The Speech Intelligibility Index: Tutorial and Applications for Children Who Are Deaf and Hard of Hearing

Abstract

Purpose:

Speech audibility refers to how much access a listener has to the acoustic cues that comprise spoken communication. The Speech Intelligibility Index (SII) is a measure of speech audibility that has had increasing use in pediatric audiology to quantify auditory access with and without hearing aids. This tutorial focuses on clinical applications of the SII in children who are deaf and hard of hearing (DHH).

Method:

We discuss the calculation of the SII and the current clinical tools that calculate aided and unaided SII. We describe practical uses of the SII throughout the steps of audiologic care, including at diagnosis of hearing loss, hearing technology candidacy, hearing aid verification, and long-term management of children who are DHH. Common questions about clinical applications of the SII are addressed throughout the tutorial. Finally, we present three case studies that illustrate the application of SII to clinical decision making and counseling with children and their families.

Conclusions:

The SII has multiple applications in pediatric audiology. Speech audibility can support device candidacy and management decisions for children who are DHH.

This tutorial describes clinical applications of the Speech Intelligibility Index (SII; American National Standards Institute [ANSI], 1997, R2024) in audiology to support clinical decision making for children who are deaf and hard of hearing (DHH). In recent years, there has been increasing interest in quantifying a child's access to the acoustic cues that are necessary for spoken language development, particularly in the context of optimizing that access for children who are DHH with residual hearing. The term “speech intelligibility” is used widely across audiology and speech-language pathology to describe both speech production and speech perception or recognition. Here, we use “Speech Intelligibility Index” to refer to audibility of the acoustic cues that comprise the speech spectrum, so as not to conflate the term with the related concepts of speech recognition or production. There are many different variations of the SII and multiple ways to calculate the measure. In this tutorial, we use the term “SII” broadly to refer to the multiple variations that have been proposed. In the clinical application discussed in this tutorial, the SII is the proportion of speech that is audible and usable to a person on a scale of 0 to 1, with 0 representing no access to audible speech and 1 representing full access to all speech sounds. Clinically, the SII is expressed as a percentage ranging from 0% to 100%. Audibility can be limited by distance, background noise, or hearing loss. Because the SII can be applied across a wide range of listening situations, this tool has shown promise for predicting speech recognition in children with normal hearing (McCreery & Stelmachowicz, 2011; Scollie, 2008) and children who are DHH (Davidson & Skinner, 2006; McCreery et al., 2015, 2019; Scollie, 2008), and as a predictor of language (Stiles et al., 2012; Tomblin et al., 2015; Walker et al., 2020), academic outcomes (Tomblin et al., 2020; Walker et al., 2023), and cognitive abilities (McCreery & Walker, 2022). In this tutorial, we provide background on the development of the SII, clinical applications of the SII, and using the SII to quantify auditory experience in children who are DHH.

Why Use the SII?

Maximizing the capacity to understand speech is a primary goal of audiologic rehabilitation for many children who are DHH, particularly those whose families choose spoken language as their primary communication mode. The SII quantifies the amount of audible acoustic speech information present in a signal, accounting for the sensation level of the speech signal, noise, and distortion (e.g., spread of masking, negative effects of high presentation levels). Clinically, the SII is often used to represent speech audibility in both unaided and aided conditions. The SII has several advantages over other measures that have been used to quantify the degree of hearing loss and associated impacts on spoken communication, such as hearing thresholds in dB HL, three- or four-frequency pure-tone average (PTA), or categorical descriptions of degree of hearing loss (e.g., mild, moderate, severe, profound). First, the SII directly estimates the effects of hearing levels on the acoustic cues that comprise speech. While the audiogram establishes hearing sensitivity across the frequencies that are most important for spoken communication, the SII weights each frequency band based on its importance for understanding speech. Second, the SII can be used to account for the effects of individual differences in ear-canal acoustics on hearing thresholds and hearing aid output. Consideration of ear-canal acoustics is especially critical for children, who experience higher sound levels during hearing assessment and hearing aid use than adults due to their smaller ear canal size. Children also experience more variable sound levels over time than adults due to ear canal growth as they age (Voss & Herrmann, 2005). Hearing measures based on dB HL from the audiogram are referenced to a 2-cm³ coupler and do not account for individual differences in ear-canal acoustics or changes in ear-canal acoustics over time. Lack of consideration for ear canal acoustics may lead to misestimation of unaided hearing and/or inaccurate hearing aid fittings. Third, the aided SII quantifies speech audibility and avoids limitations of measuring aided sound-field thresholds with pure-tone signals. Aided pure-tone thresholds are processed differently through hearing aids than speech at conversational levels and can over- or underestimate speech audibility (Stelmachowicz et al., 2002). Finally, the SII accounts for different audiometric configurations and how those configurations affect speech audibility using weighting that reflects the importance of speech information by frequency. Two children who have similar PTAs might have very different access to speech if their hearing loss is a flat versus a sloping configuration; the SII quantifies those differences (see Case Studies 1 and 2).

How Was the SII Developed?

The SII is based on research from the first half of the 20th century on sound transmission over telephone wires. At the time, researchers wanted to understand how noise and interference in the telephone system affected speech understanding. In 1947, French and Steinberg introduced the Articulation Index (AI), a method of estimating speech audibility that was a precursor to the current SII (French & Steinberg, 1947). Participants repeated nonsense consonant–vowel–consonant syllables that had been progressively low- or high-pass filtered. By analyzing the accuracy of participants' performance across filtering conditions where different frequency bands were present or absent, the researchers established 20 frequency bands of varying bandwidths that each contributed 5% importance to speech recognition. Application of frequency importance weights were incorporated into the AI to predict recognition accuracy based on the levels of speech and noise energy in each frequency band of novel stimuli. In 1969, the ANSI codified the calculation of the AI in Standard S3.5. The ANSI standard was revised in 1997 to accommodate new methods of estimating speech and noise spectrum levels (ANSI, 1997). At this revision, the name was changed from AI to SII to better capture its relationship to speech understanding.

