How Do Enriched Speech Acoustics Support Language Acquisition in Children With Hearing Loss? A Narrative Review

Abstract

Language outcomes of children with hearing loss remain heterogeneous despite recent advances in treatment and intervention. Consonants with high frequency, in particular, continue to pose challenges to affected children’s speech perception and production. In this review, the authors evaluate findings of how enriched child-directed speech and song might function as a form of early family-centered intervention to remedy the effects of hearing loss on consonant acquisition already during infancy. First, they review the developmental trajectory of consonant acquisition and how it is impeded by permanent pediatric hearing loss. Second, they assess how phonetic-prosodic and lexico-structural features of caregiver speech and song could facilitate acquisition of consonants in the high-frequency range. Last, recommendations for clinical routines and further research are expressed.

INTRODUCTION: LANGUAGE ACQUISITION WITH HEARING LOSS

Permanent hearing impairments are common sensory deficits and affect around one to six in 1000 infants worldwide (Bussé et al. 2020), with even higher prevalence rates as children reach school age (Wang et al. 2019). Most children with hearing loss (~96%) grow up with hearing parents (Mitchell & Karchmer 2004) and are thus under communicative pressure to acquire oral as opposed to sign language during early childhood. This implies that affected children often acquire their native language through the perception of an impoverished auditory signal and receive linguistic input that is perceived as quieter and distorted and/or degraded compared with what normal-hearing children perceive (Moore 2007).

Early diagnostics and treatment of children with hearing loss have advanced rapidly during the last decades, specifically with the universal newborn hearing screening (Edmond et al. 2022) and technically improved hearing devices (Lieu et al. 2020). Nevertheless, children with prelingual hearing loss continue to fall behind the abilities of their normal-hearing peers in key linguistic domains, such as vocabulary and grammar (Koehlinger et al. 2013; Tomblin et al. 2015). For example, even mild hearing loss during primary school years is associated with deficits in morphological awareness (e.g., recognizing the shared root morpheme in read-er and read-ing) and reduced listening comprehension (Walker et al. 2020). Children with hearing loss also know fewer words than their normal-hearing peers and their vocabularies are less diverse (Walker et al. 2019). These drawbacks in the acquisition of oral language due to prelingual hearing loss can have tremendous negative impacts on children’s academic skills (Marschark et al. 2015), mental health (Fellinger et al. 2012), and quality of life (Ronner et al. 2020).

Consonants as a Particular Challenge in Language Acquisition With Hearing Loss

A corner stone for linguistic development is the acquisition of speech sounds (phonemes), that is, the inventory of consonants and vowels that together form the native phonological system. In language acquisition, the successful encoding of consonants in particular and the phonotactic rules according to which speech sounds are combined to form words in a given language is one pathway into word learning (Gervain & Mehler 2010). Moreover, early consonant perception is a predictor of later language outcomes (Watt et al. 2006; Schaadt et al. 2023). Yet, consonant perception is challenging as it requires encoding of subtle and rapidly changing temporal, spectral, and intensity information (Proctor 2021). Even mild to moderate hearing loss causes aberrant processing of acoustic detail, such as frequency and temporal fine structure, in complex sounds (Halliday et al. 2019). Critically, these processing difficulties are associated with spoken language outcomes in children with hearing loss (Cabrera & Halliday 2020). Children with prelingual hearing loss are often delayed in their consonant acquisition compared with their normal-hearing peers (Wiggin et al. 2013) and impaired processing of auditory information puts children with prelingual hearing loss at risk for language problems (Blamey et al. 2001; Davidson et al. 2011; Halliday et al. 2017).

A particular linguistic challenge reported for children with hearing loss concerns consonants that carry their energy peaks at high frequencies (HF); such as voiceless fricatives (e.g., /s, f, sh/), voiceless plosives (e.g., /k, t, p/), and affricates (e.g., /ts, pf/). Peak frequencies from ~2 kHz to more than 8 kHz are reported for these consonants (Kasturi et al. 2002; Lazarus et al. 2007; Maniwa et al. 2009; Monson et al. 2014; Buss et al. 2022), thus spreading into a frequency region that is associated with reduced hearing acuity even in normal-hearing individuals (Gelfand 2017). Independently of their exact energy peak, HF consonants are characterized by short duration and low intensity, rendering them softer and less resonant than, for example, vowels or nasals which originate from lower frequencies (Proctor 2021).

HF consonants are common in consonant inventories across the world’s languages and are among the phonemes with the highest occurrence rates (i.e., /k, t, and s/, Gordon 2016; Everett 2018). In running speech, HF consonants typically occur at word and syllable boundaries, often hidden in clusters of several consonants (e.g., /stop/, /socks/, /kept/), again posing challenges to auditory perception. Crucially, HF consonants encode seminal linguistic information, they distinguish single words (e.g., /tea/ and /key/) and occur in bound morphemes for agreement inflections in many languages (e.g., /he goe-s/, /the tree-s/). Unsurprisingly, impaired perception of HF consonants due to hearing loss has been associated with negative consequences for children’s word learning (Pittman et al. 2005, 2008) and morpho-syntactic development (Hennies et al. 2012; Koehlinger et al. 2013; Tomblin et al. 2015; Penke et al. 2016; Tönjes et al. 2016; Penke & Rothweiler 2018).

