research 1.4  W o r k i n g Memory/Information Processing 1.4.1  15 16  ....  .16  1.4.2. W o r k i n g M e m o r y and Language Comprehension  18  1!4.3. Assessing W o r k i n g M e m o r y  19  1.4.4  20  1.4.5 1.5  The M o d e l . . . . . . . . '  12  W o r k i n g Memory and A g i n g , The Role o f Perception in W o r k i n g M e m o r y .  21  Temporal Cues and Speech Comprehension  26  1.5.1  Consequences o f Temporal Alterations  26  1.5.2  Measuring Temporal Processing A b i l i t y  31  1.5.2.1 Psychoacoustic Measures o f Temporal Processing  31  1.5.2.1.1 Frequency Discrimination  '.  31  1.5.2.1.2 Duration Discrimination  ':  33  1.5.2.1.3 Gap detection  34  1.5.2.1.4 Complex Stimuli  36  1.5.2.1.5 Binaural Processing  ..'  ....38  1.5.2.1.5.1 Localization  38  1.5.2.1.5.2 Binaural Unmasking  39  1.5.2.2 Speech Perception and Psychoacoustic Measures o f Temporal Processing  41  T.5.2.3 A Summary o f the Age-related Psychoacoustic Findings 1.6  ,.:  Anatomy and Physiology Related to Temporal Processing....  • 1.7  1.6.1  Temporal Precision in the Auditory System  1.6.2  A g i n g and Temporal Resolution.  Jitter  ;  1.8  Summary.:  1.9  Hypotheses  50 .,...;..:.  2.2  Experiment 1: W o r d Identification  43  .,.49  :  Objectives...:  :  47  .•  2.1  .....43  45  :  2. M E T H O D S  .53 53  ;  .,  53  2.2.1  Participants ih Experiment 1  53  2.2.2  Materials for Experiment 1  54  2.2.2.1 Preparation o f the Stimuli  54  '  •2.3  42  2.2.2.1.1  S P I N - R Sentences!  2.2.2.1.2  The Jitter....!  2.2.3  Calibration o f the Equipment for Experiment 1  2.2.4  Apparatus and Physical Setting for Experiment 1  2.2.5  Procedure for Experiment 1  54 ,55 59 ........61 61  Experiment 2: W o r d Recall  62  2.3.1  62  Participants in Experiment 2  2.3.2  .  Materials for Experiment 2  .63  2.3.2:1 Stimuli Rationale  63  2.3.2.1.1  Signal-to-Noise Ratios  63  2.3.2.1.2  Recall Set-Size..  64  2:3.3  Calibration o f the Equipment for Experiment 2  2.3.4  Apparatus and Physical Setting.for Experiment 2  2.3.5  Procedure for Experiment 2....  3. R E S U L T S . . .  :.  64 '  65 65  '  1....68  3.1 .  Introduction  ....:  ..68  3.2  Experiment 1: W o r d Identification.  68  3.2.1  Types o f Errors Made by Participants...........  68  3.2.2  Effect o f Jitter on Sentence-Final Word-Identification Scores. ...69  3.2.3 • Effect o f S / N on Sentence-Final Word-Identification Scores • 3.2.4  Effect o f Context on Sentence-Final Word-Identification Scores  ,  ..  3.3  70  71  3.2.5  Interaction Effect o f Jitter and S / N Conditions  3.2.6  Interaction Effect o f Jitter and Context Conditions  75  3.2.7  Interaction Effect o f S/N and Context Conditions  76  3.2.8  Interaction Effect o f Jitter, S / N and Context Conditions  78  3.2.9  Summary o f Results from Experiment 1  79  Experiment 2: W o r d Recall. 3.3.1  Predictability Judgements  3.3.2  Sentence-Final W o r d Identification  . . . . . : . . ..72  79 '.  80 80  via  •  3.3.2.1 Types o f Errors Made by Participants.  :  81  3.3.2.2 Effect o f Jitter on Sentence-Final Word-Identification Scores  ........81  3.3.2.3 Effect o f S/N on Sentence-Final Word-Identification Scores...:....!  82  3.3.2.4 Effect o f Recall Set-Size on Sentence-Final W o r d Identification Scores....,  83  3.3.2.5 Effect o f Context on Sentence-Final Word-Identification Scores..  :  84  3.3.2.6 Interaction Effect o f Jitter and S / N on Sentence-Final Word-Identification Scores  84  3.3.2.7 Interaction Effect o f Jitter and Recall Set-Size on Sentence-Final Word-Identification Scores  86  3.3.2.8 Interaction Effect o f Jitter and Context on SentenceFinal Word-Identification Scores  ,  86  3:3.2.9 Interaction Effect o f S / N and Recall Set-Size on SentenceFinal Word-Identification Scores  88  3.3.2.10 Interaction Effect o f S / N and Context on Sentence-Final Word-Identification Scores  88  3.3.2.11 Interaction Effect o f Recall Set-Size and Context on Sentence-Final Word-Identification Scores 3.3.2.12 Three- and Four-Way Interactions  90 91  IX  3.3.2.13 Summary o f Effects on Sentence-Final W o r d ,Identification Scores in Experiment 2. 3.3.3  91  Comparison o f Sentence-Final Word-Identification Scores from Experiment 1 and Experiment 2  93  3.3.3.1 M a i n Effects 3.3.3.2 Interactions...  :  .93  :  93  3.3.3.3 Summary o f the Comparison o f Sentence-Final W o r d Identification Scores from Experiment 1 and Experiment 2 3.3.4  Sentence-Final Word-Recall Scores.;  ..........94 !  ...95  3.3.4.1 Effect o f Jitter on Sentence-Final Word-Recall Scores.. ..95 3.3.4.2 Effect o f S / N on Sentence-Final Word-Recall Scores  96  3.3.4.3 Effect o f Recall Set-Size on Sentence-Final Word-Recall Scores '  :  97  3.3.4.4 Effect o f Context on Sentence-Final Word-Recall Scores  98  3.3.4.5 Interaction Effect o f Jitter and S / N on Sentence-Final Word-Recall Scores.  99  3.3.4:6 Interaction Effect o f Jitter and Recall Set-Size on SentenceFinal Word-Recall Scores  100  3.3.4.7 Interaction Effect o f Jitter and Context on Sentence-Final Word-Recall Scores  101  3.3.4.8 Interaction Effect o f S / N Recall Set-Size on Sentence-Final Word-Recall Scores .-.,  102  3.3.4.9 Interaction Effect o f S / N and Context on Sentence-Final Word-Recall Scores  ...103  3.3.4.10 Interaction o f Recall Set-Size and Context on SentenceFinal Word-Recall Scores  103  3.3.4.11 Three-and Four-Way,Interactions..:  :  ...105  3.3.4.12 Summary o f Sentence-Final Word-Recall Performance  '.  .105  4. D I S C U S S I O N O F R E S U L T S I N T E R M S O F H Y P O T H E S I S A N D C O M P A R I S O N T O PREVIOUS STUDIES.. 4.1  Introduction  4.2  Word-Identification."  4.3  107 ;  107 107  4.2.1  Word-Identification with Intact Sentences  108  4.2.2  Word-Identification with Jittered Sentences  109  Word-Recall.  .......  4.3.1  Word-Recall with Intact Sentences  4.3.2  Word-Recall with Jittered Sentences  5. G E N E R A L D I S C U S S I O N  112 113 :  ....114 ,116  5.1  Introduction..'  116  5.2  Word-Identification  5.3  Word-Recall  118  5.4  Final Discussion.  120  ..116  xi  5.5  Future D i r e c t i o n s . . . . . . . . . . . . .  122  REFERENCES  124  A P P E N D I X A : Experiment 1 Participants' Air-Conducted Pure-Tone Thresholds (dB H L ) for Right (R) and Left (L) Ears  134  A P P E N D I X B : Experiment 1 Participants' Characteristics  135  A P P E N D I X C : Forms o f the Revised S P I N Sentences  136  A P P E N D I X D : Spectrograms and Time-Amplitude Waveforms o f Jittered and Unjittered Speech Stimuli.  : . . . . 158  A P P E N D I X E : Connections on the Tucker Davis Technologies Modules  '....  165  A P P E N D I X F : Purpose o f Each T D T Module  166  A P P E N D I X G : Instructions for Experiment 1  : . . . . 168  A P P E N D I X H : Order o f Conditions and S P I N - R Forms for each Participant in Experiment 1  '.  ,  169  A P P E N D I X I: Experiment 2 Participants' Pure-tone Thresholds for Right (R) and Left (L) Ears  •.  A P P E N D I X J: Experiment 2 Participants' Characteristics  170 '.  A P P E N D I X K : Instructions for Experiment 2...  '.All 172  A P P E N D I X L : Order o f Conditions and S P I N - R Forms for each Participant i n Experiment 2 A P P E N D I X M : Experiment 1 W o r d Identification R a w Scores (out o f 25)  173 174  A P P E N D I X N : M e a n Percent-Correct Sentence-Final Word-Identification Scores and Standard Deviations for Each Condition i n Experiment 1 A P P E N D I X O: Predictability Judgements for Experiment 2 (out o f 50)  175 176  Xll A P P E N D I X P : . Experiment 2 W o r d Identification R a w Scores (out o f 25)  :  177  A P P E N D I X Q : M e a n Percent-Correct Sentence-Final Word-Identification Scores and Standard Deviations for the Possible Combinations o f Variables in Experiment 2 A P P E N D I X R : Experiment 2 W o r d Recall Raw Scores (out o f 25)  178 183  A P P E N D I X S: M e a n Percent-Correct Sentence-Final Word-Recall Scores and Standard Deviations for the Possible Combinations o f Variables i n Experiment 2  ...184  X1U  LIST OF T A B L E S  Table 1. M e a n percent-correct sentence-final word-identification scores and standard deviations for each Jitter X S / N condition  73  Table 2. M e a n percent-correct sentence-final word-identification score and standard deviations for each Jitter X Context condition.  76  Table 3. M e a n percent-correct sentence-final word-identification score and standard deviations for each S / N and Context condition  77  Table 4. M e a n percent-correct sentence-final word predictability score for each Jitter, S / N and Recall-Set Size condition '..  80  Table 5. M e a n percent-correct sentence-final word-identification score and standard deviations for each Jitter X S / N condition. '. ; :  85  Table 6. M e a n percent-correct sentence-final word-identification score and standard deviations for each Jitter X Recall Set-Size condition  86  Table 7. M e a n percent-correct sentence-final word-identification score and standard deviations for each Jitter X Context condition  , 87  Table 8. M e a n percent-correct sentence-final word-identification score and standard deviations for each S / N X Recall Set-Size condition  88  Table 9. M e a n percent-correct sentence-final word-identification score and standard deviations for each S / N X Context condition .90 Table 10. M e a n percent-correct sentence-final word-identification score and standard deviations for each Recall Set-Size X Context condition 90 Table 11. M e a n percent-correct sentence-final word-recall scores and standard deviations for each Jitter X S / N condition  100  Table 12. M e a n percent-correct sentence-final word-recall scores and standard deviations for each Jitter X Recall Set Size condition  100  Table 13. M e a n percent-correct sentence-final word-recall scores and standard deviations for each Jitter X Context condition.  102  Table 14. M e a n percent-correct sentence-final word-recall scores and standard deviations for each S / N X Recall Set-Size condition.  102  Table 15. M e a n percent-correct sentence-final word-recall scores and standard deviations for each S / N X Context condition  103  XIV  Table 16. M e a n percent-correct sentence-final word-recall scores and standard deviations for each Recall Set-Size X Context condition  104  XV  LIST OF FIGURES Figure 1 Temporal and Spectral Representation o f Gaussian Noise  56  Figure 2. Pure-Tones Jittered Using Three Different Jitter Parameters  58  Figure 3. M e a n Percent-Correct Word-Identification Scores (± 1 SD) in the Intact and Jittered Condition •. .  70  Figure 4. M e a n Percent-Correct Word-Identification Scores (± 1 SD) for Each Signalto-Noise Ratio Condition 71 Figure 5. M e a n Percent-Correct Word-Identification Scores ( ± 1 . SD) for Each Context Condition •.  72  Figure 6. M e a n Percent-Correct Word-Identification Scores for Each Jitter Condition in Each Signal-to-Noise Ratio.; 74 Figure 7. M e a n Percent Difference Between Word-Identification Scores i n the Intact and Jittered Conditions at Each Signal-to-Noise Ratio  74  Figure 8. M e a n Percent Difference Between Word-Identification Scores in the H i g h and Low-Context Conditions in Each Jitter Condition 75 Figure 9. M e a n Percent-Correct Word-Identification Scores for Each Context Condition at Each Signal-to-Noise Ratio  77  Figure 10. M e a n Percent Difference Between Word-Identification Scores in H i g h and Low-Context Conditions at Each Signal-to-Noise Ratio  78  Figure 11. M e a n Percent-Correct Word-Identification Scores for Each Condition at Each Signal-to-Noise Ratio  79  Figure 12. M e a n Percent-Correct Word-Identification Score (± 1 SD) i n Each Jitter Condition  82  Figure 13. M e a n Percent-Correct Word-Identification Scores (± 1 SD) in Each Signalto-Noise Ratio Condition 83 Figure 14. M e a n Percent-Correct Word-Identification Scores (± 1 SD) in Each Context Condition  84  Figure 15. M e a n Percent Difference Between Word-Identification Scores Obtained at +8 and +4 d B Signal-to-Noise Ration in Each Jitter Condition 85  xvi Figure 16. M e a n Percent Difference Between Word-Identification Scores in H i g h and Low-Context Conditions for Each Jitter Condition  87  Figure 17. M e a n Percent Difference Between H i g h - and Low-Context W o r d ,Identification Scores in Each Signal-to-Noise Ratio Condition  89  Figure 18. M e a n Percent-Correct Word-Identification Scores when Recall Set-Size=2...92 Figure 19. M e a n percent-Correct Word-Identification Scores when Recall Set-Size=8...92 Figure 20. Comparison o f M e a n Percent-Correct Word-Identification Scores for Experiment 1 and Experiment 2 with Intact Stimuli  94  Figure 21. Comparison o f M e a n Percent-Correct Word-Identification Scores for Experiment 1 and Experiment 2 with Jittered Stimuli.  ..95  Figure 22. M e a n Percent-Correct Word-Recall Scores (± 1 SD) for Each Jitter Condition  96  Figure 23. M e a n Percent-Correct Word-Recall Scores (± 1 SD) for Each Signal-toNoise Ratio Condition  97  Figure 24. M e a n Percent-Correct Word-Recall Scores (+ 1 S D ) for E a c h R e c a l l SetSize Condition  98  Figure 25. M e a n Percent-Correct Word-Recall Scores (± 1 SD) for Each Context Condition  99  Figure 26. M e a n Percent Difference Between Word-Recall Scores for Recall SetSizes o f 2 and 8 in Each Jitter Condition  101  Figure 27. M e a n Percent Difference Between High- and Low-Context Word-Recall Scores for Each Recall Set-Size  104  Figure 28. Number o f Words Recalled for Each Recall Set-Size in a Signal-to-Noise Ratio of+8 106 Figure 29. Number o f Words Recalled for Each Recall Set-Size in a Signal-to-Noise Ratio of+4 106 Figure 30. M e a n Percent-Correct Word-Identification Scores Comparing Y o u n g Participants with Intact Stimuli in Four Experiments  108  Figure 31. M e a n Percent-Correct Word-Identification Scores for Y o u n g Participants with Jittered Stimuli and O l d Participants with Intact Stimuli 110  XVll Figure 32. Number o f Words Recalled for Y o u n g Participants with Intact Stimuli, S / N = 4/5 ....113 Figure 33. Number o f Words Recalled for each Recall Set Size: Y o u n g Participants with Jittered Stimuli at S/N =4 d B , and O l d Participants with Intact Stimuli atS/N = 5dB..... < . .... '.,  114  xviii  ACKNOWLEDGEMENTS  . There are many for whom this acknowledgement is inadequate recompense. It goes without saying that this thesis could not have been completed without the assistance of Kathy Pichora-Fuller who is deserving o f much gratitude. Thanks also to Jeff Small, Carol Jaeger and Bruce Schneider for their, expertise, and advice that made this research possible. A l s o crucial were the solid collaborative team in the lab, without whom the 'thesis-experience' would have been exceedingly difficult. M a n y thanks to Trudy A d a m , Anne-Marie Roberts, Suzanne Macdonald and Terence Miranda. Appreciation must also be given to the faculty, who provided a solid knowledge base and were available for assistance, and the staff, particularly Sue Bryant, who made arriving at the school pleasurable. Most importantly, the love and support provided by family and friends, who were ignored at times and subjected to complaints at others, was unconditional and deserving o f accolades. Special thanks go to Ryan Parsons, who was particularly helpful in the frenzy o f the final moments. , This research was supported by a grant from the Natural Sciences and Engineering Research Council o f Canada.  1 1. L I T E R A T U R E R E V I E W  1.1  ,  Introduction  Older adults experience more difficulty perceiving speech in everyday listening situations than do younger adults (Bergman, 1980; C H A B A , 1988; Willot, 1991), even though under ideal quiet listening conditions this may not be apparent. W h e n the speech material is presented in a reverberant environment (e.g., Duquesnoy & Plomp, 1980; Nabelek & Robinson, 1982), or with competing noise (e.g., Duquesnoy, 1983) older adults have more trouble recognizing words. I f the speech signal itself is compromised, because, for instance, its intensity is below the most comfortable listening level ( M C L ) (Gelfand, Piper^ & Silman, 1985), or because the rate has been altered (e.g., Frisina & Frisina, 1997; Gordon-Salant & Fitzgibbons, 1995; Stuart & Phillips, 1996) the performance o f older adults with normal hearing also declines relative to younger adults. Because speech is crucial to most social situations, this problem is significant, and has thus been the focus o f a great deal o f research from numerous perspectives. A s people age, both cognitive and perceptual abilities seem to decline, each o f which could account in part for the observed language comprehension difficulties. Perceptually, the most apparent relevant deficit is the onset o f age-related hearing loss, or presbycusis. While some o f the difficulties can i n part be explained by the cochlear pathology associated with presbycusis, even those with relatively normal pure-tone thresholds experience difficulty when the listening situation becomes more challenging, suggesting that other factors play a role; other aspects o f auditory processing  could be affected by age and contribute to the noted difficulties. A n important line o f research has focused on the auditory system's ability to resolve temporal information. Spoken language by nature is an act that occurs over time; the perception o f speech necessarily requires the encoding o f temporal information. There is evidence that the auditory system is particularly adept at transmitting precise temporal cues, and that these cues can be used for example, to help in.distinguishing a desired signal from background noise (e.g., Pichora-Fuller & Schneider, 1991; 1992; 1998), in discriminating between frequencies (e.g., Hartmann, 1996), and in deciphering the small gaps that serve to differentiate some phonemes (e.g., Haubert, 1999; Strouse, Ashmead, Ohde, & Grantham, 1998). I f the integrity o f the temporal information were compromised, speech perception would also be jeopardized. Older adults tend to have more difficulty encoding temporal cues, such as those assessed by psychoacoustic tasks (e.g., A b e l , Krever & Alberti, 1990; Divenyi, 2000; Fitzgibbons and Gbrdon-Salant, 1995; 1996). It has been proposed that the neural firings necessary to precisely encode fine structure temporal information may become less synchronous as people age (e.g., Pichora-Fuller & Schneider, 1992; Pichora-Fuller, Schneider, Pass & Brown, submitted; Schneider, 1997). Physiological data also supports this contention (e.g., Boettcher, M i l l s & Norton, 1993; Frisina, 2000; Hellstrom & Schmiedt, 1990; Walton, 2000). Another line o f research has focused on the cognitive declines noted as people age and the effects these could have on language comprehension. Older adults have both a general cognitive slowing (e.g., Cerella, 1985; Salthouse, 1982) and problems retaining information (e.g., Poon, 1985; Verhaghan, Marcoen, & Goosens, 1993), both o f which  could contribute to poor language comprehension. Because there are numerous factors, it is difficult to tease out the contributions o f perception and cognition, and it is more likely that they interact and influence each other. , , , The research in the current paper takes an information-processing perspective and looks at one way in which the perceptual and cognitive systems could interact. Specifically, i f auditory processing is impaired by a reduced ability to use temporal cues, structures higher up the system receive faulty information. Top-down processes could be employed to help rescue and interpret the degraded signal, but such effort would reduce resources necessary to other processing operations. The ability to hold items i n memory, for example, may be reduced. Because language comprehension requires storing information for short periods o f time in order to integrate the items o f an entire utterance, a decrease i n this function would impair language comprehension. The following discussion w i l l review the literature relevant to both the cognitive and perceptual skills necessary to understand speech, the ways in which they interact, and the effect o f aging on both.  1.2  . Older Adults and Speech Perception A s noted, older adults tend to have a more difficult time understanding speech  when the signal is compromised. This can occur because the environment is not conducive to listening, or because the signal itself has been degraded. The following section w i l l first discuss some o f the research focussing on the contribution o f poor environments. The repercussions o f degrading the signal by increasing the rate o f presentation w i l l then be discussed.  * 1.2.1  '•  Speech Perception in Poor Environments  4 '  It has been well documented that although older adults perform similarly to younger adults i n quiet, differences emerge in noisy or reverberant environments (e.g., Dubno, Dirks, & Morgan, 1984; Duquesnoy & Plomp, 1980; Frisina & Frisina, 1997; Nabelek & Robinson, 1982; Snell & Frisina, 2000), Pichora-Fuller, Schneider and Daneman (1995) had normal-hearing young, relatively normal-hearing old, and.. presbyacusic listeners perform the revised Speech Perception in Noise (SPIN-R) test (Bilger, Nuetzel, Rabinowitz, & Rzeczkowski, 1984), which consists o f eight equivalent lists o f 50 sentences with accompanying multi-talker babble background noise. Participants are required to repeat the last word o f each sentence, half o f which are low context (They heard I called about the PET.) and half o f which are high context (My son has a dog for a PET.) at differing signal-to-noise ratios (S/N). It was found that in noise, word identification ability was worse in the older listeners, even those with near-normal hearing. These results are similar to those noted by Dubno et al. (1984) when they used an earlier version o f the S P I N test (Kalikow, Stevens & Elliot, 1977) with younger and older adults, both with and without normal word-identification in quiet. Pichora-Fuller et al. (1995) also, found that the age-related differences were not the result o f cognitive deficiencies, as the older adults were able to use contextual cues to aid in word recognition, a finding substantiated by Frisina and Frisina (1997). Stuart and Phillips (1996) conducted a study in which young and o l d participants were required to recognize monosyallabic words when continuous or interrupted noise was present. Again, no age-related difference was noted for the quiet condition. In the noise conditions, older adults performed significantly worse than younger adults, even  5  when the younger adults performed with a simulated high-frequency hearing loss. This suggests that the elderly's decreased word recognition performance in the presence o f noise does not necessarily result from concomitant high-frequency hearing losses.  1.2.2  .  i Speech Perception with Compressed Speech  . It has been found that increased rate o f speech is detrimental to older listeners attempting to comprehend spoken paragraphs (Schmitt and Carroll, 1985; Schmitt and Moore, 1989) and that this effect does not necessarily result from increased pure-tone sensitivity (LetOwski & Poch, 1996). Older adults also prefer to listen to slower speech than younger adults. When given control over the rate o f presented speech, older adults selected speech that was approximately 15% slower than that chosen by younger adults (Riensche, Lawson, Beasley, & Smith, 1979), a tendency that has been more recently confirmed by Wingfield and Ducharme (1999). Wingfield, Tun, K o h and Rosen (1999) found that normal-hearing older  •  participants (61-80 years; pure-tone average: 17.1 d B H L ) performed worse on timecompressed material. They also found, however, that when the original time was restored by inserting pauses, young adults performed at the same level as they did with unaltered speech, while the older adults still demonstrated decrements. The researchers concluded that the difficulties likely remain because o f the temporal distortion present i n the speech signal. It was proposed that older adults (with normal hearing) have more difficulty processing such degraded signals, causing perceptual stress that ultimately affects comprehension.  6  It is thus apparent that, independent o f audiome'tric pure-tone hearing loss, as people age, speech perception becomes more difficult, particularly as the signal is degraded or the listening environment is less than ideal. The next section w i l l review the underlying factors, most commonly believed to contribute to this age-related decrement i n speech comprehension.  1.3  Possible Contributors to Age-related Declines in Language  Comprehension. It is widely noted that as people age, perceptual acuity declines, as does performance on tasks designed to tap cognitive processes. These factors could account for the observed difficulties understanding spoken language. Cognitive and perceptual abilities w i l l now be outlined as they pertain to speech comprehension.  1.3.1  ' •',  Perceptual Factors  ,  •.  .  W h i l e it has been well documented that audiometric hearing loss increases with age (for a review see C H A B A , 1988), the ability to perceive degraded speech begins to decline i n the fourth decade, before audiometric hearing loss becomes clinically significant (Bergman, 1980). Although presbycusis does interfere with speech perception in that high-frequency audiometric thresholds are correlated with performance on a number o f speech tests (van Rooij, Plomp & Orlebeke, 1989), elevated pure-tone thresholds alone cannot account for the difficulty experienced by the elderly (e.g., Plomp & M i m p e n , 1979; Willot, 1991).  A number o f studies have also demonstrated that elderly listeners who perform similarly i n quiet, often vary considerably when listening to speech in noise (e.g., Duquesnoy, 1983; Plomp, 1986). This further indicates that hearing thresholds obtained in quiet laboratory conditions do not reveal the entire picture in aging listeners. It is thus surmised that perceptual deficits not identified by pure-tone audiometry exist and affect speech perception, particularly in adverse listening conditions. , M u c h research has been conducted to address this issue, with some exploring the contribution o f widened critical bands associated with cochlear hearing loss (e.g., Glasberg & Moore, 1986; Patterson & Moore, 1986; Stuart & Phillips, 1996), and some positing temporal processing difficulties resulting from more central auditory pathways (Divenyi, 2000; Stuart & Phillips, 1996). Several studies have found a relationship between reduced temporal analysis and measures o f speech perception (e.g., Dreschler & Plomp, 1985; Tyler, Summerfield, W o o d and Fernandes, 1982). These findings, combined with those that indicate an age-related deficit in the use o f temporal cues i n speech perception (e.g.:, Price & Simon, 1984), and those that correlate subjective measures o f hearing handicap with reduced temporal processing abilities (e.g. C h m i e l & Jerger, 1996) suggest that temporal resolution may play an important role in speech perception by aged listeners. This aspect o f auditory processing w i l l be further elaborated in later sections. It should be noted however, that i f auditory deficits, unrelated to pure-tone thresholds, affect speech perception, even when cognitive researchers attempt to control for hearing impairments based upon this measure alone, auditory deficits per se have not necessarily been equalized.  1.3.2  Cognitive Factors Research on aging and cognition tend to focus on two main themes. One is the  general slowing or reduction in the speed at which information can be processed (e.g., Cerella; 1985; Salthouse, 1982), and the other is a'diminished capacity to encode and recall information (for a review see Poon, 1985). Both o f these aspects o f cognition could have a profound effect on language comprehension and likely co-exist i n the elderly listener.  1.3.2.1  Rate o f Processing Age-related behavioural slowing has been demonstrated i n many studies  employing a wide range o f procedures (Cerella, 1985; Salthouse, 1985, 1996). Reaction time decreases with age; this is evident even in laboratory experiments involving a button-pressing response to a light or tone (BotwinnickJ 1971, as cited i n Cohen, 1987), but is more obvious when the tasks involve more complex processing : Elderly listeners 1  are thus disadvantaged when tasks impose increased processing demands or a heavier memory load.  .  A s noted i n ah earlier section, elderly listeners also have more difficulty when the speech rate is fast. Processing rate is crucial for the comprehension o f spoken language because the information presented is transient. I f processing is too slow, the listener falls behind and comprehension is compromised.  When, for example, participants are presented with two symbols with a given spatial relationship (i.e. + *), and a true/false statement (the cross is not to the left of the asterix), a 70 year old typically takes 30% more time than a 25-year old to respond (false) (Cohen & Faulkner, 1983). 1  9 1.3.2.2  M e m o r y Capacity Elderly people commonly complain that they have difficulties encoding and  recalling information. Cognitive researchers distinguish a number o f different memory systems (for reviews, see Craik & Jennings, 1992; Zacks, Hasher, & L i , 2000). The memory system receiving the most attention in regards to both language comprehension and agjng, is working memory, conceived o f as a construct necessary for the controlled processing o f information (van Rooij et al., 1989). The number o f items that can be held in working memory and recalled correctly is reduced as people age (see for example, Verhaeghah et al.,-1993). A s was noted with speed o f processing, the age difference is usually more pronounced as task complexity increases. In an experiment requiring participants to do mental arithmetic problems for example, young participants managed to operate with the same efficiency until the memory load exceeded five digits, while elderly participants' performance declined sharply when the load reached three digits (Wright, 1981). This form o f memory is important to language comprehension because parts o f the message must be held in mind while operations such as syntactic parsing, assigning referents to pronouns, and assigning meaning to ambiguous words are performed. I f working memory were compromised, it would likely be more difficult to integrate earlier parts o f a sentence with later parts, or within and across paragraphs over discourse. Thus, while both processing rate and working memory capacity can account in part for age-related declines in cognitive performance, there are indications that working memory becomes a better predictor for cognitive difficulties as the memory task becomes more demanding (Park, Smith, Lautenschlager, Earles, Frieske, Zwahr, & Gaines, 1996).  10 1.3.3  A n Interaction Between Perceptual and Cognitive Factors . It is clear that both auditory and cognitive processing ability decline with age;  what is perhaps less clear is the relationship between these two systems. Traditionally, perception and cognition have been treated as two separate and unrelated capacities, with perception researchers attempting to design their studies so as tb minimize the contribution o f cognitive factors, and most cognition researchers assuming perceptual acuity plays little part in their experiments. However, as Schneider and Pichora-Fuller (2000) point out, these assumptions may be valid when the listeners are young, but when the listeners are elderly, the boundaries between these two domains may be less defined. Investigators have recently become more interested in the interaction between perception and cognition, particularly in the aging population.  1.3.3.1  Correlational Evidence o f an Interaction between Perception and  Cognition A n interesting body o f research comes from the multidisciplinary Berlin A g i n g Study ( B A S E ; 1988-1993). Lindenberger and Baltes (1994) analysed the data collected from a large (N=156) heterogeneous sample o f adults ranging from 70 to 103 years (mean age: 84.9 years), stratified in six 5-year age groups (13 males and 13 females in each). Fourteen cognitive tests were conducted to measure the five cognitive abilities o f speed, reasoning, memory, knowledge, and fluency (for details on the tests, see Lindenberger and Baltes, 1994). The first three tests were assumed to reflect the fluid domain o f intelligence, the knowledge-free "mechanics". The latter two were assumed to reflect the crystallized domain o f intelligence, the knowledge-saturated "pragmatics".  , 1 1  V i s u a l acuity was measured at two different distances using two different standard reading tables, and pure-tone audiometric thresholds were obtained from 250 H z to 8000 Hz. In this exploratory analysis, the researchers found that together, vision and hearing accounted for 49.2% o f the total and 93.1% o f the age-related variance. The connection between sensory functioning and intelligence was so strong that the structural model providing the best fit was one in which auditory and visual acuity completely mediates age effects and intelligence.  <  M o r e recently, the same researchers (Baltes, & Lindenberger, 1997) performed a similar analysis on a larger sample, and extended it to include young adults in order to compare the findings across a larger age span. They collected the same sensory and intellectual measures for 516 people from 70 to 103 years (mean age = 84.9 years), and" 171 people from 25 to 69 years (mean age = 48.2 years). The findings for the older group were confirmed, while as predicted, the relationship between sensory and cognitive functioning was much lower for the younger group. Across the entire age range, however, perceptual acuity was an excellent predictor o f age differences i n cognitive. functioning. Moreover, while speed had previously been considered to mediate age and intelligence (Lindenberger, Mayr, & K l i e g l , 1993; Salthouse, 1993, 1996), in this study speed could not account for all o f the age related variance in vision and hearing, while these perceptual abilities were found to mediate the age-related variance in speed and other cognitive domains o f intelligence.  12  1.3.3.2 •  Possible Relationships Between Perceptual and Cognitive Decline with  Aging It is apparent that the role o f perceptual acuity cannot be assumed to be minimal in studies o f aging and cognition. It is however, less obvious how perception and cognition interact. Four possible relationships derived from Pichora-Fuller and Schneider (2000) w i l l now be discussed.  1.3.3.2.1  Sensory Deprivation Hypothesis  Age-related changes in cognition may co-occur with changes in sensory functioning because the impaired perceptual abilities result in a lack o f adequate sensory input. Prolonged "sensory underload" may reduce the opportunities for intellectually stimulating environmental interactions, resulting in permanent deterioration o f more central functions. This "sensory-deprivation" hypothesis assumes that age leads to sensory organ degeneration which reduces the quality o f information delivered more centrally, which in turn results in atrophy. Age-related sensory cell loss, however, does not inevitably lead to neural loss (e.g., Willot, 1991); thus weakening this hypothesis. Further evidence against this hypothesis is provided by Owsley, Berry, Sloane, Stalvey, and Wells (1998, as cited in Schneider & Pichora-Fuller, 2000) who demonstrated that following cataract removal surgery, performance on a number o f cognitive tests improves, thereby suggesting that the higher level structures remained intact despite reduced perceptual input.  13 1.3.3.2.2  Cognitive Declines  Because the cognitive system exerts top-down control over the sensory system, it is possible that a decrement in cognitive performance would affect its ability to properly direct the perceptual system. Attentional difficulties for example, could affect performance i n measures o f sensory acuity.' I f age-related changes in cognitive abilities cause the age-related degradation o f perceptual abilities, this effect should be equal across perceptual tasks so long as the cognitive demands remain the same. Schneider and Pichora-Fuller (2000) draw attention to an earlier study they conducted (Pichora-Fuller & Schneider,' 1991), i n which absolute, noise-masked monaural, and noise-masked binaural thresholds for the same pure-tones were obtained using the identical procedure. A n ageeffect was found for absolute thresholds and masked binaural thresholds, but not for masked monaural thresholds. It is unlikely that the task establishing absolute thresholds is more cognitively demanding than that establishing masked monaural thresholds (for a more detailed explanation o f this study and the various conditions, see "Binaural Unmasking", Section 1.5.2.1.5.1). Cognitive declines are therefore unable to explain such performance differences and instead perceptual abilities are implicated.  1.3.3.2.3  C o m m o n Cause Hypothesis  •  Age-related correlations between measures o f sensory functioning and cognitive performance may be the result o f a third variable with widespread consequences. Physiological changes such as neural degeneration or vascular system decrements could manifest ih an ebbing o f both cognitive and sensory functioning. While intuitively appealing, this hypothesis is difficult to test. There are a number o f ways in which to  ,  14  conceptualize this hypothesis depending on the degree to which cognition and perception are presumed to be distinct or modular. I f the perceptual and cognitive modules are •believed to interact and affect one another, as well as be similarly affected by a third variable (such as neural degeneration), it is difficult to tease out the factors contributing to the observed perceptual and cognitive decrements. I f the two were believed to be distinct modules, unable to interact, the common cause hypothesis would be supported i f a simulated perceptual deficit did not affect cognitive performance. It has been demonstrated in several studies however, that degrading an auditory signal reduces working memory capacity (e.g., Craik, Murphy, & Schneider, 1998, as cited i n Schneider & Pichora-Fuller, 2000; Pichora-Fuller et al., 1995). It has also been found however, that although there does not seeni to be a direct link between the two, there is a correlation between measures o f auditory and visual acuity and measures o f fluency and word naming (naming as many cities as possible beginning with a specific letter, for example), tasks not relying directly on perceptual acuity (Lindenberger & Baltes, 1994). This latter finding is taken as support for the common-cause hypothesis. However, although consistent with this hypothesis, it does not necessarily preclude other possibilities.  1.3.3.2.4  Perceptual Degradation  Errors in perceptual processing could disrupt cognitive processing because the information being sent to the higher centres is degraded or faulty. This would result in correlated performance on perceptual and cognitive tasks even i f the cognitive system were entirely intact. The finding noted earlier that improvements i n cognitive performance followed cataract removal, lends support to this hypothesis, as do the  findings by Craik et al. (1998, as cited i n Schneider and Pichora-Fuller, 2000) that young participants performed worse on working memory tasks when the information was presented with background babble. The former demonstrates that by restoring perceptual capacity, cognitive abilities increase, while the latter demonstrates that by simulating perceptual difficulties (by degrading the perceptual input), cognitive performance decreases. A s this theory provides some o f the foundation upon which this research is based, it w i l l be elaborated further below.  1.3.3.3  A Synthesis o f Perceptual and Cognitive Research It is difficult to disentangle the relationship and to determine the direction o f  causality, because cognitive and perceptual age-related differences are intimately linked. There are both afferent and efferent neural pathways between the auditory system and higher centres enabling the transmission o f both ascending arid descending information. Losses i n any part o f the integrated system thus affect the other parts o f the system. Increasingly, cognition and perception are modelled in this way, with the two systems conceptualized as sharing processing resources. The interaction is especially apparent in language processing because both systems are heavily relied upon. Iri language comprehension for instance, Cohen (1987) identifies eight component processes that are considered to occur simultaneously and influence each other . It is important to note that the first two o f the eight components 2  1) Analyse the acoustical signal; 2) Identify the phonemes and recognise the words; 3) Identify the meaning of the words; 4) Parse the grammatical structure of the sentences; 5) Integrate the meaning across different parts of the sentence and across different sentences; 6) Identify the mood and intentions of the speaker; 7) Relate the spoken information to previously acquired knowledge of the same topic; 8) Extract the logical inferences and underlying implications of what has been explicitly stated (p. 222)  '  !6  involve the accurate perception o f the information, while the remaining six involve cognitive processes. A l l however, are, integral to the ultimate goal o f comprehending spoken language.  1.4  Working Memory/Information Processing  ,  Central to this study is the idea that perception and cognition are interrelated, each depending upon the other to accurately process information.- Such an informationprocessing model also relies heavily on the notion that the systems share resources, and that depending on the task, resources can be allocated from one to the other as necessary. The following section w i l l explain working memory and discuss the role o f this memory system in the comprehension o f language and the ways in which working memory can be assessed. Declines in working memory associated with aging w i l l then be outlined, followed by the role o f perception i n working memory models.  1.4.1  The M o d e l W o r k i n g memory was first construed as a computational activity b y Baddeley and  H i t c h (1974, as'cited in Just and Carpenter, 1992; Baddeley, 1986) who designed tasks such that the storage and processing aspects o f comprehension would be competing. They found that sentence comprehension declined when listeners were also required to encode several digits for later recall. They proposed that a trading relation between storage and processing exists, with each drawing on a common pool o f resources. Baddeley's (1986) working memory model consists o f two components: the first o f which contains modality-specific storage and rehearsal systems for verbal representations  17 (the articulatory  loop) and visual and spatial representations; and the second o f which,  the central executive, monitors, coordinates and Controls functions during ongoing processing. . '  .  DanCman and Carpenter (1980) proposed another model o f working memory, also positing that there is both a structural storage component and a computational or processing component. Although this model o f working memory does assume a storage capacity, it does not, however, include modality-specific buffers, and instead roughly corresponds to Baddeley's (1986) proposed central executive in that coordinating o f different processes is required (Just & Carpenter, 1992). The important feature o f this model is that the amount o f activation available in working memory capacity is limited, and has a different maximum i n each individual (Carpenter, M i y a k e , & Just, 1994; 1995). Activation is the resource necessary to maintain (store) and manipulate (process) information, two functions that exist in a trading-relation. The theory proposes that all enabled processes can occur simultaneously; i f however, the amount o f activation necessary to propagate them exceeds capacity, a deallocation parameter is presumed to decrease the total amount o f resources used. Capacity is exceeded i f memory load is high (i.e. there is a large amount o f information, or the information is complex in nature), when processing time is short, or when the information must be retained for too long. Deallocation o f activation to storage results in forgetting, while deallocation o f activation to computation results in slower processing. Recall that both o f these aspects o f cognition are noted to decline with age. Working memory span has been implicated in numerous cognitive activities including the encoding o f new information to, and the retrieval o f information from,  18 long-term, memory, problem solving, syntactic processing, language comprehension, and reasoning (see, e.g., Daneman and Carpenter, 1980; Daneman and Merikle, 1996; Just and Carpenter, 1992)1 '  1.4.2.  , '  ,  Working M e m o r y and Language Comprehension  Although working memory plays a central role in all forms o f complex thinking, its function i n language comprehension is especially apparent and many studies have confirmed that speech understanding draws heavily pn working memory (e.g., Craik & Jennings, 1992; Daneman & Carpenter, 1980). Because spoken language necessarily occurs over time, in order to integrate the individual phonemes, words, phrases and , sentences, a representation o f the earlier material must be maintained as w e l l as the partial results o f the running processing. Comprehension also necessitates the concurrent manipulation o f the stored material during ongoing processing thereby engaging both the computational and storage components (Just & Carpenter, 1992). Because working memory is intimately linked to language-processing, a reduced working memory capacity manifests i n language comprehension difficulties, especially as the situation becomes more complex. W h e n syntactic complexity is increased, when the text is long such that information must be stored for a longer time, or when extrinsic memory load is increased, working memory demands are higher, thus compromising comprehension (Carpenter et al., 1994).  19 1.4:3  ' Assessing Working M e m o r y Traditional short-term memory span tests define an individual's memory capacity  as the maximum number o f items they are able to store for short-term retrieval. These tests include such measures as digit span or word span, in which the participant is required to repeat a series o f numbers or words. Such measures however, rely ,on passive storage and rehearsal o f verbal stimuli, and are not related to performance on more complex cognitive tasks; the retention o f a small number o f words, for example, does not interfere w i t h simple processing tasks (Baddeley, 1986). Since Daneman arid Carpenter's landmark paper,in 1980, researchers have tended to use more complex tasks that impose simultaneous processing and storage demands. Daneman and Carpenter's (1980) original measure, the Reading Span test, requires participants to read a set o f unrelated sentences, and subsequently recall the final word o f each sentence in the set; Such a task thus involves the usual demands o f sentence comprehension, but because the participants are required to maintain and retrieve the final words o f the sentence, an additional storage component is added. The set size is gradually increased from two to six, and the reading span measure is usually either the maximum number o f sentences for which the participant can recall all the final words or the mean number recalled from a fixed number o f sets. Consistent with the theory o f working memory capacity, reading span is generally lower for complex sentences (Just & Carpenter, 1992). A listening span measure was also devised (Daneman & Carpenter, 1980) in which participants listen to a sentence, judge whether the statement is true or false in order to ensure that the processes necessary for comprehension are engaged, and recall  ' '  •• '  '  :  ,  .  20  the last worlds o f sentence sets in a fashion similar to that o f the reading span. F o r college students reading and listening span measure are highly correlated (r = 0.80), suggesting that similar processes are involved in each.  1.4.4  Working M e m o r y and A g i n g  •  W h i l e there is relatively little decline in traditional short-term memory span tasks with age, more complex tasks that involve both processing and information storage demonstrate that there tends to be a decrement in working memory that co-occurs with aging. Carpenter et al. (1994) combined the results o f 13 tests in which the reading/listening spans o f young (mostly college students) and o l d adults (aged 60 or older) were compared. They found that older adults tend to have smaller memory spans (by 15-40 %) i n all but two studies. Correlations o f reading/listening spans and age ranged from - 0 . 4 0 to -0.70, "suggesting that age or some correlate of age is implicated in the decline o f working memory span" (Carpenter et al, 1994, p . i 102, italics added). •  Older adults have a more difficult time understanding discourse that makes large  demands on the working memory capacity o f young adults. Cohen (1979), for example, demonstrated that older adults (65-79 years old) show significantly larger deficits than younger listeners when they are required to make inferences that necessitate integrating information across sentences, although they do not demonstrate similar deficits when the questions require verbatim answers. A n interaction between age and sentence complexity was observed in a more recent study (Obler, Fein, Nicholas, & Albert, 1991) i n which four groups o f adults differing by age (one group each in their 30s, 50s, 60s and 70s) listened to sentences o f  21 varying difficulty and answered a true-false question after each. It was found that the group differences were larger for more complex sentences (e.g., The doctor who helped the patient who was sick was healthy.) than they were for simple sentences (e.g., The fierce wolf attacked the lost sheep in the woods.).  Similar findings have resulted from  other studies involving complex stimuli such as long intervening text (e.g., Cohen, 1979), dual tasks such as a secondary reaction time test (Tun, Wingfield, & Stine, 1991) and increased rate o f speech input (Stine, Wingfield & Poon, 1986). In the latter study, when sentences were presented auditorily to both young and old participants, so long as the rate was within the normal range (e.g., 200 words per minute), the listeners performed similarly, even with relatively long sentences ( 1 6 - 1 8 words). W h e n the speaking rate exceeded normal limits (e.g., 300-400 words per minute), however, an age-related difference emerged. These examples further demonstrate that when the auditory signal is degraded i n some fashion, older adults experience more difficulty.  1.4.5  The Role o f Perception in Working Memory Recall that Cohen (1987) listed eight interdependent operations necessary to the  comprehension o f spoken language. The first was "analyse the acoustical signal". I f this first level were faulty, it would inevitably cascade into processing difficulties at the levels dependant on the integrity o f the initial signal. This line o f thought is consistent with the perceptual degradation hypothesis outlined above. There is some evidence that i f perceptual differences are accounted for, working memory capacity becomes more similar for young and o l d adults.  .  2  ' Rabbitt (1968) found that young adults less accurately recalled digits and remembered fewer details o f a short passage when the materials were presented in noise compared to when they were presented in quiet. Rabbitt later (1991) proposed that effortful listening undermined the computational processes o f rehearsal, and elaborative encoding o f material presented in noise by the hearing impaired, even when perception was intact. In the framework o f the working memory model, because processing resources are allocated tb the recovery o f a degraded speech signal (due to impaired hearing, competing noise, or a corrupted signal), fewer resources are available for the other processes imperative for speech understanding. ' To further investigate the relationship between perceptual stress induced by background noise, aging, and cognitive performance, Pichora-Fuller et al. (1995) conducted a series o f tests manipulating these variables. Participants consisted o f a young normal-hearing group (mean age 23.9 years), an old group with near-normal hearing (mean age = 75.8 years; better ear thresholds < 25 d B H L from 250-3000 H z ) , and a presbycusic group (mean age 75.8 years). In the first test, participants listened to the S P I N - R test (Bilger et al., 1984). In keeping with previous studies indicating that elderly people have a more difficult time in noise, it was found that young adults identified significantly more words than older adults when the S / N decreased, and that there was no significant main-effect difference between the two old groups. It was also noted that in adverse S / N conditions, all o f the listeners were able to use contextual cues to aid in word identification, thus indicating that in difficult listening situations, processing resources can likely be reallocated to word recognition. Because older adults find word-recognition difficult across a wider range o f S/Ns, it is likely that they must  2  23 reallocate resources in this manner more often than do young listeners. L i n k i n g this to what we know about the working memory model, i f these processes tap into a limited number o f shared resources, fewer resources would be available for other cognitivelinguistic processes such as those required to integrate the individual words and store information for subsequent use. In order to determine i f this is indeed the case, the second experiment introduced a working memory task simultaneous with the word-recognition: M e m o r y span was measured by, adapting the S P I N - R test in the format o f Daneman and Carpenter's (1980) listening span tests such that listeners were required to comprehend each sentence and store each sentence-final word. A g a i n both young and old participants were tested, with each listener required to identify the sentence-final word, judge the predictability o f each sentence (to ensure comprehension), and recall all o f the sentence-final words for a given set. The S / N and the set size were varied to determine the relative effects o f each. A g a i n , the elderly group identified fewer words than the younger adults for a given S / N , regardless o f the memory load. Interestingly, the addition o f a memory load had no effect on the word identification for either age group, but in adverse S / N conditions, where presumably more resources are allocated to decipher the word using contextual cues, the listeners were less able to recall identified words. There was also a significant age-related decrement i n recall ability at all S/Ns, even though credit was given for correctly recalling misperceived words. In order to determine i f this age effect was due to changes in auditory processing or general cognitive processing, the participants were also required to participate in a reading span test, the results o f which were compared to those obtained from the listening span test in quiet. O n the reading-span task, younger  •  - •  '  •  24  and older adults performed similarly. Because the older adults recall as much read material, but less heard material than the younger adults, it seems that the difference found in the latter is due to specific age-related differences in auditory processing, rather than in working memory per se. The researchers suggest that because the elderly exhibit sub-clinical deficits (perhaps resulting from disrupted temporal processing), the act o f listening is more effortful. M o r e resources would thus be allocated to the actual perception o f the material, leaving less for other processes such as storage and retrieval. M o r e recently, Craik, Murphy, and Schneider (1998, as cited in Schneider & Pichora-Fuller, 2000) conducted a study designed to further investigate whether agerelated memory deficits could be associated with perceptual deficiencies. They tested younger and older adults in a paired-associates memory task in which the listener hears sets consisting o f five word-pairs. After each set, the first word o f the pair is presented, and the participant is required to supply the second word o f the pair. A l l participants are better able to recall the most recent pair, with performance declining for the preceding four pairs. Younger adults however, were significantly better at recalling the first three serial positions than were the older adults. The young participants also performed the task in a babble background; in this condition, performance resembled that o f older participants in quiet. It thus seems that by adding a perceptual processing load to young listeners, their recall ability becomes like that o f elderly listeners. This further suggests that the observed memory deficits in the elderly adults could result from a greater perceptual processing load induced by changes in the auditory system. To further assess whether equating perceptual stress reduces cognitive differences in young and old listeners, Schneider, Daneman, Murphy & Kwong-See (1999, as cited  25  in Schneider and Pichora-Fuller, 2000) designed tasks to explore other cognitive abilities necessary to language comprehension. Younger and older participants, screened to ensure, that all had relatively good hearing, listened to stories or lectures i n three different listening conditions (quiet, moderate level o f noise, and high level o f noise) and answered multiple-choice questions on the content.. The conditions were individually adjusted such that each person was equally likely to recognize words i n the noise. The answers to the questions required either recalling specific details from the material, or integrating information over several sentences. There was no significant age difference for the integrative questions i n any listening condition, and no difference for the detail questions in the quiet and moderate-noise conditions when S / N conditions were equalled to yield the same word-recognition performance for all listeners. W h e n the noise levels were not individually adjusted, the older adults performed significantly worse than the younger adults i n all conditions. The findings thus indicate that when testing conditions are adjusted tp create equal perceptual stress, the otherwise noted cognitive differences are drastically reduced. Taken together, the above reviewed studies suggest that, consistent with an information processing/working memory model, many apparent age-related differences in cognitive abilities are reduced when perceptual stress is equalized. It also appears that the perceptual stress incurred by the elderly is not simply assessed by pure-tone audiometry because it is more likely caused by other declines in auditory processing such as temporal processing deficits.  1.5  .  •. Temporal Cues and Speech Perception  The processing o f temporal cues is crucial to all aspects o f audition because sound consists o f rapid pressure changes over time that must be perceived. Moreover, for sounds that convey meaning such as language or music, much o f the meaning is conveyed in the temporal changes themselves. These fluctuations over time occur on different scales within the speech signal. The temporal 'envelope', which changes relatively slowly over time, is seen to modulate the rapid pressure variations that create the 'fine-structure' (Vermeister & Plack, 1993). Suprasegmental cues such as prosody, which indicates emphasis, emotion, intent (i.e., statement or question), word meaning (i.e. verb or noUn), and syllabic rhythms are all conveyed in the envelope pattern. Phonemes, occurring at rates o f up to 30 per second (Liberman et al, 1967, as cited i n Moore, 1997), also have characteristic temporal amplitude patterns arising from vocal fold vibrations, and some phonemes differ in duration o f energy or i n the presence o f energy gaps. The speech signal is thus comprised o f several temporal levels, the relative importance o f which w i l l be discussed next.  1.5.1  Consequences o f Temporal Alterations The importance o f intact temporal cues to speech perception has been  demonstrated by a number o f researchers. The dominant focus has been on the envelope o f the speech signal because numerous studies have indicated that with an intact envelope, and degraded fine structure cues, phonemes, words and entire utterances can be recognized and understood to a surprisingly high degree.  27  In several studies, V a n Tasell and colleagues (e.g., V a n Tasell, Soli, K i r b y & Wilderi, 1987; V a n Tasell, Greenfield, Logemann, & Nelson, 1992) demonstrated that crucial speech information could be obtained from the signal envelope. In the first o f these studies, the envelopes for 19 consonants (in the context /aCa/) were extracted and the fine structure for each was replaced with noise. Although the participants performed more poorly with these stimuli than with intact speech, they were able to identify the correct consonant at a level above chance . It appeared that cues associated with manner 3  of articulation and voicing were available i n the envelope, while those that indicate place of articulation were not. There was a high degree o f inter-subject variability i n both studies, and the second study demonstrated that with practice, performance improved but was not generalized to other speakers. Therefore, such temporal patterns apparently relay important information about phoneme identity, but some crucial information is missing when only the envelope is available. V e r y few studies directly compare the relative contributions o f the envelope and fine structure. Drullman and his colleagues (Drullman, Festen, & Plomp, 1994a; Drullman, Festen and Plomp, 1994b; Drullman, 1995a; Drullman, 1995b), however, not only investigated the importance o f intact envelope cues, and how various forms o f envelope distortion affect speech comprehension, but also assessed the importance o f the fine structure cues.  One study (Drullman et al., 1994a) consisted o f splitting the speech  signal into several frequency bands, keeping the fine structure intact and using a low-pass  The researchers noted that in particular three aspects of the temporal envelope were used to distinguish the phonemes from each other: the voicing envelope resultingfromthe longer duration and greater amplitude of voiced medial consonants; the amplitude envelope used to distinguish the sonorarits which have greater amplitudes and flatter peaks; and the burst envelope which consists of the abrupt, highamplitude energy resultingfromthe burst associated with voiceless stops.  28  filter to 'smear' the temporal envelopes before recombining the frequency bands. In a follow-up study, Drullman et al. (1994b) again split the speech signal into different frequency bands, but this time the envelopes were high-pass filtered to ascertain the relative importance o f the slow modulations and rapid fluctuations i n the envelope. In both cases, there was an interaction between the amount o f information available i n the envelope (macro-information) and fine structure (micro-information). W h e n the envelope was highly degraded (i.e. flattened), i f the frequency bands were broad (1 octave wide), such that there was a large amount o f fine structure available in each envelope, fine temporal information was useful: Conversely, when more information was contained within the envelope (macro-information), the frequency bandwidths could be reduced (to 1/4 octave) and a high level o f speech recognition was maintained. A later study (Drullman, 1995a) was designed to determine which aspects o f the envelope contain the most information, the peaks or the troughs, and again looked at the relationship between the envelope and fine structure. Flattening the troughs o f the envelope was found to be less detrimental than creating the same modulation reduction by adding noise, presumably because noise also interferes with the fine structure. It was concluded that the envelope is more important to speech perception than the fine structure because with intact temporal speech envelopes for twenty-four A octave bands X  with random fine structure, intelligibility was retained, whereas it was found that when the fine structure was left intact and the temporal envelopes were random, an average o f only 17% o f speech was intelligible! Shannon, Zeng, Kamath, Wygonski, and Ekelid (1995) also demonstrated the importance o f the temporal envelope for speech perception by dividing the speech signal  .29  into different bandwidths and establishing the amplitude envelope for each region using low-pass filters. Each amplitude envelope was then used to modulate band-limited noise for the same frequency range . The bandwidths were then recombined, so that only the 4  envelopes' temporal properties were present for those bands o f the speech signal. Normal hearing listeners were then asked to identify consonants and vowels and repeat words from sentences. Despite the reduced spectral content,the participants were able to correctly make voicing and manner distinctions with 90% accuracy (although correct place-of-articulation judgements remained below 70%). These dramatic findings again indicate that, a large amount o f information is conveyed by the temporal patterns o f the speech envelope, and that speech perception is enhanced by providing the listener with the envelopes from several bandwidths, as opposed to that o f the entire signal. Greenberg and A r a i (1998) further demonstrated the importance o f the envelope by dividing the speech signal into 19 frequency regions and then time-shifting the bands with respect to each other. Again, speech intelligibility remained surprisingly high. Although intelligibility decreased as the asynchrony increased, correct word identification exceeded 75% when the time shifts were 140 msec. The relative insignificance o f the fine structure cues compared to the demonstrated significance o f the envelope cues is surprising considering the precision with which the auditory system encodes periodicity. Thus far, the research reviewed has predominantly demonstrated the importance o f an intact envelope for consonant recognition, but for the most part has neglected to mention the effect o f envelope  That is, the 0-500 Hz envelope was used to modulate 500 Hz low-passed noise, the 500-1500 Hz envelope modulated bandpassed noise for this region, and the 1500-4000 derived envelope modulated highfrequency bandpassed noise (1500-4000 Hz).  