How Is the SII Measured?

Although developed for estimating communication via wired telephones, the SII has been used in a variety of ways, including to predict the speech recognition performance of normal hearing listeners given specific listening conditions and to estimate average intelligibility of speech in a specific environment. In clinical audiology practice, the SII has gained a foothold as a tool that specifies the weighted audibility of speech across frequency bands and, therefore, can be used to estimate the impact of hearing loss and amplification on speech audibility. Unlike the original AI, which was calculated from frequency bands of unequal width that contributed equally to speech intelligibility, the SII is calculated from frequency bands of equal width that contribute unequally to speech intelligibility. This adjustment in the calculation is a necessary feature for audiology applications where hearing thresholds are measured in octave frequency bands via the audiogram.

The calculation of an individual's SII requires three primary functions: the listener's hearing threshold levels (dB SPL), the stimulus and noise levels (dB SPL), and the weighted importance function based on the type of speech stimuli that will be used, all as a function of frequency.¹ The ANSI SII standard allows for the audibility and importance function to be calculated using frequency bands of varying bandwidths depending on the application (ANSI, 1997). Figure 1 shows the components of the SII calculation for an octave band method as an example. For each frequency band, the audibility function refers to the sensation levels of the speech signal above threshold or the noise level in that band, whichever is higher. The proportion of each band that is audible is calculated by dividing the sensation level in each band by 30 to represent the proportion of the full dynamic range of speech that is audible in each band. Each frequency band also has a band importance function, which represents the proportional contribution of that band to speech recognition based on the type of speech stimulus. These bands are derived experimentally and can be divided into different band importance functions (e.g., 1/3 octave, 1 octave, 20 equally contributing bands). The audibility function of each frequency band is multiplied by its band importance function such that bands with greater audibility and importance contribute to a higher SII. The product of the audibility and band importance functions for each frequency band are summed to calculate the total SII.

Figure 1.

Simplified version of steps involved in calculating Speech Intelligibility Index (SII). The audibility of speech in each frequency band is measured. The result is multiplied by the importance of that frequency band. The products of audibility and importance in each band are added together to yield the SII. SL = sensation level.

As noted above, one of the advantages of the SII for quantifying speech audibility for children is that it can be modified to incorporate individual estimates of ear-canal acoustics using the real-ear-to-coupler difference (RECD). The RECD is the difference in dB between a measurement of a soft broadband noise in a 2-cm³ coupler, which mimics the acoustic characteristics of an adult ear canal, compared to the same measurement in the child's ear canal with their earmold or an insert foam tip. The RECD is applied to thresholds measured with insert earphones and in the conversion of coupler-based hearing aid verification during simulated real ear measures. For thresholds, the RECD is used to convert dB HL thresholds measured with insert earphones to dB SPL thresholds in the ear canal. With these modifications, the unaided SII can be calculated based on dB SPL thresholds that have been converted from dB HL using the RECD. The RECD is not applied to thresholds measured with circumaural headphones or in sound field. With coupler-based hearing aid verification, the RECD is applied to the hearing aid output measured in the coupler to approximate the output of the hearing aid in the child's ear canal. The hearing aid output in the coupler plus the child's RECD are used to estimate the spectrum level of the long-term average speech spectrum for the aided SII calculation. The RECD is not applied to hearing aid outputs measured in the child's ear canal when in situ hearing aid verification is used. The incorporation of ear-canal acoustics via the RECD for unaided SII and aided SII allows for consideration of the effects of individual differences in ear-canal acoustics on thresholds measured with inserts and hearing aid outputs measured in a coupler.

What Are Different Methods for Calculating the SII?

Manually calculating the SII using the methods described in the ANSI standard is complex. Fortunately, there are several tools available to make such calculations accessible to clinicians. We present these tools from least to most useful for pediatric audiology applications. We also note below which tools calculate only the unaided SII and which tools calculate the unaided and aided SII. With the exception of the hearing aid analyzers, these methods use modifications of the ANSI standard.

The Count-the-Dot audiogram (unaided SII). This method was originally developed as a paper-based tool in 1990 and adapted in 2010 (see Figure 2; Killion & Mueller, 2010; Mueller & Killion, 1990). The audiologist plots the patient's hearing levels in dB HL on an audiogram with an overlay of 100 dots representing the unaided speech spectrum, as shown in Figure 2. The number of dots out of 100 below the patient's hearing levels (represented by blue shading) yields an estimated proportion of speech that is audible. However, there are significant limitations to the count-the-dots method for children: (a) counting each dot manually may result in human error, (b) the method does not account for individual ear canal acoustics because it is based on dB HL thresholds, (c) it does not account for masking effects or the level distortion (i.e., negative effect of high presentation levels) described in the ANSI standard, and (d) it cannot produce accurate estimates of aided SII without separately measuring insertion gain, even when pure-tone thresholds are measured with hearing aids, because speech audibility at conversational levels cannot be inferred from detection at soft input levels near threshold.

Figure 2.

Killion and Mueller Speech Intelligibility Index (SII) Count-The-Dots audiogram (described in Killon & Mueller, 2010). Dots represent speech level of 60 dB SPL. Example thresholds (Blue Xs) are plotted on the dots. Blue shaded dots below the thresholds represent audible sounds for this audiogram. Gray shaded dots above the thresholds represent inaudible sounds for this audiogram. In this example, the unaided SII would be 60 (or .60).

