2
Considerations for Evaluating Hearing Function
This chapter provides background information on the characteristics of hearing tests, presentation level, and test setup as well as defines various important terms related to the assessment of speech perception and discusses other considerations concerned with evaluating hearing function. It provides the background material necessary to understand Statement of Task items 1, 2, and 3c. Those issues also will be addressed more specifically in the chapters that follow. The chapter also provides a brief overview of the many types of speech recognition tests, which are covered in more detail in Chapter 4.
AUDITORY PROCESSING: DETECTION AND PERCEPTION
Audiologists assess at least two distinct factors with respect to auditory processing: detection and perception. Detection, also referred to as sensation, is a lower-order sensory process characterized by the perceptual recognition or awareness of an auditory stimulus. An audiogram represents detection-based processes in such a way that audiometric thresholds represent the lowest level at which an individual can reliably detect tonal stimuli across the frequency range most important for speech recognition. Perception is a higher-order sensory process that requires that the detected stimulus be transmitted to the central auditory system. As such, the assessment of speech recognition represents various aspects of perception so that a listener’s score represents first the detection (awareness) of the stimulus, then the discrimination (speech versus non-speech), identification
(particular talker), and, ultimately, comprehension of the speech stimulus (Erber, 1982).
Sentence recognition relies heavily on “top-down processing” (i.e., neurocognitive processes that access prior linguistic knowledge necessary for understanding semantic context, that use auditory working memory, and that are dependent on processing speed—all of which play critical roles in decoding a degraded auditory signal). Individuals with severe-to-profound sensorineural hearing loss—particularly those using cochlear implants—report that speech recognition is much more difficult in the presence of background noise (e.g., Donaldson et al., 2009). Despite these subjective reports of communication difficulty in noise that is characteristic of everyday listening environments, cochlear implant recipients achieve normal to near-normal detection of sounds in their environment with cochlear implant–aided audiograms in the range of 20 to 30 dB HL (decibel hearing level) from 250 through 6,000 Hz (e.g., Skinner et al., 1997, 1999). Focusing simply on detection as shown on the aided audiogram leads to over-estimating the auditory and communication potential of cochlear implant recipients, as cochlear implants do not amplify sounds to improve normal hearing; instead, they give a person a representation of sounds in the environment, which in turn helps with understanding speech (NIDCD, 2017). Consequently, the success in treating hearing loss through the use of cochlear implants is measured through functional measures of auditory processing relating to perception, namely assessments of speech recognition.
Hearing tests for those with cochlear implants are different from those used for people using acoustic hearing. The ideal test for those with cochlear implants would be reliable, highly sensitive to different conditions, and correlate well with speech perception abilities in the real world (Cullington and Aidi, 2017; Mackersie, 2002). Across-test correlation, or agreement, is also an important consideration in the choice of measures used for hearing assessment. Additional considerations for the U.S. Social Security Administration (SSA) include testing for success in the workplace and testing in children who may be uncooperative or who have experienced delays in developing language or speech skills.
Labeled cochlear implant criteria vary across manufacturers. The least restrictive labeled indications for conventional adult cochlear implant candidacy include moderate to profound sensorineural hearing loss in both ears and sentence recognition test scores ≤ 50 percent in the ear to be implanted and ≤ 60 percent in the best-aided condition (Cochlear Americas, 2020). The least restrictive labeled indications for children 2 to 17 years include severe to profound hearing loss in both ears and limited benefit from hearing aids, defined as word recognition scores ≤ 30 percent on the Multisyllabic Neighborhood Test (MLNT) or the Lexical Neighborhood Test (LNT) (Cochlear Americas, 2020).
HEARING TEST CHARACTERISTICS
Because hearing is a complex process that involves several levels of sensory processing, there are multiple considerations that an audiologist must take into account when preparing to evaluate hearing. Hearing tests can be categorized by the level of auditory skills required to complete a task (Tye-Murray et al., 2014). N. P. Erber first described the categories of auditory assessment from most basic to most complex in the 1982 manuscript Auditory Training; these categories are sound awareness, sound discrimination, sound identification, and comprehension (Erber, 1982). Sound awareness is the most basic level of auditory function and refers to the ability to detect the presence or absence of sound. Sound discrimination refers to the ability to discriminate between sounds and to recognize changes in sound over time. Sound identification is the ability of the listener to label or categorize a sound. Comprehension is the highest level of auditory skills and requires the listener to understand and interpret the sound. Functional assessment of hearing draws on tests of each of those auditory skill levels. Each higher level of auditory skill level is dependent on auditory skills lower in the hierarchy. Auditory–verbal communication with other people in occupational settings requires functional ability at all four levels.
The interdependence across levels of auditory skills has led to two approaches to functional hearing assessment related to occupational settings. One approach attempts to estimate the impact of environmental factors on individual auditory skills by testing lower-level auditory skills in realistic listening environments (Soli et al., 2018). That approach is useful for predicting functional hearing abilities at the group level, but it is less useful for predicting performance at the individual level or for specific work environments (Soli et al., 2018). The other approach directly assesses higher-level auditory skills such as identification or comprehension and may also attempt to recreate elements of a real-world listening environment (McGregor, 2003). That higher-level approach, however, can be limited by specificity. For example, if the assessment is of hearing on a radio or telephone, it cannot generalize to other occupational activities such as face-to-face communication or to auditory signals from equipment. Thus, functional hearing assessments often combine lower- and higher-level approaches to better reflect hearing abilities that are more general and tasks that are more specific to a particular occupation (NASEM, 2019).
SSA currently employs a combination of tests of varying auditory skills, including pure tone testing, speech detection via speech reception threshold, and speech recognition (namely unaided monosyllabic word recognition). However, for individuals with cochlear implants, a similar assessment cannot be completed because cochlear implant recipients hear through their implanted device transmitted via an externally worn sound processor.