4  30  alterations on vowel recognition. Because vowels are periodic signals, similar in envelope structure, and differentiated mainly by their fine harmonic and superimposed formant structure, it is unlikely that envelope alterations would have the same deleterious effect on vowel recognition. This reasoning has lead to a recent interest i n the effect o f fine-structure changes on vowel recognition (e.g., Culling & Darwin, 1993; C u l l i n g & Summerfield, 1995; de Cheveigne, M c A d a m s , & Marin, 1996; M c K e o w n & Patterson, 1995): It has been found for instance that very minor shifts (8%) in F l frequency result in perceived categorical changes (i.e. [I] to [s]) (Darwin & Gardner, 1986). Pichora-Fuller et al. (submitted) also evaluate the importance o f fine-structure cues i n running speech, and note that in Greenberg and A r a i ' s (1998) study, although an asynchrony is introduced, within each band fine-structure information remains intact. In order to disrupt the auditory system's ability to encode the fine-structure (and thus determine the importance o f this type o f information), Pass (1998) jittered the fine structure while leaving the envelope unadulterated. (For more, details on jittering, see Section 2.2.2.1.2). • The findings from the study indicate that when the fine-structure is compromised, intelligibility is drastically reduced. Furthermore, the researchers note that with the jittered speech stimuli, the performance o f young listeners resembles that o f elderly listeners with intact speech, supporting the notion that such an asynchrony may be introduced by the aging auditory system. To summarize, the research that has addressed the consequences o f altering the temporal pattern o f the speech signal, temporal information at both the level o f the envelope and the level o f the fine-structurej indicates that both types o f cues are very important to the correct perception o f speech. Even in the initial studies, which indicated  •  31  that the envelope contained more information, it was apparent that an intact envelope was hot sufficient to convey all o f the information necessary to identify a l l phonemes. The fine-structure,  too, is crucial to the perception o f speech, especially as noise levels  increase (as was demonstrated by Drullman, 1995a). Given the importance o f temporal information to speech intelligibility, it is apparent that a reduced ability to encode such information would drastically reduce speech perception.  1.5.2 • .  ,  Measuring Temporal Processing A b i l i t y  Numerous tasks have been designed to tap specific auditory temporal processing  abilities. Some o f the psychoacoustic tasks considered relevant to speech perception w i l l now be reviewed as they specifically relate to aging, followed by a discussion o f the extent to which these tasks assess the skills necessary to speech perception.  1:5.2.1."  Psychoacoustic Measures o f Temporal Processing The following discussion w i l l outline frequency discrimination, duration  discrimination arid gap detection, abilities presumed to underlie the identificatiori o f phonemes; stimuli designed to more closely replicate the complexity o f speech; and binaural processing tasks to assess the ability to locate the source o f a target sound as w e l l as to differentiate that target from background noise.  1.5.2.1.1  Frequency Discrimination  Y o u n g listeners with normal audiometric thresholds have, very good frequency discrimination. For a frequency o f 1000 H z at a moderate sound level, for instance,  listeners have difference limens o f 2 - 3 H z (Moore, 1997). It is also noted that listeners have much better frequency discrimination in the lower frequencies than they do i n the higher frequencies (Hartmann, 1996). This acuity and frequency pattern can be explained , when one takes into account the phase-locking ability o f the auditory neurons (Moore, 1997). Hartmann (1996) notes that synchrony decreases rapidly as the frequency increases from 2000 H z to 5000 H z because timing jitter is intrinsic to neural firings, and synchrony becomes more disrupted as firing increases in rate. Because phase-locking occurs for the lower frequencies arid decreases with frequency increases (Pickles, 1988), it is likely that the acuity i n the low frequericies can be attributed to discrimination in intefspike tifne differences. K o n i g (1957) found that the ability to discriminate frequencies declined with age in a linear fashion at all frequencies. Because he did not control for audiometric hearing loss, the increased difference limens at the high frequencies could be attributed to cochlear pathology. Older participants with audiometric thresholds only slightly elevated in the low frequency region (10-20 d B from 125-500 Hz), however, exhibited a frequency limen over 5 times larger than their younger counterparts at 125 H z , suggesting an age-related increase in frequency difference limens in the absence o f significant audiometric hearing loss. This finding was further supported by A b e l et al. (1990), who carefully controlled for audiometric hearing loss. They compared data collected from both young and older people (40 - 57 years old) with normal audiometric thresholds (< 20 d B H L from 500 to 4000 H z ) , and found that aging did affect the acuity in frequency discrimination at both 500Hz and 4000Hz. Moore and Peters (1992) demonstrated an age related decrement in  '•  '  ...  •  '  33  frequency discrimination in the low frequency range (100-800 H z ) , and further demonstrated that poorer frequency discrimination does not coincide with larger auditory filters. In a more recent study, He, Dubno and M i l l s (1998) found that even with closely matched audiograms, aged listeners demonstrated significantly poorer frequency  '  discrimination abilities than young listeners. Again, they found that this was most prevalent in the low frequencies and decreased as frequency increased. , Taken together, these findings contribute to a fairly clear story. It has been consistently found that the age related differences are most pronounced in the lower frequencies, the same frequencies for which precise phase-locking is crucial. It thus seems reasonable to conclude that the decrement noted i n older listeners could result from less-precise neural timing in the aged auditory system;  1.5.2.1.2  Duration Discrimination  A l s o important to the processing o f speech is the capacity to distinguish changes in the duration o f a stimulus. Such an ability would, for example, enable one to differentiate the vowel preceding a voiced and voiceless stop (i.e. 'cap' vs. 'cab') (Price & Simon, 1984). F e w studies have been conducted to establish whether age effects this aspect o f temporal processing. A b e l et al. (1990) measured duration difference limens for young normal-hearing listeners ( 2 5 - 3 5 years old) and three groups o f older listeners (40-60 years old) with different levels o f hearing sensitivity, ranging from normal to moderately-severe sensorineural hearing loss. The difference limens were measured for two different reference stimuli (20 msec and 200 msec) using 1/3 octave noise signals at 70dB S P L or  34  greater in a forced-choice experiment. The researchers found that despite their relatively young age, duration discrimination was worse in the older adults, and was unrelated to hearing loss.  ,,  •  Fitzgibbons and Gordon^Salant (1994) more recently measured duration difference limens for tone bursts as w e l l as for a silent period surrounded by tone bursts, with stimuli o f 500 H z and 4000 H z using 250 msec as the reference stimulus duration for all conditions. The participants consisted o f two groups o f young listeners (20 - 40 years old), with and without hearing impairment, and two groups o f elderly listeners (65 - 76 years old), also with (matched) and without hearing impairments. Significant decrements were found for the elderly listeners with no effects o f hearing loss, stimulus frequency or reference stimulus. These limited numbers o f studies indicate that there is an age-related deficit in this aspect o f temporal processing that is largely independent o f hearing loss.  1.5.2.1.3  Gap Detection  Gap detection, the ability to detect a pause in a stimulus, is another psychoacoustic measure that provides insight into the temporal resolving power o f the auditory system. Such an ability is considered relevant to speech understanding because some phonemes are distinguished by the presence o f a silent interval. V o i c e d and voiceless plosives for example, differ acoustically i n that for the voiceless, there is delay in the onset o f the fundamental frequency following the burst o f energy. There are a number o f recent studies investigating whether gap detection abilities decrease with age. Moore, Peters, and Glasberg (1992), for instance, used a long-duration tonal signal with  low-level notched noise masking to determine the m i n i m a l gap detection thresholds o f young normals, elderly listeners with good hearing (2000 H z and below), and an elderly group with hearing loss. They found that gap detection thresholds in the two elderly groups did not appear related to hearing loss, and that the average gap detection thresholds o f the older participants was' larger than that o f the young participants. The researchers however, attributed this difference to the inclusion o f some elderly participants with very large gap detection thresholds. Schneider, Pichora-Fuller, Kowalchuk, and Lamb (1994) had both young (mean age 23 years) and old listeners (mean age 69.2 years), with clinically normal hearing to 3000 H z , detect a gap between two Guassian-enveloped 2000 H z tone pips compared with a continuous tone o f equivalent total duration and energy i n a 2-interval forcedchoice procedure. They found that gap-detection thresholds were not correlated with audiometric thresholds for either groups, and importantly, that the gap detection thresholds o f the older adults were more variable and on average twice as large as those obtained from the young participants. This finding was further corroborated by Snell (1997) who found that older adults, with audiometric thresholds matched to those o f younger listeners, had gap-detection thresholds in short noise bursts that were 27% to 37% larger than, those o f the younger group. In an attempt to account for the variability in findings, Schneider and Hamstra (1999) investigated the effect o f marker duration. When the duration o f 2000 H z tonal markers was varied, they found that an age-effect, present when the marker durations were short, disappeared when the duration o f the marker approached 500 msec, findings that have since been replicated (Haubert, 1999). A s was proposed earlier by Price and  36 Simon (1984), this was considered to reflect a decrement in recovery from neural adaptation in older adults, and helps explain the varied results o f previous studies. A s with the previously outlined psychoacoustic tasks, it is apparent that gapdetection thresholds increase with age and that they are unrelated to audiometric hearing loss.  '  1.5.2:1.4  ' Complex Stimuli  Thus far, the psychoacoustic measures mentioned have used simple sounds presented in isolation and therefore give an estimate o f the listener's optimal capacity to resolve the temporal cues. M o r e complex stimuli, consisting o f multiple signal components, may tax the system in a way more closely resembling that o f speech. Discrimination experiments consisting o f complex sequential patterns, for example, would require "top-down" processing factors,, such as working memory, untapped by simple stimuli (Fitzgibbons & Gordon-Salant, 1996). It seems likely that the ability to correctly order two different sounds would be an important auditory ability required for processing complex forms o f stimuli such as speech. Trainer and Trehub (1989) examined whether age affected the ability to order sequences o f tones o f different frequencies in several different conditions. They found that even with practice, elderly adults (mean age: 70 years) were less accurate in the tasks that require distinguishing between tone sequences with contrasting order. Performance was also not correlated with hearing loss, further suggesting that other mechanisms may be implicated.  :•  ,  37  Humes and Christopherson (1991) compared 'young-old' listeners (65-75 years old) and 'old-old' listeners (76-86 years old) to young listeners (19-36 years old) with normal hearing, and young listeners with a simulated noise-masked hearing loss matching that o f the older adult'.' Each listener participated i n a series o f tests involving both simple tonal stimuli and more complex sequences o f tones. The two groups o f hearingimpaired elderly listeners performed significantly worse that the young participants on the frequency discrimination task, the embedded test-tone task (in which the listener is required to discriminate changes i n the duration o f a component occurring in the middle o f a 10-tone sequence), and the tasks requiring the temporal ordering o f both tones and syllables sequences. In other words, with the exception o f frequency discrimination, the tasks that tended to pose more o f a problem for the older listeners were those involving complex stimuli: O n these four tasks, there were no differences in performance between' the two young or the two elderly groups, which suggests that the performance decrement for the latter groups is age-related. Fitzgibbons and Gordon-Salant (1995) measured duration difference limens for a 4000 H z tone (250 msec in duration) embedded within a sequence o f five other tones equal in duration. The participants consisted o f young listeners with normal hearing and those with hearing impairments, and older listeners with normal hearing and with matched hearing impairments. There were four conditions designed to vary the amount o f listener certainty about the location o f the stimulus. The first condition, for example, had the same tonal stimulus pattern with the target tone always presented at a fixed middle sequence location. In the second and third conditions, the tonal sequences were random, with either fixed or random target location. Results revealed large performance  .  38  differences associated with the age o f the listener, but unrelated to hearing loss. The agerelated difference increases as the complexity o f the condition increased, perhaps because such tasks both tax the perceptual system and increase working memory load.  1.5.2.1.5  Binaural Processing  In everyday situations, the sound signals arriving at one ear are not identical to the corresponding signals at the other ear. Listeners are able to take advantage o f the interaural intensity and timing differences to first determine where the sound source is located, and second, to help extract the desired signal from background noise. It has been demonstrated that when monaural recordings o f a noisy party in a reverberant room are compared to binaural recordings, the listener is able to use the cues in the latter to concentrate on a single source: a phenomenon labelled the "cocktail party effect" (Schneider, 1997). It is thus apparent that a decrement in the ability to compare the temporal information impinging upon each ear, would have further deleterious consequences on speech comprehension in noisy situations. The following section w i l l review the literature on binaural processing as it relates to aging.  1.5.2.1.5.1  Localization  Unfortunately, very little research has been conducted examining the relationship between age and localization ability. Noble, Byrne and LePage (1994), however, found that sound localization was poorer in an elderly hearing-impaired group than it was for normal-hearing young participants and that the variation in localization ability in the  39 elderly group was not fully explained by the nature and degree o f the hearing loss. It thus seems probable that normal-hearing elderly may also exhibit impaired sound localization.  1.5.2.1.5.2  , Binaural Unmasking  Binaural unmasking refers to the ability to use interaural difference cues to help differentiate a signal from background noise. It has been found for instance, that when a noise and signal are presented monaurally ( S N ) , and the signal is adjusted so that it is m  m  just masked (i.e. at its masked threshold), and then the noise alone is simultaneously presented to the other ear (S No), the tone becomes audible once again. In other words, • m  by adding noise to the non-signal ear, the signal is made more detectable. The difference in threshold between the two described conditions is called the masking level difference ( M L D ) , and is a measure o f the system's ability to resolve binaural temporal information. The M L D can be measured with a variety o f conditions, and is simply the comparison between the thresholds o f a baseline condition in which the interaural relationships are identical for both signal and masker (diotic) and a comparison condition in which there is an interaural difference for signal and masker (dichotic). Binaural unmasking can enhance a young listener's ability to extract a low frequency signal (less than 500 H z ) i n broadband background noise by as much as 15 d B . It is thus clear that a decrement in this skill could contribute to difficulties o f understanding speech i n noisy situations. Numerous studies have demonstrated that elderly listeners have significantly smaller M L D s , even when audiometric hearing loss is factored out (e.g., Pichora-Fuller & Schneider, 1991) or the older participants have normal hearing (e.g., Grose, Poth, &  40  Peters; 1994; Pichora-Fuller and Schneider, 1992; Pichora-Fuller & Schneider, 1995; Pichora-Fuller & Schneider, 1998; Strouse et al., 19.98). Pichora-Fuller and Schneider (1991) for example, established M L D s for four dichotic conditions for young normal-hearing listeners (20-25 years old) and for old listeners (63-74 years old) in the early stages o f presbycusis (but still within normal clinical limits o f <_25 d B H L from 250 to 2000 Hz.). It was first found that the young and o l d listeners had equivalent thresholds for puretones in the dibtic condition (SoNo). Recall that in this condition the signal and noise are presented in an identical fashion to each ear. This finding that thresholds in this condition were the same indicates that there is not a general difficulty extracting a signal from noise. Comparing diotic and dichotic conditions however, the size o f the M L D was found to be significantly smaller for the older adults, even when the slight hearing losses were accounted for through an analysisof-variance. In other words, elderly listeners did not receive as much benefit from the interaural timing differences that service binaural unmasking. The researchers suggest that the age-related differences found in this study and in a later study (Pichora-Fuller & Schneider, 1992), i n which the interaural delay in the masker was varied between 0 and 5 msec, can be modelled by assuming that there is a greater degree o f timing error i n the neural firings o f older listeners' auditory system. It is further proposed that this 'jitter', or asynchrony reduces the elderly's ability to utilise interaural differences necessary to unmask a signal as well as young listeners can. ,  41  1.5.2.2  Speech Perception and Psychoacoustic Measures o f Temporal Processing It has been demonstrated that for young adults with hearing impairments,-  temporal processing ability contributes to speech recognition performance especially i n noise. Tyler et al. (1982) found that even when the effects o f pure-tone thresholds were factored out, increased temporal difference limens and longer gap-detection thresholds were significantly correlated with reduced speech intelligibility in noised Dreshler and Plomp (1986) too, found that for those with hearing impairment, poor frequency resolution and poor temporal resolution were related to impaired speech perception. Gap-detection thresholds were found to be particularly correlated with speech perception in noise, a finding that has been further corroborated by Irwin and M c C a u l y (1987) when testing recognition o f reverberant speech in noise. Furthermore, success with cochlear implants has been shown to depend on the temporal acuity o f the recipient (HochmairDesoyer, Stigl, Brunner & Wallenberg, 1984). Evidence that psychoacoustic measures o f temporal resolution are related to the perception o f degraded speech in the elderly has been more equivocal. W h i l e some studies have indicated that there is no significant correlation (e.g., Humes & Christopherson, 1991; Strouse et al., 1998), others have demonstrated an effect. M o r e recently, Gordoh-Salant and Fitzgibbons (1993) tested both young and elderly listeners. A n intact and three distorted versions o f the low-context sentences from the S P I N - R test were presented to the participants as well as a duration discrimination and gap detection test. The distortions were meant to disrupt temporal cues and consisted o f time compression; reverberation and interruption. While it was found that high frequency hearing loss was associated with poor performance on the speech tasks, age-related  42  factors other than peripheral hearing loss contributed as well. There was a significant relationship between age, gap duration discrimination and recognition o f reverberant speech.' M o r e recently, Haubert (1999) explored the possibility that some o f the discrepancy may result from the rate at which the speech stimuli are presented. A s was noted above, age-related gap-detection decrements are associated with shorter marker durations. M a n y phonemes are distinguished by the presence o f a gap i n energy. W h e n presented at a fast rate, the markers preceding the gap would be shorter, thus potentially resulting in age-related differences not noted when the stimuli is presented at a slower rate. In this study it was found that older adults were less able to identify speech contrasts when a gap served to differentiate between two words particularly for fast speech, thus supporting the notion that speech presented at a fast rate may be better correlated with psychoacoustic tasks. Psychoacoustic measures, while not perfectly correlated with performance on speech perception tasks, do seem to shed some light on the necessary auditory processes. Despite their limitations, such measures are often preferred because they allow investigators to examine the contributions o f specific abilities o f the auditory system while limiting the cognitive factors associated with language comprehension.  1.5.2.3  A Summary o f the Age-Related Psychoacoustic Findings The psychoacoustic findings clearly indicate that older adults do not process  auditory temporal information with the same acuity as younger adults. It is also apparent that this decrement is independent o f hearing impairments, as measured by pure-tone  .  '  '  '  4  3  audiometry, suggesting that something in the central auditory system, perhaps an increase in neural jitter, contributes to the difficulties experienced.  1.6  Anatomy and Physiology Related to Temporal Processing The next section w i l l look first at the anatomical structures and physiology  necessary to accurately encode temporal information, and second at the changes that occur with age.  1.6.1  .  •  Temporal Precision in the Auditory System The central auditory system in general, and the auditory brainstem i n particular, is  a highly time-sensitive system. There are many features that contribute to the high fidelity necessary to encode the temporal features o f the speech signal and reduce the accumulation o f temporal jitter. The auditory nerve, which links a single cochlear hair cell with neurons o f the cochlear nucleus, is driven by basilar membrane motion. W h e n an incoming sound stimulates the basilar membrane such that a pressure wave occurs, the haircells within the Organ o f Corti are displaced causing them to release chemical neurotransmitters. W h e n released i n sufficient quantities, the auditory nerve fibres depolarize causing them to fire, and thereby transmit the signal up the auditory pathway. Because the neurotransmitters are predominantly released when the basilar membrane is maximally displaced, the neural firing tends to occur in phase with the wave travelling along the membrane. This phase-locking which relays the temporal characteristics o f incoming sound, occurs for low frequency sounds, below approximately 4000 H z , (e.g., Greenberg, 1996; Phillips,  44  1989), beyond which the signal periodicity is too high (e g., Hartmann, 1996; Palmer & ;  Russell, 1986, as cited in Phillips, 1989). Taken together, the activation timing o f the population o f nerve fibres is thus able to convey the temporal patterns and low-frequency spectral content o f complex acoustic signals such as speech. Electrical impulses are conducted along the cochlear nerve, which projects into each cochlear nucleus and bifurcates to send branches to the three divisions o f the nuclear complex (the anteroventral [ A V C N ] , posteroventral [ P V C N ] and the dorsal [ D C N ] divisions), each o f which essentially behaves as a separate system (Greenberg, 1996). The A V C N is believed to predominantly encode for the subsequent binaural analysis o f higher centres. The A V C N neurons receive their input from the cochlear nerve by synaptic connections (the 'endbulbs o f Held') that serve to preserve temporal information by partially encapsulating the post-synaptic neuron and rapidly firing when the input neuron fires (Adams, 2000; Irvine, 1986; Phillips, 1989). These properties enable the precise temporal patterns to be transmitted to both the ipsilateral and contralateral medial superior olivary nucleus ( M S O ) . The binaural M S O neurons are thus able to compare the temporal phase input o f the low frequency signals originating from each ear. The, lateral superior olives (LSOs), receive both contralateral (via the medial nucleus o f the trapezoid body [ M N T B ] ) and ipsilateral input from the A V C N s , and are able to compare both the temporal and intensity differences (Adams, 2000). The posteroventral cochlear nucleus consists predominantly o f chopper and onset units, so labelled because o f their firing patterns. Choppers fire at a fixed rate to high frequency stimulation and phase-lock to low frequency (below 1000 Hz)'stimuli (sinusoidal or amplitude-modulated). Onset units respond with a high degree o f temporal  45. precision to the initiation o f a stimulus, after which their response diminishes. This pattern is most likely to occur with frequencies above 1500 H z , and like chopper units, they tend to phase-lock to low frequency stimuli. The high degree o f temporal precision suggests that these units are optimized to process low frequency and amplitude • modulated signals such as those found in speech (Greenberg, 1996). Ultimately, projections from the D C N , M S O , L S O and other brainstem nuclei converge on the inferior colliculus (IC), which sends ascending fibres to the medial geniculate body ( M G B ) , which in turn projects directly upon the primary auditory, cortex (Irvine, 1986; Phillips, 1989). Cortical auditory neurons are sensitive to the temporal properties o f sounds, particularly the onset o f the stimulus. The transient response at this level are precisely timed such that first-spike latencies have a jitter o f only a few hundred microseconds (Phillips, 1995). This indicates that the temporal precision is more than sufficient to encode the fine timing necessary to extract speech information such as voice onset times, consonantal bursts and aspirations (Phillips, 1995). It is thus apparent that a reduction i n the precision o f the temporal coding o f incoming stimuli at any point i n the auditory system would likely manifest in difficulties understanding such complex stimuli as speech. The next section w i l l discuss the points at which the temporal resolution could be compromised in the auditory system as a consequence o f aging.  1.6.2  A g i n g and Temporal Resolution Presbycusis is associated with a number o f cochlear changes: inner and outer  haircells and spiral ganglion cells are lost; the basilar membrane thickens and calcifies;  .  ,  1  •  46  the spiral ligament shows degeneration; and the stria vascularis, responsible for maintaining electro-chemical potentials in the cochlear fluids, degenerates (for reviews, see Schneider, 1997; Willot, 1991). It is thus apparent that age-related peripheral damage contributes to hearing impairments among the elderly. These changes however, cannot adequately account for decrements in temporal processing and speech comprehension in degraded conditions.  ,  The auditory nerve, an essential component i n me transmission o f temporal information, also appears to be affected by age. These changes include, among others, a loss o f neurons, changes in neuronal nuclei, changes in the presence o f proteins critical for optimal neural firing, a disappearance o f dendrites, and alterations i n myelination (Frisina, 2000; Schmiedt & Lang, 2000; Walton, 2000; Willot, 1991), all modifications that would affect the fidelity o f neural timing. A n i m a l studies have revealed some physiological changes in auditory neurons. W i l l o t (1991) notes that there is evidence that in mice, the neurons become 'sluggish' with age, and are thereby potentially less effective at phase-locking to incoming stimuli. Hellstrom and Schmiedt (1990) studied young and aged quiet-reared gerbils and found that the amplitude o f the compound action potential in the older group was significantly less than that for the younger group. Because this measure relies on synchronous neural firings, such a reduction i n magnitude suggests that a decrease in discharge- synchrony occurs in the elderly gerbils. This explanation is consistent with the findings o f Boettcher et al., (1993), who measured auditory brainstem responses ( A B R s ) in quiet-reared young and aged gerbils. They too found that the amplitude o f the A B R i n older subjects is significantly reduced compared to young subjects. Because aged gerbils with minimal  •  47  threshold shifts, had a reduction in A B R magnitude o f 50% or more, these changes were not a direct result of.auditory sensitivity loss and more likely resulted from changes to the auditory nerve. A n age-related loss i n neural synchrony would effect higher-order processes dependant on temporal resolution. It has already been noted that there is a decrement i n the ability to perform the psychoacoustic tasks thought to tap the auditory processes necessary to speech perception (frequency discrimination, gap or stimulus duration discrimination, localization, and binaural unmasking). It appears that both age-related changes to the compound action potentials o f the auditory nerve, and behavioural changes as measured by psychoacoustic tasks are consistent with a loss o f neural synchrony.  