Development of HF Consonant Perception in Normal and Impaired Hearing

Already shortly after birth, normal-hearing infants perceive phonemic differences between speech sounds and discriminate most native as well as nonnative consonants (see Chládková & Paillereau 2020, for an overview). For example, 1- and 4-month-old infants discriminate the stop consonants /b/ and /p/ based on duration differences in the interval between the plosive burst and the beginning of vocal fold vibration (i.e., voice-onset time; Eimas et al. 1971). At around 8 months, infants encode the order of consonants within and across syllables (i.e., phonotactic rules; Sundara et al. 2022), information that can also assist infants in recognizing words in running speech (Graf Estes et al. 2011). For fricatives, as typical HF consonants, however, the evidence of early perception is mixed. Some studies describe infants’ discrimination, for example, between /fa/ and/ θa/ syllables (Levitt et al. 1988), already within the first 2 months of life. Yet, other studies report infants’ difficulties in discriminating fricatives, for example, /sa/ versus /za/, around the same age (Eilers 1977), and even persisting across preschool years (Nittrouer 1996, 2001). The particular perceptual difficulties regarding fricatives during development have been attributed to these phonemes’ relatively low perceptual salience (Nam & Polka 2016) and children’s immature phonetic processing abilities (Nittrouer 2002).

What is already a challenge for normal-hearing children, turns out to be extremely difficult for infants and children with hearing loss. These latter children experience perceptual problems affecting the establishment of the native consonant system, with consequences on speech production and later language development (Blamey et al. 2001; Davidson et al. 2011; Wiggin et al. 2013; Halliday et al. 2017; Cabrera & Halliday 2020). Broadly speaking, the more severe the degree of children’s hearing loss, the more substantial the consequences are on their speech perception and production (Blamey et al. 2001).

As the negative effects of hearing loss on infants’ phonological development can be remedied by early application of hearing devices (Sininger et al. 2010), the clinical assessment of speech perception after fitting hearing aids or cochlear implants (CIs) is currently recommended as part of the pediatric minimum speech test battery (Uhler et al. 2017). For infants with hearing loss, however, there is a lack of standardized tests available to clinically assess prelinguistic speech perception (Morini et al. 2017). The only technique feasible at this young age, namely, visual reinforcement infant speech discrimination, seems to be rarely implemented in clinical practice (Uhler et al. 2017). Relatedly, there is a dearth of clinical studies on consonant perception under aided and unaided conditions, especially regarding those children with mild to moderate degrees of hearing loss. The first studies emerging so far report the discrimination of consonant contrasts to be more difficult than vowel contrasts for both infants with and without hearing loss (Martinez et al. 2008; Uhler et al. 2022).

To improve the perception of consonants and particularly HF consonants, current hearing aids already selectively compress or transpose the HF region (Alexander 2019) and use extended frequency bandwidths (Hunter et al. 2020). The timely prescription of devices, ideally within infants’ first months of life, has benefitted from the introduction of the universal newborn hearing screening (Yoshinaga-Itano et al. 2021). Nevertheless, HF lowering is often accompanied by spectral distortion (Souza et al. 2013) and the perceptual benefit is dependent on individual device settings, the kind of auditory input, and the position of the respective sound in the syllable (Kimlinger et al. 2015; Brennan et al. 2017; Glista & Scollie 2018). Moreover, the application of frequency compression, transposition, and extended high-frequency bandwidth in hearing devices requires elaborate fitting routines. As there is currently no standard way of controlling for the optimal implementation of these techniques in young children, they remain nonstandard in treating children with hearing loss (Akinseye et al. 2018; Lough & Plack 2022).

For children with hearing loss wearing a CI, current devices continue to have relatively poor frequency and intensity resolution, while temporal information is transmitted more reliably (Winn & Nelson 2021). As CIs provide HF information within designated frequency bands, HF consonants in the speech stream are in principle accessible for listeners with CIs. However, consonant place and manner of articulation are transmitted via a combination of frequency and intensity cues (Raphael 2021). Therefore, listeners with CIs often confuse consonants sharing place or manner of articulation, regardless of whether these consonants mainly occur in the lower or higher frequency region (e.g., /t/ perceived as /k/; /m/ as /n/; Rødvik et al. 2018; Winn & Nelson 2021). Given that consonants, especially HF consonants, are essential for children’s lexical and morpho-syntactic development (see studies reviewed earlier), there is an urgent need for alternative ways to improve perception of HF consonants in children with hearing loss.

Development of HF Consonant Production in Normal and Impaired Hearing

Studies on speech production in normal-hearing infants and toddlers reveal that even for this healthy population some HF consonants are difficult to acquire (McLeod & Crowe 2018). At around seven months of age, normal-hearing infants typically start articulating canonical consonant-vowel syllables, for example, /bababa/ (Oller 2000). The most frequently occurring consonants produced at this canonical-babbling stage are stop consonants with labial and alveolar places of articulation, such as /b/, /p/, and /d/ (Irwin 1947; de Boysson-Bardies et al. 1991). Other HF consonants, however, namely fricatives and affricates (e.g., /f/ and /s/) are only fully acquired between 3 and 5 years of age (McLeod & Crowe 2018) and are generally more error-prone in children’s speech (Rvachew 2021). The protracted course of acquisition for fricatives and affricates is most likely attributable to the high articulatory precision required for their production (Kent 1992).