Furthermore, as mentioned in the introduction, the characterization of aided detection on the audiogram for a cochlear implant recipient does not accurately reflect functional performance on measures of speech recognition, which are critical for effective communication. As such, SSA currently relies on a single, higher-level test—the Hearing in Noise Test (HINT) sentences (Nilsson et al., 1994)—that assesses speech recognition in a quiet background with sentences presented in the sound field1 via a loudspeaker placed at 0o azimuth (i.e., directly in front of the listener) (SSA, 2010).
PRESENTATION LEVEL AND TEST SETUP
The standard unit of measurement used to express the level of a sound is the decibel (dB). However, a description of the sound level in dB alone is not sufficient to characterize the magnitude of a signal because that description is completely dependent on the reference level against which it was compared. Sound pressure level (SPL) is a common measure of sound level, with units in dB, that describes the displacement of air molecules with reference to 20 micropascals (20 μPa). Calculation of dB requires a reference. dB SPL is the reference that most people are familiar with, and it is in reference to the pressure of the measured displacement of air molecules relative to the surrounding or ambient air pressure, the latter of which is 20 μPa. dB SPL can be measured with a single measurement microphone. For dB SPL, the calculation is 20 log10(p/po), where p is the physical sound pressure of the signal measured in Pa, and po is 20 μPa (the reference sound pressure measurement in air—ambient air pressure). So, if the physical measurement is 20 μPa, then the calculated level in SPL = 0 dB SPL.
Due to the structural, mechanical, and resonant characteristics of the human auditory system, sound detection (sensitivity) is not equivalent in dB SPL across the frequency range of human hearing. Specifically, humans are most sensitive in the mid-frequency range from 2,000 to 6,000 Hz, with detection thresholds being much higher (i.e., poorer hearing) for sounds that are lower or higher in frequency than sounds in this middle ground. In an effort to provide a norm-referenced scale of hearing detection across the audiometric frequency range—typically 250 through 8,000 Hz—the dB HL scale was created. In this dB HL scale the zero reference level is defined in such a way that mean audiometric detection thresholds in dB SPL for individuals with normal hearing are set to be 0 dB HL across all frequencies.
The dB HL scale is used to express audiometric thresholds for pure tones and for spondees2 used to determine speech reception thresholds and
___________________
1 A defined space containing sound waves; in this context, the room that the listener is in.
2 Spondees refer to compound words consisting of two independent words spoken with equal stress such as “airplane,” “birthday,” “popcorn,” etc.
also to characterize the presentation level of monosyllables for unaided word recognition. In other words, the use of the dB HL scale is limited to diagnostic audiology practices. The dB HL scale is not used as the unit of measurement or calibration for speech recognition assessment in the sound field because such assessment should be accomplished at speech levels typically encountered in everyday communicative environments, which are measured and expressed in dB SPL.
In the mid-1970s, the U.S. Environmental Protection Agency funded a study to characterize speech and environmental sound levels encountered in everyday listening environments (Pearsons, 1977). That study found that average “normal” conversational speech at average conversational distances (~1 meter) was at 60 dB SPL and that higher speech levels representative of raised (65 dB SPL), loud (74 dB SPL), or shouted (84 dB SPL) voices could not be sustained for extended periods of time—particularly in the presence of background noise—given the vocal effort required of the talker (Olsen, 1998; Pearsons, 1977). Given that 60 dB SPL represents average conversational speech levels, best practices recommendations, as included in the Minimum Speech Test Battery (Auditory Potential, 2011) and pediatric Minimum Speech Test Battery (MSTB) (Uhler et al., 2017) manuals, specify the use of recorded speech stimuli presented at 60 dB SPL for the assessment of speech recognition performance for both pre- and post-implant assessment. While it is possible to characterize speech levels in the sound field using the dB HL scale, accounting for across-frequency audibility in the free field would transform 60 dB SPL to 40 dB HL. However, because sound field calibration uses a free-field microphone attached to a sound level meter and sound level meters are not equipped with a dB HL reference, presentation levels for speech recognition testing obtained in a sound field are characterized in dB SPL. As such, a presentation level recommendation referencing dB HL for sound field assessment is not possible and would therefore be inappropriate.
Calibration of Speech Stimuli in the Sound Field
When an acoustic signal is presented in the sound field, the sound can be characterized in the near field and in the far field relative to the listener. In the near field, large changes in SPL are associated with small changes in distance from the loudspeaker. To avoid rapid changes in SPL at the ear should the listener move, the listener in sound field testing is placed between the near- and far-field boundaries. To achieve that, the loudspeaker is placed at approximately 1 meter from the listener (Dirks et al., 1976) and should not be situated close to the walls of the test booth or other reflective surfaces.
Speech stimulus calibration for sound field testing focuses on output-based calibration. Output-based calibration in the audiology clinic is
achieved by varying the audiometer dial setting, in dB HL, to achieve the desired presentation level in the sound field, in dB SPL, as measured with a sound level meter (SLM). The SLM with an attached free field microphone is generally placed on a stand and placed at the position of the listener’s head when seated in a chair. This height is approximately 39 inches (86 cm) from the floor, which is also the recommended height of the loudspeaker (Auditory Potential, 2011). For sound field calibration, a calibration noise is required as the use of a calibration tone would result in standing waves in the free field. The calibration noise is normalized to the root mean square level of the accompanying speech stimuli. Thus, the clinician adjusts the audiometer dial in 1 dB increments to achieve a level of 60 dB SPL as shown on the SLM. The numerical reference on the audiometer dial setting (in dB HL) will likely not match the numerical level of the dB SPL reading. This mismatch between dB SPL and dB HL reading on the audiometer is expected and is related to the input level of the stimulus as saved to a compact disc or, if speech stimuli saved to an attached computer are being played, is related to the sound card specifications.
Most clinicians are familiar with the use of calibration tones, which are critical for input-based calibration completed for unaided speech audiometry, such as with spondees and monosyllables. However, it is not necessary to complete input-based calibration for sound field testing except to ensure that the input stimulus and audiometer sensitivity setting (EXT A or B)3 are not resulting in the stimulus being clipped. That would be evident by a peaking response on the VU meter4 of the audiometer. Note that if the input sensitivity for EXT A or B or the computer master volume is adjusted in any way after output calibration has been completed in the sound field, then the output-based calibration must be repeated. That is a quick and simple step to ensure that the audiometer’s dial setting yields 60 dB SPL in the sound field.