1.7  •  Jitter From the above discussion it should be clear that there is compelling evidence  that a deficit in temporal processing ability tends to co-occur with age. The idea o f a neural asynchrony effectively resulting in a temporal 'jitter' has now been introduced in a number o f the above sections. It was noted as a possible explanation for the decline in performance on a number o f the psychoacoustic tasks, and physiological evidence supports the hypothesis. It appears, however, that until recently no studies looked at this aspect o f temporal processing. Pass (1998) focussed on the importance o f periodicity coding by introducing temporal asynchrony to speech stimuli while keeping the temporal envelope intact. Based upon a model originated by Durlach (1972) and refined by Pichora-Fuller and Schneider (1992), this study presumed that the imposition o f external  temporal jitter models loss o f internal neural synchrony. The jittered speech was created by a computer program designed by Bruce Schneider (1997), the algorithms o f which were used i n the present study and w i l l be further explained i n section 2.2.2.1.2. , Pass (1998) and Pichora-Fuller et al. (submitted) created three jitter conditions varying both the standard deviation o f the jitter and the rate at which the distribution .occurred . Temporal jitter was applied to the S P I N - R test (Bilger et al., 1984) and 5  presented to young listeners (19-35 years) with normal hearing at two signal to noise ratios (+4 and +8 d B ) . It was found that, although the temporal envelope was unaffected, intelligibility was reduced. In the easier S/N, the performance on the predictable sentence-final words in the most difficult jitter condition was 22% below that o f the intact condition; and when the signal to noise ratio and predictability were least advantageous, there was a 54% difference. Thus, complementary to previous findings (e.g., Shannon et al., 1995; Greenberg and A r a i , 1998), it seems that when the spectral fine structure is distorted, even i f the envelope remains intact, speech intelligibility is greatly compromised.  •  '  •  Pichora-Fuller et al. (submitted) note that the difficulties displayed by the young normal hearing participants when presented with externally jittered stimuli resembles the difficulties displayed by the elderly listening to intact stimuli. It is thus possible, that jitter o f this type mimics the internal neural asynchrony believed to occur with age.  It should be noted that the way in with the jitter was applied in this experiment resulted in a high noise-floor that affected the high frequencies more than the lower frequencies, thereby effectively masking the higher frequencies. This problem was eliminated in the present study, and will be further elaborated in Section 2.2.2.1.2. . '  • 1.8  •  •  49  Summary , Older adults tend to have more difficulty'understanding speech than younger  adults. It has been proposed that these difficulties result, not from cochlear pathology that can be predicted from elevation o f audiometric thresholds, but from processing deficits that may occur i n the peripheral and/or central auditory system. Specifically, the encoding o f temporal information seems'to be compromised, contributing to the observed decrements in the perception o f speech. It has been further suggested that as people age, the timing o f auditory neural firings becomes less precise dr synchronous. This hypothesis has been supported with both behavioural and physiological evidence. I f the elderly auditory system introduces asynchrony i n the fine structure o f speech, older adults would be constantly processing a less stable signal. Because the perceptual information is compromised, in order to obtain the same level o f comprehension, top-down processes must be employed to a greater degree than they are for young people processing unadulterated information. From an information-processing perspective, i f older adults must deploy more resources to simply perceive spoken stimuli, fewer resources would be available for other processes necessary to comprehend language. Storage capacity for instance, would be compromised resulting i n memory breakdowns. The experiments o f this study further explore the perceptual degradation hypothesis explained above in section 1.3.3.2.4. Schneider and Pichora-Fuller (2000) noted that the various hypotheses regarding the relationship between age-related declines in perceptual and cognitive processing could be investigated by simulating aging i n young adults. I f neural jitter causes perceptual difficulties that cascade into a reduced  50 ability to perform higher level processing, it seems reasonable to assume that i f this jitter could be simulated in young adults, similar higher level deficits would be evoked. The first experiment in this series refined the jitter used by Pass (1998) and introduced the new version to S P I N - R sentences (see Section 2.2.2.1.2 for more details on the jitter and the refinements included i n the current experiments). These sentences were then presented to young normal-hearing participants and word-identification performance was evaluated. In the second experiment, a memory load was incurred in the fashion o f Pichora-Fuller et al. (1995) i n order to determine i f the external jitter introduced to the speech signal affected working memory capacity.- Ultimately the findings from these experiments were compared to the results obtained from older adults with intact stimuli.  1.9  Hypotheses  N u l l Hypothesis 1. The introduction o f low frequency jitter w i l l have no effect on wordidentification performance. Accompanying Research Hypothesis and Prediction. The refined jitter w i l l have a significant effect on sentence-final word-identification performance in young adults with normal hearing, with their performance being reminiscent o f the performance o f elderly listeners to intact speech. Because the auditory system relies on neural synchrony up to about 1000 to 1200 H z , the introduction o f low pass jitter w i l l significantly compromise speech perception. However, because the high frequencies w i l l not be masked, performance w i l l not be as poor as that obtained by Pass (1998) with the same jitter parameters.  51 N u l l Hypothesis 2. The use o f contextual cues w i l l be the same regardless o f a) S / N and b) jitter condition.  '  Accompanying Research Hypothesis and Prediction: Top-down processing w i l l be relied upon more as the speech signal is compromised by jitter and by increased background , noise. The difference in performance between low-context and high-context sentences w i l l thus be less when the S / N is greater and stimuli are intact than when either the S / N is less or the stimuli are jittered. N u l l Hypothesis 3. In the. word recall task, word recall w i l l hot be affected by a) S/N, b) contextual cues, or c) jitter. Accompanying Research Hypothesis. A s the signal becomes more difficult to perceive and higher-order processes are allocated to the perception o f the signal, less resources w i l l be available Tor other cognitive functions such as information storage. Recall abilities w i l l thus be compromised. , N u l l Hypothesis 4. A n y effect of jitter on word-recall performance w i l l be the same whether participants are required to remember 2 or 8 words. Accompanying Research Hypothesis. Because remembering two items w i l l not greatly tax the system, there w i l l be little difference between the conditions when the recall sets contain two items. Eight items, however, do induce a significant memory load, and w i l l thus likely require more resources for storage. I f the resources are being otherwise allocated (to perceiving the signal), there w i l l be a difference in recall performance for recall sets containing eight items compared to sets containing two items. N u l l Hypothesis 5. W o r d recall ability in young adults with normal hearing with jittered stimuli w i l l not resemble that o f elderly listeners with intact stimuli.  ,  •  52  Accompanying Research Hypothesis: Because the artificially jittered stimuli is meant to resemble the jitter applied to the speech signal by asynchronous auditory nerve firings, young people with normal hearing should have recall difficulties reminiscent o f those experienced by elderly people in the same task with intact stimuli especially when context is low. W h e n the context is high, differences may occur because the elderly have been shown to be better able to make use o f contextual cues (Pichora-Fuller et al., 1995).  2. M E T H O D S  2.1  Objectives This study was designed to determine i f participants' ability to recall sentence-  final words was compromised by an increase in the level o f background babble, and more importantly-by the introduction o f temporal asynchrony to the speech signal. T w o separate studies were conducted in order to meet these objectives. The first objective was to determine the effect o f jittering applied only below 1200 H z on word identification. The second objective was to determine i f reduced temporal asynchrony had an effect on working memory span. Both studies are outlined in detail in the following sections.  2.2  Experiment 1: W o r d Identification  ^  The first study was conducted in order to determine i f introducing low-frequency jitter to the speech signal significantly affects the word identification performance o f young listeners with normal hearing.  2.2.1  ,  Participants in Experiment 1 The listeners consisted o f 12 normal-hearing young participants (mean age = 24.7,  standard deviation = 3.2), recruited from the university community and paid for their participation. A l l participants had pure-tone thresholds o f 20 d B H L or better bilaterally from 250 to 8000 H z . See Appendix A for audiometric thresholds o f the participants, and Appendix B for participant characteristics.  2.2.2  Materials for Experiment 1 Each participant completed a consent form, a Hearing and Language History  questionnaire and the M i l l H i l l vocabulary test (Raven, ,1938). The first four o f the eight forms from the Revised test o f Speech Perception in Noise (SPIN-R) (Bilger et a l , 1984) sentences were used for the first session and the last four forms were used for the second session. Each form consists o f 50 sentences, 25 o f which are high context (the last word is predictable based on the preceding content) and 25 o f which are l o w context (the last word is not predictable). Intact and jittered sentences were presented to the right ear at , two signal-to-noise ratios in each session!  ,2.2.2.1  Preparation o f the Stimuli  2.2.2.1.1  S P I N - R Sentences  The S P I N - R test is available in clinical settings to evaluate clients' abilities to understand speech i n the presence o f background noise. T h e original S P I N test ( K a l i k o w et al, 1977) consists o f 10 sets o f 50 tape-recorded sentences presented with background 'babble' noise consisting o f 12 people simultaneously reading passages aloud. Listeners are required to repeat the last word o f each sentence. In'each sentence set, 25 o f the sentence-final words are predictable based on the context o f the sentence, and 25 are not predictable (Refer to Appendix C for a list o f the S P I N - R forms). In the high-context sentences, the listener is able to use contextual cues to assist in word-identification, whereas i n the low-context sentences the listener must rely solely on'acoustic cues. The S P I N test thus yields two scores, one each for high-context and low-context sentences, the difference between which presumably reveals the participant's ability to anticipate the  55 last word based on contextual cues. The S P I N test therefore indicates both how w e l l listeners are able to recognize words in the presence o f background noise without context, and how well listeners take advantage o f context to assist them i n identifying words ih speech. The levels o f both the speech and the background babble can be manipulated to alter the signal-to-noise ratio to determine the listeners' abilities in different conditions. The revised version o f this test (Bilger et al., 1984), which was standardized on hearing-impaired listeners, differs slightly from the original in that it consists o f eight rather than ten sets o f sentences; two were discarded because they were found to deviate from the others. The remaining eight sets were then rearranged in order to insure that they were equivalent at a signal-to-noise ratio o f +8 d B . Different also, is that the babble in the revised S P I N test (the SPIN-R) consists o f eight people reading rather than twelve. This test was chosen so that the interaction between jitter and noise could be investigated, and the results compared to the findings o f Pass (1998) and Pichora-Fuller et al. (1995).  2.2.2.1.2  , The Jitter  .  The original jitter program created in house by Bruce Schneider (1998) was based on the assumption that internal jitter results in random temporal changes to an incoming signal. Visualizing the speech signal as a graph portraying changes in amplitude over time, jitter can be seen to cause slight alterations in the timing o f each point. Schneider's program enabled the experimenter to determine both the range o f changes (positive and negative delays) as well as the rate at which the delays were applied. W i t h i n these parameters the alterations occur randomly based on the Gaussian distribution o f bandlimited white-noise. Figure 1, taken from Green (1976) displays a temporal  56 representation o f noise like that from which the values for jittering were derived. I f the bandwidth o f this noise were higher in frequency, more rapid changes in the same time sample would be depicted, and conversely, i f it were lower the changes would occur more slowly. For each time sample in the sound-file (20 k H z sampling rate), a delay value is selected by referring to the distribution o f noise, determining the amplitude o f the noise at the corresponding point i n time and then using this amplitude as a delay value. The delay value then determines the position o f the time sample in the original speech waveform  PROB. DISTRIBUTION OF SAMPLE Xi  L  o  TIME-  TEMPORAL REPRESENTATION OF NOISE  Figure 1. Temporal and spectral representation o f Gaussian noise. The instantaneous pressure value follows a Gaussian distribution as illustrated on the right side o f the temporal representation. Reprinted from Green, 1976, p. 44.  whose amplitude value is to be substituted in for the value at the time position in question. In the event that a delay o f zero is used then the original amplitude value is kept for that time position. I f a delay o f +2 is used then the amplitude value from 2 time samples beyond the position in question is substituted in'for'the original amplitude value etc. The chosen bandwidth o f the low-passed noise determines the rate at which the delay values w i l l change and the standard deviation alters the amplitude (or root-meansquare) o f the noise signal thereby increasing the range o f delay values that can be selected. This program was used by Pass (1998) to create three different jittered versions o f the S P I N - R forms. For this project, a new version o f the jitter program was created by Carol Jaeger to allow the jittering o f up to three different frequency bands using different jitter parameters for each band. In this refined program, a Fast Fourier Transform is used to separate the incoming signal into its component frequencies. The signal is then divided into a maximum o f four frequency bands according to values provided by the experimenter and are then converted back to the time domain using an Inverse Fast Fourier Transform. The lower three bandlimited signals can then be jittered according to the values provided by the experimenter using the same algorithms as in the original program. It should be noted that it is a valid option to apply no jitter to a band. Finally, the individual bandlimited signals are added together to produce the final jittered signal. For the purpose o f the present study, only two frequency bands were created: one from 0 to 1200 H z , and one from 1200 H z to the Nyquist frequency for the signal (10,000 H z in this case). Jitter was applied only to the lower frequency band. 1200 H z was chosen as the ceiling for this study because, as opposed to the higher frequencies i n  58  which the splatter effectively masks the signal, Figure 2 displays that frequencies below 1200 Hz introduce a level of splatter that is at least 10 dB below the level of the signal, well below the signal-to-babble levels used. This cut-off also seemed reasonable because for frequencies above this, phase locking is not considered to be as necessary for the representation of frequency, and thus would not be as sensitive to neural asynchrony.  PQ  0.4 RMS BW  f  k H z  = 0.25 = 100  Hz  = 0.05 = 500  Hz  = 0.25 = 500  Hz  =  O.B  kHz  f  =  1.2  kHz  = 1.6  f  kHz  m »  >  w -20  RMS BW  ms  PH  CJ  W  PH  Ul  -20  >  -40  <  -60  W  K  RMS BW  i  i  m  i  i  i  0 1 2 3 4 1 2  I  I T  3  i  i  i  I  i  i  r  4 1 2 3 4 1 2 3 4  FREQUENCY IN kHz  Figure 2. Pure-tones jittered using three different jitter parameters. The bottom row displays the jitter used in the current experiment. It can be seen that for pure-tones of 1200 Hz and below, energy at the pure-tone frequency rises well above the noise floor.  59 The jitter parameters were determined based upon Pichora-Fuller and Schneider (1992) i n which they derived jitter parameters for both young and old adults based upon their performance in binaural unmasking tasks. It was proposed that the elderly auditory system introduces a jitter with a standard deviation (the range o f timing displacement) o f 0.25 msec. This standard deviation was used by Pass (1998) in two o f her jitter conditions (coupled with bandwidths o f 100 and 500 H z ) , and incidentally, it was these two conditions that resulted in significant performance decrements. The standard deviation chosen for the creation o f the present jitter, was thus 0.25 msec. The rate at which the jitter is applied must also be established by the experimenter when creating the stimuli. Because the combination o f a 500 H z bandwidth and a standard-deviation o f 0.25 msec resulted in the most significant performance differences in the original experiment, 500 H z was chosen for the current experiments as well: A l l o f the jittered stimuli were thus created by having each S P I N - R form run through the program with the parameters such that the jitter was applied below 1200 H z at an average rate o f 500 H z (once every two msec), and an average range o f 0.25 msec. Appendix D displays some spectrographs and time-amplitude waveforms o f a S P I N - R sentence both intact and jittered i n the way described above.  2.2.3  Calibration o f the Equipment for Experiment 1 The R M S (root-mean-square) amplitude values for the S P I N sentences have  already been established and are considered equal for each sentence (Bilger et al., 1984). These values were incorporated into the in-house computer program used to run the initial experiment by Pass (1998). A s the current study was simply a refined version o f .  60  one already, conducted in the same laboratory, calibrations consisted o f ensuring that the new jitter program did not affect the amplitude o f the S P I N - R sentences, and establishing that the Tucker Davis Technology system was still calibrated in accordance with the values used i n the initial study. R M S values were calculated for both the original and current stimuli using an i n house program (rms_spch.exe designed by K i m Y u e , Erindale College, University o f Toronto). The values for five random sentences were compared and found to be essentially equal; the largest difference between versions was 0.10 d B . This slight difference is to be expected, and is not considered likely to affect word recognition. Calibration o f the equipment was conducted in the method suggested by Wilber (1994) by passing a 1 k H z calibration tone, originally created in C S R E 4.5 'ecosgen' program by Pass (1998), through the Tucker Davis Technology ( T D T ) system with the modules connected exactly as they would be for the experiment, but without any attenuation. A 1kHz tone was chosen because it is considered to approximate the peak intensity o f speech. The intensity o f the calibration tone was measured from the right T D H - 3 9 headphone located in a double-walled sound-attenuating I A C booth. The sound pressure level was found t o b e 116.60 d B A , almost exactly the same as that measured in 1998 (116.70 d B A ) and incorporated into the in-house program by Hollis Pass. The intensity o f the calibration tone was then attenuated by 20 d B by the S M 3 module to ensure that peak-clipping did not contaminate the original measurement. A s the resulting signal decreased by essentially 20 d B (resulting in 96.50 d B A ) , it can be concluded that the original measurement was valid.  61 2.2.4  1  ' i Apparatus and Physical Setting for Experiment 1 A l l experimental sessions occurred with the participant seated in a double-walled,  sound-attenuating I A C booth. The S P I N - R sentences were presented through T D H - 3 9 P lOW'headphones. The experimenter sat outside the booth and listened to the participant through an O B 802 audiometer. The experimenter controlled the S P I N - R form number, jitter condition, S / N , and presentation level o f the stimuli through an in-house computer program on an I B M compatible personal computer. The digital signals were routed through the Tucker Davis Technologies D D I , F T 5 , P A 4 , S M 3 (with both gain dials set to - 2 0 ) , and H B 5 modules before reaching the listener's earphones. See Appendix E for a diagram o f the T D T setup and Appendix F for a description o f each modules' purpose.  2.2.5  Procedure for Experiment 1 Air-conducted pure-tone thresholds were obtained for each participant before  proceeding to the experimental conditions. Instructions for the S P I N - R task were then read to each participant; refer to Appendix G for the specific instructions given to each participant. W h i l e facing the participant in the sound booth, the experimenter then read the first 5 practice sentences and had the participant practice the task by repeating the sentence-final words; refer,to Appendix C for a list o f the S P I N - R practice sentences. The participants then listened to the S P I N - R sentences through their right ear at 70dB S P L , chosen because it is considered to be a typical conversational level. The experiment occurred in two sessions. In the first, participants were presented with both intact and jittered stimuli at signal-to-noise ratios of+8 and +4 d B . In the second session,  62 participants again listened to intact and jittered stimuli, but in this session, the signal-tonoise ratios were more difficult at Oand - 4 d B . In each session, the order i n which the conditions were presented remained constant across participants and progressed from easiest to most difficult. This order was chosen because a declination in performance could then be attributed to the condition parameters rather than to a difficulty with the requirements o f the task. The order o f the first four S P I N - R forms was balanced across the first session, and the order o f the last four forms was balanced across the second session . See Appendix H for the order o f conditions and S P I N - R forms presented to 6  each participant. Time was provided for breaks throughout the testing, and most participants completed each session in about 70 minutes. See the Results section for detail on the findings from this experiment.  •2.3  Experiment 2: W o r d Recall  2.3.1  Participants in Experiment 2 There were sixteen participants i n the second experiment. A s with Experiment 1,  all who participated were young (mean age= 26.8; standard deviation = 1.78), had clinically normal hearing (< 20 dB H L from 250 H z to 8000 H z ) and were native , speakers o f English. For air-conducted pure-tone thresholds, refer to Appendix I, and for participant characteristics refer to Appendix J.  It should be noted that one participant was unable to return for the second session, and instead a new participant completed all eight conditions of the experiment in one session. This subject listened to the conditions in a slightly different order, although this did not appear to affect the results. 6  63 2.3.2  Materials for Experiment 2 A s with Experiment 1, participants were required to complete an in-house hearing  and language questionnaire followed by the 20-item M i l l H i l l vocabulary test (Raven, 1938). •  '  The second experiment also used intact and jittered S P I N - R sentences with multitalker background babble. Refer to section 2.2.2 "Materials for Experiment 1" for a more detailed description o f the materials used iri the Experiment 2.  ' 2.3.2.1  Stimuli Rationale  •  Because there are only eight S P I N - R forms, it was necessary to choose the stimuli such that there were only eight conditions. Because both intact and jittered stimuli were necessary, only two signal-to-noise ratios and two recall set sizes could be used.  2.3.2.1.1  Signal-to-Noise Ratios  Pichora-Fuller et al. (1995) had young and old participants participate in the recall tasks at four different signal-to-noise ratios. There was no significant difference in performance for either age group in silence compared to in a signal-to-noise ratio of+8 dB. Because the S P I N - R test is standardized for +8 d B , and because the stimuli had already been verified at this ratio, +8 dB was chosen as the easier signal-to-noise condition. A signal-to-noise ratio o f +4 d B was chosen for the more difficult noise condition, again because the stimuli had been verified at this ratio. These levels also allow for comparisons with the data obtained by Pichora-Fuller et al. (1995).  •  2.3.2.1.2  .  64  Recall-Set Size  A g a i n , an easy set-size was necessary i n order to obtain a baseline. In PichoraFuller et al. (1995), it was found that there were no significant age-related differences when the set size was two at any signal-to-noise ratio, whereas a set size o f four resulted in age differences i n lower signal to noise ratios (0 and +5 dB). A set size o f two was thus chosen as the easier recall set size for the present experiment. ' The set-size chosen for the more difficult conditions must be adequate to stress memory capacity i n order to demonstrate age-related differences. Pichora-Fuller et al. (1995) found that age differences became significant when the set size was four or more for low signal-to-noise ratios (0 and +5 dB) and six or more for higher signal to noise ratios (+8 d B and oo). Because the interaction was most significant with a set size o f eight ( p O . 0 1 as opposed to p<0.05 at the lower signal-to-noise ratios), the present experiment employed a set size o f eight for the more difficult conditions, in order to ensure that the working memory system was maximally taxed, and thus most likely to reveal differences resembling those correlated with age.  2.3.3  Calibration o f the Equipment for Experiment 2 A s the identical stimuli and equipment were to be used, calibration from the initial  experiment was carried over into the second. For more information regarding calibration o f the equipment, refer to section 2.3.2.  2.3.4  Apparatus and Physical Setting for Experiment 2 The equipment was set up in an identical fashion to that o f the first experiment.  In this experiment it was also necessary for the experimenter to stop the presentation o f S P I N - R sentences' i n order t o a l l o w the participant to recall the items. This too, was controlled from outside the booth using the in-house program. See section 2.2.3 for a more detailed description o f the experiment's physical setting. •  2.3.5  •  Procedure for Experiment 2  •  •  A g a i n participants first partook in an air-conducted pure-tone hearing assessment. They were then instructed in the requirements for the experiment and had an opportunity to practice the test. Appendix K contains the instructions for this experiment and Appendix C has the practice S P I N - R sentences. The participants again listened to the S P I N - R sentences through the right headphone and repeated the sentence final word. They were further required to make a predictability judgment to ensure that they were attempting to comprehend the entire sentence rather than simply attending to sound o f the final word. The task was stopped by the experimenter after either two or eight sentence-final words had been repeated and judged, at which time the participant was required to recall as many words as possible from the just completed set. The set sizes remained constant for a given S P I N - R form and the listener was told ahead o f time whether they were required to remember words from a set o f two or eight. For a set size o f two, for example, the participant listened to the sentence, repeated the final word, then said either 'yes' or ' n o ' depending on whether they thought the word was predictable from the context o f the sentence. They would then  66  do the same for a second sentence, after which they would repeat the words from both sentences. It did not matter in which order the words were repeated, and credit was given for correctly recalling a word that had been incorrectly identified. It should be noted that when the set size was eight, there were six sets, and two extra sentences at the end. These remaining sentences were included for word identification and predictability scores, but were hot used in recall measures. Because in Experiment 1 it was determined that jitter created more difficulty than a reduced signal to noise ratio, rather than dividing them by signal-to-noise ratio, unjittered conditions were presented before jittered condition. The first half o f the experiment consisted o f intact S P I N - R forms, at both S/Ns and both recall set sizes, progressing from easiest to most difficult. The second half consisted o f the same order o f conditions, but the S P I N - R forms were jittered. A s in the first experiment, the S P I N - R forms were balanced across conditions and participants. See Appendix I for the order o f conditions and presentation o f the S P I N - R forms. In total the session took almost three hours; participants were able to take as many breaks as they wanted, and were encouraged to complete each half on separate days. Those who chose to complete all conditions in one day were required to take an hour break between the two halves, i n order to ensure that a decrement in performance was not due to fatigue.  The  experimenter sat outside the booth and listened to the verbal responses o f the participants through headphones. The S P I N - R forms were used, and a checkmark was placed on the score line for the correct identification o f the sentence-final word; any misidentified words were written down for further analysis. A ' y ' or an ' n ' was marked next, to record the predictability judgement. For each word that the participant correctly recalled  67  another checkmark was placed directly next to the written word. A g a i n , credit was given for the correct recall o f a misidentified word. See the Results section for the findings o f this study.  68 ""'  3.1  •.  3. R E S U L T S  :  Iritfoduction , '  The following chapter w i l l describe the results of the present series o f experiments, first outlining the word-identification findings o f Experiment 1, followed by the word-identification results from Experiment 2. The findings o f the two experiments w i l l then be compared to each other, and finally, the word-recall performance from the second experiment w i l l be reported. In the following chapter, the findings o f the present experiment w i l l be compared to those Pichora-Fuller et al. (1995). '  3.2  Experiment 1: W o r d Identification The raw scores for this experiment are displayed in Appendix M .  