In general, consonants that develop late in normal-hearing children, such as /s/ and /z/, occur even later in speech production of children with hearing loss, and this delay increases with hearing loss severity (Moeller et al. 2007; Wiggin et al. 2013; Ambrose et al. 2014). Specifically, many fricatives and affricates are only produced beyond the age of 7 years, and thus significantly later than in normal-hearing children (Wiggin et al. 2013). Consonant inventories of children with mild to profound degrees of hearing loss grow in similar yet slower trajectories than in normal-hearing peers, between 1 and 4 years of age (Wiggin et al. 2013). Regarding consonant accuracy, 2-year-olds with mild to severe degrees of hearing loss have been shown to produce more consonant errors than an age-matched control group, but their errors overall resembled the production patterns typically observed in younger normal-hearing children (Ambrose et al. 2014). Similarly, 3-year-olds with moderate hearing loss showed overall similar phoneme accuracy levels as a group of younger normal-hearing children, again suggesting a general delay in consonant production (Teveny & Yamaguchi 2023). Thus, within the group of children with hearing loss, those with less severe degrees of hearing loss show more accurate consonant production and less delay than more severely affected children (Ambrose et al. 2014).

Apart from hearing loss severity, other audiological variables have been shown to influence consonant development: 10-month-old infants with moderate hearing loss who consistently use their hearing aids every day had larger consonant inventories than their peers with fewer hours of hearing-aid use. Crucially, hearing-aid use at this young age was also positively associated with consonant proficiency at 36 months (Persson et al. 2022). Moreover, 2-year-olds with mild to moderate hearing loss were shown to achieve age-appropriate consonant production when fitted with hearing aids within their first half year of life (Ambrose et al. 2014). Similarly, even older, 3- to 4-year-old children with severe to profound degrees of hearing loss can reach standard scores for speech and language measures and produce consonant clusters with similar accuracy as their age-matched normal-hearing peers, once they receive early and optimal amplification, consistently use their hearing aids, and regularly participate in auditory-verbal therapy (Fulcher et al. 2012, 2014).

In summary, for children with hearing loss, heterogeneous trajectories of consonant development have been reported, depending on, for example, the severity of the hearing loss, age at intervention, and the duration of hearing-device use (see McDaniel & Gifford 2020 and Zussino et al. 2022, for reviews). The role of other, specifically socio-environmental factors as moderators of consonant development in children with hearing loss, will need to be addressed by future research. For example, past studies attributed maternal education a potential role in consonant development (e.g., Fulcher et al. 2012; but see Moeller et al. 2007 and Ambrose et al. 2014).

On the lexical level, children with hearing loss produce the typical error patterns also produced by children with normal hearing, but their errors are often more persistent for words with HF consonants: children with hearing loss tend to omit HF consonants, especially at word-endings, or replace them with more audible phonemes (Baudonck et al. 2010; Moeller et al. 2010; Penke et al. 2016; Tönjes et al. 2016). When children with hearing loss actually produce HF consonants, they often show aberrant articulation patterns (Grandon & Vilain 2020), which in turn negatively impacts speech intelligibility (Tobey et al. 2003). Moreover, even children with mild hearing loss, who are provided with early amplification and have used their hearing aids for several years, still experience difficulties in applying standard morphological rules (e.g., plural-s) to novel words (Davies et al. 2020). In conclusion, despite years of technical intervention and speech-language therapy in children with prelingual hearing loss, their particular problems with HF consonants are persistent into school age (Koehlinger et al. 2013; Wiggin et al. 2013; Davies et al. 2020).

THE ROLE OF ENRICHED SPEECH FOR LANGUAGE ACQUISITION IN CHILDREN WITH NORMAL AND IMPAIRED HEARING

For children with normal hearing, language acquisition is geared by the interactions with their caregivers as well as the speech input overheard in their home environment and wider social context. For children with impaired hearing, who grow up without contact with native signers, this first fundamental immersion into spoken language is limited due to their hearing loss. Nevertheless, hearing caregivers of children with hearing loss engage their children in meaningful interactions from birth and high quantity and quality of these early interactions have been identified as major contributors to gains in language outcomes in these children (Ambrose et al. 2015; Holzinger et al. 2020; Yoshinaga-Itano et al. 2020). Accordingly, the World Health Organization, in their world report on hearing (WHO 2021), as well as national guidelines (DGPP 2013; Moeller et al. 2013; JCIH 2019) recommend early family-centered interventions focusing on responsive caregiver behavior and the creation of a language-rich environment immediately after the diagnosis of hearing loss. Ideally, caregivers’ speech input for their children with hearing loss hits a sweet spot that is challenging enough to stimulate further language development as well as sufficiently accessible to ensure a contingent response of the child interlocutor (Rowe & Snow 2020). In the following, we will review phonetic-prosodic as well as lexico-structural enrichment of caregiver speech to infants and young children and discuss to what degree these features as well as the quantity of speech input per se vary according to infants’ hearing status and age. With the term enrichment, we generally refer to acoustic and structural modifications in caregiver speech when addressing children versus adults (see e.g., Cristià 2013, for a similar notion). Caregiver speech enrichment (i.e., modifications in comparison to adult-directed speech) may affect all linguistic levels, for example, the phonetic expression of speech sounds as well as more higher-level aspects, such as word order or syntactic features. The available empirical evidence will be reported and discussed in a narrative review. While a systematic review, combined with a meta-analysis, would be of obvious value with respect to early intervention, such analyses are not currently feasible, given the limited number of studies, small to medium participant samples, and the variability of reported structural and acoustic features. Note as well that tactile and visual features of caregiver interactions are also highly relevant for children with hearing loss but are beyond the scope of this article. For children with normal hearing the available literature on caregiver speech focuses on children’s first 2 years of life, hence termed infant-directed speech. For children with hearing loss, many studies on caregivers’ speech cover the first months and years after technical intervention, when children already reached toddlerhood. To accommodate this wider age range, we will use the more general term child-directed speech (CDS), which includes speech to young infants (see e.g., Bernstein Ratner 2013; Schwab & Lew-Williams 2016, for similar approaches).