Ideally, speech stimuli presented in the sound field should be calibrated every day prior to each sound field assessment of speech recognition; however, unless a clinic owns a SLM dedicated for sound field calibration, that is not a realistic expectation. The next best option would be daily calibration, because sound field assessment of speech recognition performance without prior calibration might result in an inaccurate description of a listener’s auditory performance at the desired presentation level. Furthermore, without calibration the tester could potentially under- or over-estimate speech recognition for patients as the tester could be presenting at a much lower or higher presentation level than intended. That is particularly true if
___________________
3 EXT A or B refers to the external input setting or the auxiliary input for the audiometer.
4 VU meter refers to a volume unit meter, which is a device that displays a representation of the signal level in audio equipment.
a clinic is using digitized/computerized speech stimuli, as the master volume on the computer might be accidentally adjusted between assessments. For audiology clinics where daily access to a SLM is not possible, clinicians can access a variety of SLM applications from any smartphone. More important than the specific app itself is that the clinician tests the accuracy of the SLM app using the smartphone-integrated microphone against the calibrated SLM with a free-field microphone.
Speech Recognition in Quiet Versus Noise
Improvements in cochlear implant technology and the expansion of adult implant indications have produced increasing levels of speech recognition in quiet, to the point that unilateral cochlear implant recipients with post-lingual onset of deafness are routinely achieving 60 percent open-set5 word recognition, on average (e.g., Buchman et al., 2020; Holden et al., 2013). That outcome is essentially double what was reported for adults with the first generation cochlear implant system (Skinner et al., 1994). Indeed, an increasingly higher proportion of adult and pediatric implant recipients demonstrate at or near ceiling-level performance for sentence recognition in quiet (Dunn et al., 2020; Gifford et al., 2018). Despite that great success, most cochlear implant recipients exhibit significant communication difficulty in the presence of background noise, with mean scores dropping by 30 to 40 percentage points at +5 dB signal-to-noise ratio (SNR)6 as compared with scores in quiet (e.g., Dunn et al., 2020; Gifford et al., 2018). In everyday listening environments such as restaurants, train stations, and department stores, +5 dB is the most common SNR experienced, providing evidence for the ecologic validity of speech recognition in noise assessments (Pearsons et al., 1977; Smeds et al., 2015). For that reason, both the adult and pediatric MSTB7 have recommended assessment of speech recognition
___________________
5 Open set versus closed set: Closed-set tests are designed with a fixed number of possible responses such as are encountered in a multiple-choice exam or a speech recognition task for which the listener must pick the correct word from a fixed set of words provided to the listener. Open-set tasks are designed without limitations on the possible responses, such as essay questions on an exam or speech recognition tasks for which the listener must indicate their responses without any prompts.
6 Although the term “signal-to-noise ratio (SNR)” is commonly used in audiology and hearing science to describe the level difference between the target signal (typically speech) and the background noise, it is actually a misnomer. For example, if the background noise level is 60 dB SPL and the target speech signal is 65 dB SPL, the SNR would be described as 5 dB. The term “ratio” is a misnomer because a mathematical ratio is a quotient, whereas, because decibels are defined in terms of logarithms, an SNR is actually a difference score (although it does correspond to a ratio).
7 MSTB is the Minimum Speech Test Battery (Auditory Potential, 2011), designed to document word recognition in bilaterally hearing impaired cochlear implant individuals.
in noise using a +5 dB SNR to describe receptive communication abilities for patients in realistic listening scenarios (Auditory Potential, 2011; Uhler et al., 2017).
Hearing Configuration: Unilateral and Bilateral Cochlear Implants
A number of studies have compared bilateral and unilateral cochlear implant recipients using open-set sentence tests including the Bamford-Kowal-Bench Speech-in-Noise Test (BKB-SIN), the HINT, and the Maryland consonant–nucleus-consonant (CNC) word test (Dorman et al., 2011). The primary findings are briefly described below as well as in Table 2-1, which provides additional information.
Speech recognition performance significantly increased over the first 6 months of cochlear implant use for all conditions tested (ear cochlear implant alone and bilateral cochlear implant) (Buss et al., 2008; Koch et al., 2010; Litovsky et al., 2006).
On average, speech recognition in the bilateral cochlear implant condition was significantly higher than either cochlear implant alone for words (Buss et al., 2008; Koch et al., 2010; Litovsky et al., 2006), sentences (Buss et al., 2008; Koch et al., 2010; Litovsky et al., 2006), and sentences in noise (Buss et al., 2008; Koch et al., 2010; Litovsky et al., 2004, 2006).
Speech recognition performance for each cochlear implant alone (Eapen et al., 2009) and bilateral cochlear implant (Chang et al., 2010; Eapen et al., 2009) was stable over 4–6 years following cochlear implant activation.
Verification of Cochlear Implant Function
The American Academy of Audiology’s Guidelines for the Audiologic Management of Adult Hearing Impairment specifies that verification of hearing aid function should be completed via probe microphone measurements in the ear canal to ensure that the prescribed gain and output are achieved using a validated prescriptive fitting formula (Valente et al., 2006). In fact, hearing aid audibility is typically measured via sound pressure level (SPL) in the ear canal at various input signal levels over the range typical of conversational speech, from casual (55 dB SPL) to average (60 dB SPL) and up to raised/loud speech (70 dB SPL) (Olsen, 1998; Pearsons et al., 1977). However, for cochlear implant verification, it is difficult to directly measure a physical output level because cochlear implants provide a transcutaneous transmission of the incoming stimulus to the implanted system using a radio frequency signal. That requires the use of behavioral responses to acoustic stimuli presented in the sound field as a measure of low-level audibility in cochlear implant users.
