Kent Article Flashcards
Clinical applications of auditory judgments are based on what assumptions about listeners?
- common understanding of perceptual labels like hoarse, nasal, rough, monoloud, excess and equal stress, stuttering
- use essentially the same verbal descriptors and associated scale values to assess speech/voice
- can isolate for judgment one perceptual dimension from several co-occurring dimensions
- have uniform reliability in judging various dimensions that give complete clinical portrait of speech/voice disorders
- can make perceptual judgments for which interjudge differences are smaller than the differences needed for clinical classification/to discern changes in clinical status
What are problems with perceptual judgment?
- judges don’t appear to have equivalent definitions of dimensions to be rates
- specialists fail to reach consensus on which perceptual dimensions should be rated for a given disorder
- perceptual ratings of various dimensions are intercorrelated – values obtained for one dimension may be influences by co-occurring dimensions of a disorder
- various perceptual dimensions aren’t rated with uniform reliability
T/F
Listeners can make finer auditory discriminations than they can label using available identification responses.
True
What is the phonemic restoration effect?
An effect that results when listeners fail to detect that a speech sound has been replaced by a nonspeech sound. It’s an example of hearing something that is not there and reflects the activation of the listener’s top-down strategies, which make hypotheses about the semantic and syntactic features of speech.
What is the verbal transformation effect?
Listeners hear a changing phonetic pattern for an unchanging acoustic stimulus, such as a word that is replayed repeatedly. As the stimulus is replayed, listeners report changing percept, often hearing an entirely different word. They may be continually advancing hypotheses of an incoming speech signal.
T/F
Listeners’ attempts to make linguistic sense of speech signal leads to potential errors in perception.
True
What strategies do listeners use to make linguistic sense of speech?
- listen for stress and intonation patterns
- derive phrase structure
- try to recognize words
- pay special attention to stressed vowels
2 likely sources of error in the perception of consonants are:
Segmental substitutions, place of articulation
Some speech perception errors can be attributed to which 2 phonological processes?
- listener doesn’t realize that a simplification rule has been used by the speaker and fails to recover the reduced segment or syllable
- listener believes that a rule was used in production and hears a spurious segment or syllable
What is phonemic false evaluation?
mistaken recognition of phonemes that were not produced by the talker
T/F
With regards to phonemic false evaluation, perceptual errors for normal speakers largely arise at levels of below that of motor commands.
False – above the level of motor commands
What are 3 primary limitations in the traditional perceptual classification of speech errors?
- listener normalization is a regular part of the perceptual process that may override detection of subtle errors
- errors at a fine phonetic level are difficult to detect reliably, especially when listener is attending to verbal content
- suitable transcription techniques are lacking for highly or subtly anomalous sounds
How can lexical status affect phonetic categorization?
- lexical status of a sound can affect the phonetic boundary for a constituent feature, such as VOT
- clinician’s lexical biases may influence ability to detect allophonic differences in sound production (lexical identification shift)
- semantic content of a sentence can influence the perception of acoustically ambiguous words
What is an equivalence class?
A category of related sounds
What happens when sounds from 2 different languages are acoustically similar?
the sound from the non-native language tends to be assimilated to the sound in the native language
What does the Speech Learning Model do?
- explains how phonetic elements of a non-native language are acquired
- emphasizes perceptual representation as a key factor in learning a second language’s phonological system
What are the 5 patterns of assimilation?
- 2-category contrasts
- single-category contrasts
- uncategorizable contrasts
- category goodness difference contrasts
- non-assimilated contrasts
Describe the 2-category contrast pattern of assimilation.
- non-native phone is assimilated to a different native phoneme category
- excellent discrimination is expected because each phone is attached to a familiar category
Describe the single-category contrast pattern of assimilation.
- both non-native phones are equally assimilable to same native phoneme category
- poor discrimination is expected given that the listener is making a within-category discrimination
Describe the uncategorizable contrast pattern of assimilation.
- involves sounds that are heard as speech but neither phone can be assimilated to a native phoneme category
- discrimination is expected to be poor but better than single-category contrast
Describe the category goodness difference contrast pattern of assimilation.
- both non-native phones are assimilated to the same phoneme
- discrimination is expected to be moderate to good
Describe the non-assimilated contrast pattern of assimilation.
- involves non-native phones, both of which are outside the native phonetic space and may not even be heard as speech
- discrimination is expected to be moderate to good
What is the McGurk effect?
- visual and auditory info can interact in phonetic decision-making; visual info may override or complement auditory info
- eg /da/ in auditory-only presentation may be heard as /ba/ when accompanied by simultaneous video info
Describe the prosodic influences on phonetic categorization.
- as speaking rate is changed, acoustic boundaries of phonetic segments can change as well
- slowing of sentential speaking rate shifts locations of VOT ranges
- sentence-level info alters internal perceptual structure of a phonetic category
T/F
There is no interaction between talkers, listeners and utterances in decision-making about intelligibility and talker identification.