CDS as Enriched Input for Children With Normal and Impaired Hearing

Phonetic-Prosodic Features of CDS

CDS to infants with normal hearing is a characteristic speech register that is acoustically enriched by higher pitch, larger pitch variability, shorter utterances, more pauses, a slower speech rate, and, at least for some languages, longer and exaggerated vowels compared with adult-directed speech (see reviews by Fernald et al. 1989; Cristià 2013; Cox et al. 2022). These modifications are expressed by female as well as male caregivers, with only limited gender-specific acoustic variation (Ferjan Ramírez 2022). Phonetic-prosodic enrichment mainly affects suprasegmental speech information, thereby improving the signal to noise ratio (Smith et al. 2024) and possibly contributing to better transmission and intelligibility of the speech signal (Smiljanic 2022). Phonetic-prosodic enrichments of caregiver speech occur most strongly in interaction with infants until they are around 2 years old. Note though that there are dynamic changes in enrichment strength for some acoustic features (Cox, et al. 2022): while overall pitch, speech rate, and vowel duration become more like adult-directed speech over time, higher pitch variability and vowel space exaggeration remain stable throughout infants’ first years.

The phonetic-prosodic features of CDS to infants and children with hearing loss have been extensively reviewed by Lovcevic et al. (2022). The elevated pitch and exaggerated pitch excursions as well as the slower speech rate reported for CDS toward normal-hearing infants are also present in interactions with children with hearing loss. Yet, the acoustical properties of CDS vary as a function of, for example, the degree of hearing loss and the type of hearing device. For example, for children with hearing aids, pitch in CDS has been reported to vary to similar degrees as for age-matched normal-hearing children (Lovcevic et al. 2020). For children wearing CIs, however, pitch characteristics as well as speaking rate of CDS are rather scaled-down compared with speech directed to hearing-age-matched (younger) normal-hearing peers (Kondaurova et al. 2013; Vanormelingen et al. 2016).

The way in which CDS enrichment affects the acoustic characteristics of vowels is currently debated. There are accounts proposing an extended vowel space for CDS (i.e., vowel hyper-articulation), resulting in enhanced phonetic contrasts, potentially facilitating infants’ vowel acquisition (Kuhl et al. 1997). Alternative accounts, however, report reduced phonetic clarity (i.e., vowel hypo-articulation, e.g., Martin et al. 2015; Smith et al. 2024). Concerning the expression of vowels in CDS toward children with hearing loss, Wieland et al. (2015) observed clearer maternal vowel articulation in infant-directed versus adult-directed speech toward both normal-hearing children as well as children wearing CIs and hearing aids. It is important to note that for some acoustic measures (vowel dispersion) these differences in speech register were larger for children wearing hearing aids than in age- and hearing-age-matched control children (i.e., normal hearing). Other studies, however, find no differences in vowel articulation in CDS for infants with versus without hearing loss (Dilley et al. 2020; Lovcevic et al. 2020), or even report reduced vowel hyper-articulation toward a child with hearing loss compared with his normal-hearing twin brother (Lam & Kitamura 2010) and in the case of simulated hearing loss (Lam & Kitamura 2012).

Lexico-Structural Features of CDS

Lexico-structural features of CDS refer to, for example, the actual topics of conversation (abstract versus concrete), richness of vocabulary, and utterance complexity. Typically, CDS toward infants contains a limited, repetitive vocabulary within short and syntactically simple utterances (Soderstrom 2007; Rowe & Snow 2020). As children grow older, lexical as well as syntactic complexity increases (Rowe 2012), albeit presumably not in a linear fashion (Genovese et al. 2019).

Current recommendations for caregivers of children with hearing loss encompass the use of facilitative language techniques, such as open-ended questions and imitations. These techniques form no fixed inventory, yet there seems to be consensus on the following approach (Ambrose et al. 2015; Dirks et al. 2020; Holzinger et al. 2020): First, during infancy and toddlerhood, low-level techniques seem most appropriate to stimulate the use of oral language in children with hearing loss. These include linguistic mappings (i.e., reformulating what a child is trying to express), directives (e.g., “Say elephant!”), close-end questions (e.g., “Do you want a banana?”) and imitations. Second, once children are older and have reached a more mature developmental stage, high-level techniques can be used, such as open-ended questions, expatiation (i.e., expanding on the child’s utterance), and recasts (i.e., creating a question from the child’s utterance) (Rowe & Snow 2020). The use of high-level techniques, in particular, has been associated with better language outcomes in children with various degrees of hearing loss (Cruz et al. 2013; Ambrose et al. 2015; Holzinger et al. 2020).