Computer-based audiometers (unaided SII). The SII calculation has been integrated into computer-based audiometers. For instance, the GSI AudioStar Pro calculates the unaided SII based on dB HL hearing thresholds using the octave frequency band method described in the ANSI (1997) standard (Prigge, 2025). Although less prone to human error than the manual Count-the-Dots approach, the limitations of these systems are similar because these systems do not account for individual ear canal acoustics and do not produce estimates of aided SII.

Online calculators (unaided SII). The website https://sii.to/ provides software code that calculate the SII; however, these methods do not incorporate estimates of individual ear canal acoustics based on the RECD at the time of this writing. The Knowledge and Implementation in Pediatric Audiology (KIPA) group provides a web-based interface (see Figure 3; http://kipagroup.org/charts) where users enter dB HL hearing thresholds and obtain ear-specific unaided SII for speech stimuli of different intensity levels (Knowledge and Implementation in Pediatric Audiology, 2022). The KIPA calculator has the added advantage of including ear canal acoustics in its calculations by using chronological age-based average RECDs or by entering previously measured RECD values into the program. Online calculators are useful in environments where hearing aid analyzers are often unavailable, such as in auditory brainstem response (ABR) testing suites and school settings. This calculator uses slightly different assumptions than those used in the ANSI standard (e.g., different assumptions for speech peaks and valleys). Thus, values may differ slightly compared to systems that use the ANSI standard to calculate SII (e.g., hearing aid analyzers).

Figure 3.

Unaided Speech Intelligibility Index (SII) for average level speech input (65 dB SPL) for a 6-year-old boy calculated using the Knowledge and Implementation in Pediatric Audiology (KIPA) online calculator. RECD = real-ear-to-coupler differences; LTASS = long-term average speech spectrum.

Hearing aid analyzers (unaided and aided SII). Probe microphone verification via a hearing aid analyzer is the only available method for easily calculating SII for pediatric hearing aid verification. Most hearing aid analyzers calculate both unaided SII, based on the unamplified long-term average speech spectrum, and aided SII based on the amplified long-term average speech spectrum from hearing aid verification measures completed either in the ear (in situ) or in a coupler (simulated real-ear measures). Hearing aid analyzers allow the measurements of SII at a range of input levels of the speech spectrum, most commonly soft (50 or 55 dB SPL), average or conversational (60 or 65 dB SPL), and loud speech (70 or 75 dB SPL; see Figure 4). Measuring speech across different input levels allows clinicians to ensure hearing aids are providing hearing across a dynamic range, such that soft speech is audible, and loud speech is comfortable. Ensuring the audibility of soft speech is important for hearing at a distance, overhearing, and incidental language learning. Hearing aid analyzers are the only available method for calculating SII that can calculate aided SII and that can measure individual RECDs.

Figure 4.

Example of hearing aid analyzer (Verifit2) measurements. Hearing aid output is measured in response to average-level speech stimuli (65 dB SPL). Hearing aids were matched to Desired Sensation Level (DSL) prescriptive targets for a 6-year-old child using age-average real-ear-to-coupler differences (RECDs). SII = Speech Intelligibility Index; RMSE = root-mean-square error.

What Are Clinical Applications of SII Measurements?

Counseling During Diagnosis

The SII can be incorporated into the clinical decision pathway for children who are DHH, from hearing loss diagnosis through provision of amplification and modifying intervention. At the time of the hearing loss diagnosis, parents may experience feelings of confusion, grief, helplessness, and disbelief upon learning that their infant or young child has hearing loss (Luterman, 2004; Scarinci et al., 2018). They also must make critical decisions about their child's intervention as they process these feelings and before they may have developed any true hearing concerns for their child. For children with hearing loss in the mild to moderately severe range, strong and observable behavioral responses to some environmental sounds, such as head turning, may compound these challenges and make accepting the diagnosis and recognizing the benefits of intervention difficult (Fitzpatrick et al., 2016; Holte et al., 2012). The most common descriptive tools leave room for variable interpretations of the seriousness of hearing loss. Categorical hearing loss classifications based on the audiogram such as “slight” and “mild” can minimize the perceived developmental impact of hearing loss (Haggard & Primus, 1999; Manchaiah & Freeman, 2011; Vogel et al., 2007). These categorial classifications can also be challenging for parents to understand and to remember (Watermeyer et al., 2012). Alternatively, clinicians can opt to use audibility-based counseling or counseling using ear-specific SII values. Ear-specific SII provides a numerical estimate that reflects the proportion of speech that is audible to the child that can complement the categorical descriptions of hearing loss based on the audiogram. The unaided SII may provide a more meaningful estimate to communicate the challenges of listening with a hearing loss than pure-tone threshold levels from the audiogram. In cases of mild or moderate levels of hearing loss, unaided SII measurements can be helpful to validate a parent's report that their child is responding to some sounds but also highlight that the amount of audibility may not be sufficient to support development of spoken language.

Descriptions of hearing loss based on the unaided SII are associated with increased levels of parent concern compared to traditional categorical descriptions based on the audiogram using words such as slight and mild (Sapp et al., 2023). Because of the direct link to communication access, the unaided SII has the potential to motivate appropriate next steps after diagnosis. Two children's audiograms may both fall in the mild range but lead to distinctly different developmental outcomes based on their configurations and require different intervention approaches (see Case Studies 1 and 2, for an example). Likewise, children categorized with mild hearing loss have unaided SII values that range from .50 to .95 (McCreery et al., 2015), which represents a wide range of speech audibility for the same audiometric category of hearing loss. If implications of these descriptions are not accurately conveyed to parents, they may delay seeking intervention or experience uncertainty about decision regarding device use. Alternatively, a parent may aggressively seek a hearing technology solution for a child who will not experience benefit from amplification due to a high unaided SII.