TABLE 2-1 Comparison of Open-Set Sentence and Monosyllabic Word Recognition in Adults with Simultaneous Bilateral Cochlear Implants
Outcome Measure | Scoring Method | Reference (Sample Size) | Results |
---|---|---|---|
HINT sentences in quiet | % correct |
|
(1) Performance improved from 1 to 6 months of cochlear implant use. (1 and 2) No significant difference between individual ear scores at any time point from preoperation to 6 months post-operation; binaural summation at all time points (10 percentage points, on average); average bilateral HINT score reached 90 percent correct by 6 to 8 months. (2) Performance significantly improved from 6 to 8 months of cochlear use. |
CNC words in quiet | % correct |
|
(1–3) No significant difference between individual ear scores at any time point; binaural summation at all time points (10 percentage points, on average). (1–5) Average bilateral CNC score was 60 percent by 6–12 months of cochlear implant use. (4 and 5) Bilateral CNC scores were stable from 6 months up to 4–6 years. |
BKB-SIN | SNR-50 (dB) S0N0 |
|
(1 and 2) 2–3 dB bilateral benefit (binaural summation) over either unilateral cochlear implant alone. (1) Some listeners exhibited significant differences across ears, despite no interaural differences for speech in quiet. |
NOTE: CNC = consonant–nucleus–consonant test; dB = decibel; HINT = Hearing in Noise Test; S0N0 = speech and noise both presented at 0° azimuth; SNR = signal-to-noise ratio.
Cochlear implant–aided sound field thresholds are critical for programming verification to ensure the audibility of low-level stimuli, such as soft speech. Skinner and colleagues first discussed the importance of sound field audiometric thresholds in the range of 20–30 dB HL for determining the minimum audibility available for implant recipients (Skinner et al., 1997, 1999), and there are now a number of published papers demonstrating that aided thresholds in the range of 20–25 dB HL are associated with significantly higher speech recognition outcomes for adult and pediatric cochlear implant recipients (e.g., Davidson et al., 2009; de Graaff et al., 2020; Holden et al., 2013, 2019). Thus, it is recommended that prior to assessing speech recognition abilities for cochlear implant recipients, aided
sound field thresholds to warbled (frequency-modulated) pure tones are documented in the range of 20–30 dB HL from 250 through 6,000 Hz.
If aided thresholds are higher (i.e., poorer hearing) than the 20–30 dB range, the two most likely causes are implant processor microphone fidelity or suboptimal cochlear implant programming. For patients showing elevated aided thresholds, particularly in the high-frequency region, microphone fidelity may be suspect. It is recommended that a technician or audiologist listen to all implant processor microphones to check for static, low-level humming, or any evidence of compromised sound quality. Microphone issues often can be resolved via the replacement of microphone covers or filters on the implant sound processor. Less often, microphone issues will require processor repair or replacement. In cases where aided sound field thresholds are poor and microphone issues are ruled out, the most likely solution will be additional cochlear implant programming focusing first on lower stimulation levels (often called threshold or T levels). Increasing lower stimulation levels will generally result in better aided detection thresholds, although the exact levels at which T levels are set varies by each of the implant manufacturers. Note that each implanted ear should be verified independently to verify appropriate programming for each ear. Also, unilateral cochlear implant recipients who have some acoustic hearing in the non-implanted ear should have that ear occluded via foam plug or by a completely occluding earmold so that the implant ear is isolated for sound field testing. It is necessary to verify proper function of the cochlear implant(s) prior to assessing any speech reception in a sound field to ensure that the wearer’s best performance is captured.
Aided Versus Unaided Speech Recognition Assessment
As mentioned above, audiologists routinely complete both unaided and aided speech recognition testing. Unaided speech recognition is most common in audiology practice as it involves threshold-based measures such as a speech reception threshold (SRT) and unaided word recognition testing at a suprathreshold level—typically in the range of 20 to 40 dB above SRT (Guthrie and Mackersie, 2009). Threshold and suprathreshold speech assessments should always include the use of recorded materials for standardized assessment across clinics and test administrators and to allow accurate tracking of performance over time. However, it has been reported that more than two-thirds of audiologists routinely use monitored live voice (MLV) for speech recognition assessments (Martin et al., 1998; Medwetsky et al., 1999). Roeser and Clark (2008) assessed word recognition obtained via MLV and recorded stimuli for 16 adult participants (32 ears) and found that word recognition scores for MLV and recorded stimuli were significantly different for 23 of the 32 ears (72 percent of the sample).
Furthermore, the difference between the MLV and recorded speech recognition scores was as high as 80 percentage points. In a similar study of 29 pediatric cochlear implant recipients aged 4 to 17 years, Uhler et al. (2016) assessed word and sentence recognition via MLV and recorded stimuli. They observed a 13 percentage point difference (range: 0 to 28 percentage points) between MLV and recorded stimulus presentation, which was found to be statistically significant.
There is a place for both unaided and aided assessments of suprathreshold speech recognition. As mentioned above, unaided assessments are typically conducted at levels 20–40 dB above the SRT, in dB HL. As an example, for a listener with a moderate to severe hearing loss exhibiting a 60 dB HL SRT, unaided word recognition would be tested in the range of 80–100 dB HL, corresponding to a range of approximately 100–120 dB SPL. Although unaided word recognition testing does not apply frequency-specific amplification to compensate for a listener’s hearing loss, testing at such high presentation levels often results in better speech recognition scores than one would obtain in everyday listening scenarios where speech levels range from 55 to 70 dB SPL (Olsen, 1998; Pearsons, 1977). To gain a comprehensive profile of an individual’s auditory skills in various communicative environments, unaided speech recognition should be paired with aided speech recognition—both in quiet and in noise—at an everyday conversational level, usually 60 dB SPL in quiet and 65 dB SPL in the presence of noise (Pearsons, 1977; Uhler et al., 2017).
INTRODUCTION TO SPEECH TESTS
Assessment of speech recognition taxes the central auditory system by requiring perception-based processing that involves discrimination, identification, and comprehension to various degrees (Erber, 1982). There are many available speech tests that are used to assess auditory function in individuals with cochlear implants. An overview of these tests can be found in Table 2-2. For additional information on each of the tests, please see Chapter 4.