3.2.1  Types o f Errors Made by Participants M o s t errors occurred when the sentences were low-context, when the S / N was  less favourable, and when the sentences were jittered. When the S / N was +8 d B , errors predominantly occurred at the end o f the words; word-final consonants were replaced or omitted, or a plural /s/ was added. The substitutions tended to retain a feature o f the original phoneme, such as one voiceless stop for another (e.g., cap becoming cat). This pattern was similar in both the intact and jittered conditions, with more errors occurring when the sentences were jittered. When the S / N was +4 d B , the errors were similar to those made in the higher S / N conditions, again consisting mainly o f word-final consonant errors, but also including a small number o f vowel substitutions. The substitutions  *  • "  ,  •  69  tended to maintain a feature o f the original vowel (e:g., one back vowel for another, as in  loot for logs). A s the S / N dropped to 0 and - 4 d B , errors became more abundant. W h e n the sentences were left intact, the incorrect phoneme usually maintained one or more feature in common with the target sound.' In these more difficult conditions, it was common for more than one phoneme to be substituted (e.g.,  courts replacing thorns, duck  replacing gum). When the sentences were jittered, many o f the errors were like those described above, but there were also many more complete word-substitutions; usually a word more common in the English language substituted for one that is less common, but again the substituted words tended to have some sounds in common (e.g., bed replacing  peg, charts replacing chunks, dust replacing loot). A number o f times, the participant . repeated back a word from ah earlier part o f the sentence (e.g.,  escape, in the sentence  The burglar escaped with the loot.). In summary, although the number o f errors increased as the conditions became more difficult, the nature o f the errors remained similar, with the vowels comprising the most robust o f the phonemes, and the word-final consonants the most vulnerable.  3.2.2  Effect o f Jitter on Sentence-Final Word-Identification Scores Overall, word-identification was poorer when S P I N - R sentences were jittered  (57% ± 7.8) than when they were intact (68% ± 8.8), as shown in Figure 3. This observation was confirmed by an analysis o f variance, with a significant main effect o f jitter, ( F ( l , l 1)=262.5,2<0.001).  70  Figure 3 Mean Percent-Correct Word-Identification Score (+/-1 SD) in the Inact and Jittered Condition  Intact  Jittered Condition  3.2.3  Effect o f S/N Condition on Sentence-Final Word-Identification Scores Word-identification performance was better when the S / N was higher as can be  seen in Figure 4, which displays the mean percent-correct sentence-final wordidentification scores for each S / N condition collapsed over condition and effect. A main effect o f S / N was confirmed by an analysis o f variance, (F(3,33) = 424.1, p<0.001).  71  Figure 4 Mean Percent-Correct Word-Identification Scores.(+/-1 SD) for Each Signal-to-Noise Ratio Condition  4 0 Signal-to-Noise Ratio (dB)  The figure also illustrates that the largest performance differences seem to occur in the lower S / N conditions, an observation confirmed by a Student Neuman Keuls test o f multiple comparisons which indicated that the score was significantly worse in the - 4 d B S / N condition than i n the 0 d B S / N condition, which was in turn worse than in the two higher S / N conditions which did not differ significantly from each other.  3.2.4  Effect o f Context on Sentence-Final Word-Identification Scores Figure 5 displays that context, too, had an effect on participants' ability to identify  the sentence-final words. The mean-percent correct score for high-context sentences,  72  including both intact and jittered sentences as well as all S/Ns, was 74% (± 14). F o r lowcontext sentences, the mean-percent correct score was 50.6% (± 10.9). Again, an analysis o f variance confirmed a significant main effect for context, ( F ( l , l 1) = 416.8, p<0.001).  Figure 5 Mean Percent-Correct Word-Identification Scores (+/-1 SD) for Each Context Condition  High  Low Context  3.2.5  ,  Interaction Effect o f Jitter and S / N Conditions In Table 1, which lists the mean-percent correct scores and corresponding  standard deviations for the intact and jittered conditions for each S / N (combining both high and low contexts), it can be seen that the decrement in performance associated with a reduced signal to noise ratio was larger in the jittered conditions. Figure 6 displays the  .  .  73  mean percent-correct sentence-final word-identification scores for each jitter condition i n each S / N condition, while Figure 7 displays the mean percent difference between identification scores obtained with intact and jittered stimuli for each S / N condition. It is apparent that as the S / N was decreased from 8 to 0 d B , the decrement resulting from listening to jittered sentences becomes more pronounced. W h e n the S / N is decreased to -4 d B , the difference, although still apparent, is less pronounced. A n analysis o f variance confirmed a significant interaction between S / N and jitter,(FY3,33) = 6.7, p<0.001).  It is  also apparent from Figure 7 that the presence o f jitter has the greatest effect when the S / N is 0 d B ; a Student Neuman-Keuls test o f multiple comparisons confirmed that the difference between word-identification scores in the intact versus jittered conditions was greater i n the 0 d B S / N condition than in any other S / N condition.  Jitter Condition  S/N  N o Jitter N o Jitter N o Jitter N o Jitter Jittered Jittered Jittered Jittered  8 4 0 -4 8 4 0 -4  M e a n percentcorrect sentencefinal wordidentification scores 91.1 87.2 64.4 28.8 83.7 75.5 46.2 20.9  Standard Deviation  5.17 6.42 10.08 13.34 • 7.62 , 6.36 8.49 8.70  Table 1. M e a n percent-correct sentence-final word-identification-scores and standard deviations for each jitter X S / N condition.  74  Figure 7 Mean Percent Difference Between W o r d Identification Scores in the Intact and Jittered Conditions at Each Signal-to-Noise Ratio 40  8  4  0  Signal-to-Noise Ratio (dB)  -4  75 3.2.6  .  Interaction Effect o f Jitter and Context Conditions  Figure 8 Mean Percent Differences BetweenWordIdentification Scores in the High- and Low-Context Conditions in Each Jitter Condition  Intact  Jittered Condition  Table 2 lists the mean percent-correct sentence-final word-identification scores and corresponding standard deviations for the intact and jittered condition for both high and l o w context sentences. Each mean includes both S/Ns. Figure 8 plots the differences between mean performance in high-context and low context-sentences. It is apparent that the difference in performance between high and low context conditions is quite similar regardless o f whether the stimulus is jittered or intact. There thus does not appear to be a large interaction between jitter and context, an observation confirmed by an analysis o f variance, (F(l,11) = 0.16,p>0.1).  76  Jitter Condition  Context Condition  M e a n PercentCorrect SentenceFinal W o r d Identification Scores  N o Jitter N o Jitter Jittered Jittered  High Low High Low  79.8 56.0 68.5 45.2  Standard Deviation  '  6.57 10.91 7.48 8.12  Table 2. M e a n percent-correct sentence-final word-identification scores and standard deviations for each Jitter X Context condition  3.2.7  Interaction Effect o f S / N and Context Conditions Table 3 lists the mean percent-correct word-identification scores and  corresponding standard deviations for each S / N for each context condition, with both intact and jittered conditions combined. Figure 9 illustrates the mean percent-correct sentence-final word-identification score in each S / N and context condition, and Figure 10 shows the differences between scores for high- and low- context items in each S / N . It can be seen that as the S / N is decreased from 8 to 0 d B , the benefit obtained from context becomes more apparent. When the S / N is - 4 d B , however, this difference is less pronounced, because performance is poor even with high-context items. A n analysis o f variance confirmed that S / N and context do interact significantly, (F(3,33) = 20.6, 2<0.001), and a Student Neuman-Keuls test o f multiple effects confirmed that the differences are significant at each S / N .  77  S / N Condition (dB)  . Context Condition  8 8 ' 4 ' 4 0 0 -4 ' -4  •  ."  High Low • High Low High Low High Low  Mean Percent^ Correct SentenceFinal Word- . ' Identification Score 97.2 , '79.5 92.8 69.8 •74.8 35.8 31.8 18.0  Standard Deviation  . 3.43 9.35 3.18 9.61 9.64 8.93 . 11.84 10.20  Table 3. M e a n percent-correct sentence-final word-identification scores and standard deviations for each S / N and Context condition.  Figure 9 Mean Percent-Correct Word-Identification Scores for Each Context Condition at Each Signal-to-Noise Ratio  8  4 0 Signal-to-Noise Ratio (dB)  -4  m  High  I I  Low  78  Figure 10 Mean Percent Difference Between WordIdentification Scores in High- and Low-Context Conditions at Each Signal-to-Noise Ratio  8  4  0  -4  Signal-to-Noise Ratio (dB)  3.2.8  Interaction Effects o f Jitter, S / N , and Context Conditions There is no 3-way interaction between jitter, S / N and context conditions, as is  evidenced by an analysis o f variance, (T(3,33) = 2.53, p > 0.05). Appendix N lists the mean percent-correct score and corresponding standard deviation for both intact and jittered conditions i n each context and S / N condition.  79 3.2.9  Summary o f Results from Experiment 1 Figure 11 displays the results from this experiment. It can be easily seen that  performance was significantly worse when the stimuli were jittered, when the S / N was lower, and when contextual cues were absent. It is also apparent that both jitter and context have more o f an effect when the S / N is +4 or 0 d B , illustrating the noted interactions.  Figure 11 Mean Percent-Correct Word-Identification Scores for Each Condition at Each Signal-toNoise Ratio 100  Intact, High —B— Intact, Low Jittered, High Jittered, Low  0 4 Signal-to-Noise Ratio (dB)  3.3  Experiment 2: Word-Recall Because participants were scored for predictability judgements, sentence-final  word-identification performance and sentence-final word recall, the tasks w i l l be  80 discussed separately. After the word-identification results have been rep6rted, they w i l l be compared to the results o f Experiment 1 before reporting the results from the wordrecall task.,  3.3.1  Predictability Judgements Table 4 lists the mean percent-correct predictability judgement scores for each  condition. It can be seen that the ability to judge whether the sentence-final word was predictable based on contextual cues never dropped below a mean o f 94%. Therefore it seems reasonable to assume that listeners were comprehending the sentences rather than simply attending to the sentence-final word. See Appendix O for each participant's predictability scores i n each condition.  Jitter Condition  N o Jitter N o Jitter, N o Jitter N o Jitter Jittered Jittered Jittered Jittered  S / N Condition (dB)  ,  '  Recall SetSize  8,  '8 • 4 . 4 8 8 4 • 4-  2  ?  2 . 8 2 8 2 8  M e a n PercentCorrect SentenceFinal W o r d Predictability Score 97.9 . 96.8 97.1 96.8 95.6 97.4 94.9 94.0  Table 4. M e a n percent-correct Sentence-Final Word-Predictability Score for each Jitter, S / N , and Recall Set-Size condition.  3.3.2  Sentence-Final Word-Identification The raw scores from this experiment are displayed in Appendix P.  3.3.2.1  Types o f Errors made by Participants Sentence-final word-identification errors in this experiment were similar to those  made in the same S/Ns in thefirst experiment! A g a i n we see that errors occur mainly at the end o f the word, with word-final consonants replaced with another consonant maintaining a feature o f the original, the addition o f plural /s/, or the omission o f the final consonant. Again, the vowels tend to be robust, and even when words are replaced entirely, the incorrect phonemes almost always have at least one feature i n common with the target sound. See Section 3.2.1 for more details on the types o f error made by participants in Experiment 1.  3.3.2.2  Effect o f Jitter on Sentence-Final Word-Identification Scores Figure 12 displays the mean percent-correct sentence-final word-identification  and standard deviations for both the intact and jitter conditions i n this experiment. S / N , recall set-size and context conditions have been collapsed. The mean percent-correct sentence-final word-identification score for the intact conditions was 90.8% (± 5.5), whereas the mean score for the jittered conditions was 82.5% (+ 7.1). A n analysis o f variance confirmed that this difference is significant, (F(l,15)=86.4,p<0.001).  82  Figure 12 Mean Percent-Correct Word-Identification Score (+/-• 1 SD) in Each Jitter Condition 10090-  I 8 e Qc  80706050403020100Intact  Jittered Condition  3.3,2.3,  Effect o f S/N on Sentence-Final Word-Identification Scores A s was found in Experiment 1, sentence-final word-identification performance  was better when the S / N was higher. Figure 13 shows that when the S / N was +8 d B , the mean percent-correct word-identification score was 90% ( ± 5 . 1 ) , whereas the mean score was 83% ( ± 8 . 1 ) when the S / N was +4 d B . This difference is significant as was confirmed by an analysis o f variance, (F(l,15)=70.2,p<0.001).  83  Figure 13 Mean Percent-Correct Word-Identification Score (+/-1 SD) in Each Signal-to-Noise Ratio Condition  8  4 Signal-to-Noise Ratio (dB)  3.3.2.4  Effect o f Recall Set-Size on Sentence-Final Word-Identification Scores Increasing memory load by increasing the recall set-size results i n essentially no  difference in mean percent-correct scores for word identification. W h e n the recall setsize was 2, the word-identification mean percentage was 86.7% (± 6.90), and for the recall set size o f 8 participants scored 86.5% (± 6.21). A n analysis o f variance confirmed that there is no significant main effect o f set size on word-identification score, (F(l,15)=0.16,p>0.1).  84 3.3.2.5  Effect o f Context on Sentence-Final Word-Identification Scores A s i n Experiment L context had a main effect on word-identification (Figure 14).  When the sentence-final word was predictable based on the context, participants correctly identified the word 84.4% (± 3.8) o f thetime, yet when the final words were not predictable performance was on average 76.6% (± 9.3). A n analysis o f variance confirmed that the main effect o f context is significant,(F(l,15)=330.5, _p<0.001).  Figure 14 Mean Percent-Correct Word-Identification Scores (+/- 1 SD) in Each Context Condition 1009080-  8 70-  fccj 60-  1  1e 8 50s<u  cca 40-  OH  <u  302010-  •  0High  Low Context Condition  3.3.2.6  Interaction Effect o f Jitter and S / N on Sentence-Final Word-Identification  Scores Table 5 lists the mean percent-correct scores for word-identification and the corresponding standard deviations for the intact and jittered sentences at both S/Ns.  •  ,  85  Recall set-size and context conditions have been combined. In Figure 15, it can be seen that S / N makes a slightly larger difference in word-identification performance for jittered sentences than it does for intact sentences. Although this suggests that the effect o f S / N is more deleterious for jittered than intact stimuli, an analysis o f variance indicated that it failed to reach significance, (F(l,15)=l.l,p>0.10).  Jitter Condition  N o Jitter , N o Jitter Jittered • Jittered  S/N  8 4 8 4  M e a n PercentCorrect sentencefinal wordidentification scores 93.9 87.6 86.4 78.5  Standard deviation  3.75 7.33 . 6.37 8.75  Table 5. M e a n percent-correct sentence-final word-identification scores and standard deviations for each Jitter X S / N condition.  Figure 15 Mean Percent Difference Between WordIdentification Scores Obtained at +8 and +4 dB Signal-to-Noise Ratios in Each Jitter Condition % 8 a o J  40  Intact  Jittered Condition  86  3.3.2.7  Interaction Effect o f Jitter and Recall Set-Size on Sentence-Final W o r d -  Identification Scores  >  "  In Table 6, which lists the mean percent-correct sentence-final word-identification scores and corresponding standard deviations for intact and jittered conditions by recall set-sizes, it can be seen that performance in both the intact and jittered conditions remains similar when the set size is changed from 2 to 8. That is, set-size seems to have little effect on word-identification in either jitter condition. This observation o f no significant interaction effect o f jitter and set-size was confirmed by an analysis o f variance, (F(l,15)=0.6,p>0.10).  Jitter Condition  N o Jitter N o Jitter Jittered Jittered.  Recall Set-Size  2 • 8 2 8  M e a n PercentCorrect SentenceFinal W o r d Identification Scores 91.2 90.4 82.3 ' 82.6  Standard Deviation  5.71 5.35 8.06  7.07  Table 6. M e a n percent-correct sentence-final word-identification scores and standard deviations for each Jitter X Recall-Set Size condition.  3.3.2.8 ,  Interaction Effect o f Jitter and Context on Sentence-Final W o r d -  Identification Scores Table 7 lists the mean percent-correct word-identification scores and corresponding standard deviations for both intact and jittered sentences for each context condition. Both S/Ns and set-sizes have been included in the averages. Figure 16 shows , the differences between performance on high-context sentences and performance on lowcontext sentences for each jitter condition. It can be seen that contextual information  !  ,  87  made more o f a difference when the participants were listening to jittered sentences than it did with the intact stimuli., A n analysis o f variance confirmed that this finding is significant, (F(l,15)=7.1, p<0.05), and a Student Neuman-Keuls o f multiple comparisons indicates that each, o f the mean percent-correct sentence-final word-identification scores listed i n Table 7 is significantly different from the others.  Jitter Condition  Context Condition  N o Jitter N o Jitter Jittered , Jittered  High Low High ... L o w  Mean PercentStandard Deviation Correct SentenceFinal W o r d Identification Scores 99.1. 1.71 , 82.4 9.36 94.1 " , 5.92 70.8 9.24  Table 7. M e a n percent-correct sentence-final word-identification scores and standard deviations for each Jitter X Context condition.  Figure 16 Mean Percent Difference Between WordIdentification Scores in High- and Low- Context Conditions for Each Jitter Condition • 40  Intact  Jittered ' Condition  88 3.3.2.9  Interaction Effect o f S / N and Recall Set-Size on Sentence-Final W o r d -  Identificatidn Scores A s can be seen from the numbers in Table 8, which lists the mean percent-correct , sentence-final word-identification scores for recall set sizes by S / N , as w e l l as the corresponding standard deviations, recall set-size does not seem to interact with S / N . When the signal to noise ratio is +8 d B , the difference in mean percent-correct word identification scores between the two set sizes is negligible, and the same is found for a S / N o f +4 d B . The differences are only 0.1% and 0.6% respectively. A n analysis o f variance confirmed that this is not a significant interaction, (F( 1,15)=0.217, p>0.10).  S/N  Recall Set-Size  +8 +8 +4 +4.  2 8 2 . 8  "  '  '  Mean PercentCorrect SentenceFinal W o r d Identification Score  Standard Deviation  90.1 90.2 83.4 82.8  5.28 4.84 8.52 7.58  Table 8. M e a n percent-correct sentence-final word-identification scores and standard deviations for each S / N X Recall Set-Size condition.  3.3.2.10  Interaction Effect o f S/N and Context on Sentence-Final'Word-  Identification Scores Table 9 contains the mean percent-correct word-identification scores and corresponding standard deviations for each context condition at each S/N. Both recall set-sizes and jitter conditions are included in the averages. It can be seen in the table and in Figure 17 that context makes more o f a difference when the S / N is low. W h e n the S / N was +4 d B , the mean identification score for high-context items was 24.2% higher than it  89  Figure 17 Mean Percent Difference Between High-and LowContext Word-Idendification Scores in Each Signal-to-Noise Ratio Condition .40-r  •  —  :  •  — ,  - i  no  a 0)  35 'Vi  Signal-to-Noise Ratio (dB)  was for low-context items, whereas when the S / N was +8 d B , context resulted in an increase o f only 15.8%. The significance o f this interaction was confirmed by an analysis o f variance, (F(l,15)=24.3, p<0.001). A Student Neuman-Keuls test o f multiple comparisons indicated that the scores obtained for low-context sentences at each S / N were significantly lower then the score for high-context sentences at the same S / N .  90  S/N  . Context Condition  +8 ' +8 +4 +4  High. Low High Low  M e a n PercentCorrect SentenceFinal W o r d Identification Scores 98.0 85.2 95.2 71.0  Standard Deviation  2.54 . 7.58 5.10 11.00  Table 9. M e a n percent-correct sentence-final word-identification scores and standard deviations for each S / N X Context condition.  3.3.2.11  Interaction Effect o f Recall Set-Size and Context on Sentence-Final W o r d  Identification Scores In Table 10, which lists the mean percent-correct word-identification scores and corresponding standard deviations for each recall set-size and context condition, it can be seen that the size o f the memory load has virtually no effect on the benefit obtained from context. This observation was confirmed by an analysis o f variance, (F(l,15)=0.1, p>0.10). W h e n the recall set was 2, context improved scores an average o f 20.1%, and when the recall set-size was 8, context resulted in a score 19.8% better than that obtained without context.  Recall Set-Size  2 2 8 :• 8'  Context Condition  High Low High Low  M e a n PercentCorrect SentenceFinal WordIdentification Scores 96.8 76.7 96.4 76.6  Standard Deviation  4.05 9.74 3.57 8.85  Table 10. M e a n percent-correct sentence-final word-identification scores and standard deviations for each Recall Set-Size X Context condition.  3.3.2.12  Three- and Four-Way Interactions  There were no significant three or four-way interactions. See Appendix Q for tables containing lists o f mean percent-correct sentence-final word-identification scores and standard deviations for the possible combinations o f variables.  3.3.2.13  Summary o f Effects on Sentence-Final Word-Identification Scores in  Experiment 2 A s in Experiment 1, the presence o f jitter, S / N and context conditions had a significant effect on sentence-final word-identification scores. This can be seen i n Figure " 18 and 19. B y comparing these two figures, which differ only in recall set-size, it is apparent that set-size had little effect on word-identification scores and did not interact with the other variables! Context was seen to interact significantly with both jitter and S/N. Not surprisingly, in each instance, participants benefited from the presence o f context more i n the more difficult conditions, when the sentences were jittered and when the S / N was low.  •  92  Figure 18 Mean Percent-Correct Word-Identification Scores when Recall Set-Size = 2  20 H 10-  o-|  1  4  8  Signal-to-Noise Ratio (dB)  Figure 19 Mean Percent-Correct Word-Identification Scores when Recall Set-Size = 8  3.3.3  Comparison o f Sentence-final word-identification scores from E x  1 and Experiment 2 When the results from Experiment 1 and Experiment 2 are compared, it is found that there is no significant difference between the participant groups, (F(l,26)=1.8, p_>0.10). Because only two S/Ns were used in Experiment 2, the comparison o f results w i l l be confined to findings obtained when the S/Ns were +4 d B and +8 d B . 7  3.3.3.1 •  •  M a i n Effects  W h i l e the results for Experiment 2 have been detailed for the two S/Ns in question, it should be noted that when the conditions are confined to S/Ns o f +4 d B and +8 d B i n the Experiment 1, main effects for jitter ( F ( l , l l)=123.6,p<0.001), S / N ( F ( l ' , l 1)=13.2, p<0.005), and context ( F ( l , l 1)=163.8, p<0.001) were also found. The main effects were thus the same for the two experiments. A s would be expected, when the results i n the two experiments are combined, the same main effects are significant: jitter (F(l,26)=171.7, p<0.001);.S/N (F(l,26)=52.2, p O . 0 0 1 ) ; and context ,(F(l,26)=432.5,p<0,001).  3.3.3.2 ,  • .  Interactions  Although interactions were noted for Experiment 1 in the above section, when only the +4 and +8 d B S/Ns are considered, the interactions are no longer significant. However, i n the second experiment, which has a larger number o f participants, significant interactions were found for S / N X context and for jitter X context. W h e n the  Because there is no significant effect of set-size, the results from each recall set were averaged in order to obtain the scores used from Experiment 2.  94  results are combined, the same interactions found in the Experiment 2 remain significant: S / N X context (F( 1,26)= 17.2, p O . O O l ) and jitter X context (F( 1,26)= 10.5, p<0.001).  3.3.3.3  Summary o f the Comparison o f Sentence-Final Word-Identification  Scores F r o m Experiment 1 and Experiment 2 Figures 20 and 21 display the mean percent-correct sentence-final wordidentification scores from the two experiments at S/Ns o f +4 d B and +8 d B respectively. The proximity o f the lines indicates that the results from the two experiments are similar, and that the differences between the participant groups are minimal.  Figure 20 Comparison of Mean Percent-Correct WordIdentification Scores for Experiment 1 and Experiment 2 with Intact Stimuli 100-  Experiment 1, High  90-  —B— Experiment 1, Low  £ 80-| 8 8 70-1 t 60o u 50i  uu u  0-  Experiment 2, High -A-  40302010H  0-  1  4  Signai-to-Noise Ratio (dB)  8  Experiment 2, Low  95  Figure 21 Comparison of Mean Percent-Correct WordIdentification Scores from Experiment 1 and Experiment 2 with Jittered Stimuli  Experiment 1, High —B— Experiment 1, Low Experiment 2, High  - e - Experiment 2, Low  4  8 Signal-to-Noise Ratio (dB)  3.3.4  Sentence-Final Word-Recall Scores The raw scores from this experiment are displayed in Appendix R.  3.3.4.1  Effect o f Jitter on Sentence-Final Word-Recall Scores When the S P I N - R sentences were jittered, participants had a more difficult time  recalling the sentence-final words when required to do so after listening to either 2 or 8 sentences. Figure 22 shows the mean percent-correct word-recall scores for each jitter condition. Each column contains both S/Ns, both recall set-sizes and both context conditions. W h e n the sentences were intact, participants correctly recalled an average o f 84.8% (±6.5) o f the words, whereas when jitter was applied to the sentences, recall  96  averaged 79% (± 7-2). A significant main effect o f jitter as was confirmed by an analysis ofvarianee,(F(l,15)=43;718,p<0.001).  ,  Figure 22 Mean Percent-Correct Word-Recall Scores (+/- 1 SD) for Each Jitter Condition 100 90 80 -«-»  o  4> . fc:  Uo  70 60  t:  o 50 <5 ' O H '40  § 4>  •  30 20 10 0  Intact  Jittered Condition  3.3.4.2  Effect o f S / N on Sentence-Final Word-Recall Scores Figure 23 displays the mean percent-correct sentence-final word-recall scores for  each S / N with jitter condition, recall set-size and context condition collapsed. It can be seen that recall performance i n each S / N is very similar, with a mean score o f 81.2% (± 6.5) for a S / N of+8 d B , and a mean score o f 82% (± 7.0) for a S / N of+4 d B . A n analysis o f variance confirmed that there is no main effect o f S/N, (F( 1,15)=0. Vp<0.10). Recall that S / N did have a significant effect on word-identification.  97  Figure 23 Mean Percent-Correct Word-Recall Scores (+/-1 SD) for Each Signal-to-Noise Ratio Condition  Signal-to-Noise Ratio (dB)  3.3.4.3  Effect o f Recall Set-Size on Sentence-Final Word-Recall Scores It can be seen in Figure 24 that an increased memory load resulted in lower mean  percent-correct sentence-final word-recall scores. S/N, context and jitter condition have been collapsed so that only recall set-size is compared. W h e n the recall set-size was 2, participants correctly recalled on average 98.6% (± 2.1) o f the sentence-final words, whereas when the set-size was 8, mean percent-correct recall scores dropped to 65.2% (± 11.7). This effect o f set-size is significant, as was established by an analysis o f variance (F(l,15)=286.7,p<0.001).  98  Figure 24 Mean Percent-Correct Word-Recall Scores (+/-1 SD) for Each Recall Set-Size Condition 1009080-  8 70-  6  60-  § -50OH  i 0)  40302010-  Recall Set-Size  3.3.4.4  Effect o f Context on Sentence-Final Word-Recall Scores Word-recall was better for the sentences in which the final word was predictable  from the context. Figure 25 displays the mean percent-correct sentence-final word-recall scores for both high- and low-context items, and indicates that when the final words were not predictable based on the context, mean percent-correct score was 79.0% (± 8.1), and when the final words were predictable the mean percent correct score was 84.4% (± 5.6). A n analysis o f variance confirmed that there is a main effect o f context on mean wordrecall scores (F(l,15)=53.6, p<0.001).  99  Figure25 Mean Percent-Correct Word-Recall Scores (+/-1 SD) for Each Context Condition 1009080o  70fc o 60: OJ  O -»-» c 500>  o  IH  OH  C CS  403020: 100High  Low Context  3.3.4.5  Interaction Effect o f Jitter and S / N on Sentence-Final Word-Recall Scores Table 11 lists the mean percent-correct sentence-final word-recall scores for each  jitter and S / N combination, with recall set-size and context collapsed. It is apparent that a higher S / N does not provide much benefit, and that this lack o f benefit is consistent across the jitter conditions. A n analysis o f variance confirms that there is no significant interaction between S / N and jitter (F(l,15)=0.9, p>0.10).  100  Jitter Condition  N o Jitter,, N o Jitter Jittered Jittered .  S/N(dB)  8 4 8 4  Mean PercentCorrect SentenceFinal Word-Recall Scores 84.3 85.2 79.3 78.7  Standard Deviation  6.05 6.94 7.50 6.97  Table 11: M e a n percent-correct sentence-final word-recall scores and standard deviations for each Jitter X S / N condition.  3.3.4.6  Interaction Effect o f Jitter and Recall Set-Size on Sentence-Final W o r d -  Recall Scores The mean percent-correct sentence-final word-recall scores for each recall set-size and jitter condition are listed in Table 12 with both S / N and context conditions collapsed. Figure 26 displays the difference in performance when the set size is 2 as opposed to 8 for each jitter condition. It appears that set size has more o f an impact when the sentences are jittered than when they are intact, an observation supported by an analysis o f variance (F(l,15)=35.3, p<0.001). A Student Neuman-Keuls test o f multiple comparisons indicated that recall was significantly poorer for jittered than intact materials when the set-size was 8 and that both o f these scores were significantly poorer than for the conditions with set-size 2, which did not differ significantly from each other.  Jitter Condition  Recall Set-Size  M e a n PercentStandard Deviation Correct Sentencefinal word-recall scores N o Jitter 2 99.2 1.50 N o Jitter 8 70.3 11.49 Jittered 2 98.0 2.66 Jittered , 8 60.0 11.82 Table 12. M e a n percent-correct sentence-final word-recall scores and standard deviations for each Jitter X Recall Set-Size condition  101  Figure 26 Mean Percent Difference Between Word-Recall Scores for Recall Set-Sizes of 2 and 8 in Each Jitter Condition  Intact  Jittered Condition  3.3.4.7  Interaction Effect o f Jitter and Context Condition on Sentence-Final  Word-Recall Scores Table 13 lists the mean percent-correct sentence-final word-recall scores and corresponding standard deviations for both high- and low-context sentences for intact and jittered sentences. It can be seen that benefit from context is similar for both intact and jittered sentences. A n analysis o f variance confirmed that there is no significant interaction effect between jitter and context conditions (F(l,15)=0.5, p>0.10).  102  Jitter Condition  Context Condition  N o Jitter N o Jitter • Jittered • Jittered  High Low High Low  M e a n PercentCorrect SentenceFinal Word-Recall scores 88.0 81.5 81.5 76.5  Standard Deviation  .  5.02 7.96 6.22 8.24  Table 13. M e a n percent-correct sentence-final word-recall scores and standard deviations for each Jitter X Context condition.  3.3.4.8  Interaction o f S / N and Recall Set-Size on Sentence-Final Word-Recall  Scores Increasing memory load had a very similar effect at both S/Ns. Table 14 lists the mean percent-correct sentence-final word-recall scores and corresponding standard deviations for each recall set-size i n each S/N. It can be seen that the benefit obtained from a smaller set-size is similar regardless o f the S / N . There is no significant interaction between S / N and recall set-size, an observation confirmed by an analysis o f variance (F(l,15)=0.074,p>0.10).  S/N  . Recall Set-Size  8 8 4 . .  