For children old enough to complete speech perception testing, the SII can help distinguish between scores that may be appropriate given the hearing loss configuration and scores, which may be worse or better than expected. The SII by itself does not indicate the proportion correct of speech recognition for a given hearing loss, but the SII can be related to speech recognition accuracy using a transfer function. Transfer functions quantify the accuracy of speech recognition across the range of the SII and have been developed for a range of different conditions, including varying linguistic context in the stimulus (nonwords as in Pavlovic & Studebaker, 1984, sentences as in Studebaker et al., 1987), listener age (McCreery & Stelmachowicz, 2011), or the listener's hearing status (Scollie, 2008; Sherbecoe & Studebaker, 2003). Scollie (2008) derived a transfer function for consonant recognition on the UWO-Distinctive Features Test for children with normal hearing at various ages who listened to stimuli that were masked at incrementally louder levels to create a reduced SII (as mentioned above, noise reduces the audibility index in its corresponding band in the calculation of the SII). For example, a normal hearing 5-year-old with an SII of .80 in response to conversational speech input (60 dB SPL) presented in quiet is expected to get ~75% correct, whereas a 15-year-old would get ~95% correct (see Figure 5). This age effect demonstrates that the ability to use limited auditory cues to distinguish speech information matures over time in a predictable way. Scollie also found that children with hearing loss fitted with hearing aids using the Desired Sensation Level (DSL) prescription were able to perform at similar levels to children with normal hearing with similar audibility levels.

Figure 5.

The effects of the age-related proficiency correction factor on the fitted transfer functions for children with normal hearing of various ages. Transfer function describe the predicted level of speech recognition accuracy across the range of audibility for specific listening conditions or listener characteristics. Reprinted from S. D. Scollie (2008), “Children's Speech Recognition Scores: The Speech Intelligibility Index and Proficiency Factors for Age and Hearing Level,” Ear and Hearing, 29(4), 543–556, https://doi.org/10.1097/AUD.0b013e3181734a02, with permission from Wolters Kluwer Health, Inc. SII = Speech Intelligibility Index; y = years.

The final component of audibility-based counseling at the time of diagnosis includes how we record this information as part of the audiology medical record. Pediatric audiology records at the time of hearing loss diagnosis are regularly disseminated to a variety of professionals including otolaryngologists, speech-language pathologists, early interventionists, pediatricians, and other medical specialists who participate in the care of a child who is DHH. When audiologists report unaided SII values alongside the audiogram and traditional descriptors such as degree of hearing loss and configuration, all recipients can benefit from the interpretation of the findings on a meaningful percentage scale and appreciate the expected range for improved access with hearing aids. In follow-up fitting and follow-up hearing aid reports, audiologists can contrast aided and unaided SII values and track aided audibility benefit over time.

Hearing Technology Candidacy

Unaided thresholds or PTA from the audiogram have long been the gold standard for hearing aid candidacy. Audiologists must use their clinical judgment to determine the level of hearing loss that would support the fitting of a hearing aid on a pediatric patient. A lack of consensus of this level has led to inconsistent recommendation of hearing aids across audiologists and clinics, particularly for children with mild hearing loss (Fitzpatrick et al., 2016; Walker et al., 2020). This uncertainty has resulted in an emerging shift in hearing aid candidacy from PTA/threshold-based to audibility-based candidacy using the unaided SII (McCreery et al., 2020; Walker et al., 2020).

Audiologists can use unaided SII to help guide clinical decisions about hearing aid candidacy in children. Research on children with mild hearing loss that helps inform these clinical decisions has come from the longitudinal research project called “Outcomes of Children with Hearing Loss” (OCHL), a study that has followed around 300 children with hearing loss over the past decade. Research from the OCHL study suggests that children who have an unaided SII of less than .80 in response to average-level speech input (60 or 65 dB SPL) in quiet may benefit from the additional audibility provided by a hearing aid. McCreery et al. (2020) and Walker et al. (2020) both examined spoken language outcomes of school-age children with bilateral hearing loss. McCreery et al. found that children who either did not receive hearing aids or never wore their hearing aids with unaided SII values at or below .80 fell below the median scores of age-matched hearing peers on receptive vocabulary, expressive vocabulary, and expressive grammar. Walker et al. found that children with an unaided SII of less than .80 needed to wear hearing aids more frequently than children who had an unaided SII of more than .80 to achieve the same levels of listening comprehension. Both studies provide outcomes-based evidence for using unaided SII to guide hearing aid candidacy decisions in children who are hard of hearing.

The emergence of audibility-based hearing aid candidacy based on data from children with bilateral hearing loss has naturally raised questions about using a similar criterion to make decisions about hearing aid provision for children with unilateral hearing loss. Like children with mild bilateral hearing loss, children with unilateral hearing loss may experience delay in hearing aid fitting or not be fitted at all (Fitzpatrick et al., 2017). However, there is accumulating evidence that altered bilateral input during development compromises language development (Fischer & Lieu, 2014; Fitzpatrick et al., 2019; Gordon et al., 2015; Lieu et al., 2010). In the absence of clinical guidelines and empirical evidence to support the level of audibility that is needed to support the development of binaural hearing, children with unilateral hearing loss who have an unaided SII less than .80 for an average-level speech input level in their poorer ear might also be considered candidates for hearing aids as their peers with bilateral hearing loss, rather than provide amplification only after auditory, language, or academic deficits are apparent.

In addition to considering unaided SII, hearing aid candidacy for children with unilateral hearing loss should also consider the upper limit of aidable hearing, as children with unilateral hearing loss with severe or profound thresholds may have poor hearing aid benefit compared to children with unilateral thresholds in the mild or moderate range (McKay et al., 2008). Audiometric hearing aid candidacy for children with unilateral hearing loss often defines the upper limit of hearing aid candidacy around 70 dB HL (British Columbia Early Hearing Program, 2019). The unaided SII is not ideal for establishing a corresponding upper limit of hearing aid candidacy for children with unilateral hearing loss because many children with pure-tone averages greater than 55–60 dB HL have unaided SII equal to 0. This means that children in the upper range of moderate unilateral hearing loss will have the same unaided SII as a child with severe or profound unilateral hearing loss who would be considered unaidable.