Scoring Speech-in-Noise Tests
Clinical speech-in-noise tests are commonly scored using one of two methods. The most common scoring method for cochlear implant recipients is “percent correct based,” reporting the percentage of words repeated correctly across a list of sentences presented at a fixed SNR, such as +5 dB SNR. In that particular example, the speech stimuli would be presented 5 dB higher than the background noise. The second most common scoring method for cochlear implant recipients is “threshold based,” reporting the
TABLE 2-2 Commonly Used Speech Perception Tests
Speech Perception Test | Target Population | General Overview | Scoring | Reference |
---|---|---|---|---|
Word Tests | ||||
WIN (Words in Noise) | Adults | Monosyllabic words presented at an adaptive level in fixed multi-talker babble | SNR-50 | Wilson, 2003; Wilson and Burks, 2005 |
Digit Triplet | Adults | Digit triplets adaptively presented in fixed noise | Speech reception threshold | Smits et al., 2013 |
MLNT (Multisyllabic Lexical Neighborhood Test) | Children | Multisyllabic words presented at fixed level | Percent correct | Kirk et al., 1995 |
LNT (Lexical Neighborhood Test) | Children | Monosyllabic words presented at fixed level | Percent correct | Kirk et al., 1995 |
CNC (Consonant–Nucleus–Consonant) | Adult | Monosyllabic word lists presented at fixed level | Percent correct | Causey et al., 1984 |
NU6 (Northwestern University Test No. 6) | Adult | Monosyllabic word lists presented at fixed level | Percent correct | Tillman and Carhart, 1966 |
PBK (Phonetically Balanced Kindergarten) | Children | Monosyllabic word lists presented at a fixed level | Percent correct | Haskins, 1949 |
Sentence Tests | ||||
HINT (Hearing in Noise Test) | Adults | Sentences presented at an adaptive level in fixed multi-talker babble | SNR loss/SNR-50 | Nilsson et al., 1994 |
Speech Perception Test | Target Population | General Overview | Scoring | Reference |
---|---|---|---|---|
HINT-C (Hearing in Noise Test-Children) | Children | Sentences presented at an adaptive level in fixed multi-talker babble | SNR loss/SNR-50 | Nilsson et al., 1996 |
BKB-SIN (Bamford-Kowal-Bench Speech in Noise Test) | Adults and children | Sentences presented in four-talker babble with up to eight predetermined SNR levels | SNR loss/SNR-50 | Etymōtic Research, 2005 |
QuickSIN (Quick Speech in Noise) | Adults | Sentences presented in multi-talker babble at six predetermined SNR levels | SNR loss/SNR-50 | Killion et al., 2004 |
AzBio (Arizona Biomedical) | Adults | Sentences presented in 10-talker babble at a fixed SNR | Percent correct | Spahr et al., 2012 |
Pediatric AzBio | Children | Sentences presented in 10-talker babble at a fixed SNR | Percent correct | Spahr et al., 2014 |
CID Sentences (Central Institute for the Deaf) | Adults | Sentences presented at a fixed level, in quiet or noise | Percent correct | Silverman and Hirsh, 1955 |
CUNY Sentences (City University of New York) | Sentences presented at a fixed level, in quiet or noise | Percent correct | Boothroyd et al., 1985 |
NOTE: SNR = signal-to-noise ratio; SNR-50: “threshold-based” scoring reporting the SNR needed to obtain 50 percent correct recognition, measured in decibels.
SNR needed to obtain 50 percent correct recognition, measured in dB. In other words, during the test, the background noise (steady state noise or multi-talker babble) is increased to the point at which the listener can correctly identify 50 percent of the words or sentences, and the difference between the signal and noise is recorded as the listener’s threshold in dB (e.g., Donaldson et al., 2009; Wilson et al., 2007).
The threshold-based method for scoring speech recognition in noise was originally employed using a psychophysical “staircase” procedure in which the level of the background noise is adaptively varied for each sentence based on the listener’s prior response (Leek, 2001). For example, if a listener correctly repeats a sentence, the noise is increased for the following sentence. Conversely, if the patient cannot accurately repeat all words in a sentence for a given SNR, the noise is decreased for the following sentence. In the traditional adaptive staircase procedure, the SNR yielding 50 percent correct is calculated as the average SNR for the last 6 to 8 “reversal points” as it captures the SNR around which the listener achieves approximately 50 percent correct (e.g., Leek, 2001; Nilsson et al., 1994).
The adaptive staircase procedure for threshold-based scoring of speech recognition in noise is not clinically feasible because of time constraints and the need for specialized equipment and software. Thus, for clinical administration of threshold-based measures of speech recognition in noise, the 50 percent point is calculated using the Spearman-Karber equation (Finney, 1952), which can be done across various speech tests in noise, including the digit triplet test (Smits et al., 2004), the Words in Noise test (WIN; Wilson et al., 2007), the QuickSIN (Killion et al., 2004), and the BKB-SIN (Etymōtic Research, 2005). The Spearman-Karber equation (Finney, 1952) is 50 percent = i + ½(d) – (d)(# correct) / (w), where i = the initial presentation level (dB S/B8), d = the attenuation step size (decrement), and w = the number of items per decrement (Wilson et al., 2007).
Sentence Tests and Word Tests
SSA currently uses HINT sentences presented in a quiet background to assess hearing in individuals with cochlear implants at the Listing of Impairments (the Listings) level. The HINT corpus is a list of 25 everyday sentence lists spoken by a single male speaker.
The HINT is an adaptive speech recognition task (i.e., the level of each sentence is adjusted based on the response of the listener). The speech presentation level is decreased after each correct response, which raises the level of difficulty for the next sentence on the list. Conversely, the presentation level is increased after each incorrect response, which reduces the
___________________
8 Signal-to-babble ratio.
difficulty for the following sentence. The level of the noise is held constant, thus adapting the presentation level of the sentences results in assessment at various SNRs.