4  '  2 8 2 8  Mean PercentCorrect SentenceFinal Word-Recall Scores 98.6 65.9 98.6 65.3  Standard Deviation  2.20 11.35 1.95 11.96  Table 14. M e a n percent-correct sentence-final word-recall scores and standard deviations for each S / N X Recall-Set-Size condition.  • 3.3.4.9  \  "  103  Interaction Effect o f S / N and Context Condition on Sentence-Final W o r d -  Recall Scores,,  :  •  '  The provision o f context appears to be o f almost equal benefit in each S / N condition. Table 15 lists the mean percent-correct sentence-final word-recall scores and corresponding standard deviations for both high- and low-context sentences in each S / N . It can be seen that regardless o f S/N, the benefit obtained from context, improves the word-recall score by a similar amount. A n analysis o f variance confirmed that this interaction is hot significant (F(l,15)=0.3, p>0.10).  S/N.  Context Condition  8 8 4 4  High Low High Low  Mean PercentCorrect SentenceFinal Word-Recall Scores 84.5 ' 79.3 85.1 78.9  Standard Deviation  ' ' 5.32 ' 8.22 5.93 . 7.98  Table 15. M e a n percent-correct sentence-final word-recall scores and standard deviations for each S / X Context condition.  3.3.4.10  Interaction Effect o f Recall Set-Size and Context Condition on Sentence-  . Final Word-Recall Scores Context seems to have more o f an effect when the memory load is increased. Table 16 lists the mean percent-correct sentence-final word-recall scores and corresponding standard deviations obtained in each context condition for each recall setsize. In Figure 27, it can be seen that the benefit from context was much higher when the recall set-size was 8 than when it was 2. A n analysis o f variance confirmed that this interaction is significant (F(l,15)=30.1, p<0.01). A Student Neuman Keuls test o f  104  multiple comparisons indicated that the significant interaction occurs between the highand low-context scores obtained when the recall set-size is 8, and that there is no significant difference'between the context conditions when the recall set-size is 2.  '  Recall Set-Size  2 . 2 8' 8 ..  Context Condition  '  High Low High Low.  '•  M e a n PercentCorrect SentenceFinal Word-Recall Scores 99.3 98.0 70.3 60.0  Standard Deviation  1.69 5.92 9.64 13.67  Table 16. M e a n percent-correct sentence-final word-recall scores and standard deviations for each Recall Set-Size X Context condition.  Figure 27 Mean Percent Difference Between High- and Low-Context Word-Recall Scores for Each Recall Set-Size  Recall Set-Size  105  3.3.4.11  Three- and Four-Way Interactions  There were no significant three- or four-way interactions i n this experiment. See Appendix S for tables containing the mean percent-correct sentence-final word-recall scores and standard deviations for each combination o f variables.  3.3.4.12  Summary o f Sentence-Final Word-Recall Performance  ' Figures 28 and 29 display the mean number o f words correctly recalled i n each condition. B y comparing the squares (intact) and circles (jittered) it can first be seen quite clearly that the ability to correctly recall items from a list o f S P I N - R sentences is significantly affected by the presence o f jitter. It is also clear that participants are better able to remember the sentence-final words when the words are predictable from the context o f the sentence, and when the set-size is 2 as opposed to 8. The interactions are also apparent in these figures: jitter and context both have more o f an effect when the set size is 8. Ceiling effects reduce the possibility o f observing differences for set-size 2. O f course, the purpose o f testing set-size 2 was to demonstrate how participants perform when memory load is not taxed. S / N did not have a significant main effect, and did not interact with the other variables in a significant manner, such that 28 and 29 are very similar. Note, however that the two S / N conditions tested were relatively favourable compared to the other two levels tested in Experiment 1 when word-identification was measured.  106  Figure 28 Number of Words Recalled for Each Recall Set-Size in a Signal-to-Noise Ratio of +8 dB  -m-Intact, High  8-  7-  —B— Intact, Low  I" T36  Jittered, High  O  ai  5-  Jittered, Low  tn  I" 4  2 3h  1 0 Recall Set-Size  Figure 29 Number of Words Recalled for Each Recall SetSize in a Signal-to-Noise Ratio of +4 dB 8-  -»-  7-a6  Intact, High  —B— Intact, Low  I'  -•-  o  -e-  06 5-|  •a 3210Recall Set-Size  Jittered, High Jittered, Low  107  4. D I S C U S S I O N O F R E S U L T S rN T E R M S O F H Y P O T H E S E S A N D C O M P A R I S O N T O P R E V I O U S "STUDIES  4.1  Introduction  '  ,  This section w i l l compare the results o f the current experiments with those o f previous studies (Pass, 1998; Pichora-Fuller et al.; 1995), as well as address the hypotheses proposed in the first chapter. First, word-identification with both intact and jittered stimuli w i l l be discussed, followed by word-recall performance with both types o f stimuli. The performance o f young listeners exposed to jitter w i l l be compared to the performance o f old listeners to evaluate the extent to which the jitter manipulation mimics the effects o f aging.  4.2  -  '  W o r d Identification In both o f the current experiments, participants were required to repeat the  sentence-final words o f the S P I N - R sentences at a number o f different S / N conditions. Because the findings from each were similar, they w i l l be discussed together. The performance with intact sentences w i l l first be reviewed and compared to previous research on young and old listeners, followed by a discussion o f the results from the jittered stimuli and the ways i n which this resembles the performance o f elderly listeners.  108 4.2.1  W o r d Identification with Intact Sentences Consistent with previous research (e.g., Pichora-Fuller et al., 1995), the  participants in the current study were better able to correctly identify the sentence-final words when the S / N was higher, and when the context o f the sentence provided cues. Figure 30 displays the mean percent-correct sentence-final word-identification scores for young participants in the current experiments at S/Ns o f +4 d B and +8 d B , as w e l l as for those from Pichora-Fuller et al. (1995) and Pass (1998) in the same conditions.  Figure 30 Mean Percent-Correct Word-Identification Scores Comparing Young Subjects with Intact Stimuli in Four Experiments  Exp I, High  -e-  -•-  Exp  1, Low  Exp 2, High Exp 2, Low Pass, High Pass, Low Pichora-Fuller, High Pichora-Fuller, Low  4  8  Signal-to-Noise Ratio (dB)  109  A l l participants scored close to 100% when the S / N was +4 d B or +8 d B and the sentences provided context. W h e n context was low, although the participants i n the current experiments performed slightly better than those in the Pass (1998) study, they still did not obtain scores as high as those in the study o f Pichora-Fuller et al (1995). In the 1995 experiment, the participants were 'experienced' listeners in that they regularly participated in listening tasks. It is possible that this practice had an effect, and that the participants in the current experiments would improve their scores given more experience. The participants in the second experiment performed better than those i n the first, lending support to this notion. In Experiment 2, the added memory task necessitated twice as many sentences for each condition, thereby allowing the participants more time to adjust to the listening situation. The stimuli were also slightly different in the experiment run by.Pichora-Fuller et al. (1995). For that experiment, the S P I N - R sentences were simply run from a cassette tape through the audiometer into the sound-attenuating booth. F o r the experiments run by Pass and those o f the current study, the S P I N - R sentences were first digitized, stored on computer disks and fed through the Tucker-Davis Technologies modules where they were converted back into an analog signal. It is possible that this iatter method affected the sound quality and contributed to the reduced scores.  4.2.2  W o r d Identification with Jittered Sentences Again, the findings from the two current experiments were similar when the  sentences were jittered. Both found that jitter significantly affected a participant's ability  110 to identify the sentence final word, thereby refuting N u l l Hypothesis 1, which stated that jitter would not have such an effect. The results are also remarkably similar to those obtained from o l d listeners with normal pure-tone thresholds in the Pichora-Fuller et al. (1995) study. In Figure 31, the mean percent-correct sentence-final word-identification scores for jittered sentences i n Experiment 1 as well as the scores for the elderly listeners in the Pichora- Fuller et al. (1995) study can be compared.  Figure 31 Mean Percent-Correct Word-Identification Scores for Young Participants with Jittered Stimuli and Old Participants with Intact Stimuli  For low-context sentences (empty symbols), the resemblance is immediately striking. The young listeners in the current experiment are able to identify close to the same number o f sentence-final words when presented with jittered sentences as older  '  111  people with normal audiometric thresholds are able to when presented with intact sentences. Thus, the jitter manipulation seems to mimic the effect o f age i n wordidentification, at least when contextual cues are minimal. N u l l Hypothesis 2 stated that contextual cues would not be used more when the stimuli were jittered or the S / N was lower. In the first word-identification experiment, it was found that when the stimuli were jittered, participants relied upon the contextual cues to a greater extent than when the stimuli were intact, thereby refuting the first component o f the N u l l Hypothesis. Although there was no interaction between S / N and context i n Experiment 1, the results from the word-identification task in Experiment 2 indicate that participants benefited from contextual cues when the listening condition was worsened either by the introduction o f jitter or the reduction o f S/N. , Recall that performance on the low-context S P I N - R sentences is believed to represent auditory skill, whereas increased performance on high-context sentences additionally represents the listener's ability to make use o f contextual cues. PichoraFuller et al. (1995) found that older adults were better able to make use o f context than were younger adults, especially as the listening conditions became less favourable. In Figure 31, it can be seen that when the S / N is +8 d B , both the old and the young listening to jittered stimuli perform close to 100% when the context is high, and that when the S / N is - 4 d B , both perform quite poorly regardless o f context. A t these two extremes, the elderly participants are therefore not better able to make use o f context, in the first case because both groups are near ceiling, and in the latter case because both are close to floor. A performance difference occurs when the S / N is moderately difficult, +4 d B and 0 d B , listening conditions that likely mimic the more difficult ones encountered i n every-day  .112  life. C H A B A (1988) supplies data concerning the S/Ns encountered in typical environments. Inside the home, S / N tends to be 9-14 d B , whereas outdoors it drops to 58 d B . The S / N at a cocktail party has been calculated to be from - 2 to +1 d B , and i n public transportation systems, -2 d B . Clearly the moderately difficult S/Ns encountered in these experiments are representative o f real listening-situations, and i n conditions such as these, elderly people are better able to use context to recover information degraded in perceptual processing. Older adults thus come to the laboratory with a set o f skills not replicable in an experimental setting, such that younger adults with imposed perceptual deficits are not able to perform at the same level as older adults experienced with such difficulties. In Experiment 2, an increased memory load had no effect on word-identification ability. This suggests that resources are allocated to the perception o f stimuli before they are allocated to their encoding and retention, a finding also noted in Pichora-Fuller et al. (1995).  43  W o r d Recall The word-recall findings o f Experiment 2 w i l l now be compared to those o f  Pichora-Fuller et al.'s (1995), with the results obtained with intact stimuli compared to those o f the younger listeners in the 1995 study, and those obtained in young listeners with jittered stimuli compared to the results from the elderly participants i n the 1995 study.  113  4.3.1  W o r d Recall with Intact Sentences When the stimuli were intact, the participants in the current study performed  similarly to the young participants in Pichora-Fuller et al. (1995). Because different S / N conditions and recall set sizes were used i n the 1995 experiment, the findings cannot be directly compared . It is still possible to demonstrate the trends with the data available. Figure 32 displays the mean number o f words recalled when the S / N was + 4 d B in the current experiment, and when the S / N was +5 d B in the earlier experiment.  Figure 32 Number of Words Recalled for Young Participants in Two Studies with Intact Stimuli, S/N = 4/5 dB Intact, High —B— Intact, Low 1995, High 1995, Low  Recall Set-Size  In the 1995 experiment, when the S/N was +8 dB the largest recall set-size was 6, whereas in the current experiment the recall set-size was +8 dB. Although both experiments used a S/N of +8 dB, the current experiment also used 4 whereas the 1995 experiment used +5 dB.  114  For high-context sentences, it can be seen that when the recall set-size was 2, recall was almost perfect for both groups, and when set-size was increased to 8, performance decreased but was similar for both groups. For low-context sentences, a slight difference can be seen, but again it should be noted that the participants in the 1995 study were listening at a slightly higher S/N. It thus seems that the young participants with intact stimuli performed quite similarly in both experiments. This finding indicates that the performance o f the young adults in the current experiment is representative and thus provides a baseline from which to gauge the deficits in recall ability observed when jitter was applied to the stimulus.  4.3.2  W o r d Recall with Jittered Sentences Figure 33 displays the number o f words correctly recalled for the young people in  the current study when presented with jittered stimuli o f both high- and low-context Figure 33 Number of Words Correctly Recalled for Each Recall Set Size: Young Participants with Jittered Stimuli at S/N=4 dB, and Oid Participants with Intact Stimuli at S/N=5 dB  -*-  Jittered, High  —B— Jittered, Low  -•-  Old, High Old, Low  Recall Set-Size  115  materials (Jittered, H i g h ; Jittered, L o w ) in a S / N o f +4 d B . A l s o displayed are the results obtained from elderly listeners to intact stimuli o f both high- and low-context (Old, H i g h ; Old, L o w ) in a S / N o f +5 d B . B y comparing the word-recall performance o f the young in this condition to those in Figure 32, it is apparent that when presented with jittered stimuli, word-recall ability declines when the set-size is 8. This finding refutes N u l l Hypothesis 3, which stated that the introduction o f jitter would not affect recall ability., N u l l Hypothesis 4, which proposed that set-size would not interact with jitter is also refuted, because regardless o f whether the stimuli was jittered or intact, when set-size was 2, recall performance was close to 100%, and when the set-size was 8, performance dropped from an average o f 84.8% to an average o f 79.0% with the introduction of jitter. Figure 33 also demonstrates that the performance o f both groups is clearly affected by context, thereby refuting N u l l Hypothesis 3b. Y o u n g participants with jittered stimuli performed quite similarly to old participants with intact stimuli, thereby addressing N u l l Hypothesis 5, which stated that the application of jitter would not result in a resemblance between the word-recall performance o f young and old adults. The implications o f the findings w i l l be discussed next in the General Discussion.  •  '  .  116  5. G E N E R A L D I S C U S S I O N  5.1  Introduction This series o f experiments had two related purposes: The first was to garner  further evidence for the hypothesis that an increase in neural asynchrony may account for some o f the speech perception difficulties reported by older adults. The second purpose was to explore the possibility that this proposed age-related neural asynchrony causes listening to be more effortful and requires more processing resources for audition thereby leaving less resources for manipulating and retaining heard information. These hypotheses were explored by attempting to "simulate perceptual deficits i n younger participants to mimic those found in older participants" (Schneider & Pichora-Fuller, 2000; pp. 185-186). For this purpose, jitter, or asynchrony believed to resemble that imposed by the aged auditory system was introduced to alter the speech signal presented to young adult listeners.  5.2  W o r d Identification  '  In regard'to the first o f the hypotheses, it was found that when the low-frequency jitter was applied to frequency components below 1200 H z , word-identification performance became almost identical to that obtained from elderly participants for the low-context S P l N - R sentences. Because these sentences are believed to assess auditory ability while minimizing cognitive input, this result supports the hypothesis that a disruption to the fine-structure o f speech, as might result from neural asynchrony, had deleteriously affects on speech perception. This finding corroborates that o f Pass (1998)  who applied various forms o f jitter to the speech signal, and also found that it significantly reduced word-identification performance for young adults with normal hearing. Recall that Pass (1998) also jittered frequency components above 1200 H z such that high-frequency masking resulting from the jittering could have accounted for the . reduction i n word recognition that she observed. This possibility was ruled out by the present experiment, and the match to the data o f the older adult was actually improved. When the results from the 1995 study were compared to those o f the present experiment, it was found that for high-context S P I N - R sentences, elderly participants are better able to identify sentence-final words when the S / N is moderately difficult than are the young participants with jittered sentences. Because some S / N conditions that are easy for young adults are typically difficult for older adults, older adults are forced to depend on context more often i n everyday listening situations to recover information lost i n perceptual processing. Gordon-Salant and Fitzgibbons (1997) found that elderly listeners, even those with hearing loss, are able to take advantage o f contextual cues to surmount speech-understanding difficulties. Wingfield and Stine-Morrow (2000) note that declines in bottom-up processing associated with aging are i n part off-set by use o f contextual cues, and Pichora-Fuller et al. (1995) found that old participants benefit more from context than do young participants. Overall, these results are consistent with the view espoused by Craik (1983) that older adults require and benefit more from supportive context.  The findings o f this study thus provide further evidence that for older adults,  top-down processes are likely employed more often to assist in the interpretation o f incoming perceptual information than they are for younger adults. This aspect o f the word identification ability o f older listeners was not so well matched when younger  118 adults heard jittered input, presumably because the'young continue to use their usual, less successful, top-down processing strategies.  5.3  W o r d Recall In the second experiment, the word-identification results from the first experiment  were replicated, and it was found that when an asynchrony was introduced, the wordrecall ability in young participants declined significantly. This finding is consistent with the working-memory model in which there is a finite amount o f available resources to be allocated to various processes; the allocation o f resources to one process reduces the amount o f resources left for the operation o f other processes (Carpenter et al., 1994; 1995). In this case, the jittered speech signal made perception difficult and thus required listeners to engage more top-down resources for perceptual processing. Such allocation would leave fewer resources for the deep encoding, manipulation, and storage o f information, as is evidenced by the reduced word-recall scores. The results o f this experiment are consistent with the literature, which has demonstrated that increasing processing demands reduces storage capacity (e.g., Craik et al., 1998, as cited i n Schneider & Pichora-Fuller, 2000; Gordon-Salant & Fitzgibbons, 1997; Pichora-Fuller, 1996; Rabbitt, 1968). Furthermore, it appears that when presented with jittered stimuli, the word-recall ability o f younger adults became similar to that o f older adults who heard intact stimuli. It has been proposed that because o f age-related deterioration i n the auditory system, the very act o f listening necessitates more top-down processing to rescue the perceptual signal. The use o f contextual cues, for example, consumes more o f the central pool o f re-  119 allocable resources, leaving less for memory processes (e.g., Pichora-Fuller et al., 1995; Schneider, 1997; Schneider & Pichora-Fuller, 2000). The finding that the introduction o f jitter decreased word-recall ability in young participants in a manner similar to the decreased ability in older adults supports the hypothesis that external jitter resembles the internal jitter inherent i n the aged auditory system, and results i n similar processing demands. For both older adults with intact stimuli and younger adults with jittered stimuli, fewer resources are available for remembering. The findings from Experiment 1 suggest that by jittering the stimuli in this manner, the perceptual processing demands experienced by older adults were replicated. The fact that there was a slight difference in recall performance between elderly adults and young adults with jittered stimuli suggests that although the increased resources allocated to the perceptual channel may explain a large part o f the word-recall deficits, something else further contributed to the noted recall difficulties. This too, is in keeping with a large body o f research that proposes that the memory difficulties experienced by older adults have a cognitive component. A n age difference is often noted i n working memory as the task complexity increases (e.g., Obler et al., 1991; Wright, 1980). In the comparison between the present experiment and the Pichora-Fuller et al. (1995) study, the noted difference occurred when the memory load was high (recall set-size o f 8), and when the listening conditions were moderately difficult (S/N o f 4 or 5 d B ) . It is thus not surprising that a slight difference occurred and, in fact, this finding further supports the notion that perceptual and cognitive factors cannot be considered i n isolation, particularly in the aged population.  . 1 2 0  Although it appears that both perceptual and cognitive deficiencies contribute to the noted word-recall difficulties in older adults, the findings o f the current study and those o f Pichora-Fuller et al. (1995) suggest that perceptual processing receives priority in resource allocation. In both o f these experiments, word-identification performance was not affected by an increased memory load, even though increased perceptual stress reduced recall.. F r o m the perspective o f an information-processing model, it makes sense that priority is given to perceptual processing; preserving resources for memory and other cognitive processes would be o f little benefit i f the cost were a loss or reduction i n quality o f incoming information. Similar asymmetries o f effects have been reported by L i , Lindenberger, Freund, and Baltes (2000), in a study involving motor skills and memory tasks, i n which older adults were found to prioritize walking over memorizing.  5.4  Final Discussion This simulation o f the neural jitter that is thought to disrupt the ability to phase-  lock to lower frequencies i n the aging auditory system was successful. The applied jitter was determined to. affect young adults' performance such that it resembles that o f older adults i n two different tasks. Such a finding helps explain why there tends to be an agerelated decline i n the ability to perceive speech, particularly in the presence o f background noise, when no hearing loss is evidenced by standard clinical pure-tone audiometry. The findings also help explain some aspects o f working-memory problems experienced when information is presented auditorily under challenging listening conditions. ,  ' ' .  ,  121 Furthermore, this series o f experiments provides evidence for the interrelationship between perceptual (bottom-up), and cognitive (top-down) channels. It was demonstrated that degraded perceptual processing affects processes, such as memory, which are traditionally believed to be cognitive. Cognitive processes, such as the use o f available semantic cues to help one predict content, were also shown to be employed to rescue signals degraded at the perceptual level. A s Gordon-Salant and Fitzgibbons note, "the performance o f elderly listeners... is influenced by a combination o f auditory processing factors, memory demands, and speech contextual information" (1997, p. 423). This experiment clearly indicates that the importance o f perceptual acuity cannot be ignored when discussing what has hitherto been conceived o f as a strictly cognitive process, especially in the aging population. The present findings also have implications for the everyday lives o f older adults. Although this study attempted to explain the auditory processing difficulties witnessed in older adults who have audiometrically normal hearing, the majority o f older adults also have other aspect o f presbycusis, such as cochlear hearing loss that predominantly affects the high frequencies. This loss typically reduces the intensity o f incoming sounds and adds additional distortion to the speech signal. Most older adults, therefore, have to deal with a combination o f factors that corrupt both the integrity o f an incoming signal and the fidelity with which it is transmitted up the auditory pathway. A s was evidenced in the current study, perceptual impairment cascades and affects higher-level processes, such that older people not only have a more difficult time identifying the signal initially, but they also have a more difficult time manipulating, integrating, and storing it. In turn, comprehension and discourse performance is compromised. A s an increase in the  .122  percentage o f the population approaches old age, this reduction in language comprehension and conversational abilities has serious social implications, especially given the fact mat most everyday listening situations are less than ideal. The results o f this study underscore the importance o f enhancing the communication environment tb accommodate older adults, even i f they have audiometrically normal pure-tone thresholds. Such strategies include increasing S/N, by reducing background noise and minimizing the detrimental effects o f reverberation by avoiding conversation areas in which there are many reflective surfaces. Conversational partners should also be aware o f the benefit obtained from contextual cues, and attempt to orient the listener to the topic.  5.5  .  Future Directions Most people note that their speech perception difficulties are increased when they  do not have accompanying visual cues, either because the conversation is held over the telephone, or because the speaker's face is obstructed. V i s u a l cues are even more important when auditory cues are compromised; Pichora-Fuller (1996) demonstrated that word-recall ability was enhanced when visual cues were provided with the S P I N - R sentences. In a similar format to that followed in the present set o f experiments, young participants were required to recall the sentence-final words i n a number o f S / N conditions. It was found that when the S / N was very difficult (-8 d B ) , participants were significantly better at recalling words when they also had access to the cues provided by . the speaker's face. Because most conversations do occur face-to-face, it would be interesting to determine the benefit provided by visual cues when the speech signal is  123 degraded by jitter. A recent study (Sommers & Tye-Murray, 2000) reported that older adults with poor binaural gap-detection abilities were poorer integrators o f auditory and visual information. In light o f the current findings regarding auditory jitter, it would be interesting, td further tease out the relationships between perceptual acuity and language • comprehension. Although numerous behavioural tasks have been'devised to assess auditory temporal processing abilities', the physiological studies have mainly been confined to animal research. It would be interesting to garner further physiological support for the notion o f neural asynchrony through such tools as auditory brainstem responses, which rely on neural synchrony, to replicate in humans that which has been demonstrated in animals (Boettcher et al., 1993; Hellstrom & Schmiedt, 1990). Recently, in an exciting use o f available technology, Frisina (2000) conducted P E T imaging experiments to study the changes i n human brain activity during the processing o f speech in background noise. M o r e research i n this direction would be an asset to hearing science. Perhaps one o f the more important things to be derived from the current series o f experiments is the importance o f collaborative research. Maintaining the dichotomy between perceptual and cognitive research results in a limited understanding o f both; i n contrast, an integration o f both yields a more accurate picture o f language . comprehension. Furthermore, the holistic perspective gained from an increased emphasis on collaborative and centripetal research not only fosters an appreciation o f the complex inter-relationships between areas o f research, but also benefits specific research projects.  124  REFERENCES  A b e l , S . M . , Krever, E . M . , & Alberti, P.W. (1990). Auditory detection, discrimination and speech processing in aging, noise-sensitive and hearing-impaired listeners. Scandinavian Audiology, 19, 43-54 • , A d a m , T . J (2000). Signal generation in the Lateral Superior Olive: The rate code o f interaural disparities o f sound. P h . D . thesis completed at the University o f British Columbia. Baddeley, A . D . ; (1986). Working memory. N e w Y o r k : Oxford University Press. Baltes, P.B., & Lindenberger, U . (1997). Emergence o f a powerful connection between sensory and cognitive functions across the adult life span: A new window to the study o f cognitive aging? Psychology and A g i n g , 12,' 12-21: Bergman, M . (1980). A g i n g and the Perception o f Speech. Baltimore: University Park Press. , . Bilger, R . C . , Nuetzel, M . J . , Rabinowitz, W . M . , & Rzeczkowski, C . (1984). Standardization o f a test o f speech perception in noise. Journal o f Speech and Hearing Research, 27, 32-48. Boettcher, F . A . , M i l l s , J.H., & Norton, B . L . (1993). Age-related changes in auditory evoked potentials o f gerbils: I. Response amplitudes. Hearing Research, 71, 137145, Carpenter, P . A . , Miyake, A . , & Just, M . A . (1994). W o r k i n g memory constraints in comprehension: Evidence from individual differences, aphasic and aging. In M . A . Gernsbacher (Ed.), Handbook o f psycholinguistics (pp. 1075-1122). San Diego, C A : Academic Press. Carpenter, P . A . , Miyake, A . , & Just, M . A . (1995). Language comprehension: Sentence and discourse processing. Annual Review o f Psychology, 46, 91-120. Cerella, J. (1985). Information processing rates in the elderly. Psychological Bulletin, 98, 67-83. C H A B A (Committee on Hearing, Bioacoustics, and Biomechanics). (1988). Speech understanding and aging. Journal o f the Acoustical Society o f America, 83, 859895. Chmiel, R., & Jerger, J. (1996). Hearing aid use, central auditory disorder, and hearing handicap i n elderly persons. Journal o f the American Academy o f Audiology, 7, 190-202.  125 Cohen, G . (1979). Language comprehension in old age. Cognitive Psychology, 11,412-429.  .226  .  Cohen, G . (1987). Speech Comprehension. British Journal o f Audiology, 21, 221'  Cohen, G . , & Faulkner, D . (1983). A g e differences in two information processing tasks: strategy selection or processing capacity? Journal o f Gerontology, 38, 447-454. Craik, F . I . M . (1983). O n the transfer o f information from temporary to permanent memory. Philosophical Transactions o f the R o y a l Society o f London, 302, 341-359. Craik, F.I.M,' & Jennings, J . M . (1992): Human Memory. In F . I . M Craik & T. Salthouse (Eds.) The Handbook o f A g i n g and Cognition, (pp.51-110). N e w Jersey: Lawrence Erlbaum Associates. , ' Culling, J.F., & Darwin, C . J . (1994). Perceptual and computational separation o f simultaneous vowels: Cues arising from low-frequency beating. Journal o f the Acoustical Society o f America, 95, 1559-1569. Culling, J.F. & Summerfield, Q. (1995). Role o f frequency modulation i n the perceptual segregation o f concurrent vowels. Journal o f the Acoustical Society o f America, 98, 837-846. Daneman, M , & Carpenter, P . A . (1980). Individual differences ih working memory and reading. Journal o f Verbal Learning and Verbal Behaviour, 19, 450-466. ;  Daneman, M . , & Merikle, P . M . (1996). Working memory and language comprehension: A meta-analysis. Psychonomic Bulletin and Review, 3, 450-466. Darwin, C.J. & Gardner, R . B . (1986). Mistuning a harmonic o f a vowel: grouping and phase effects on vowel quality. Journal o f the Acoustical Society o f America, 79, 838-845. de Cheveigne, A . , M c A d a m s , S., & M a r i n , C . M . H . (1997). Concurrent v o w e l identification. II. Effects of phase, harmonicity and task. Journal o f the Acoustical Society o f America, 101, 2848-2856. D i v e n y i , P . L . (2000). Beyond presbyacusis: Non-hearing loss-related temporal processing deficits i n the elderly. Journal o f the Acoustical Society o f America, 107, ,2796. Dreschler, W . A . , & Plomp, R. (1985). Relations between psychophysical data and speech perception for hearing impaired subjects. II. Journal o f the Acoustical Society o f America, 78, 1261-1270.  126 Drullman, R . (1995a). Temporal envelope and fine structure cues for speech intelligibility. Journal o f the Acoustical Society o f America, 97, 585-592. Drullman, R . (1995b). Speech intelligibility i n noise: Relative contribution o f speech elements above and below the noise level. Journal o f the Acoustical Society o f America, 98,1796-1798. Drullman, R., Festen, J . M . , & Plomp, R. (1994a). Effect o f temporal envelope smearing on speech reception. Journal o f the Acoustical Society o f America, 95, 10531064Drullman, R., Festen, J . M . , & Plomp, R . (1994b). Effect o f reducing slow temporal modulations oh speech perception. Journal o f the Acoustical Society o f America, 95, 2670-2680. Dubno, J.R., Dirks, D . D . , & Morgan, D . E . (1984). Effects o f age and m i l d hearing loss on speech recognition i n noise. Journal o f the Acoustical Society o f America, 76, 8796. Duquesnoy, A . J . (1983). The intelligibility o f sentences in quiet and noise i n aged listeners. Journal o f the Acoustical Society o f America, 74, 1136-1144. Duquesnoy, A . J . & Plomp, R. (1980). Effect o f reverberation and noise on the intelligibility o f sentences i n cases o f presbycusis. Journal o f the Acoustical Society o f America, 68, 537-544. Durlach, N . I . (1972). Binaural signal detection: Equalization and cancellation theory. In J . V . Tobias (Ed.), Foundations o f Modern Auditory Theory, V o l . 2 . N e w Y o r k : Academic Press, 374-362. Fitzgibbons, P.J., & Gordon-Salant, S. (1994). A g e effects on measures o f auditory duration discrimination. Journal o f Speech and Hearing Research, 37, 662-670. Fitzgibbons, P.J., & Gordon-Salant, S. (1995). A g e effects on duration discrimination with simple and complex stimuli. Journal o f the Acoustical Society o f America, 98,3140-3145. Fitzgibbons, P.J., & Gordon-Salant, S. (1996). Auditory temporal processing i n elderly listeners. Journal o f the Acoustical Society o f America, 7, 183-189. Frisina, R . D . (2000). Neural substrates for presbycusis: Anatomy, chemistry, and neuroimaging o f the central auditory system. Journal o f the Acoustical Society of, America, 107, 2798. Frisina, D.R., & Frisina, R . S . (1997). Speech recognition i n noise and presbycusis: relations to possible neural mechanisms. Hearing Research, 106, 95-104.  127  Gelfand, S.A., Piper, N . , & Silman, S. (1985). Consonant recognition in quiet as a function o f aging among normal hearing subjects. Journal o f the Acoustical Society o f America, 78, 1198-1206. Glasberg, B . R . , & Moore, B . C . J. (1986). Auditory filter shapes i n subjects with unilateral and bilateral cochlear impairments. Journal o f the Acoustical Society o f America. 55, 297-303. Gordon-Salant, S., & Fitzgibbons, P.J. (1993). Temporal factors and speech recognition performance in young and elderly listeners. Journal o f Speech and Hearing Research, 36, 1276-1285. Gordon-Salant, S., & Fitzgibbons, P.J.' (1995). Recognition o f multiply, degraded speech by young and elderly listeners. Journal o f Speech and Hearing Research, 38, 1150-1156. .. . . Gordon-Salant, S., & Fitzgibbons, P.J. (1997). Selected cognitive factors and speech recognition performance among young and elderly listeners. Journal o f Speech, Language, and Hearing Research, 40, 423-431. Greene, D . M . (1976). A n Introduction to Hearing. N e w Y o r k : Lawrence Erlbaum Associates. Greenberg, S. (1996). Auditory processing o f speech. In N . J . Lass (Ed.), Principles o f Experimental Phonetics (pp. 362-407). St. Louis: Mosby. Greenberg, S., & A r a i , T. (1998). Speech intelligibility is highly tolerant o f crosschannel spectral asynchrony, Journal o f the Acoustical Society o f America, 103, 3057. Grose, J.J., Poth, E . A . , & Peters, R . W . (1994). M a s k i n g level differences for tones and speech i n elderly listeners with relatively normal audiograms. Journal o f Speech and Hearing Research, 37,422-428. Hartmann, W . M . (1996). Pitch, periodicity and auditory organization. Journal o f the Acoustical Society o f America, 100, 3491-3502. Haubert, N . (1999). Temporal perception in older listeners. M . S c . thesis completed at the University o f British Columbia. He, N . , Dubno, J.R., & M i l l s , J . H . (1998). Frequency and intensity discrimination measured in a maximum-likelihood procedure from young and aged normal-hearing subjects. Journal o f the Acoustical Society o f America, 103, 553-565. Hellstrom, L . I . , & Schmiedt, R . A . (1990). Compound action potential input/output functions in young and quiet-aged gerbils. Hearing Research, 50, 163-174.  128  Hochmair-Desoyer, I.J., Stiglbrunner, H . K . , & Wallenberg, F . L . (1984). Relationships between speech understanding and temporal processing abilities i n cochlear implant patients. Journal o f the Acoustical Society o f America, Suppl. 1., 76, S48. Humes, L . E . , & Christopherson, L . (1991). Speech identification difficulties o f hearing-impaired elderly persons: the contributions o f auditory processing deficits. Journal o f Speech and Hearing Research, 34, 686-693. . Irvine, D . R . F . (1986) The Auditory Brainstem: A Review o f the Structure and Function o f Auditory Brainstem Processing Mechanisms. In Ottoson, D . (Ed.) Progress i n Sensory Physiology V I I . (pp. 279). N e w York, N . Y . : Springer-Verlag. Irwin, R.J., 8i McCauley, S.F. (1987). Relations among temporal acuity, hearing loss and the perception o f speech distorted by noise and reverberation. Journal o f the Acoustical Society o f America, 81, 1557-1565. Just, M . A . , & Carpenter, P . A . (1992). A capacity theory o f comprehension; Individual differences in working memory. Psychological Review, 99, 122-147. K a l i k o w , D . N . , Stevens, K . N . , & Elliott, L . L . (1977). Development o f a test o f speech intelligibility in noise using sentence materials with controlled word predictability. Journal o f the Acoustical Society o f America, 61, 1337-1351. K o n i g , E . (1957). Pitch discrimination and age. A c t a Otolaryngologica, 48, 475489. ' Letowski, T., & Poch, N . (1996). Comprehension o f time-compressed speech: Effects o f age and speech complexity. Journal o f the American Academy o f Audiology, 7,447-457. L i , K . Z . H . , Lindenberger, U . , Freund, A . M . , & Baltes, P . B . (2000). W a l k i n g while memorizing: A g e differences in external aid use and task priority as compensation during concurrent task performance. Cognitive A g i n g Conference, Atlanta. Lindenberger, U . , & Baltes, P . B . (1994). Sensory functioning and intelligence i n old age: A strong connection. Psychology and A g i n g , 9, 339-355. Lindenberger, U . , M a y r , L i . , & K l i e g l , R. (1993). Speed and intelligence in o l d age. Psychology and A g i n g , 8, 207-220. M c K e o w n , J.D., & Patterson, R . D . (1995). Time course o f auditory segregation: Concurrent vowels that vary i n duration. Journal o f the Acoustical Society o f America, 98,1866-1877.  129 Moore, B . C . J . (1997). A n Introduction to the Psychology o f Hearing, (4 Academic Press: San Diego.  th  ed.).  Moore, B . C . J . , & Glasberg, B . R . (1988). Gap detection with sinusoids and noise in normal, impaired, and electrically stimulated ears. Journal o f the Acoustical Society o f America, 83, 1093-1101. Moore, B . C : J . , Peters, R . W . , & Glasberg, B . R . (1992). Detection o f temporal gaps in sinusoids by elderly subjects with and without hearing loss. Journal o f the Acoustical Society o f America, 92, 1923-1932. Nabelek, A . K . , & Robinson, P . K . (1982). Monaural and binaural speech perception in reverberation for listeners o f various ages. Journal o f the Acoustical Society o f America, 71, 1242-1249. Neils, J., Newman, C . W . , H i l l , M . , & Weiler, E . (1991). Effects o f rate, sequencing, and memory on auditory processing in the elderly. Journal o f Gerontology, 46,P71-P75. Noble, W . , Byrne, D . , & LePage, B . (1994). Effects on sound localization o f configuration and type o f hearing impairment. Journal o f the Acoustical Society o f America, 95, 992-1005. Obler, I.K., Fein, D . , Nicholas, M . , & Albert, M . L . (1991). Auditory comprehension and aging: Decline in syntactic processing. Applied Psycholinguistics, 12, 433-452. Park, D . C . , Smith, A . D . , Lautenschlager, G . , Earles, J.L., Frieske, D . , Zwahr, M . , & Gaines, C L . (1996). Mediators o f long-term memory performance across the life span. Psychology and A g i n g , 11, 621-637. Parker, S., & Schneider, B . A . (1994). The stimulus range effect: Evidence for topdown control o f sensory intensity in audition. Perception and Psychophysics, 56, 1-11. Pass, H . (1998). Importance o f Temporal Synchrony i n W o r d Recognition. M . S c . thesis completed at the University o f British Columbia. Patterson, R . D . , & Moore, B . C . J . (1986). Auditory filters and excitation patterns as representations o f frequency resolution, In B . J . C Moore (Ed.), Frequency Selectivity in Hearing. London: Academic, pp 67-83. Phillips, D.P. (1989). Neurobiology relevant to some central auditory processing disorders. Journal o f Speech-Language Pathology and Audiology, 13, 17-41. Phillips, D.P. (1995). Central auditory processing: A view from auditory neuroscience. American Journal o f Otology, 16, 338-352.  130  Pichora-Fuller, M . K . (1996). Speechreading and working memory. In D . G . Stork . & M . E . Hennecke (Eds.), Speechreading by humans and machines: Models, systems, and applications, (pp. 256-274). Berlin, Germany: Springer-Verlag. Pichora-Fuller, M . K . , & Schneider, B . A . (1991). Masking-level differences i n the elderly: a comparison o f antiphasic and time-delay dichotic conditions. Journal o f Speech and Hearing Research, 34,1410-1422. , Pichora Fuller, M . K . , & Schneider, B-.A. (1992). The effect o f interaural delay o f the masker on masking-level differences in young and old adults. Journal o f the Acoustical Society o f America, 91, 2129-2135. :  Pichora-Fuller, M . K . , & Schneider, B . A . ,(1998). Masking-level differences i n older adults: the effect o f the level o f masking noise. Perception and Psychophysics, 60, 1197-1205. , Pichora-Fuller, M . K . , Schneider, B . A . , & Daneman, M . (1995). H o w young and old adults listen to and remember speech in noise. Journal o f the Acoustical Society o f America, 97, 593-607. Pichora-Fuller, M . K . , Schneider, B . A . , Pass, H . , & Brown, S. (submitted). Temporal jitter disrupts speech intelligibility. , Pickles, J.O. (1988). A n introduction to the physiology o f hearing ( 2 Diego, C A : Academic Press.  nd  ed.). San  Plomp, R. (1986). A signal-to-noise ratio model for the speech reception threshold o f the hearing impaired. Journal o f Speech and Hearing Research, 29, 146-154. Plomp, R., & M i m p e n , A . M . (1979). Improving the reliability o f testing the speech reception threshold for sentences. Audio logy, 18, 43-52. Poon, L , W . (1985). Differences in human memory with aging: Nature, causes and clinical implications, In J.E. Birren & K . W . Schaie (Eds.), Handbook o f the Psychology o f A g i n g , (pp.427-462). N e w Y o r k : V a n Nostrad Reinhold. Price, P.J., & Simon, H.J., (1984). Perception o f temporal differences by 'normalhearing' adults: Effects o f age and intensity. Journal o f the Acoustical Society o f America, 76, 405-410. Rabbitt, P . M . (1968). Channel capacity, intelligibility and immediate memory: Quarterly Journal o f Experimental Psychology, 20, 241-248.  131 Rabbitt, P . M . (1991). M i l d hearing loss can cause apparent memory failures which increase with age and reduce with IQ. A c t a Otolaryngologica, 476 (suppl.), 167176. Raven, J.C. (1938). The M i l l H i l l Vocabulary Scale. London: Lewis. • Reinsche, L . , Lawson, G . , Beasley, D.S.,' & Smith, L . L . (1979). A g e and sex differences on preferred listening rates for speech. The Journal o f Auditory Research, 19, 91-94. Salthouse, T . A . (1982). Adult Cognition: A n Experimental Psychology o f Human A g i n g , (pp.111-114). Springer-Verlag, N e w Y o r k . Salthouse, T . A . (1985). Speed o f behavior and its implications for cognition. In J.E. B i r r i n & K . W . Schaie (Eds.), Handbook o f the Psychology o f A g i n g . N e w Y o r k : V a n Nostrand Reinhold, pp. 400-426. Salthouse, T . A . (1993). Speed mediation o f adult age differences in cognition. Developmental Psychology, 29, 722-738. Salthouse, T . A . (1996). The processing speed theory o f adult age differences in cognition. Psychological Review, 103, 403-428. Schmiedt, R . A . & Lang, H . (2000). Functional changes in the ear with o l d age: A review. Journal o f the Acoustical Society o f America, 107, 2797. Schmitt, J.G., & Carroll, M . R . (1985). Older listeners' ability to comprehend speaker-generated rate alteration o f passages/Journal o f Speech and Hearing Research, 28,309-312. Schmitt, J.F., & Moore, J.R. (1989). Natural alteration o f speaking rate: The effect on passage comprehension by listeners over 75 years o f age. Journal o f Speech and Hearing Research, 32, 445-450. Schneider, B . A . (1997). Psychoacoustics and aging: Implications for everyday • , listening. Journal o f Speech-Language Pathology and Audiology, 21, 111-124. Schneider, B . A . , & Hamstra, S J . (1999). Gap detection thresholds as a function o f tonalduration for younger and older listeners. Journal o f the Acoustical Society o f America, 106, 371-380. Schneider, B . A . , Pichora-Fuller, M . K . (2000). Implications o f perceptual deterioration for cognitive aging research. In F . I . M Craik & T . A . Salthouse (Eds.), The Handbook o f A g i n g and Cognition, 2 Edition, N e w Jersey: N e w Erlbaum Associates, pp. 155-219. n d  132 Schneider, B . A . , Pichora-Fuller, M . K . , Kowalchuck, D . , & L a m b , M . (1994). Gap detection and the precedence effect in young and o l d adults. Journal o f the Acoustical Society o f America, 95, 980-991. Shannon, R . V . , Zeng, F . , Kamath, V . , Wygonski, J., Elelid. (1995). Speech recognition with primarily temporal cues. Science, 270, 303-304. Snell, K . B . (1997). Age-related changes in temporal gap detection. Journal o f the Acoustical Society o f America, 101, 2214-2220. Snell, K . B . , & Frisina, D.R. (2000). Age-related differences in temporal processing and speech perception in noise. Journal o f the Acoustical Society o f America, 107,2796. Sommers, M . S . , & Tye-Murray, N . (2000). Auditory-visual integration in younger and older adults. Cognitive A g i n g Conference, Atlanta. Stine, E . A . L . , Wingfield, A . & Poon, L . W . (1986). H o w much and how fast: Rapid processing o f spoken language i n later adulthood. Psychology and A g i n g , 1, 303311. Strouse, A . , Ashmead, D . H . , Ohde, R . N . , & Grantham, D . W . (1998). Temporal processing in the aging auditory system. Journal o f the Acoustical Society o f America, 104,2385-2399. ' Stuart, A . , & Phillips, D . P . (1994). W o r d recognition in continuous arid interrupted broadband noise by young normal-hearing, older normal-hearing, and presbyacusic listeners. Ear and Hearing, 17, 478-489. Trainer, L . J . , & Trehub, S.E. (1989). A g i n g and auditory temporal sequencing: Ordering the elements o f repeating tone patterns. Perception and Psychophysics, 45, 417426. . Tun, P . A . , Wingfield, A . , & Stine, E . A . L . (1991). Speech-processing capacity in young and older adults: A dual-task study. Psychology and A g i n g , 6, 3-9. Tyler, R.S., Summerfield, Q., Wood, E.J., & Fernandes, M . A . (1982). Psychoacoustic and phonetic temporal processing i n normal and hearing-impaired listeners. Journal o f the Acoustical Society o f America, 72, 740-752. van Rooij, J . C . G . M . , Plomp, R.> & Orlebeke, J.F. (1989). Auditive and cognitive factors in speech perception by elderly listeners. I: Development o f test battery. Journal o f the Acoustical Society o f America, 86, 1294-1309.  133  ' Van Tasell, D J . , Soli, SID;, K i r b y , V M . . & Widen, G . P . (1987). Speech waveform envelope cues for consonant recognition. Journal o f the Acoustical Society o f America, 82, 1152-1161. V a n Tasell, D.J., Greenfield, D . G . , Logemann, J.J., & Nelson, D . A . (1992). Temporal cues for consonant recognition: Training, talker generalization, and use i n evaluation o f cochlear implants. Journal o f the Acoustical Society o f America, 92, 12471257. \ • • Verhaeghen, P., Marcoen, A . , & Goosens, L . (1993). Facts and fiction about memory aging: A quantitative integration o f research findings. Journal o f Gerontology: Psychological Sciences, 48, P157-P171. Viemeister, N . F . , & Plack, C J . (1993). Time analysis, In W . A . Yost, A . N . Popper & R.R. Fay (Eds.), Human Psychophysics (pp.116-154). N e w Y o r k : Springer Verlag. Walton, J.P. (2000). Physiological correlates o f presbycusis: Single-neuron recordings in awake animals. Journal o f the Acoustical Society o f America, 107, 2797. Wilber, L . A . (1994). Calibration, Puretone, Speech and Noise Signals. In J. K a t z (Ed.), Handbook o f Clinical Audiology (4 ed., pp.83). Baltimore: W i l l i a m s & Wilkens. th  Willot, J.F. (1991). A g i n g and the Auditory System: Anatomy, Physiology, and Psychophysics..San Diego: Singular. Wingfield, A . , & Ducharme, J.L. (1999). Effects o f age and passage difficulty on listening-rate preferences o f time-altered speech. Journal o f Gerontology: Psychological Sciences, 54B, 199-202. Wingfield, A . , & Stine-Marrow, E . A . L . (2000). Language and speech, In F.I. M Craik & T . A . Salthouse (Eds.), The Handbook o f A g i n g and Cognition, 2 Edition, (pp. 359-416). N e w Jersey: Lawrence Erlbaum Associates. n d  Wingfield, A . , Tun, P . A . , K o h , C . K . , & Rosen, M . J . (1999). Regaining lost time: Adult aging and the effect o f time restoration on the recall o f time-compressed speech. Psychology and A g i n g , 14, 380-389. Wright, R . E . (1981). A g i n g , divided attention and processing capacity. Journal o f Gerontology, 36, 605-614. Zacks, R.T., Hasher, L . , & L i , K . Z . H . (2000). Human Memory. In F . I . M . Craik & T . A . Salthouse (Eds.), The Handbook o f A g i n g and Cognition, 2 Edition, (pp. 293357). N e w Jersey: Lawrence Erlbaum Associates. n d  APPENDIX A Experiment 1 Participants' Air-Conducted Pure-Tone Thresholds (dB H L ) for Right (R) and Left (L) Ears  Test Frequency (Hz) Participant  250  500  .  R  L  R  L  1  -10  -5  -5  -5  2  5  0  5  10  20  4  10  5 .6  1000 R  2000  4000  L  R  L  5  -5  -5  0  -5,  •5  5  10  0  5  5  10  15  10  20  0  -5  0  5  0  -5  5  10  5  5  5  5  10  5  0  5  5  0  0  10  5  -5  7 •.  5  0  10  5  10  5  8  5.  5  0  0  0  0  0  0 -5  .3  .9  .  ,-5 . •  10  5  0  5  11  10  -5  -5  12  5•  0  5  8000  L  R  L  -5  -5  0  10  15  10  '5  5  15  20  10  0  0  20  5  5*  -5  0  0  -5  -5  0  5  -5  20  0  0  0  0  5  5  0  0  0  -5  10  0  0  -5  10  -5  -5  -5  -5  10  5  0  5  0  5  -5  5  -10  -5  0  -10  5  0  5  10  5 '•  0  -5 • 5 0  ,  5  -5. . 5  . 5 0 -5 • 5  135  APPENDIX B Experiment 1 Participants' Characteristics  Pure-Tone Average  Vocabulary  (dB H L )  Score  Participant  Years o f Age,  3  Education  b  Handedness  (of 20) R  L  •  •  a  b  ''  '  .  •  1  -1.7  6.7  12  20  16  R  2  3.3  6.7  14  23  19  R  3  6.7  10.3  12  26  15  R  4  33  ' 5  12  20  15  ' 5 .  .5 '  1.7  15  6  1.7  7  6.7  . 8  •  0  I  R R  7  16  '28  20  R  3.3  15  21  16  R  0  0  13  25  21  9  -3.3  1.7  15  27  19  R  10  5  -1.7  14  25  19  R  11.  -1.7  -5  12'  16  R  12  5  18  R  Mill Hill, Raven (1938)  3.3 •  this score includes kindergarden  •  , 3 0  -  ' 15  . '  2  1  30  . - '  '  R  APPENDIX C Forms o f the Revised S P I N Sentences  Practice S P I N - R Sentences 1  She had spoken about the S C A R .  2  Camels store water in their F I U M P S .  3  M r . Smith might discuss the M I L L .  4  I built the model from a K I T .  5.'  The loud noise made h i m jump with F R I G H T .  6  Ruth couldn't know about the S H R I M P .  7  W e were discussing the G A S .  8  She's been considering the C O M B .  9  The winning card was an A C E .  10  T o m wore a tie and a white S H I R T .  11  , I ' m glad T o m asked about the S C O U T .  12  Her skin broke out in a R A S H .  13  Paul heard they asked about the R I C E .  14  The child threw bread crumbs to the D U C K S .  15  Paul chopped down the tree with his A X E ;  16  They want to speak about the T R A S H .  17  H e ' s as stubborn as a M U L E .  18  Paul is considering the D R I L L .  19  . The woman talks about the B E A C H .  ( S P I N - R Practice Sentences continued) 20  Hospitals should be free o f G E R M S .  21  Y o u should discuss the F L U T E .  22 '  Let's slide down the hill on a SLED...  23  He weighed the meat on a S C A L E .  24  Paul wants to think about the P A L M .  25  She was interested in the S O A P .  26  Y o u lead them on a merry C H A S E .  27  The swimmer was attacked by S H A R K S .  28  M i s s White spoke about the S A U C E .  29  Stainless steel w i l l never R U S T .  30  He has considered the I T C H .  31  She had a problem with the P L U M S .  32  The robber committed a C R I M E .  33  The secret message was i n C O D E .  34  B r o i l the steak on a charcoal G R I L L .  35  H e ' s been discussing the A C H E .  36  Ruth sat on the living room C O U C H . '  37  B i l l should have considered the H O O D .  38  A bee can give a painful S T I N G .  39  They are discussing the C U R L .  40  Nancy had known about the C O U G H .  41  Jane heard he called about the B U S .  (Practice S P I N - R Sentences continued) 42  The soap washed away the D I R T .  43  H e is discussing the G I F T .  44  Jane was dressed ih a skirt and B L O U S E . •  ,45  She is considering the S H E L F .  46  Animals are kept i n the Z O O .  47  Harry was discussing the M A I L .  48  His face was concealed by a M A S K .  49  H a l f a quart is a P I N T .  50  Ruth is considering the C L O W N .  SPIN-R Form 1 1  .His plans meant taking a big R I S K .  2  Stir your coffee with a S P O O N .  3  M i s s White won't think about the C R A C K .  4  He would think about the R A G .  5.  The plough was pulled by an O X .  6  The o l d train was powered by S T E A M .  7  The o l d man talked about the L U N G S . .  8  I was considering the C R O O K .  9  Let's decide by tossing a C O I N .  10  The doctor prescribed the D R U G .  11  B i l l might discuss the F O A M .  (SPIN-R Form 1 continued) 12'  Nancy didn't discuss the S K I R T .  13  H o l d the baby in your L A P .  14  Bob had discussed the S P L A S H .  15  The dog chewed on a B O N E . ,  16  Ruth hopes he heard about the H I P S .  17  The war was fought with armored T A N K S .  18  She wants to talk about the C R E W .  19 .  They had a problem with the C L I F F .  20  They drank a whole bottle o f G I N .  21  Y o u heard Jane called about the V A N .  22  The witness took a solemn O A T H .  23  W e would consider the F E A S T .  24  B i l l heard we asked about the H O S T .  25  They tracked the lion to his D E N .  26  The cow gave birth to a C A L F .  '27 •  I had not thought about the G R O W L .  28  The scarf was made o f shiny S I L K .  29  The super highway has six L A N E S .  30  He should know about the H U T .  ,31  For dessert he had apple P I E .  32  The beer drinkers raised their M U G S .  33  I ' m glad you heard about the B E N D .  .140 ( S P I N - R F o r m 1 continued) 34  Y o u ' r e talking about the P O N D .  35  The rude remark made her B L U S H .  36  Nancy had considered the S L E E V E S .  37  W e heard about the ticking o f the C L O C K .  38  He can't consider the C R A B .  39-  He killed the dragon with his S W O R D .  40  T o m discussed the H A Y .  41  M a r y wore her hair in B R A I D S .  42 '  She's glad Jane asked about the D R A I N .  43  B i l l hopes Paul heard about the M I S T .  44  W e ' r e lost so let's look at the M A P .  45  N o one was injured in the C R A S H .  46  W e ' r e speaking about the T O L L .  47  M y son has a dog for a P E T .  48  He was scared out o f his W I T S .  49  W e spoke about the K N O B .  50  I've spoken about the P I L E .  SPIN-R Form 2 .1  M i s s Black thought about the L A P .  2  The baby slept in his C R I B .  3  The watchdog gave a warning G R O W L .  ( S P I N - R Form 2 continued) 4  M i s s Black would consider the B O N E .  5  The natives built a wooden H U T .  6  B o b could have known about the S P O O N .  7  U n l o c k the door and turn the K N O B .  8  He wants to talk about the R I S K .  9  H e heard they called about the L A N E S .  10  W i p e your greasy hands on the R A G .  11  She has known about the D R U G .  12  I want to speak about the C R A S H .  13.  The wedding banquet was a F E A S T .  14  I should have considered the M A P .  15  PauL hit the water with a S P L A S H .  16  The ducks swam around on the P O N D .  17.  Ruth must have known about the PIE.  18  The man should discuss the O X .  19  B o b stood with his hands on his H I P S .  20  The cigarette smoke filled his L U N G S .  21  They heard I called about the P E T .  22  The cushion was filled with F O A M .  23  Ruth poured the water down the D R A I N .  24  B i l l cannot consider the D E N .  25  This nozzle sprays a fine M I S T .  .  (SPIN-R F o r m 2 continued) 26 .  The sport shirt has short S L E E V E S .  27  She hopes Jane called about the C A L F . .  28  Jane has a problem with the C O I N .  29  She shortened the hem o f her S K I R T .  .30  Paul hopes she called about the T A N K S .  31  The girl talked about the G I N .  32  , The guests were welcomed by the H O S T .  33  M a r y should think about the S W O R D .  34 .  Ruth should have discussed the W I T S .  35  The ship's Captain summoned his C R E W .  36  Y o u had a problem with the B L U S H .  37  The flood took a heavy T O L L .  38  The car drove off the steep C L I F F .  39  W e have discussed the S T E A M .  40  The policeman captured the C R O O K .  41  The door was opened just a C R A C K .  42  T o m is considering the C L O C K .  43  The sand was heaped in a P I L E .  44  Y o u should not speak about the B R A I D S .  45.  Peter should speak about the M U G S .  46  Household goods are moved in a V A N ,  47  He has a problem with the O A T H .  143 (SPIN-R F o r m 2 continued) 48  F o l l o w the road around the B E N D .  49  T o m won't consider the S I L K .  50  The farmer baled the H A Y .  SPIN-R Form 3 1  K i l l the bugs with this S P R A Y .  2  M r . White discussed the C R U I S E .  3  H o w much can I buy for a D I M E ?  4  M i s s White thinks about the T E A .  5  W e shipped the furniture by T R U C K .  6  He is thinking about the R O A R .  7 ,  She's spoken about the B O M B .  8  M y T. V . has a twelve-inch S C R E E N .  9  That accident gave me a S C A R E . '  10  Y o u want to talk about the D I T C H .  11  The king wore a golden C R O W N .  12  The girl swept the floor with a B R O O M .  13  We're discussing the S H E E T S .  14  The nurse gave h i m firs A I D .  15  She faced them with a foolish G R I N .  16  Betty has considered the B A R K .  17 •  Watermelons have lots o f S E E D S .  •  ( S P I N - R F o r m 3 continued) 18  Use this spray to k i l l the B U G S .  19  T o m w i l l discuss the S W A N .  20  The teacher sat on a sharp T A C K . ,  21-.  Y o u ' d been considering the G E E S E .  22  The sailor swabbed the D E C K .  23  They were interested in the S T R A P .  24  He,could discuss the B R E A D .  25  He tossed the drowning man a R O P E .  26  Jane hopes Ruth asked about the S T R I P E S .  27  Paul spoke a bout the P O R K .  28  The boy have the football a K I C K .  29  The storm broke the sailboat's M A S T .  30  Mr. Smith thinks about the C A P .  31  W e are speaking about the P R I Z E .  32  M r . B r o w n carved the roast B E E F .  33  The Glass had a chip on the R I M .  34  Harry had thought about the L O G S .  35  Bob could consider the P O L E .  36  Her cigarette had a long A S H .  37  Ruth has a problem with the J O I N T S .  38  He is considering the T H R O A T .  39  The soup was served in a B O W L .  (SPIN-R F o r m 3 continued) 40  W e can't consider the W H E A T .  41  The man spoke about the C L U E .  42  The lonely bird searched for its M A T E .  43  Please wipe your feet on the M A T .  44  D a v i d has discussed the D E N T .  45  The pond was full o f croaking F R O G S .  46  He hit me with a clenched FIST.  47  B i l l heard T o m called about the C O A C H .  48  A bicycle has two W H E E L S .  49  Jane has spoken about the C H E S T .  50  M r . White spoke about the F I R M .  SPIN-R Form 4 1  The doctor X-rayed his C H E S T .  2  M a r y had considered the S P R A Y .  3  The woman talked about the F R O G S .  4  The workers are digging a D I T C H .  5  M i s s B r o w n w i l l speak about the G R I N .  6  B i l l can't have considered the W H E E L S .  7  The duck swam with the white S W A N . .  8  Y o u r knees and your elbows are J O I N T S .  9  M r . Smith spoke about the A I D .  '  (SPIN-R F o r m 4 continued) 10  H e hears she asked about the D E C K .  11  Raise the flag up the P O L E .  12  Y o u want to, think about the D I M E .  13  Y o u ' v e considered the S E E D S .  14,  The detectives searched for a C L U E . •  15  Ruth's grandmother discussed the B R O O M .  16  The steamship left on a C R U I S E .  17  M i s s Smith considered the S C A R E .  18  Peter has considered the M A T .  19  Tree trunks are covered with B A R K .  20  The meat from a pig is called P O R K .  21 ,.  The old man considered the K I C K .  22  Ruth poured herself a cup o f T E A .  23  W e saw a flock o f w i l d G E E S E .  24  Paul could not consider the R I M .  25  H o w .did your car get that D E N T ?  26  She made the bed with clean S H E E T S .  27  I've been considering the C R O W N .  28  The team was trained by their C O A C H .  29  I've got a cold and a sore T H R O A T .  30  W e ' v e spoken about the T R U C K .  31  She wore a feather in her C A P .  (SPIN-R Form 4 continued) 32  The bred was made from whole W H E A T .  33  M a r y could not discuss the T A C K .  34  Spread some butter on your B R E A D .  35"  The cabin was made o f L O G S .  36  Harry might consider the B E E F .  37  W e ' r e glad B i l l heard about the A S H .  38  The lion gave an angry R O A R .  39  The sandal has a broken S T R A P .  40  Nancy should consider the FIST.  41  He's employed by a large F I R M .  42  They did not discuss the S C R E E N  43  Her entry should w i n first P R I Z E .  