Quantity of CDS Exposure

High quantity of speech in direct child-caregiver interactions, as opposed to overheard speech, is associated with higher gains in language development in children with normal hearing (Hoff & Naigles 2002; Rowe 2012; Weisleder & Fernald 2013). For children with hearing loss, these associations are attested as well (Nittrouer et al. 2020; Yoshinaga-Itano et al. 2020). For example, caregiver language-input to children with hearing loss during preschool years predicts these children’s language outcomes even at around 10 years of age (Nittrouer et al. 2020). These findings are the more important, given that within the first year(s) after technical intervention, children with hearing loss seem to be exposed to a highly variable amount of caregiver speech. Some studies report similar number of words, utterances, and/or conversational turns from caregivers of children with and without hearing loss (Vandam et al. 2012; Vanormelingen et al. 2016; Smith & McMurray 2018; Dirks et al. 2020; Nittrouer et al. 2020). Yet, other studies attest that especially within the population of children wearing CIs, the amount of CDS uttered in close proximity to the child can vary drastically (Su & Roberts 2019; Arjmandi et al. 2021), and the number of conversational turns and received number of words can be particularly low for these children (Ambrose et al. 2015; Kondaurova et al. 2022). The particular findings for children with CIs might be due to reduced intelligibility of child utterances (Su & Roberts 2019) and/or a decrease in verbal caregiver responsiveness (Kondaurova et al. 2022).

Songs as a Special Type of Enriched Speech Input

Similar to CDS, singing is a universal part of caregiving routines across the world (Mehr et al. 2019; Hilton et al. 2022) and provides a powerful means to modulate children’s arousal state (Cirelli et al. 2019). In addition to their socio-emotional appeal, songs should be considered a source of native language-input for young children. Caregiver singing is modulated in a slow, rhythmic, and repetitive manner (Trehub & Trainor 1998; Mehr et al. 2019), thus conveying native language information within an attractive auditory scaffold. Infants with normal hearing are strongly attracted to caregiver singing (Lense et al. 2022) and encode single words (Snijders et al. 2020), rhymes and phrases (Hahn et al. 2020, 2021), syllable order (Thiessen & Saffran 2009), and vowels in songs (Falk et al. 2021) - all being linguistic units that build the stepping stones for further language acquisition. Moreover, several studies have indicated a benefit of at-home musical enrichment, such as singing and playing musical instruments, for the development of receptive vocabulary (Franco et al. 2021; Papadimitriou et al. 2021) and gesture use (Gerry et al. 2012) in children with normal hearing. To what extend infants and children with hearing loss are also able to process linguistic information in songs has not been systematically investigated so far. Yet, songs are a standard component of speech-language therapy and recent evidence suggests a multitude of beneficial effects of musical activities for auditory rehabilitation (Torppa & Huotilainen 2019; Pesnot-Lerousseau et al. 2020). In families of children with hearing loss, musical activities and caregiver singing seem to be equally common as in other families (Looi et al. 2019), indicating that hearing loss is no contraindication for singing and other forms of musical engagement in families of children with hearing loss. Caregiver singing might thus complement early family-based intervention for children with hearing loss.

Socio-Emotional and Attentional Benefits of CDS Enrichment

Enriched CDS also modulates children’s attention and socio-emotional processing. Specifically, enriched CDS expresses positive caregiver affect (Trainor & Desjardins 2000), thereby strengthening the emotional bond between child and caregiver and contributing positively to children’s emotional, physical, and intellectual development. Furthermore, CDS strongly attracts and maintains infants’ attention and infants prefer to listen to CDS over adult-directed speech (Frank et al. 2020). Enriched CDS thus draws children’s attention to the speaker and, presumably, to what is being said. For children with hearing loss, the enrichment present in CDS seems to be accessible despite the imposed sensory limitations. Specifically, both infants with hearing aids as well as CIs prefer listening to CDS carrying the earlier mentioned enrichment, as opposed to adult-directed speech that is acoustically less prominent (Wang et al. 2017a, 2018). This preference for CDS declines as children grow older. For example, at 17 months, children with hearing aids no longer show a preference for CDS anymore, which bears similarity to normal-hearing children, who lose their preference for CDS over the course of their first 2 years of life (Wang et al. 2018). Interestingly, the strength of the CDS preference among 2-year-olds with CIs predicts language outcomes up to eleven years after CI activation (Wang et al. 2017b), suggesting that a CDS preference might come with advantages for children’s language development.

WHAT IS THE FUNCTION OF CDS ENRICHMENT?