The adaptive nature of the HINT ensures that the listener will approach a 50 percent correct response rate. Note that the HINT sentences can also be administered without noise in order to assess sentence recognition in quiet. In that case, a threshold for sentence recognition is obtained. If the test is administered in noise, adapting the SNR allows for the estimation of a SNR threshold for speech recognition in noise. As the SNR score decreases, the listening conditions become more difficult. For instance, if a hard-of-hearing listener is able to understand speech at an SNR of –3 dB in an omnidirectional hearing aid setting, then that individual may be able to understand speech at an SNR of –6 dB when the microphone setting has been changed to be directional. The change in microphone allows the listener to understand speech under more adverse conditions, thus improving his or her speech recognition ability in noise. However, everyday sentences inherently contain contextual cues for identification of individual words. Indeed, the HINT sentences contain five words per sentence, on average, and the language is consistent with a first-grade reading level, as evidenced by Flesch-Kincaid grade level assessment (Flesch, 1948; Kincaid et al., 1975).
An obvious concern with the use of speech tests is the extent to which context clues influence overall performance, which is dependent on the interaction between these context clues and a listener’s knowledge of the language. Such an interaction might mean, for example, that subjects who are native speakers of the test language would yield norms that are inappropriate for nonnative speakers (NRC, 2005). The current Listing for hearing addresses this issue by noting,
If you are not fluent in English, you should have word recognition testing using an appropriate word list for the language in which you are most fluent. The person conducting the test should be fluent in the language used for the test. If there is no appropriate word list or no person who is fluent in the language and qualified to perform the test, it may not be possible to measure your word recognition ability (2.00B4).
A second concern with the use of speech tests is the method of scoring sentence recognition performance, although most contemporary sentence tests assess accuracy based on the recognition of keywords. A third concern is whether the test has been standardized in quiet, noise, or both environmental conditions (NRC, 2005). As mentioned previously, the HINT sentence test was both developed and validated as a measure to be presented in an adaptive, staircase procedure in the presence of steady-state
noise (Nilsson et al., 1994). Thus, the use of HINT sentences in a quiet background—or even in the presence of a fixed-SNR noise—is not consistent with the test’s standardization and normative data. Furthermore, as discussed previously, the high level of specificity might not apply to the working environment of the individual taking the test.
Thus, sentence tests, including the HINT, the BKB-SIN, and the QuickSIN, all provide varying degrees of context cues to the listener (Wilson et al., 2007). By contrast, word tests using monosyllabic or bisyllabic words test a listener’s ability to recognize discrete phonemes. These tests are typically conducted in quiet (Martin et al., 1998). Each test involves presenting full lists of the recorded materials (usually 50 words). Performance is quantified by a percentage correct score (NRC, 2005). Additionally, mono- and bisyllabic tests benefit from not being influenced by linguistic context or memory (Massa and Ruckenstein, 2014; McArdle et al., 2005). Those issues will be discussed further in Chapter 3.
REFERENCES
Auditory Potential. 2011. Minimum Speech Test Battery for adult cochlear implant users. http://www.auditorypotential.com/MSTBfiles/MSTBManual2011-06-20%20.pdf (accessed October 17, 2020).
Boothroyd, A. 1985. Evaluation of speech production of the hearing impaired: Some benefits of forced-choice testing. Journal of Speech and Hearing Research 28(2):185–196.
Buchman, C. A., R. H. Gifford, D. S. Haynes, T. Lenarz, G. O’Donoghue, O. Adunka, A. Biever, R. J. Briggs, M. L. Carlson, P. Dai, C. L. Driscoll, H. W. Francis, B. J. Gantz, R. K. Gurgel, M. R. Hansen, M. Holcomb, E. Karltorp, M. Kirtane, J. Larky, E. A. M. Mylanus, J. Thomas Roland, Jr., S. R. Saeed, H. Skarzynski, P. H. Skarzynski, M. Syms, H. Teagle, P. H. van de Heyning, C. Vincent, H. Wu, T. Yamasoba, and T. Zwolan. 2020. Unilateral cochlear implants for severe, profound, or moderate sloping to profound bilateral sensorineural hearing loss a systematic review and consensus statements. JAMA Otolaryngology—Head and Neck Surgery 146(10):942–953.
Buss, E., H. C. Pillsbury, C. A. Buchman, C. H. Pillsbury, M. S. Clark, D. S. Haynes, R. F. Labadie, S. Amberg, P. S. Roland, P. Kruger, M. A. Novak, J. A. Wirth, J. M. Black, R. Peters, J. Lake, P. A. Wackym, J. B. Firszt, B. S. Wilson, D. T. Lawson, R. Schatzer, P. S. C. D’Haese, and A. L. Barco. 2008. Multicenter U.S. bilateral MED-EL cochlear implantation study: Speech perception over the first year of use. Ear and Hearing 29(1):20–32.
Causey, G. D., L. J. Hood, C. L. Hermanson, and L. S. Bowling. 1984. The Maryland CNC test: Normative studies. International Journal of Audiology 23(6):552–568.
Chang, S. A., R. S. Tyler, C. C. Dunn, H. Ji, S. A. Witt, B. Gantz, and M. Hansen. 2010. Performance over time on adults with simultaneous bilateral cochlear implants. Journal of the American Academy of Audiology 21(1):35–43.
Cochlear Americas. 2020. Nucleus cochlear implants: Physician’s package insert. https://www.accessdata.fda.gov/cdrh_docs/pdf/P970051S172C.pdf (accessed September 2, 2020).
Cullington, H. E., and T. Aidi. 2017. Is the digit triplet test an effective and acceptable way to assess speech recognition in adults using cochlear implants in a home environment? Cochlear Implants International 18(2):97–105.
Davidson, L. S., M. W. Skinner, B. A. Holstad, B. T. Fears, M. K. Richter, M. Matusofsky, C. Brenner, T. Holden, A. Birath, J. L. Kettel, and S. Scollie. 2009. The effect of instantaneous input dynamic range setting on the speech perception of children with the Nucleus 24 implant. Ear and Hearing 30(3):340–349.
de Graaff, F., B. I. Lissenberg-Witte, M. W. Kaandorp, P. Merkus, S. T. Goverts, S. E. Kramer, and C. Smits. 2020. Relationship between speech recognition in quiet and noise and fitting parameters, impedances and ECAP thresholds in adult cochlear implant users. Ear and Hearing 41(4):935–947.