For children with unilateral hearing loss who have an unaided SII equal to 0, it may be beneficial to evaluate their aided SII to determine if the hearing loss is aidable. An examination of data from McCreery et al. (2013) suggests that 95% of children with pure-tone averages less than 70 dB have an aided SII greater than .50 in response to average-level speech input (60 or 65 dB SPL) in quiet if their hearing aids are fitted in proximity to prescriptive targets. Therefore, a using an audibility-based criteria of aided SII ≥ .50 when unaided SII = 0 may approximate current audiometric hearing aid candidacy criteria for children with unilateral hearing loss. This approach should be empirically validated and could provide an approach for estimating the upper limit of hearing aid candidacy for this underserved subpopulation of children with hearing loss.

Hearing Aid Verification

As noted above, audiologists use probe microphone measurements to verify that hearing aid output meets prescriptive targets and to calculate aided SII. These measurements ensure that children have access to speech through their hearing aids for soft, average, and loud speech. Normative aided SII values by pure-tone average are available for soft and average input levels based on research from University of Western Ontario (UWO) for fittings in proximity to DSL targets. These normative values can help audiologists check that hearing aids are providing access to speech at expected levels given a child's unaided hearing/PTA. These normative values are now incorporated into some hearing aid analyzers (see Figure 4), making it simple to use aided SII as a method of hearing aid verification. Other metrics related to audibility, such as deviation from prescriptive target (root- mean-square error [RMSE]), can be used in conjunction with SII to verify that a hearing aid output is optimal for a child given their individual unaided hearing thresholds and individual ear canal acoustics (McCreery et al., 2013, 2017; Wiseman, Walker, et al., 2023). Aided SII and RMSE should be used as tools to confirm optimal audibility via fitting hearing aids to prescriptive targets, not to maximize aided SII by increasing gain above prescriptive targets.

What happens to SII when we use frequency lowering signal processing? Although the audible bandwidth of hearing aids has improved over the last 2 decades, some hearing aids still have insufficient bandwidth to match prescriptive targets at 4000 Hz and above. Frequency lowering signal processing may be needed in those instances to improve audible bandwidth by lowering high frequencies to lower frequencies in the hearing aid output (Scollie et al., 2016). If frequency lowering is activated, the frequency characteristics of the input to the hearing aid and the output will not match, as higher frequencies are relocated to lower frequencies using signal processing. The aided SII assumes that the distribution of speech weighting in the input to the hearing aid and output are aligned in frequency. This lack of alignment when frequency lowering is used can lead to an underestimation of speech audibility because the lowered signal often contains more speech information than the input signal without frequency lowering, but there is not currently a method for remapping the frequency bands of the SII to accommodate frequency lowering. Previous studies have suggested that speech recognition increases as estimated audible bandwidth increases with frequency lowering (McCreery et al., 2013). However, the magnitude of changes in speech recognition with frequency lowering are not proportional to predicted changes based on remapping the frequency bands of the aided SII (McCreery et al., 2014). Some children had increases in speech recognition with frequency lowering that were proportional to the increase in audibility with frequency lowering, whereas others had no change or even decreased speech recognition when lowering increased audibility. Further research is needed to assess the impact of frequency lowering on the SII, but for the time being, audiologists can keep in mind that an aided SII with frequency lowering underestimates audibility in most cases and that may cause the aided SII to fall outside of normative values.

Why should I use SII instead of functional gain to verify my hearing aid fittings? Prior to the introduction of SII into clinical practice, audiologists used functional gain (i.e., aided sound-field testing) to verify device fittings. When hearing aids ran on analog linear circuits, this approach offered valid estimates of gain. Currently, aided SII is the evidence-based method for hearing aid verification due to limitations of functional gain (Academy of Audiology, 2013; Hawkins et al., 1987, Stelmachowicz et al., 2002). Given complex digital signal processing systems in modern hearing aids, there are several problems with functional gain that limit its utility for clinical use (as Pat Stelmachowicz aptly titled her 2002 article, “Is Functional Gain Really Functional?”). First, functional gain relies on a warble-tone stimulus at very low input levels that cannot accurately estimate how a hearing aid responds to a more real-world speech stimulus at conversational levels. This is in part because hearing aids will provide more gain to a warble-tone stimulus at a soft input level than for a speech signal at conversational levels (Henning & Bentler, 2005). For instance, Stelmachowicz et al. (2002) found widely varying aided thresholds in two hearing aids that differed in processing (linear, nonlinear) but had similar audibility for speech. This finding suggests that hearing aid processing can impact functional gain. Second, it is challenging to interpret functional gain results because there are no standards for what aided thresholds are appropriate for a child. Finally, obtaining aided warble-tone thresholds from a child can be time consuming and labor intensive on top of obtaining unaided responses. Thus, calculating aided SII may in fact be less time intensive than measuring functional gain in children.