The evidence discussed so far gives rise to three main functions of enriched CDS: communicating positive emotions, regulating child attention, and facilitating language acquisition (for reviews, see e.g., Soderstrom 2007; Golinkoff et al. 2015). Indeed, it has been argued that the phonetic-prosodic and lexico-structural modifications in caregiver speech result in a speech signal that directly benefits linguistic processing, for example, by enhancing acoustic contrast between phonetic categories (Kuhl et al. 1997). The prosodic bootstrapping account (Gleitmann & Wanner 1982) describes the alignment of speech acoustics and linguistic units that enables infants to infer linguistic units such as words and phrases from the speech stream (Cutler 1994; Morgan & Demuth 1996; Soderstrom et al. 2003). A recent intervention study and a meta-analysis indicate that specifically, the prosodic features of CDS contribute positively to attentional as well as prelinguistic and linguistic abilities such as vocal responsiveness and vocabulary size (Spinelli et al. 2017; Ferjan Ramírez et al. 2019). Some forms of enrichment may thus facilitate language learning by guiding attention to relevant linguistic entities in the speech signal. However, the current state of evidence questions this appealing account and calls for a more nuanced discussion of enrichment benefits. For example, hyperarticulated vowels and consonants are no universal feature of CDS, as will be further reviewed later. Enrichment effects on speech clarity are highly variable and might rather be secondary to attentional and socio-emotional communicative aspects (Benders 2013; Cristìa & Seidl 2014; McMurray et al. 2013; Kalashnikova et al. 2017). For caregivers of children with hearing loss, another nuance needs to be added to the discussion, because for them a facilitation of linguistic development of their child is probably of greater importance than for caregivers of unaffected children. Alongside the psychosocial aspects involved in enriching their CDS, these caregivers might in-fact, whether consciously or unconsciously, strive for clearer speech input for their child.

Interim Summary

Taken together, caregivers of both children with and without hearing loss enrich their speech, thereby affecting linguistic as well as nonlinguistic processing of their child. During infancy, CDS is characterized by simplified structure and repetitive vocabulary as well as exaggerated prosody. As children grow older, lexico-structural features increase in complexity, while prosodic features become less prominent. The specific domain and degree of enrichment can be actively modulated in quantity and quality by caregivers, for example, in accordance with their child’s age and hearing status. Thereby, caregivers can positively contribute to the language development of their child.

Current recommendations for caregivers of children with hearing loss entail creating a language-rich environment in a stimulating, responsive mutual exchange (Moeller et al. 2013; JCIH 2019; WHO 2021). These recommendations encompass singing as well as book reading and other communicative activities at home. So far, these recommendations are empirically validated mostly for children during toddlerhood and preschool years. Given that children with hearing loss are diagnosed at an increasingly earlier age, with the potential of timely intervention, the role of caregiver speech expressed within the first months and years of children’s lives, becomes increasingly important. In particular, the role of CDS for the acquisition of speech sounds as the most essential components of native language acquisition, requires further research. For vowels, intriguing evidence of associations between maternal articulation and language outcomes of children with hearing loss is emerging (Dilley et al. 2020; Lovcevic et al. 2020). These associations might as well apply for the acquisition of HF consonants, a linguistic area of explicit difficulty for children with hearing loss.

ENRICHED SPEECH TO SUPPORT CONSONANT ACQUISITION IN CHILDREN WITH IMPAIRED HEARING?

HF Consonants in CDS

The general acoustic setting of enriched CDS might be advantageous for HF consonant transmission, as CDS is typically expressed in close proximity to the child with direct gaze (Çetinçelik et al. 2021) and often at raised intensity levels overall (Hilton et al. 2022). This communication setting most likely reduces loss of HF energy between caregiver and infant (Monson et al. 2012a). Moreover, CDS is universally associated with high pitch (Cox et al. 2022), and there is a predominance of female caregivers, with speech from female voices carrying more energy in the higher versus lower frequency spectrum (Monson et al. 2012b).

Given the importance of HF consonant perception for later language development, surprisingly little is known about their occurrence frequencies and acoustic realization in CDS. Many of the most frequent words from English caregivers (e.g., /this, that, they/, Lee et al. 2010) and German caregivers (e.g., /das, ist, was/, Stärk et al. 2021) contain HF consonants. Within these frequent words, HF consonants occur in mono-syllabic words with simple syllable structures, potentially rendering single consonants more audible.

The few studies investigating HF consonant articulation in CDS provide evidence for both improved learnability due to hyper-articulation as well as increased acoustic variance among HF consonants (i.e., more versus less clear articulation). Crucially, these studies only consider speech toward typically developing children (see e.g., Cristià 2010; Dilley et al. 2019; for reviews). Hyper-articulation of CDS consonants has been reported for some fricatives and plosives. For example, the acoustic contrast in spectral energy of /s/ versus /∫/ (e.g., /seat/ versus /sheet/) is enhanced in speech to 1-year-old, but not to 6-month-old English-learning infants (Cristià 2010), and individual mothers’ expression of these sibilants predicts how well their child is able to discriminate /s/ from /∫/ (Cristià 2011). The contrast in spectral energy between /f/, /s/, and /x/ is enhanced toward 11-month-old, but not 15-month-old Dutch-learning infants (Benders 2013). Voice-onset time contrasts for stop consonants toward 3- to 6-year-olds are slightly enhanced when mothers produce these sounds under high versus low background noise (Smith et al. 2024). Last, in phonological contexts where /t/, can be easily confused with other speech sounds, it is articulated more canonically (Dilley et al. 2019; Khlystova et al. 2023). However, hypo-articulated consonant contrasts in CDS are attested as well, for German (Baer-Henney & von Kries 2021), Japanese (Martin et al. 2015), and Nepali (Benders et al. 2019) speech samples. Moreover, in a large English corpus, /t/ and /s/ are not produced more canonically in CDS compared with adult-directed speech (Khlystova et al. 2023). The variation in caregiver HF consonant acoustics within and across languages and especially for speech directed at populations of different ages and needs clearly requires further research.