Dirks, D. D., C. Kamm, and S. Gilman. 1976. Bone conduction thresholds for normal listeners in force and acceleration units. Journal of Speech and Hearing Research 19(1):181–186.
Donaldson, G. S., T. H. Chisolm, G. P. Blasco, L. J. Shinnick, K. J. Ketter, and J. C. Krause. 2009. BKB-SIN and ANL predict perceived communication ability in cochlear implant users. Ear and Hearing 30(4):401–410.
Dorman, M. F., W. A. Yost, B. S. Wilson, and R. H. Gifford. 2011. Speech perception and sound localization by adults with bilateral cochlear implants. Seminars in Hearing 32(1):73–89.
Dunn, C. C., J. Oleson, A. Parkinson, M. R. Hansen, and B. J. Gantz. 2020. Nucleus Hybrid S12: Multicenter clinical trial results. Laryngoscope 130(10):E548–E558.
Eapen, R. J., E. Buss, M. C. Adunka, H. C. Pillsbury, and C. A. Buchman. 2009. Hearing-in-noise benefits after bilateral simultaneous cochlear implantation continue to improve 4 years after implantation. Otology and Neurotology 30(2):153–159.
Erber, N. P. 1982. Auditory training. Washington, DC: Alexander Graham Bell Association for the Deaf.
Etymōtic Research. 2005. BKB-SIN user’s manual. https://www.etymotic.com/downloads/dl/file/id/260/product/160/bkb_sintm_user_manual.pdf (accessed October 17, 2020).
Finney, D. J. 1952. Statistical method in biological assay. London, UK: C. Griffen.
Flesch, R. 1948. A new readability yardstick. Journal of Applied Psychology 32(3):221–233.
Gifford, R. H., L. Loiselle, S. Natale, S. W. Sheffield, L. W. Sunderhaus, M. S. Dietrich, and M. F. Dorman. 2018. Speech understanding in noise for adults with cochlear implants: Effects of hearing configuration, source location certainty, and head movement. Journal of Speech, Language, and Hearing Research 61(5):1306–1321.
Guthrie, L. A., and C. L. Mackersie. 2009. A comparison of presentation levels to maximize word recognition scores. Journal of the American Academy of Audiology 20(6):381–390.
Haskins, H. 1949. A phonetically balanced test of speech discrimination for children. Evanston, IL: Northwestern University.
Holden, L. K., C. Brenner, R. M. Reeder, and J. B. Firszt. 2013. Postlingual adult performance in noise with HiRes 120 and ClearVoice Low, Medium, and High. Cochlear Implants International 14(5):276–286.
Holden, L. K., J. B. Firszt, R. M. Reeder, N. Y. Dwyer, A. L. Stein, and L. M. Litvak. 2019. Evaluation of a new algorithm to optimize audibility in cochlear implant recipients. Ear and Hearing 40(4):990–1000.
Killion, M. C., P. A. Niquette, G. I. Gudmundsen, L. J. Revit, and S. Banerjee. 2004. Development of a Quick Speech-in-Noise Test for measuring signal-to-noise ratio loss in normal-hearing and hearing-impaired listeners. Journal of the Acoustical Society of America 116(4 I):2395–2405.
Kincaid, J. P., R. P. Fishburne, Jr., R. L. Rogers, and B. S. Chissom. 1975. Derivation of new readability formulas (Automated Readability Index, Fog Count and Flesch Reading Ease Formula) for Navy enlisted personnel. Research branch report 8-75. Millington, TN: Institute for Simulation and Training.
Kirk, K. I., D. B. Pisoni, and M. J. Osberger. 1995. Lexical effects on spoken word recognition by pediatric cochlear implant users. Ear and Hearing 16(5):470–481.
Koch, D. B., S. D. Soli, M. Downing, and M. J. Osberger. 2010. Simultaneous bilateral cochlear implantation: Prospective study in adults. Cochlear Implants International 11(2):84–99.
Leek, M. R. 2001. Adaptive procedures in psychophysical research. Perception and Psycho-physics 63(8):1279–1292.
Litovsky, R., A. Parkinson, J. Arcaroli, and C. Sammeth. 2006. Simultaneous bilateral cochlear implantation in adults: A multicenter clinical study. Ear and Hearing 27(6):714–731.
Mackersie, C. L. 2002. Tests of speech perception abilities. Current Opinion in Otolaryngology and Head and Neck Surgery 10(5):392–397.
Martin, F. N., C. A. Champlin, and J. A. Chambers. 1998. Seventh survey of audiometric practices in the United States. Journal of the American Academy of Audiology 9(2):95–104.
Massa, S. T., and M. J. Ruckenstein. 2014. Comparing the performance plateau in adult cochlear implant patients using HINT and AzBio. Otology and Neurotology 35(4):598–604.
McArdle, R. A., R. H. Wilson, and C. A. Burks. 2005. Speech recognition in multitalker babble using digits, words, and sentences. Journal of the American Academy of Audiology 16(9):726–739.
McGregor, A. 2003. Fitness standards in airline staff. Occupational Medicine 53(1):5–9.
Medwetsky, L., D. Sanderson, and D. Young. 1999. A national survey of audiology clinical practices. The Hearing Review 11:24–32.
NASEM (National Academies of Sciences, Engineering, and Medicine). 2019. Functional assessment for adults with disabilities. Washington, DC: The National Academies Press.
NIDCD (National Institute on Deafness and Other Communication Disorders). 2017 (March 6). Cochlear implants. https://www.nidcd.nih.gov/health/cochlear-implants (accessed August 19, 2020).
Nilsson, M., S. D. Soli, and J. A. Sullivan. 1994. Development of the Hearing in Noise Test for the measurement of speech reception thresholds in quiet and in noise. Journal of the Acoustical Society of America 95(2):1085–1099.