There are two situations where aided sound-field testing may be necessary in device verification for children who are DHH: cochlear implants (CIs) and bone-conduction hearing aids. While the same limitations in signal processing of warble tone signals apply when testing CIs and bone-conduction devices, there is currently no way for aided SII to be accurately calculated for CIs. Aided SII for hearing aids is calculated by measuring the acoustic output of a hearing aid in dB SPL at individual frequency bands in response to a speech signal. Cochlear implants do not produce acoustic output; therefore, we cannot measure the output of a CI speech processor using a microphone like we would with a hearing aid. Furthermore, SII measures that have been developed with CIs in mind are not predictive of speech recognition in CI users (Lee et al., 2019). Therefore, SII cannot be used to verify CIs at this time. Aided sound-field testing and detection of Ling Sounds remain the best verification options for CI users who cannot yet complete speech recognition testing with their CI. Evidence shows that aided sound-field thresholds are related to speech recognition outcomes in adults and children with CI, with thresholds of 20–35 dB HL resulting in the best performance (de Graaff et al., 2020; Firszt et al., 2004; Geers et al., 2013; Holden et al., 2019). Bone-conduction hearing devices can be verified via hearing aid analyzer in the test box via a skull simulator (Hodgetts & Scollie, 2017); however, this approach has not been widely adopted into clinical practice. Moreover, the calculation of SII for bone-conduction devices has yet to be validated against speech recognition outcomes in children who use these devices, as the relationship between speech recognition outcomes and the SII for bone-conduction devices has yet to be established in children. Skull simulator measures of the aided SII may not be predictive of bone-conduction devices worn on a soft band or head band. Functional gain remains standard clinical procedure in most clinics for verifying bone-conduction hearing devices.

Long-Term Management

Device use counseling. Children who are DHH can only fully realize the benefits of amplification for developing spoken language when they are able to use their devices during spoken language learning opportunities. SII-based counseling can be helpful as a scaffold for conversations with families as they work toward full-time use. By regularly referencing the aided audibility benefit from the hearing aids, professionals may gain buy-in from parents who feel uncertain about whether their child needs hearing aids. Counseling with the aided SII can also help parents understand the benefit a child receives from their hearing aids and reinforce the goals of their family. If a family's goals include listening skills or age-appropriate spoken language benchmarks, full-time use of hearing aids is an important step toward achieving them (Tomblin et al., 2015; Walker et al., 2015). Figure 6 gives examples of SII-based counseling that can be used to reinforce device benefit.

Figure 6.

Example of Speech Intelligibility Index (SII)–based counseling for hearing aid fitting and device use provided to the parent of a young child with a hearing aid. DSL = Desired Sensation Level.

As a child's ear canal acoustics change with age or if their hearing loss progresses over time, revisiting aided SII values provides the opportunity to continually check that auditory access remains satisfactory. With each follow-up hearing exam and repeat hearing aid verification, families have the opportunity to receive updated information about the continued benefit of the daily hours spent wearing hearing aids. When there are lengthy gaps between appointments and necessary updates to hearing aid programming are delayed, the audiologist may review the mismatch between previous hearing aid settings and optimized settings based on the most current pure-tone thresholds. This can be achieved by measuring the output of the hearing aids at previous user settings based on an updated audiogram and updated RECD values. If there are large deviations from prescriptive targets or aided SII values that are significantly below the expected range, this provides evidence that a child has been functioning with suboptimal audibility for spoken language and helps families understand the need for regular follow-up visits.

Changes to hearing aid fitting and technology. Aided SII can also be used when considering further adjustments to a child's hearing aid fitting, hearing technology, or other aspects of intervention. Even if a child falls within the normative range of aided SII values, it is possible they may still lack adequate access to audible speech to support language development. Children need a certain amount of auditory access to develop age-appropriate spoken language skills. For example, Stiles et al. (2012) found a relationship between aided SII and language skills, suggesting that children needed an aided SII of at least .65 (speech input = 65 dB SPL in quiet) to achieve age-appropriate language outcomes. Wiseman, McCreery, and Walker (2023) examined audiometric measures that signal risk for language delay and found that children who had an aided SII less than .68 in response to average-level speech input (60 or 65 dB SPL) in quiet were more likely to be at risk for spoken language delay and children with an aided SII less than .53 had insufficient audibility to support speech and language development. It is likely that these children need a change in hearing technology intervention, such as hearing aid reprogramming, fitting a more powerful hearing aid, or referral for CI candidacy, or additional support to improve access to language.

Should I expect a child's unaided SII to change over time? As children age, their dB HL hearing thresholds may change over time. Long-term changes in dB HL thresholds may be related to changes in hearing acuity, changes in ear-canal acoustics related to growth, or a combination of these two factors. Changes in ear canal size relative to the 2-cm³ coupler that is used to calibrate audiometers can lead to the appearance of worsening thresholds over time, as the sound that was once played in a smaller space (infant ear canal) is now being played in a larger space (child ear canal), reducing the sound pressure in the ear. The unaided SII can decrease over time because dB SPL thresholds also will decrease as the ear canal grows. However, because dB SPL thresholds separately quantify contributions of ear-canal acoustics using the RECD, audiologists can determine if the decrease is related to a change in the RECD or a change in threshold. Changes in the unaided SII related to ear-canal growth tend to be gradual and small in magnitude. Sudden, large changes in thresholds and the unaided SII are not generally related to ear-canal acoustics outside of tympanic membrane perforations, tympanostomy tubes, or surgical modification of the ear canal. If there is a significant change in the unaided SII, it often represents a change in hearing acuity rather than changes in other factors that might influence hearing thresholds measured in dB HL. The unaided SII is very sensitive to changes in hearing threshold within the average speech spectrum (~40–55 dB HL). When hearing loss exceeds 55 dB and the unaided SII is zero, continued progression in hearing loss will not be evident in the unaided SII.

What is the role of SII in language development for children who are DHH? Children who are DHH have restricted access to auditory and linguistic information. This restricted access to speech sounds compromises spoken language development. In the case of children with profound deafness, unaided SII is zero and even aided SII is extremely limited. Most children who are DHH have hearing levels in the mild to severe range, however, and these children vary greatly in their unaided and aided SIIs. An important clinical and research question is the degree to which aided audibility contributes to developmental outcomes and growth in language and literacy skills. This question is critical because it addresses how hearing aids act as a protective factor for reduced linguistic experience (Moeller & Tomblin, 2015) and whether SII is more of a representative estimate of functional hearing abilities than aided audiograms or PTA.