44  The old man thinks about the M A S T .  45 .  Paul wants to speak about the B U G S .  46  The airplane dropped a B O M B .  47  Y o u ' r e glad she called about the B O W L .  48  A zebra has black and white S T R I P E S .  49 '  M i s s B l a c k could have discussed the R O P E .  50  I hope Paul asked about the M A T E .  S P I N - R F o r m ,5 Betty knew about the N A P . 2  The girl should consider the F L A M E .  3  It's getting dark, so light the L A M P .  4  To, store his wood he built a S H E D .  5  They heard I asked about the B E T .  6  The mouse was caught in the T R A P .  7  M a r y knows about the R U G .  8  The airplane went into a D I V E .  9  The fireman heard her frightened S C R E A M .  TO  H e was interested in the H E D G E .  11  H e wiped the sink with a S P O N G E .  12  Jane did not speak about the S L I C E .  13  M r . B r o w n can't discuss the S L O T .  14  The papers were held by a C L I P .  15  Paul can't discuss the W A X .  16  M i s s B r o w n shouldn't discuss the S A N D .  17 ,  The chicks followed the mother H E N .  18  D a v i d might consider the F U N .  19  She wants to speak about the A N T .  20  The fur coat was made o f M I N K .  21  The boy took shelter in a C A V E .  22  H e hasn't considered the D A R T .  (SPIN-R F o r m 5 continued) 23  Eve was made from A d a m ' s R I B .  24  The boat sailed along the C O A S T .  25  W e ' v e been discussing the C R A T E S .  26  , The judge is sitting on the B E N C H .  27  W e ' v e been thinking about the F A N .  28  Jane didn't think about the B R O O K .  29  Cut a piece o f meat from the R O A S T .  30  Betty can't consider the C R I E F .  31  The heavy rains caused a F L O O D .  32  The swimmer dove into the P O O L .  33  Harry w i l l consider the T R A I L .  34  Let's invite the whole G A N G .  35  The house was robbed by a T H I E F .  36  Tom is talking about the F E E .  37  Bob wore a watch on his W R I S T .  38  T o m had spoken about the P I L L .  39  T o m has been discussing the B E A C H .  40  The secret agent was a S P Y .  41  The rancher rounded up his H E R D .  42  T o m could have thought about the S P O R T .  43  M a r y can't consider the T I D E .  44  A n n works in the bank as a C L E R K .  (SPIN-R F o r m 5 continued) 45,  A chimpanzee is an A P E .  46  He hopes T o m asked about the B A R .  47  W e could discuss the D U S T .  48  The bandits escaped from J A I L .  49  Paul hopes we heard about the L O O T .  50  The landlord raised the R E N T .  SPIN-R Form 6 1  Y o u were considering the G A N G .  2  .  The boy considered the M I N K .  3 /  Playing checkers can be F U N .  4 '  The doctor charged a low F E E .  5  He wants to know about the R I B .  6  The gambler lost the B E T .  7  '  Get the bread and cut me a S L I C E .  8  She might have discussed the A P E .  9  The sleepy child took a N A P .  10  Instead o f a fence, plant a H E D G E .  11  The old woman discussed the T H E I F .  12  Drop the coin through the S L O T .  13  They fished in the babbling B R O O K .  .14  , Y o u were interested in the S C R E A M .  (SPIN-R Form 6 continued) 15 16.  W e hear they asked about the S H E D . . The widow's sob expressed her G R I E F .  17  The candle flame melted the W A X .  18  I haven't discussed the S P O N G E .  19  He was hit by a poisoned D A R T .  20  Ruth had a necklace o f glass B E A D S .  21  Ruth w i l l consider the H E R D .  22  The singer was mobbed by her F A N S .  23  The old man discussed the D I V E .  24  The class should consider the F L O O D .  25  The fruit was shipped in wooden C R A T E S .  26  . I ' m talking about the B E N C H . '  27  Paul has discussed the L A M P .  28  The candle burned with a bright F L A M E .  29  Y o u know about the C L I P .  30  She might consider the P O O L .  31.  W e swam at the beach at high T I D E .  32  Bob was considering the C L E R K .  33  W e got drunk in the local B A R .  34  A termite looks like an A N T .  35  The man knew about the S P Y .  36  The sick child swallowed the P I L L .  (SPIN-R F o r m 6 continued) 37  The class is discussing the W R I S T .  38  The burglar escaped with the L O O T .  39  They hope he heard about the R E N T .  40  M r . White spoke about the J A I L .  41  He rode off i n a cloud o f D U S T .  42  M i s s B r o w n might consider the C O A S T .  43  B i l l didn't discuss the H E N .  44  The bloodhound followed the T R A I L .  45  The boy might consider the T R A P .  46  O N the beach we play on the S A N D .  47  He should consider the R O A S T .  48  M i s s B r o w n spoke about the C A V E .  49  She hated to vacuum the R U G .  50  Football is a dangerous S P O R T .  '  SPIN-R Form 7 1  W e ' r e considering the B R O W .  2 .  Y o u cut the wood against the G R A I N .  3.  I am thinking about the K N I F E .  4  They've considered the S H E E P .  5  The cop wore a bullet-proof V E S T .  6  H e ' s glad we heard about the S K U N K .  (SPIN-R F o r m 7 continued) 7  H i s pants were held up by a B E L T .  8  Paul took a bath in the T U B .  9  The girl should not discuss the G O W N .  10  Maple syrup is made from S A P .  11  M r . Smith knew about the B A Y .  12  They played a game o f cat and M O U S E .  13  The thread was wound on a S P O O L .  14  W e did not discuss the S H O C K .  15  The crook entered a guilty P L E A .  16  M r . Black has discussed the C A R D S .  17  A bear has a thick coat o f F U R .  18  M r . Black considered the F L E E T .  19  To open the jar, twist the L I D .  20  W e are considering the C H E E R S .  21  Sue was interested i n the B R U I S E .  22  Tighten the belt by a N O T C H .  23  The cookies were kept in a J A R .  24  M i s s Smith couldn't discuss the R O W .  25  I am discussing the T A S K .  26  The marksman took careful A I M .  27  I ate a piece o f chocolate F U D G E .  28  Paul should know about the N E T .  (SPIN-R F o r m 7 continued) 29  M i s s Smith might consider the S H E L L .  30  John's front tooth had a C H I P .  31  A t breakfast he drank some J U I C E .  32  Y o u cannot have discussed the G R E A S E .  33  I did not know about he C H U N K S .  34  Our cat is good at catching M I C E .  35  I should have known about the G U M .  36  M a r y hasn't discussed the B L A D E .  37  The stale bread was covered with M O L D .  38  Ruth has discussed the P E G .  39  H o w long can you hold your B R E A T H ?  40  His boss made him work like a S L A V E .  41  W e have not thought about the H I N T .  42  A i r mail requires a special S T A M P .  43  The bottle was sealed with a C O R K .  44  The old man discussed the Y E L L .  45  They're glad we heard about the T R A C K .  46  Cut the bacon into STRIPS.  47  Through out all his useless J U N K .  48  The boy can't talk about the T H O R N S .  49  B i l l won't consider the B R A T .  50  The shipwrecked sailors built a R A F T .  SPIN-R Form 8 1  Bob heard Paul called about the STRIPS.  I  M y turtle went into its S H E L L .  3  Paul has a problem with the B E L T .  4  I cut m y finger with a K N I F E .  5  They knew about the F U R .  6  W e ' r e glad A n n asked about the F U D G E .  7  Greet the heroes with loud C H E E R S .  8  Jane was interested in the S T A M P .  9  That animal stinks like a S K U N K .  10  A round hole won't take a square P E G .  11  M i s s White would consider the M O L D .  12  They want to know about the A I M .  13  The A d m i r a l commands the F L E E T .  14  The bride wore a white G O W N .  15  The woman discussed the G R A I N .  16  Y o u hope they asked about the V E S T .  17  I can't guess so give me a H I N T .  18  Our seats were i n the second R O W .  19  W e should consider the J U I C E .  20  The boat sailed across the B A Y .  21  The woman considered the N O T C H .  (SPIN-R Form 8 continued) 22  That job wan an easy T A S K .  23 .  The woman knew about the L I D .  24  Jane wants to speak about the C H I P .  25  The shepherd watched his flock o f S H E E P .  26  Bob,should consider the M I C E .  27  D a v i d wiped the sweat from his B R O W .  28  Ruth hopes she called about the J U N K .  29  I can't consider the P L E A .  30  The bad news came as a S H O C K .  31  A spoiled child is a B R A T .  32  Paul was interested in the S A P .  33  The drowning man let out a Y E L L .  34  A rose bush has prickly T H O R N S .  35'  He's glad you called about the J A R .  36  The dealer shuffled the C A R D S .  37  M i s s Smith knows about the T U B .  38  The man would not discuss the M O U S E .  39  The railroad train ran o f f the T R A C K .  40  M y j a w aches when I chew G U M .  41  A n n was interested in the B R E A T H .  42  Y o u ' r e glad they heard about the S L A V E .  43  He caught the fish in his N E T .  (SPIN-R F o r m 8 continued) 44  B o b was cut by the jack-knife's B L A D E .  45  The man could consider the S P O O L .  46  T o m fell down and got a bad B R U I S E .  47  Lubricate the car with G R E A S E .  48  Peter knows about the R A F T .  49  Cut the meat into small C H U N K S .  50  She hears Bob asked about the C O R K .  "' •  ,  '158  APPENDIX D Spectrograms and Time-Amplitude Waveforms of Jittered and Unjittered Speech Stimuli  The following images were obtained using C S R E 4.5 and contain the first S P I N - R sentence His plans meant taking a big risk sampled at a rate o f 20 kHz. The first image is a sprectrogram o f the unjittered form o f the sentence-final word (risk), and the second is the low-frequency jittered (standard deviation = 0.25 msec, bandwidth = 500 H z ) version of the same word. Note the bottom right window "Cross Section". This image conveys the frequency amplitudes for the same point in time in each version (within the vowel [I]). When the two images are compared, it is apparent that although there is more highfrequency noise in the jittered version, the formant energy remains the same when jittered. The third image is the time-amplitude unjittered waveform o f the word risk and the fourth image is the low-frequency jittered form o f the same word. Note the slight changes to the temporal fine-structure, and the similarity o f the temporal envelopes i n the two images. The fifth and sixth images are the unjittered and jittered time-amplitude waveforms o f the entire sentence (His plans meant taking a big risk). Note that that the temporal envelope has not been altered.  159 Unjittered risk  HiaBiiiSBiiEaHGgiErjana -4U  -sa -5B  -63  | SPECTROGRAM Size : 497.25 ms CURSOR AT Time : 6001.4 ms Freq : N/A Mag. : N/A  -67  FILE PARAHETERS Proc. AC Wind. 512 pts Bands 256 50 Z Ovlp. Func, HANNING 20.00 kHz SF 98.0 Z Pre. 15 Order  -72  -77  -81  -HB  -9B  SPJ0_01-Uaveform  TT!  n P  S . B l B.B  o-5.B •t  B.B 5911  5961  BB1B  BBBB  BI 1 B T I M E  SPJO.Ol-Data Processed  IF  BIBB  E21H  631  625B  (fll)  SPJO.Ol-Cross-Section  B.B  P  A  5.Hi  A  HARKER AT F(Hz)  °-5.B •t  -93  B.B 6 H 0 U B B B B 6 B 0 9 BB1 2 BB1 il BB1 7 BB1 9 6 B 2 2 6 0 2 H B S  T i H E <Ml>  l . B  2.B  5 PEAKS F<Hz) 546.0 1796.0 3984.0 5664.0 N/A  3.B  U.B  5.B  Fl-equcncH  B.B (kHz)  7.B  B.B  9 . B 1 B |  M(dB> -58.60 -58.90 -72.00 -85.20 N/A  M<dB>  160 Jittered risk  m11 m*\mr®  I SPECTROGRAh Size : 495.28 ms CURSOR AT Time : 6001.4 ms Freq : N/A Mag. : N/A  [3 • • •  SPINl.Ol-Waveform ! a.a P 5.B i B.B o.5. B  1  B.B 5911  5961  5B1B  6B6B  61 I B TIME (hi)  SPIN1_01-Data Processed ! B.B M P S.B  61 6 B  621 B  626B  63  6310  SPINl.Ol-Cross-Secti on  i B.B B  O-5.0  PARAHETERS : AC : 512 pts : 256 : 50 Z : HANNING : 20.00 kHz : 98.0 Z ; 15  5 PEAKS F<Hz> 585.0 1835.0 4023.0 5898.0 7031.0  M<dB> -60.80 -59.30 -74.20 -78.40 -75.50  MARKER AT F(Hz> M<dB)  77  + B.B 6 B B 4 6 B B 6 6 B B 9 6B1 H 6B1 4 6 B 1 7 6B1 9 6 B 2 2 6 B 2 4 6 B Ti M e (Mi)  FILE Proc. Wind, Bands Ovlp. Func. SF Pre. Order  1.0  2 . B  3.0  4.0 5.0 Fl-equnncH  6.0 7.0 (kHz)  B.B  7.0  IB]  161 Unjittered risk  SPJO.Ol--Haveform 1B.B  •CQI^fj|g|t?|i||^|fflB||BEi^|p|oN<j|®  II" \ > | « | »  • RECORD/CUT/SCftLE I Size : 368.45 ms CURSOR AT Time : 5970.68 m Next Pt : -213 Next Pt : -0.0650 V  2S.B  p U B.B o  r-5.0 X  -1D.B  SB7B  _ _  5915  5952  59B9  SPJO 01-UavefoTrMprOl^C]|»>\«\»  -Result Waveform  BB2B  TjMe  CMi)  BBB3  BB99  6 1 3 B 6 1 7 3  SPJ001-Ua^ef^rr7Mpnr3|^c]|>>  ii 1  \<j\»  FILE Full Full Disp Disp SF AScl  PARAMETERS ; 219331 pts : 10966.55 ms ; 7384 pts : 369.20 ms : 20.00 kHz : OFF  INSIDE Begin t End : Diff. : RMS : •Peak : -Peak :  MARKERS -0.05 ms -0.05 ms 0.00 ms 0.0000 V 0.0000 V 0.0000 V  162 Jittered risk  SPINl.Ol-Uaveform • © l ^ l t l E l ^ l ^  |B M^|SB|E|^ |P|<&|<Cl|® |l>"  |t»  1B.B  \«\» I  Size : 370,03 ms CURSOR AT Time : 5967.20 m Next Pt : -2582 Next Pt : -0.7880 V  S5.B  -1B.B  RECORD/CUT/SCftLE |  5B76  SPINi.Ol-ldaveform  -Result Uaveform  591H  5951  59BB  BB25 Tine (Mi)  6B62  BB99  SPINljDl-Uaveform  fil  36  BI 73  62, '  FILE Full Full Disp Disp SF ftScl  PARAHETERS : 219331 pts : 10966.55 ms : 7416 pts : 370.80 ms : 20.00 kHz : OFF  INSIDE MARKERS Begin : -0.05 ms End : -0,05 ms D i f f . : 0.00 ms RMS : 0.0000 V +Peak : 0.0000 V -Peak : 0.0000 V  163 Unjittered His plans meant taking a big risk.  SPJ0_01  I  RECORD/CUT/SCALEl  Size : 1988.81 ms  CURSOR AT  (J5.B  n p U  Time : 4816.15 m Next Pt : 1205 Next Pt : 0.3677 V  B.B  0 I t.S.B  -1B.B 5301  531B T I M E  5619  SPJ0_01-Uaveform  SPJO.Ol-Uaveform  1B.B  1B.B  2S  Full Full Disp Disp SF AScl  « 5 . B  H  n>rrrrfhkhh-•  B.B  U  B.B  0 1 t.S.B  t.S.B  s  s  -IB.a  -1B.B  4 B 4 7 4 B 9 6 41 4 6 4 1 9 6 4 2 4 5 4 2 9 5 4 3 4 4 4 3 9 4 4 4 4 4 Tide ( M i )  44  W  f  .  6 B 3 3 6 B B 3 61 3 2 6 1 B 2 6 2 3 2 6 2 B 1 Tine ( H i )  6331  6301  643B  64  •  5.B  p  illHH'i  t-5.B  -1B.B  199  397  794  993  TiME ( M l )  1192  : 219331 pts : 10966.55 ms : 61874 pts : 3093.70 ms ; 20.00 kHz : OFF  INSIDE HARKERS  BIIEEBSHMUB 03 D9 • •  S$CSRE$$-Result Waveform  £  FILE PARAHETERS  6B  B  n p U  6Z3B  ( M I )  15B9  17B7  191  Begin End Diff. RMS •Peak -Peak  : : : : : :  4245.20 ms 6231.65 ms 1986.45 ms 0.5087 V 1.8552 V -2.7673 V  164  Jittered His plans meant taking a big risk.  SPINl.Ol-Uaveform •(!)l^illillHi^IIHl^l^lfflRlfflFl^ 1 £l€>l<dl(R)llM It* l « l » I RECORD/CUT/SCALE I i s . a  Size ; 2047.07 ms CURSOR AT Time : 4833.25 m Next Pt : 540 Next Pt : 0.1G48 V  gs.B U B.B  t.S.B  -1B.B 4167  4433  469B  4964  5229  5495  5760  6026  6291  55  Tine ( H i )  SPINl_01-Waveforn«PhD'hC]|^|^|^  SPINl.Ol-Uaveform •  1B.B  1B.B  2  2  5.B  5 . B  p  n P U B.B  U B.B  0  o  1 t.S.B  1 t-5.B  s  X  -1B.B  4 B 2 1 4 B 7 2 4 1 2 4 41 7 5 4 2 2 6 4 2 7 7 4 3 2 0 4 3 7 9 4 4 3 0 4 4  PlC'IOll*  "J-.-  w  -1B.B  Tine ( M I )  6B67 611B 6169 6220 6272 6323 6374 6425 6475 55  Tine (Mx)  WSREW-Result Waveform 1B.B  " 5 . 0  n  p  gB.BpTipFt-5.B  •la.a  B1B  1 B23  Tine (Mx)  1 227  1B41  20i  FILE Full Full Disp Disp SF AScl  PARAHETERS : 219331 pts : 10386.55 ms : 53101 pts : 2655.05 ms : 20.00 kHz : OFF  INSIDE Begin : End : Diff. : RMS : +Peak : -Peak :  HARKERS 4225.90 ms 6271.70 ms 2045.80 ms 0.4991 V 1.8066 V -2.8760 V  APPENDIX E Connections on the Tucker Davis Technologies Modules  166  APPENDIX F Purpose of. Each T D T Module  The Tucker Davis Technologies system consists o f hardware modules that are controlled v i a a computer. The following T D T hardware modules were used in the present studies. DDI This module is a digital-to-analog and analog-td-digital converter. The digital speech and babble signal were converted into analog signals through this module. FT5 This anti-aliasing filter reduces the high frequencies that accompany recorded signals, and smoothes the digital signal. The signal that reaches the listener thus sounds more natural. PA4 This module is an attenuator that can be programmed to adjust the presentation level o f signals routed through it. For this experiment two such modules were used, one each for the S P I N - R sentences and one for the background babble. For each presentation o f a S P I N - R sentence, the R M S values for that sentence and accompanying babble file were sent to the P A 4 modules which attenuated according to the values set by the experimenter for that set. This ensured that all sentence and babble files were presented at the same R M S levels to each listener.  . ..  167  SM3 • This module serves two purposes, the first o f which is to remove the high-frequency hiss generated by the T D T during the presentation o f signals. Second, this module functions as a sound mixer; that is, two separate signals can be combined so that the listener can hear both signals through one headphone. For this experiment, the S P I N - R sentences were combined with the babble signal and combined through this module before being sent to only one earphone. HB5  '  ,  This module is used to relay the speech signal from the T D T to the headphones i n the sound-attenuated booth.  168  APPENDIX G  .  Instructions for Experiment 1  In this task you w i l l hear some sentences that are presented with noise i n the background. This noise w i l l sound much like the background chatter at a cocktail party. A l s o , some o f the sentences may sound unusual. Y o u w i l l only hear the sentences i n your right ear, and some o f them may sound distorted. Y o u r task is to repeat the last word o f each sentence. I f you are not sure what the last word was, take a guess. It is important that you give a response each time, so it's better to guess than say nothing. The sentences w i l l be presented in groups o f 50. Let's practice a few sentences first, so you can get used to the task.  169  APPENDIX H Order o f Conditions and S P I N - R Forms for each Participant in Experiment 1  Session 2  Session 1 Participant  •  N o Jitter  S/N=8  Jittered  S/N=-4 , S/N=0  S/N=-4  S/N=4  S/N=0  2  3  4  5  6  7  8  2  4  1  3  6  7  8  5  3  2 •  3  6  5  4 •  1  0  7  4  4  3  2  1  5  6  7  8  2 .  3  4  8  5  6  7  1  2,  5 '  S/N=4  N o Jitter  S/N=8  1  '  Jittered  :  1  •  .  '  •  1  6  2  4  1  3  7  8  5  6  7  3  1  4  2  6  7  8  5  1  5  6  , 7  8  3  4  6  7  8  5  1  3  .8  5  6  7  8  .  4  2  ' 3 ' •  9  1  2  10  2 ,  4  .  '  11  . -3  1  4  2  7  8  ' .5 .  6  12  4  3  2  1  8  5  6  7  170 APPENDIX I Experiment 2 Participants' Pure-tone Thresholds for Right (R) and Left (L) Ears  , Test Frequency (Hz) Participant  •  .  250  500  1000  2000  4000  8000  R  L  R  L  R  L  R  L  R  L  R  L  .1  0  •5  -5  -5  0  0  0  0  0  0  0  0  2  5  10  5  5  -5  10  0  5  0  0  5  5  ' P.  0  0  0  -5  0  5  0  -10  0  10  10  •4  5  5  5  5  5  5  5 ' -5  0  -5  5  5  -5  0  -5  0  0  0  -5  -5  -5  0  0  6  20  10  20  5  15  0  10  0  0  -5  10  0  7  0  0  0  5  0  5  5  5  0  5  0  15  8  5  5  10  5  0  5  ' 5  -5  0  5  5  5  9 •  0  5 ,,.  0  .5  10  -5  -5  10  0  10  15  10  5  10  5  10  5  15  5  15  5  5  10  10  11  5  10  5  5  0  0  0  0  .0  0  0  5  12  0  5  5  5  5  0  0  0  0  0  15  15  13  5  5  5  .5  5  5  5  0  -10  0  0  5  14  10  5  10  0  -5  5  15  10  15  10  15  15  ' 5  0  -5  -5  '-5  -5  0  -5  -10  5  5  5  16  -5  0  -5  -5  0  0  0  0  -5  -5  10  10  3  • -5" t  '  "  5  10  ,  '  "  171  APPENDIX J Experiment 2 Participants' Characteristics  Pure-tone • average  a  (dB H L ) .  Participant ,  Vocabulary , Score  Years o f Education  Age  3  6  Handedness  (of 20) R  L  1  -1.7  ' -1.7.  2  0  3 4  18  R  19  R  23  19  R  14  27  18  R  -3.3  16  24  19  R  15  1.7  15  27  19  L  ' 7  1.7  5  13  13  R  8  5 •  1.7  17  •29  15  R  9  1.7  1.7  16  28  17  R  10  5  13.3  14  27  18  R  11  1.7  1.7  15  29  19  R  • 12  3.3  1.7  14  21  16  R  13  5  3.3  14  28  17  R  14  5 ' '  6.7 ,  14  26  19  R  5• 6  a  b  17  27 •  6.7  15  29  0  0  19  5  5  -1.7  •  . 2 6  •  15  -3.3  -5  15  29  17  R  16  -1.7  -1.7  14  28  22  R  M i l l Hill, Raven (1938) this score includes kindergarden  172  APPENDIX K Instructions for Experiment 2  In this task you w i l l hear some sentences that are presented with noise in the background. This noise w i l l sound much like the background chatter at a cocktail party. A l s o , some o f the sentences may sound unusual and distorted. Y o u are to repeat the last word o f each sentence. I f you are not sure what the last word was, take a guess. It is important that you give a response each time, so it's better to guess than say nothing. Some o f the words w i l l be predictable based on the context o f the sentence, and some w i l l not be predictable. Y o u are to simply say 'yes' i f it is predicable, or ' n o ' i f it is not after you have repeated the word. After a set o f sentences - o f either two or eight - 1 w i l l stop the task, at which time you w i l l repeat the sentence-final words in that set. Y o u w i l l be told ahead o f time whether you w i l l be required to remember two or eight words. Each set contains 50 sentences. Let's practice first, so you can get used to the task.  173  , APPENDIX L Order o f Conditions and S P I N - R Forms for each Participant in Experiment 2  N o Jitter Participant  SS=8 ,  a  1 •  S/N=4  S/N=8  SS=2  SS=8  . SS=2  SS=8  SS=2  SS=8  1  3  5  7  2  4  6  8  .8'  1  3  5  7  2  4  6  8  1,  3  5  7.  2  6  8  1  3  • 5  7  2  4  6  8  1  3  5  7  7  2  4  6  8  1 ,  3  5  7  5  7  • 2  4  6  8  1  3  8  3  5  7  2  4  6  8  1  1  3  5  7  2  4  6  8  10  8  1  3  5  7  2  4  6  11  6  8  ,1  3  5  7  2  4  12  4  6  8  1  3  5  7  2  13 .  2  4  6  8  1  3  5  7  14  7  2  4 ..  6  8  1  3  5  15  '5  7  2  4  6  8  1  3  16  ' ' 3  5  7  2  4  6  8  1  2  6  • •3 '4  , 4  5 6 '  S/N=4  S/N=8 SS =2  ;  Jittered  9,  ,  *  SS = Recall Set-Size  .  •  '  ,  '.  4 •  2  174  APPENDIX M Experiment 1 W o r d Identification R a w Scores (out o f 25)  Session 2  Session 1 Participant  N o Jitter S / N = +8 H  a  L  b  N o Jitter  Jittered  S / N = +4  S / N = +8  S/N = +4  Jittered  S/N = 0  S / N = -4  S/N = 0  S / N = -4  H  L  H  L  H  H  L  , H  L  H  L  H  L  L  1  25  21  25  "23  23  23  21  19  24  12  8  4  15  8  9  2  2  25  21  24  18  23  16  22  18  22  10  4  5  12  7  1  1  3  25  24  25  20  25  . 19  24  14  21  13  14  3  19  7  8  2  23 •  16,  24  15  •21  14  23  13  11  6  6  8  11  7  7  4  5  25  19  24  21  23  18  23  17  24  11  10  5  16  8  5  4  6  25  21  24  16  24  20  21  15  •22  10  10  3  16  7  8  2  7  25  21  24  16  24  20  22  16  18  16  6  5  14  5  6  3  8  25 ,  22  25  22  25  19  24  16  22  12  15  15  20  9  12  6  9  25  19  25  22  25  18  22  18  21  12  9  7  17  6  8  2  10  25  22  25  18  24  16  23  13  23  6  7  2  17  7  6  5  11  25  21 ,  25  17  24  14  23  18  21  11  10  6  16  5  9  6  12  25  22  25  20  24  20  24  14  24  9  7  4  17  11  6  4  4  ,  a  b  H=High Context L=Low Context  1  175  , A P P E N D I X N " ••" _ M e a n Percent-Correct Sentence-Final Word-Identification Scores and Standard Deviations for Each Condition in Experiment 1  Jitter Condition  S / N Condition ,. (  N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter • N o Jitter N o Jitter' Jittered Jittered Jittered Jittered Jittered Jittered Jittered Jittered  d B  )  8 ' 8 4' ' 4 0 0 -4 ' • -4.. • 8 8 4 4 , 0 ". 0 -4 -4  Context Condition  High Low High Low High Low High Low High Low High , Low High Low High. Low  M e a n percentcorrect Sentence-final wordidentification score 99.3 83.0 98.3 - . 76.0 86.3 42.6 35.3 22.3 95.0 72.3 87.3 66.0 63.3 29.0 28.3 13.6  Standard Deviation  !  .  2.31 8.02 2.06 10.79 8.94 11.23 12.97 13.69 4.55 10.71 4.29 8.44 10.35 6.63 10.71 6.71  . 176  APPENDIX O Predictability, Judgements for Experiment 2 (out o f 50)  . Participant  ,  N o Jitter  S / N = +8'  L o w pass jittered  , ' S / N = +4  S/N = +8  S/N = +4  SS =2 a  SS=8  SS=2  SS=8  SS-2  , SS=8  . ' 1 •  50  48  '49  47,  49  50  50  49  2  ' 47  46  47  49  46  48  46  45  3  50  49  50  49  49  47  46  , 42  4  50  48  44  47  43  46  42  43  5  . 48  50  . 50  49  49  50  50  48  6  50  49 '  48  47  50  49  46  50  7•  48  47  49.  50  49  50  50 •  48  8  49  49  49  49  48  49  48  46  9  46  47  47  45  44  47  45  49  10  50  49  49  49  . 50  49  50  49  11 ,  48  45  47  47 ,  47 .  48 ,  47  43  12 '  50  .49  50  47  47  49  46  48  13  48  49  50  50  47  50  48  45  14  50  49  50  49  49  50  49  50  15  49  50  48  50  50  49  47  49  16  50  50  50,  50  48  50  49  48  '  SS=Recall Set Size  , SS=2  SS=8  177  APPENDIX P Experiment 2 W o r d Identification R a w Scores (out o f 25)  N o Jitter ., Partic-  S / N = +8  ipant  ,  a  b  0  '  SS' '=2  SS =8  H  H  b  L  c  L  L o w pass jittered  S / N ,= +4  SS - 2  S / N = +8  SS =8  SS =2  S / N = +4  SS =8  H  L  •H.  L  H  L .  H  24  16  25  13  25  19  22,  25  19  25  18  23  25 20  25  13  25  17  23  12  L  SS =2  SS==8  H  L  H  L  24 20  25  17  24  17  19  23  20  22  18  22 20  2 4 ' 16  24  18  21  17  18  14  2 4 , 13  21  14  25  16  17  12  21  16  1  25  20  2 .  25  21  3  25  23  4  25 21  25  •5'  25 23  25 24  2 5 ' 24  24  19  25  17  25 21  25  16  23  13  6  25 22  25 23  25 ' 20  24 23  25  19  25  17  23  18  25  18  7 . ' 25 22  25 23  25 23  25  18  24 20  25  17  24  14  24  18  8  24  25 24  25 20  25 22  24 22  25 20  25  15  •22  17  9  25 22  24 24'  25 23  24 20  25 21  23 21  24  13  24  13  10  25 24  25 23  25 22  24 20  25 ' 18  25 20  25 25  24  19  11.  25' 22  25 - 22  25 21  25 21  ;24 20  23 20  23  21  21  21  12  24  19  25 20  25 20  23 20  24  19  25 20  22  18  23  18  13  25 25  25 21  25. 21  25 21  24  19  23  21  25  16  21  16  14  25 21  25 24  24  24 20  22 21  25  15  23  17  23  15  15  25 21  25 23  25 20  25  18  24 22  24  16  24  11  23  18  16  25 22  25 21  25  25  19  23 23  25 20  24  16  25  15  •  25  SS = Recall Set-Size H = High Context L = Low Context  25 20 ,25  19  19  19  178  APPENDIX Q M e a n Percent-Correct Sentence-Final Word-Identification Scores and Standard Deviations for the Possible Combinations o f Variables i n Experiment 2  M e a n percent-correct sentence-final word-identification scores and standard deviations for Jitter, S / N and Recall Set-Size Jitter condition  •  .  S/N  Recall Set-Size  8 8 4 4• 8 8 4 4  2 8 2 8 2 8 .2 8  •  N o Jitter N o Jitter N o Jitter N o Jitter Jittered Jittered Jittered Jittered  1  •  '  ..  Mean Percent • Correct Sentence final WordIdentification Score ' 93.9 94.0 . 88.5 86.7 86.4 86.4 78.2 78.7  Standard Deviation  3.60 3.88 7.85 6.81 6.95 5.73 9.22 .8.32  179  Sentence-Firial Word-Identification Scores and Standard Deviations for Jitter, S / N and Context Conditions  Jitter condition  N o Jitter N o Jitter N o Jitter N o Jitter Jittered Jittered , Jittered Jittered  , S/N  8 8 4 4 8  • 8, 4 4  Context Condition  •  High Low High Low High Low High Low  M e a n PercentCorrect Sentence-Final Word Identification Scores .99.7 88.1 . 98.5 76.8 96.3 76.4 91.9 65.2  Standard Deviation  0.5 6.48 2.42 12.25 4.02 8.70 7.78 9.74  180  Sentence-Final Word-Identification Scores and Standard Deviations for Jitter, Recall SetSize and Context Conditions  Jitter condition  N o Jitter N o Jitter N o Jitter N o Jitter Jittered Jittered Jittered Jittered  Recall Set-Size  ,  . 2 2 8 8 2 . 2 8 8  .  Context , Condition  High Low High Low High Low High Low,'  M e a n PercentCorrect Sentence-Final WordIdentification Score 99.4 83.0 98.9 81.9 94.3 70.4 94.0 71.25 •  Standard Deviation  •  1.65 9.76 1.77 8.92 6.44 9.71 5.36 8.74  181  Sentence-Final Word-Identification Scores and Standard Deviations for S / N , Recall SetSize and Context Conditions  S / N Condition  8 8, 8 8 4 4 4  ,  4  ' .  Recall Set-Size  2 2 ' 8 . 8 2 2 8 8  Context Condition  ,  High Low High Low High Low High Low  Mean PercentCorrect Sentence Final Word Identification Score 97.6 82.6 98.5 75.9 96.0 70.75 94.4 71.3  Standard Deviation  2.78 7.74 2.25 7.42 2.66 11.74 4.89 10.26  182  Sentence-Finai Word-Identification Scores and Standard Deviations for all Conditions i n Experiment 2  Jitter Condition  S/N Condition,  Recall SetSize  Context Condition  N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter Jittered Jittered Jittered Jittered Jittered Jittered Jittered Jittered  8 8 8 8 4 4 4 4 8 8 8  2 2 8 8 2 2 8 8 , 2 •2 8 8  High Low High Low High Low High Low High Low High Low High Low High Low  .  • 8 4 4 4 ', 4  •2 2 S ,  '  8  '  Mean PercentCorrect Sentencefinal W o r d Identification Score 99.7 88.0 99.7 88.2 99.0 78.0 98.0 75.7 95.5 77.3 97.3 75.5 93.0 63.5 90.8 67.0  Standard Deviation  1.00 6.20 1.00 6.77 2.31 13.39 2.53 11.11 4.59 9.32 3.49 8.12 8.33 10.11 7.26 9.41  183 APPENDIX R Experiment 2 W o r d Recall R a w Scores (out o f 25)  N o Jitter Partic-  ; S / N == +8  ipant .  • SS=2 '•  S/N = +4  SS =8 '  H  L  H  1  25  25  19  2' '  25  25 .  20  3  25 ' 25  4  , . 2 4 ' 25  Jittered  L  SS =2  S / N = +8  SS==8  S / N = +4  SS =2 • . SS==8  SS =2  SS =8  H  L  H  L  H  L  H  L  H  L  H  L  18  25  24  18  15  25  25  12  10  25  23  16  17  15  25  24  22  15  24  23  17  15  24  23  14  11  17' 20  25  2 5 , 20  15  24  24  16 . 15  25  24  13  16  18  .19  25  24  23  17  23  22  17  15  24  22  13  16  5  25  25  19  9  25  25'  18  13  25  24  16  18  25  25  15  9  6  25  24  21  18  25 • 25  16  21  24  25'  14  16  25  25  18  12  • 7  .25  24  15  16  24  24  19  19  25  24  17. 12  25  24  17  10  8  25  25  18  14  25  25  18  13  24  24  12  15  25  24  14  14  9 ,  25  •25  15  15  25  25  17  11  25  25  15  8  25  24  14  9  10  25  25 .  20  16  24  25  20  14  25  25 • 16  13  25  25  16  15  11  25  25  20  18  25  24  24  20  25  23  19  13  25  24  16  13  12  25  25  18  16  25  24  17  20  25  24  17  13  24  24  14  16  13  24  24  17  15  25  25  19  15  25  25  18  10  25  25  14  17  14  25  25  20  18  25  25  19  23  25  25  20  19  24  24  22  15  15  25  25;  1,9  14  25  25  20  12  24  25  13 ' 13  25  25  17  10  16  25  25  23  22  25  25  23  20  25 • 25  21  25  25  22  18  19  184 APPENDIX S M e a n Percent-Correct Sentence-Final Word-Identification Scores and Standard Deviations for the Possible Combinations o f Variables in Experiment 2  M e a n Percent-Correct Sentence-Final Word-Recall Scores and Standard Deviations for Jitter, S / N and Recall Set-Size  Jitter Condition  N o Jitter N o Jitter N o Jitter N o Jitter Jittered Jittered Jittered Jittered  S / N Condition  8 8 ... ' '  4 4 8 8 4 4  Recall Set-Size  2 8 2 , 8 2 8 2 • 8  M e n Percentcorrect Sentence final W o r d Recall Scores 99.5 69.1 . 99.0 71.5 99.7 60.9 98.2 59.1  Standard Deviation  1.31 10.79 1.68 12.18 3.10 12.50 2.21 11.72  185 Sentence-Final Word-Recall Scores and Standard Deviations for Jitter, S / N and Context Conditions  Jitter condition  S / N Condition '  N o Jitter N o Jitter N o Jitter N o Jitter Jittered Jittered. . Jittered Jittered  .-  8 • 8• , ' 4 4 8 8 •.  Context Condition  4  4  High Low High Low High Low High Low  M e a n PercentCorrect Sentence-Final Word-Recall , Score 87.5 . 81.1 . 88.6 81.9 81.5 77.1 81.5 75.9  Standard Deviation  5.58 7.51 5.46 8.40 6.07 8.92 . 6.39 7.55  186  Sentence-Final Word-Recall Scores and Standard Deviations for Jitter, Recall Set-Size, and Context Conditions  Jitter Condition  N o Jitter N o Jitter N o Jitter N o Jitter Jittered Jittered Jittered Jittered  .  Recall Set-Size  2 2' • 8 8 2 2 8 8  . Context Condition  • High Low High Low High Low High Low  M e a n percentcorrect Sentence-final word-recall score • 99.6 98.9 76.5 64.1 98.9 97.1 64.1 55.9  Standard Deviation  1.18 1.81 8.86 14.10 2.04 3.27 10.40 13.20  187  Sentence-Final Word-Recall Scores and Standard Deviations S / N , Recall Set-Size and Context Condition  S / N condition  '  8, 8 • 8 8 4 4 4 4 .  Recall Set-Size  2 2 8 8 2. 2 ' 8 8  Context Condition  High Low High Low High Low High Low  Mean Percentcorrect Sentence-Final Word-Recall Score 99.1 98.1 69.9 60.1 99.4 97.9 70.7 59.9  Standard Deviation  1.74 2.67 8.92 13.76 1.49 2.40 10.36 13.55  .  188  Sentence-Final Word-Recall Scores and Standard Deviations for Jitter, S / N , Recall SetSize and Context Condition  Jitter Condition ,  S/N  N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter N o Jitter Jittered Jittered Jittered Jittered Jittered Jittered Jittered Jittered  8 .' 8 8 • ' 8 . • 4 4 4 4 8 " 8 8 8 ,  Recall SetSize  4 4 4 4  "  •.  •2 2 8 8 2 ' 2 8 8 , 2 ' 2 8 8 ' 2 2 8 8 ..  • Context Condition  High Low High Low High. Low High Low High ' Low High Low High Low High Low  Mean Percent Correct ' SentenceFinal W o r d Recall Score , 99.7 99.2 75.2 63.0 99.5 98.5 77.8 65.3 98.5 97.0 64.5 57.3 99.3 97.3 63.8 54.5  Standard Deviation ,  1.00 1.61 8.16 13.42 1.37 2.0 9.57 14.80 2.48 3.72 9.67 14.14 1.61 2.82 11.17 12.30  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0089735/manifest

Comment

Related Items