HF Consonants in Caregiver Singing

Singing is in general expressed with larger jaw movements, increasing the amplitude of the signal (Livingstone et al. 2015). Female voices, in particular, increase in intensity at frequencies above around 5 kHz (Monson et al. 2012b). These level changes at HFs are detectable for adult listeners and contribute to their correct HF consonant discrimination (Monson et al. 2012a, b; Vitela et al. 2015). The acoustic features of caregiver singing, with natural HF amplification, might in fact be beneficial for speech perception of children with hearing loss. For example, high-pitched singing can improve infants’ vowel discrimination (Falk et al. 2021). However, Trainor and Desjardins (2002) report reduced vowel discrimination in infants for high versus low-pitched artificial vowels. Other seminal studies suggest improved word learning in children with hearing loss after improving audibility of the HF region compared with low-pass filtered speech (Pittman et al. 2005; Pittman 2008). HF consonants with labial and dental place of articulation (e.g., /f, θ/) might be particularly easy to derive from songs, given that singing attracts more infant gazes to the singer’s mouth (Alviar et al. 2022). More generally speaking, caregiver singing during infancy is expressed within designated rituals in quiet surroundings (Falk & Audibert 2021) and involves the same limited set of songs that is repeated across occasions (Kragness et al. 2021). Caregiver singing is also a highly captivating auditory stimulus and efficient in attracting and maintaining infants’ attention (Franco et al. 2021; Alviar et al. 2022).

Like the acoustic enrichment provided in CDS, we hypothesize for songs that children with hearing loss benefit from the enhanced acoustics and the ritualized communicative context in their perception of HF consonants. We also propose a direct relation between children’s frequent exposure to songs in their everyday environment, from their own as well as from caregiver singing, to this speech-perception benefit. Several studies already indicate that children with hearing loss benefit from musical activities, including singing, for different linguistic abilities, for example, speech-in-noise perception (Torppa & Huotilainen 2019; Pesnot-Lerousseau et al. 2020). This body of research mainly focuses on school-age children with hearing loss, who are taking part in formal music classes or who play a musical instrument. For example, one study reports a positive effect of music classes on phoneme discrimination (including HF consonants) in school children with profound hearing loss (Rochette et al. 2014). Other studies observe better speech-in-noise perception, stress perception, verbal intelligence, and word detection in 5- to 13-year-olds wearing CIs who sing about five times per week and whose parents sing on a regular basis (Torppa et al. 2014, 2018, 2019). This promising body of research needs to be complemented by future studies with younger children with more variegated degrees of hearing loss to systematically investigate the role of caregiver singing for consonant acquisition in children with hearing loss.

Interim Summary

The pronounced acoustic features of CDS and songs, in particular the raised intensity overall and at the HF region, together with the close spatial proximity between child and caregiver, might enable children with hearing loss to gain improved access to the otherwise difficult-to-perceive HF consonants. Enriched speech input in the form of CDS and songs is thus proposed to facilitate HF consonant acquisition for infants and children with hearing loss. While there is first encouraging evidence, this proposal clearly requires corroboration by further research, suggestions of which are outlined in more detail later.

RECOMMENDATIONS FOR CLINICAL PRACTICE AND FURTHER RESEARCH

Current recommendations for family-centered at-home interventions for children with hearing loss encourage caregivers to create a language-rich environment for their children and adopting a responsive style of communication (Moeller et al. 2013; JCIH 2019; WHO 2021). The communicative strategies so far mainly target speech quantity and lexico-structural enrichment of caregiver speech, such as the use of high-level facilitative language techniques during interactions from around 2 years onward (Nittrouer et al. 2020; Curtin et al. 2021). With decreasing age of diagnosis, however, at-home intervention strategies need to be adapted for younger children. According to current recommendations (JCIH 2019; WHO 2021; Yoshinaga-Itano 2023), pediatric hearing loss should ideally be diagnosed at around the age of 1 month and treated by 6 months of age. At this young age, the phonetic-prosodic, rather than the lexico-structural features of caregiver speech are of greater perceptual relevance, as infants use the former to infer linguistic structure from the speech stream (Gleitmann & Wanner 1982). We here recommend more strongly incorporating CDS as well as songs into the communicative repertoire of caregivers of infants and children with hearing loss as a form of at-home intervention immediately after early hearing loss diagnosis. During infants’ first months of life, enriched speech in the form of CDS and songs might stimulate various domains of linguistic development, such as HF consonant development. For example, by repeatedly singing the same captivating child-song to their preverbal infant, caregivers implicitly familiarize their child with the sound structure of their native language. Even before hearing devices are fit, caregivers can thus positively contribute to language development of their children with hearing loss.