False
T/F
Segmental-phonetic decisions can be influences by accompanying prosodic information.
True
Describe the parallel-contingent mode of speech perception.
- acoustic info is extracted from the speech signal to supply 2 info paths: one for segmental-phonetic info and the other for suprasegmental (prosodic, paralinguistic) info
- although paths for phonetic and suprasegmental info are parallel, phonetic decisions are based on contingent info from suprasegmental path, such as speaking rate and speaker identity
Listener confidence in assigning ratings and actual interjudge agreement are high or relatively high for what dimensions?
- interjudge agreement is high for pitch, loudness and rate of speech, and relatively high for vocal variability and age
T/F
There is a lack of convergence on dimensions of voice quality, which is a barrier to the standardization of voice ratings.
True – out of 27 terms, there has only been convergence on 2 (hoarse, nasal)
T/F
Studies showed that judgments of breathiness depended on judgments of concomitant roughness.
False – Judgments of roughness depended heavily on breathiness, but not vice-versa.
T/F
Studies showed that judgments of vocal pitch interacted with roughness.
True
How were vowels that were moderately to severely dysphonic judged compared to vowels that were clear or mildly dysphonic?
Moderately to severely dysphonic – matched with significantly lower pitch
T/F
Improvement in interjudge ratings of stuttering was noted when observers were given behavioural definitions of stuttering, when both visual and auditory info was available to them, or when rate of speech was slowed electronically.
False
What types of dysfluencies are most likely to be judged as stuttering?
sound, syllable, part-word repetitions
Reliable perception is noted for which types of dysfluencies?
elongations, blockages, repetitions, interjections
Unreliable perception is noted for what dimensions of stuttering?
length of pauses and quality of breathing during phonation
T/F
In Zyski and Weisinger’s study, grad students rated deviant dimensions of dysarthria better than experienced SLPs, reflecting the need for training with reference samples.
True
Out of the ataxic dysarthria dimensions of imprecise consonants, excess and equal stress, irregular articulatory breakdown, distorted vowels and harsh voice, agreement has been lowest for which dimensions?
Irregular articulatory breakdown, harsh voice
T/F
Perceptually, speech dimensions are independent and are rated as such.
False – there’s high intercorrelation between rated dimensions, and rating values that clinicians assign to apparently separate dimensions may reflect the overall perception of a number of concurrent, salient speech characteristics
______ is a good indicator of overall speech impairment.
Intelligibility
T/F
Listeners have difficulty distinguishing between monopitch, monoloudness and monoduration.
True
T/F
When listening to dysarthric speech, listeners may rely on higher-level processing to retrieve the intended message.
True
The intelligibility of dysarthric speech may vary with what 2 factors?
- predictability of its syntactic and semantic content
- listeners’ familiarity with dysarthric patterns
How do phonemic paraphasias impact hearer consistency?
likely to be a subtle articulatory disruption in fluent aphasiacs that affects the acoustic picture, which leads to hearer inconsistency and to false evaluation of phonemic intentions of the fluent aphasic speakers
AOS is characterized by what types of errors?
Substitution errors
Subtle errors that perception misses
T/F
AOS operates exclusively at the level of phonemic mechanisms.
False
There is _____ reliability and a _____ range of scores in audio-recorded vs live conditions.
Lower, greater
T/F
Judgments of misarticulation are equally reliable across sound classes.
False (eg more reliable for /s/ than /r/)
T/F
Reliability of narrow transcription is worse than that of broad transcription.
True – narrow transcription provides more information about speech articulation, but its complexity exceeds the auditory working memory of the listener and leads to greater uncertainty
The coding system developed for typical adult speech has ____ reliability and ____ validity when used with developing or disordered speech.
Lower, uncertain
T/F
The choice of rating scale can affect judge’s ability to discern reliable differences along a dimension of interest.
True
A given dimension may be _____ or ______.
Prothetic, metathetic
Define prothetic.
- Varies in magnitude or quantity; sometimes described as additive, quantitative continuum. Perceptual judgment is one of determining if more or less of an attribute is present (eg loudness)
- scaled with direct magnitude estimation scale (DMES)
Define metathetic.
- varies in terms of a change in quality; sometimes described as substitutive, qualitative continuum (eg pitch )
- can be scaled with either an equal-appearing interval scale (EAIS) or direct-magnitude estimation scale (DMES)
T/F
The Mayo Clinic system combines protethic and metathetic ratings.
True
T/F
Auditory-perceptual assessments may be biased by certain speaker characteristics, such as physical appearance and history.
True
How does the auditory salience of speech impact perception?
- clinicians’ judgments of severity may depend on slowly varying components of temporal pattern of speech and less on rapidly varying temporal features such as stop bursts and voice onset times
- ratings of voice quality are more reliable for vowel-onset and whole-vowel stimuli than for post-onset stimuli, apparently because cues for voice quality are less salient in the most stable portion of vowel phonation
T/F
In terms of listener characteristics, the listener’s familiarity with the speaker is of particular importance.