In recent years, a number of studies have demonstrated the significant impact of aided audibility (as measured by the aided SII) on individual differences in speech recognition, language, and literacy skills for children who are DHH (McCreery et al., 2015; Stiles et al., 2012; Tomblin et al., 2015; Walker et al., 2020). The earliest of these studies was by Stiles et al. (2012), which showed that higher levels of aided audibility were associated with better lexical ability in children who are hard of hearing. The researchers also found that aided SII was a stronger predictor of word recognition, nonword repetition, and receptive vocabulary than better ear PTA, which suggests that SII is a more valid estimate of a child's capacity to acquire language using audition than unaided thresholds measures.

Similar findings have been seen in more recent studies. Tomblin et al. (2015) utilized a comprehensive test battery that included measures of articulation, vocabulary, grammar, and pragmatics. These measures were combined to create a composite language score. Results showed that aided SII levels were associated with different language growth trajectories, even after controlling for unaided hearing. Children who had higher aided SIIs showed steeper growth in language development, whereas children with reduced auditory access displayed slower rates of language growth. In another study with the same cohort of children, aided SII had the largest effect on variance in vocabulary depth and breadth, after accounting for factors such as chronological age, maternal education level, and age at hearing aid fitting (Walker et al., 2020). Taken together, these results support the hypothesis that aided audibility has a substantial impact on language ability, above and beyond the contribution of unaided pure-tone thresholds. While other factors beyond audibility also contribute to individual variation in language outcomes in children, these findings highlight the critical importance of audiologists using best practices for hearing aid verification to ensure that children who are DHH have optimal access to the speech spectrum.

Case Studies

The following cases show how limited audibility due to early hearing loss impacts a child's language learning opportunities and the role that intervention plays over time. Each case highlights a specific application of the SII for clinical decision making and parent counseling.

Case Study 1: Alice

Alice was first seen by an audiologist at 4 years of age when her parents noticed that her language skills seemed delayed relative to her sister, who was a year older. Alice had passed her newborn hearing screening with otoacoustic emissions (OAEs) after several repeated screens, although her parents reported that both ears were initially a “refer.” She did not have any history of chronic middle ear issues prior to the audiology appointment.

Clinical data. Results from a pure-tone audiogram indicated a bilateral hearing loss with thresholds within normal limits out to 1000 Hz sloping to mild to moderate levels out to 8000 Hz in both ears, with hearing levels slightly better in the left ear (see Figure 7). Her left ear four-frequency PTA (500, 1000, 2000, 4000 Hz) was 22.5 dB HL, and her right ear PTA was 31.25 dB HL. Without amplification, Alice had an unaided SII of .71 in the left ear and .59 right ear in response to conversational speech input (65 dB SPL) in quiet. Following the diagnostic hearing assessment, the audiologist recommended bilateral hearing aids for Alice. After she was fitted with bilateral hearing aids, her left ear aided SII was .93 and her right ear was .90.

Figure 7.

Case 1 audiometric profile and hearing aid verification.

Within 2 months of receiving hearing aids, a speech-language pathologist administered a battery of speech and language measures. For receptive vocabulary and articulation skills, Alice performed almost 1 SD below average compared to her same-age hearing peers (standard scores of 86 for both measures with an average standard score of 100 and one standard deviation equal to 15). Morphosyntax appeared to be a particular area of weakness, with a scaled score of more than 2 SDs below average for her age (scaled score of 3 with an average scaled score of 10 and one standard deviation equal to 3).

After 2.5 years of hearing aid use, Alice was reevaluated by a speech-language pathologist on the same battery of speech and language measures. Her articulation standard score increased from 86 to 103 (an increase of 1 SD), her receptive vocabulary standard score increased from 86 to 121 (an increase of 2 SDs), and her grammar scaled score increased from 3 to 12 (an increase of 2 SDs). Because standard scores and scaled scores are norm-based measures that take into account a child's chronological age, it is unlikely that a child would demonstrate that level of improvement in speech and language scores based on maturation alone.

Case summary. Alice's case highlights several important features related to the SII. First, the unaided SII in both ears helps to quantify the amount of access to speech sounds a child has in quiet environments at a conversational level. When background noise, soft speech, or distance are introduced into the listening situation, audibility will decline even further, making it difficult to understand speech and develop language at the same rate as peers who have full access to the speech spectrum. Second, the difference between unaided and aided SII values shows the benefit that amplification can provide, even when the degree of hearing loss is in the mild range. In Alice's case, several years of hearing aid use and improved audibility facilitated steep growth in her speech and language skills, so that she was able to effectively close the gap with her hearing peers by first grade and achieve her full potential. This growth in speech and language occurred in the absence of any special education or related services (e.g., early intervention, speech-language therapy, or deaf education); indeed, she was evaluated for an individualized education plan (IEP) just before turning 5 years old and did not qualify. However, it is important to note that many children may require the addition of speech-language therapy services in combination with optimized audibility to achieve a level that is commensurate with their same-age peers.

Case Study 2: Beatrix

Beatrix was identified with hearing loss at birth after referring in both ears on her newborn hearing screen. A diagnostic ABR at 2 months of age revealed a flat mild bilateral hearing loss. Based on the audiologist and physician's recommendations, Beatrix's parents decided not to pursue amplification following diagnosis.