Our suggestion complements best practice recommendations of family-centered early intervention for children with hearing loss (Moeller et al. 2013; JCIH 2019; WHO 2021; Yoshinaga-Itano 2023). Yet, caregivers should be provided with evidence-based strategies for early communication, rendering further research urgently needed. Specifically, caregiver enrichment of their speech needs to meet the cognitive, socio-emotional, as well as communicative needs of their children (Rowe & Snow 2020). For children with normal hearing, there is a natural decline in CDS enrichment as children grow older and more mature in their language development (Cox et al. 2022), yet changes in CDS complexity do not seem to follow a linear trajectory: for example, some lexico-structural features of Italian CDS become more simplistic when infants are around nine months old and increase in complexity toward the end of the first year of life (Genovese et al. 2019). For children with hearing loss, future research will need to evaluate which form and degree of speech enrichment is beneficial during specific communicative situations at different developmental stages, while also taking into account children’s individual hearing loss configuration. Promising studies indicate already now that caregiver-coaching programs can be successfully implemented to increase the amount and quality of CDS toward typically developing children (Jeong et al. 2021) and children with preterm birth, autism, or intellectual disabilities (Bang et al. 2020), with a current meta-analysis suggesting that these programs have small to medium-sized effects on language development in typically developing children (Jeong et al. 2021). Crucially, caregiver-coaching can alter speech input already during early infancy (Ferjan Ramírez et al. 2020), the age at which pediatric hearing loss should ideally be diagnosed and treated within family-centered intervention. Clinical routines for infants and children with hearing loss should thus be expanded to the time immediately after hearing loss diagnosis. These procedures should specifically target the use of enriched CDS, caregiver singing, and other forms of informal musical activities.

We also recommend a detailed examination of the expression of HF consonants in caregiver speech to infants and children with hearing loss in different age groups. Specifically, future research will need to clarify, whether CDS enrichment improves transmission of the speech signal in general, and/or whether there are phonetic benefits for HF consonants, for example, due to an increase in acoustic distance between different consonants or a reduction of acoustic variability within the same consonants. HF consonants vary in a multitude of acoustic features which in turn are shaped by external factors, such as sound position, grammatical function, lexical stress, speaking rate, speaker gender, and underlying emotion (Hsieh et al. 1999). Identifying the acoustic features that render an acoustic-phonetic enhancement and whether this enhancement leads to a perceptual benefit, in combination with the technical specifications of hearing devices and children’s developing auditory processing abilities, is a complex research effort. For example, fricatives carry most of their acoustic information within the HF region, as illustrated earlier. If CDS enrichment indeed improves audibility of the HF region, this might be especially beneficial for fricative perception of children wearing hearing aids, given that their device typically transmits HF information with less precision. For children wearing CIs, in turn, the discrimination of voiceless HF plosives, such as /p/ and /t/, from voiced counterparts (/b/ and /d/) might be improved by CDS enrichment. These speech sounds differ in their voice-onset-time, an acoustic feature that is transmitted via temporal variation in the low-frequency region (Proctor 2021). CIs typically transmit low-frequency information only poorly, while temporal information is preserved (Winn & Nelson 2021). The slow articulation rate of CDS (McMurray et al. 2013; Cox et al. 2022) might thus be beneficial to remedy these device-specific shortcomings.

Future research should also target potential associations between children’s HF consonant perception, production, overall language outcomes, and caregivers’ speech. Ideally, these future studies feature a mix of large-scale instruments, such as parental questionnaires, and more fine-grained analyses of experimental studies and recordings of natural caregiver-child interactions. These investigations should also address in more depth the role of child-internal and socio-environmental factors (for a discussion, see Houston 2022). Specifically, individual susceptibility to an enriched speech benefit might be moderated by children’s cognitive abilities, related to attention or nonverbal intelligence (Halliday et al. 2008) as well as the families’ socioeconomic status or at-home musical activities (Torppa & Huotilainen 2019). For songs, in particular, future studies will need to elucidate whether they can directly facilitate HF consonant perception, as suggested earlier, or rather indirectly, for example via overall improved auditory processing abilities (see e.g., Franco et al. 2021, for the mediating role of attention to songs). Last, mishearing song lyrics is a common phenomenon in adults (Collister & Huron 2008) and children (Planchou et al. 2015). Yet, so far it is unclear whether reduced intelligibility of song lyrics also affects HF consonants in caregiver singing. Our hypothesis of a perceptual benefit for HF consonants from enriched CDS also entails a potential high-voice-benefit for female caregivers. While we acknowledge this fact given psychophysical parameters, fathers’ contribution to HF consonant acquisition deserves its own scientific inquiry. Fathers as well as mothers naturally engage in enriched CDS and are essential in their roles as active caregivers (Fernald et al. 1989; Ferjan Ramírez 2022). Last, the role of caregiver communication before technical intervention for children with hearing loss should be studied in more detail, especially concerning the acquisition of HF consonants. Right after hearing loss diagnosis and while waiting for hearing-device fitting, affected families might benefit from the intuitive caregiving routines that enriched CDS and songs provide. This is even more important given that the period after hearing loss diagnosis is commonly accompanied by high uncertainty and emotional distress in affected families (Holzinger et al. 2022).

CONCLUSION

CDS and songs provide a natural enrichment of the spectral and amplitude features essential for the perception of a particular set of consonants. Infants and children with hearing loss could make use of this enrichment in their consonant acquisition once their caregivers provide them with sufficient quantity and quality of enriched speech input. We recommend incorporating this perspective more strongly into future research and early intervention programs for infants and children with hearing loss.

ACKNOWLEDGMENTS

Preparation of this manuscript was supported by the Marga und Walter Boll-Stiftung (grant awarded to C.M. and D.M.). The authors thank Samira Zabaneh, Linda Müller, Dörte Pollex, and Franziska Schmidt at the Department of Audiology and Phoniatrics, Charité -Universitätsmedizin Berlin for their professional insights into their clinical work with children with hearing loss and their families. A preprint of this article has been published at https://osf.io/preprints/osf/qwtkp.