True
How can a lack of perceptual differentiation of sounds affect an SLP’s judgments of the speech of a child with a phonological disorder?
SLP may conclude that the child has a collapsed phonetic contrast when they’re really producing a contrast that SLP isn’t detecting
How does higher-level information impact phonetic judgments?
When judges know the meanings of utterances, their transcriptions conform more closely to expected adult forms of utterances
Perceptual judgments are ____ robust than instrumental judgments.
More
Advantages of the auditory-perceptual method are:
convenience, economy, usefulness for outcome assessment, robustness
- ability to understand speech under various conditions of interference (masking, temporal interruptions, filtering) and to assess disorders of speech and voice over a range of severity
What is the problem with auditory perception only categorizing errors into phonemic categories?
- there may be subtle phonetic (allophonic) differences among items placed in a single category
- biases reduce listener’s sensitivity to subphonemic errors/irregularities
What is a common weakness of many perceptual ratings, and what is a solution to this?
- not suited to multidimensional nature of speech and voice
- measurement techniques have been developed that provide means of analyzing multidimensionality of these signals and determining perceptual independence/nonindependence of the dimensions that constitute overall quality impression (based on principles of signal detection theory and Thurstonian scaling)
What is the trace-context theory?
- posits that 2 types of noise determine the resolution of judgments on a decision axis
- has the potential of delineating dimensions that constitute a multidimensional signal such as speech
Describe the two types of noise implicated in the trace-context theory.
- sensory noise – associated with inherent neural noise
- memory noise – varies with memory load experienced by subject in different experimental paradigms (single interval tasks, pairwise discrimination tasks)
Perception is organized with respect to a ______ model.
Source-filter
T/F
Listeners perform more poorly in spectral slope discrimination when fundamental frequency is varied
True – because both spectral slope and fundamental frequency are source properties, their covariation hindered listeners’ judgments
T/F
Listeners also perform poorly in their discrimination of spectral slope when slope shape (filter property) is varied.
False
What are the implications of the source-filter model on perception of voice disorders and dysarthria?
- because vocal pathologies frequently involve simultaneous changes in 2 or more source characteristics, it would be expected that listeners’ judgements of voice disorders would reflect this limitation on discrimination for any given source characteristic
- would be expected that simultaneously varying filter (vocal tract) characteristics, which are likely to occur in dysarthria, would present difficulties for discrimination of any single filter characteristic
T/F
The high degree of nonindependence among perceptual judgments is justification for the simplification of many multidimensional systems
True
T/F
It’s not important to include summaries of the listening environment in your clinical reports documenting assessming/treatment of speech sound disorders.
False
What is the problem with trying to overcome perceptual errors by supplementing perceptual judgment with instrumental measures?
- only weak associations between acoustic measures and perceptual ratings
- a fixed set of acoustic measures will not necessarily correlate highly with perceived severity across a range of vocal abnormalities and across judges
An important step in improving correlations between perceptual and acoustic data is:
to use appropriate frequency and intensity scales in acoustic analysis
What are 2 alternatives to the Bark scale?
- equivalent-rectangular bandwidth rate (ERB-rate) scale (may be superior to the Bark scale for the analysis of F1 of high vowels and intonation; may serve as psychophysically based scale that’s suited to both low-frequency energy of intonation and higher-frequency info of vowel formants and consonant noise energy)
- wave collation visual speech display (makes both fundamental frequency and formant frequency info available on same display)
What might be a useful way of understanding individual differences in perception?
Self-organizing systems
A good example of an acoustic measure that can be used to validate perception is:
VOT
An example of a new approach to the acoustic analysis of speech is:
statistical analyses of a power density spectrum
What can be used to classify noise spectra for stops and fricatives in both normal and disordered speech?
- coefficients for 3 spectral moments (mean, skewness, kurtosis)
- formant frequencies of the first 3 formants (derived by automatic LPC formant tracking), fundamental frequency, rms amplitude
What are the advantages of statistical analyses of a power density spectrum?
- could be useful supplement to perceptual judgments of both segmental and suprasegmental properties of speech
- analyses are sensitive to changes in speech that occur with disease progression (specifically ALS) and with clinical treatment (palatal lift prosthesis)
- automatic analysis would provide fast and reliable data with minimal human effort
It may be helpful to conceptualize acoustic-perceptual relation as polar extremes. What are these extremes?
- many-to-one relation (several acoustic variables are associated with a single perceptual attribute, such as hoarseness)
- one-to-one relation (perceptual decision can be mapped against a single acoustic dimension or against a weighted average of 2 or more acoustic dimensions; eg consonant voicing contrasts)
Acoustic analysis of _________ could be used to define physical correlates of perceived severity
Slowly varying temporal properties
Two possible explanations for the limitations of expert perceptual judgment are:
- characterized by intuitive rather than analytic performance
- current understanding of perceptual decision rules is inadequate to permit an adequate computer-modelling of the process by which humans reach perception judgments
