development is very wide, particularly for younger children. Results from Hustad and colleagues (Hustad, Mahr, & Rathouz, 2020) indicate that there is an intelligibility advantage for single word production prior to 41 months of age; after 41 months of age there is an intelligibility advantage for multiword production. Otherwise, the range of variability for typical children is similar for single word utterances and multiword utterances. Notably, typically developing children are not 100% intelligible as indicated by objective measures at 4 years of age. In addition, ongoing preliminary work comparing intelligibility development in children who speak different native languages is revealing important convergences. These results suggest that there may be some intelligibility development universals across languages. Such a finding would have critical implications for early identification of functional speech deficits in children.
A key issue is the ability to differentiate between children whose intelligibility falls within the range of age‐level expectations from those whose performance is delayed or disordered with regard to age‐level milestones. For many children, intelligibility reductions beyond age‐expectations may have a significant detrimental impact on functional communication and on social participation, leading to important negative educational consequences, since speech is a primary modality through which children in the early grades demonstrate their learning. Accurate differential diagnosis of intelligibility deficits, early identification, and treatment to improve intelligibility is critical for these children.
Studies are currently underway that seek to identify cut points for typical intelligibility development and to validate the diagnostic accuracy of intelligibility cut points for separating children who have mild or subtle speech motor disorders from those who are in the lower percentiles of typical development. Recent work employing receiver operating characteristic (ROC) curves (DeLong, DeLong, & Clarke‐Pearson, 1988) suggests that intelligibility scores differentiate between children with cerebral palsy (CP) who have speech motor impairment and typically developing peers with nearly perfect certainty (area under curve = .99). These data further suggest that at 5 years of age nearly all typically developing children had intelligibility scores above 87%, while the vast majority of children with CP and speech motor impairment had intelligibility below 72% at the same age (Hustad, Sakash, Broman, & Rathouz, 2019). These findings are consistent with earlier work suggesting that the range of intelligibility between 75% and 85% may represent a “gray area” for determining whether a 5‐year‐old child was performing at an age‐appropriate level (Hustad, Oakes, & Allison, 2015). The determination of cut points of this nature for children across the age span is currently ongoing, with a focus on children who have CP and are thus at considerable risk for speech motor disorders.
In other work focused on intelligibility development in children, studies have shown that children with CP experience their greatest growth in single word intelligibility between the ages of 36–60 months, but that intelligibility is still developing through 8 years (Hustad, Sakash, Natzke, Broman, & Rathouz, 2019). Results indicate that growth is impacted by speech and language profile characteristics, such that children with CP who do not show evidence of speech motor involvement have their greatest growth at earlier ages and reach higher intelligibility levels at 8 years of age than those who do have speech motor involvement. Further, children with comorbid receptive language impairment and speech motor impairment lag behind their peers who do not have receptive language impairment but do have speech motor involvement (Hustad, Mahr, Broman, & Rathouz, 2020). Clearly, many attributes of the speaker impact intelligibility and its development.
4.5 Perceptual Learning
Historically, there has been a strong emphasis on the study of the speech signal and on disorder characteristics of the speaker to understand and explain intelligibility deficits. This is observed in the clinic, with efforts to manage intelligibility impairments focused largely on the speaker. For example, for clients with dysarthria and an associated intelligibility deficit, the target is often to improve the speaker’s production of speech or give them strategies and/or technology to compensate for the impairment (Duffy, 2012). However, the past decade has seen a surge in studies that examine listeners and their contribution to intelligibility and potential remediation of impairment. Focusing on the role that listeners play in the construct of intelligibility may lead to new interventions that adjunct traditional speaker‐focused treatments or intervention alternatives when the speakers themselves are not viable treatment candidates.
When listeners encounter speech that is difficult to understand, they rapidly adapt their perceptual systems, mapping the non‐canonical acoustic cues onto linguistic categories stored in memory (i.e., cue‐to‐category mapping). This is known as perceptual learning. According to theoretical models, the cue‐to‐category mapping process is driven by signal predictability (Kleinschmidt & Jaeger, 2015). So long as there is some level of regularity in the degraded speech signal, perceptual learning should transpire. Listeners can and do adapt to the speech of speakers with dysarthria, even when significant intelligibility impairment exists. Empirical studies report statistically and clinically significant intelligibility improvements, ranging from 8 to 20 percentage points, for listeners who receive a brief familiarization experience with audio recordings of a speaker with dysarthria (e.g., Borrie & Schäfer, 2015; Borrie et al., 2012; Liss et al., 2002). This rapid perceptual adaptation to a speaker with dysarthria has been primarily studied with young, normal‐hearing adult listener populations. However, preliminary evidence indicates that older adults, including those with impaired hearing, also experience intelligibility improvements following familiarization with a speaker with dysarthria (Lansford, Luhrsen, Ingvalson, & Borrie, 2018). Further, while existing research in this area has heavily focused on how listeners learn and adapt to the speech of a specific speaker with dysarthria, recent evidence shows that familiarization with one talker with dysarthria can generalize to intelligibility improvements in a novel talker with dysarthria, with the level of intelligibility improvements regulated by the similarity of speech deficit (Borrie, Lansford, & Barrett, 2017a). Although additional studies are warranted, the theoretical and clinical implications of perceptual learning generalization across speakers who share similar patterns of impairment are many. While there is strong collective evidence that listeners perceptually adapt to the impaired speech signals of speakers with dysarthria, there may be a small subset of speakers, those with largely unpredictable speech features (i.e., the hyperkinetic speech patterns associated with the involuntary movements of Huntington’s disease) that listeners simply cannot learn (Borrie, Lansford, & Barrett, 2018; Lansford, Borrie, & Barrett, 2019).
4.6 Individual Differences
While still an emerging area of investigation in speech pathology, it would be a disservice to the discussion of intelligibility impairment if we did not touch upon individual differences. Traditional models of intelligibility and learning with normal‐hearing listeners are routinely developed on group averages. Acknowledgment of the variability in individual listener performance has, historically, been minimal. Further, differences have often been treated as a type of error or as justification to remove the data from the analysis. This, and the fact that intelligibility of adults without communication disorders generally hovers around ceiling, has contributed to a false sense of homogeneity among normal‐hearing listeners.
There is mounting evidence that listeners vary widely in their ability to decipher speech across populations and across the age span. Research findings clearly show there is considerable variability in the performance of different listeners who hear the same speaker (Hustad & Cahill, 2003; Hustad, Schueler, Schultz, & DuHadway, 2012; Lam & Tjaden, 2013; Pennington et al., 2013). In fact, studies report standard deviations within a given speaker across listeners of up to 17% for individuals with the most severe dysarthria (Hustad & Cahill, 2003). Other contexts that seem to heighten individual differences in intelligibility performance include the presence of background noise and unfamiliar accent. Findings suggest that some listeners are simply better equipped to tackle the demanding task of understanding