Clinical data. Beatrix was followed regularly by an educational audiologist during early childhood. By age 6 years, her thresholds had remained stable over time. Audiometric results obtained through behavioral testing showed a symmetric mild hearing loss in both ears, with a 4-frequency PTA of 26 dB HL in the left ear and 25 dB HL in the right ear (see Figure 8). Beatrix's unaided SIIs were .94 in both ears at a 65 dB SPL speech input in quiet. Norm-referenced standard scores at 6 years of age indicated performance within the average range for pragmatics, syntax, listening comprehension, with a total composite language standard score of 96. Word reading and reading comprehension also appeared to be on par for Beatrix's age and grade level. Beatrix's mother reported no concerns with her academic performance in kindergarten or first grade.

Figure 8.

Case 2 audiometric profile.

Case summary. Beatrix presents an interesting contrast to Alice. While their PTAs are similar and they both present with mild bilateral hearing loss, Beatrix's unaided SII values of .94 demonstrate that she has good access to conversational speech in both ears in quiet environments, whereas Alice's unaided SII values are significantly lower, even in her better hearing ear. The differences in the SII values between the two children are primarily due to Alice's poorer thresholds from 2000 to 6000 Hz in both ears. As noted earlier in this tutorial, the higher frequency region of the long-term average speech spectrum is weighted more heavily in the SII calculation than low-frequency information because that frequency band contributes more to the intelligibility of speech. Thus, Alice's poorer thresholds in the 2000- to 6000-Hz range result in lower unaided SIIs in both ears compared to Beatrix.

Another important point to keep in mind regarding case management and audiologic habilitation recommendations for Beatrix is that hearing aids would not provide her with significant benefit in terms of speech audibility, compared to her unaided SII. At the same time, it is critical to continue closely monitoring Beatrix's hearing on a regular basis to ensure that thresholds and unaided SII levels remain stable over time and that she is making appropriate progress in her communication and academic performance relative to her peers.

Case Study 3: Charlie

Charlie was identified with bilateral hearing loss at 4 months of age after referring on the newborn hearing screening via automated ABR. He received hearing aids at 5 months and had intermittent audiology care until about 4 years of age. He had one younger brother who also had bilateral hearing loss.

Clinical data. Results from his most recent audiogram at 11 years of age indicated a bilateral sensorineural hearing loss with thresholds ranging from mild to severe in both ears (see Figure 9). His four-frequency PTA was 72.5 dB HL in both ears. Without amplification, Charlie had an unaided SII of .08 for the left ear and .04 for the right ear at a 65 dB SPL input. His fitting showed a significant aided audibility benefit and produced aided SII levels at .55 in the left ear and .56 in the right ear at a 65 dB SPL input. Hearing aids were fitted to target with < 3 dB RMSE in each ear (0.8 and 0.9 dB in the left and right ears, respectively).

Figure 9.

Case 3 audiometric profile and hearing aid verification.

When Charlie was 11 years old, a speech-language pathologist administered a battery of speech, language, and auditory skill measures. Standardized test scores indicated moderate to severe delays across multiple domains of language and listening comprehension. Norm-referenced standard scores were 1–2 SDs below average compared to his same-age hearing peers (standard scores of 69 for auditory comprehension, 71 for expressive communication, and 66 for a total composite language score).

Charlie was in a general education classroom with an IEP in place that included a remote microphone system. He was an excellent hearing aid user and consistently showed datalogging that exceeded 12 hr of use per day. Even with full-time use of well-fitted devices, however, his ability to access speech information was limited, as shown by his restricted aided SII values. Under ideal, quiet listening conditions at a conversational input level, we would expect that Charlie would miss almost half of the speech spectrum. After using his limited aided audibility measurements as a counseling tool, his family received a referral for a cochlear implant evaluation.

Case summary. Charlie's case illustrates the shifting focus between unaided and aided SII-based counseling depending on the level of hearing loss and the variable limitations in auditory access. For children with slight or mild hearing loss like Alice, SII-based counseling focuses on how the underlying hearing loss limits access to important speech and environmental sounds without amplification. In these cases, counseling supports the decision to pursue amplification and describes why this recommendation improves access for the child. For children such as Charlie who already use hearing aids consistently, discussing the unaided SII levels for average conversational speech is less meaningful. Specifically, these values hover near zero: levels that warrant intervention even to those unfamiliar with hearing loss. In Charlie's situation, SII-based counseling for hearing technology focuses on his aided audibility and the degree to which hearing aids can or cannot improve auditory access. As hearing levels become poorer, the ability of a hearing aid to support a family's language development goals becomes more restricted (Ching et al., 2015; Wiseman, McCreery, & Walker, 2023). When a well-fitted hearing aid is not able to restore sufficient auditory access, a referral for a cochlear implant evaluation may be warranted.

Conclusions

Audibility, measured using the SII, is a foundational concept that links the diagnosis of early hearing loss and subsequent management for pediatric patients and their families. This tutorial describes the ways in which the SII can be applied clinically by audiologists who work with children and provides background on the role of audibility as a driving construct for motivating hearing technology decisions and positive outcomes in spoken language domains. The SII has applications as a criterion for hearing technology candidacy and other management decisions (e.g., additional academic accommodations), a metric to support verification of hearing aids, and a counseling tool for discussing the impact of hearing loss on early auditory experience in children.

Data Availability Statement

No data are available because data were not generated or analyzed as part of this tutorial.

Funding Statement

The tutorial was supported by the National Institute on Deafness and Other Communication Disorders of the National Institutes of Health under Award Numbers R01DC019081 (Principal Investigator [PI]: Elizabeth Walker) and R01DC013591 (PI: Ryan W. McCreery). The content of this project is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Deafness and Other Communication Disorders. All authors contributed to the conceptualization, writing, editing, and reviewing of this tutorial. Several people provided support, assistance, and feedback at various points in the project, including Allison Ocel, Melissa Henry, Steven Gianakas, Amberlee Haggerty, and Wendy Fick.