Nilsson, M. J., S. D. Soli, and D. J. Gelnett. 1996. Development of the Hearing in Noise Test for Children (HINT-C). Los Angeles, CA: House Ear Institute.
NRC (National Research Council). 2005. Hearing loss: Determining eligibility for Social Security benefits. Washington, DC: The National Academies Press.
Olsen, W. O. 1998. Average speech levels and spectra in various speaking/listening conditions: A summary of the Pearson, Bennett, & Fidell (1977) report. American Journal of Audiology 7(2):21–25.
Pearsons, K. S. 1977. Effect of tone/noise combination on speech intelligibility. Journal of the Acoustical Society of America 61(3):884–886.
Roeser, R. J., and J. L. Clark. 2008. Live voice speech recognition audiometry: Stop the madness! Audiology Today 20(1):32–33.
Silverman, S. R., and I. J. Hirsh. 1955. CX problems related to the use of speech in clinical audiometry. Annals of Otology, Rhinology and Laryngology 64(4):1234–1244.
Skinner, M. W., G. M. Clark, L. A. Whitford, P. M. Seligman, J. S. Staller, D. B. Shipp, J. K. Shallop, C. Everingham, C. M. Menapace, P. L. Arndt, T. Antogenelli, J. A. Brimacombe, S. Pijl, P. Daniels, C. R. George, H. J. McDermott, and A. L. Beiter. 1994. Evaluation of a new spectral peak coding strategy for the Nucleus 22-channel cochlear implant system. American Journal of Otology 15(Suppl 2):15–27.
Skinner, M. W., L. K. Holden, and T. A. Holden. 1997. Parameter selection to optimize speech recognition with the nucleus implant. Otolaryngology—Head and Neck Surgery 117(3 I):188–195.
Skinner, M. W., L. K. Holden, T. A. Holden, and M. E. Demorest. 1999. Comparison of two methods for selecting minimum stimulation levels used in programming the Nucleus 22 cochlear implant. Journal of Speech, Language, and Hearing Research 42(4):814–828.
Smeds, K., F. Wolters, and M. Rung. 2015. Estimation of signal-to-noise ratios in realistic sound scenarios. Journal of the American Academy of Audiology 26(2):183–196.
Smits, C., T. S. Kapteyn, and T. Houtgast. 2004. Development and validation of an automatic speech-in-noise screening test by telephone. International Journal of Audiology 43(1):15–28.
Smits, C., S. Theo Goverts, and J. M. Festen. 2013. The Digits-in-Noise Test: Assessing auditory speech recognition abilities in noise. Journal of the Acoustical Society of America 133(3):1693–1706.
Soli, S. D., C. Giguère, C. Laroche, V. Vaillancourt, W. A. Dreschler, K. S. Rhebergen, K. Harkins, M. Ruckstuhl, P. Ramulu, and L. S. Meyers. 2018. Evidence-based occupational hearing screening I: Modeling the effects of real-world noise environments on the likelihood of effective speech communication. Ear and Hearing 39(3):436–448.
Spahr, A. J., M. F. Dorman, L. M. Litvak, S. Van Wie, R. H. Gifford, P. C. Loizou, L. M. Loiselle, T. Oakes, and S. Cook. 2012. Development and validation of the AzBio sentence lists. Ear and Hearing 33(1):112–117.
Spahr, A. J., M. F. Dorman, L. M. Litvak, S. J. Cook, L. M. Loiselle, M. D. DeJong, A. HedleyWilliams, L. S. Sunderhaus, C. A. Hayes, and R. H. Gifford. 2014. Development and validation of the Pediatric AzBio sentence lists. Ear and Hearing 35(4):418–422.
SSA (U.S. Social Security Administration). 2010. Listing of Impairments—Adult listings (part A). In The blue book. Washington, DC: U.S. Social Security Administration.
Tillman, T. W., and R. Carhart. 1966. An expanded test for speech discrimination utilizing CNC monosyllabic words: Northwestern University Auditory Test no. 6. Technical report SAM-TR-66-55. Brooks Air Force Base, Texas: USAF School of Aerospace Medicine. https://pdfs.semanticscholar.org/008e/54f6708d34231a350af7e7b585b53a8c048c.pdf (accessed January 15, 2021).
Tye-Murray, N., S. Hale, B. Spehar, J. Myerson, and M. S. Sommers. 2014. Lipreading in school-age children: The roles of age, hearing status, and cognitive ability. Journal of Speech, Language, and Hearing Research 57(2):556–565.
Uhler, K., A. Biever, and R. H. Gifford. 2016. Method of speech stimulus presentation impacts pediatric speech recognition monitored live voice versus recorded speech. Otology and Neurotology 37(2):e70–e74.
Uhler, K., A. Warner-Czyz, and R. Gifford. 2017. Pediatric Minimum Speech Test Battery. Journal of the American Academy of Audiology 28(3):232–247.
Valente, M., D. Benson, T. Chisolm, D. Citron, D. Hampton, A. Loavenbruck, T. Ricketts, H. Solodar, and R. Sweetow. 2007. Guidelines for the Audiologic Management of Adult Hearing Impairment Task Force members. https://audiology-web.s3.amazonaws.com/migrated/haguidelines.pdf_53994876e92e42.70908344.pdf (accessed January 16, 2021).
Wilson, R. H. 2003. Development of a speech-in-multitalker-babble paradigm to assess word-recognition performance. Journal of the American Academy of Audiology 14(9):453–470.
Wilson, R. H., and C. A. Burks. 2005. Use of 35 words for evaluation of hearing loss in signal-to-babble ratio: A clinic protocol. Journal of Rehabilitation Research and Development 42(6):839–851.
Wilson, R. H., R. A. McArdle, and S. L. Smith. 2007. An evaluation of the BKB-SIN, HINT, QuickSIN, and WIN materials on listeners with normal hearing and listeners with hearing loss. Journal of Speech, Language, and Hearing Research 50(4):844–856.
This page intentionally left blank.