Babble Boot Camp for Infants With Down Syndrome: Piloting a Proactive, Caregiver-Led Intervention Designed to Boost Earliest Speech and Language Skills

Abstract

Purpose:

Down syndrome (DS) is associated with lifelong difficulties with verbal communication, beginning in infancy when vocalizations are sparse and first words emerge late. Because DS is diagnosed at or even before birth, these difficulties can be anticipated, yet there have been limited developments of systematic, proactive interventions. The purpose of the pilot study described here was to investigate feasibility and potential benefits of such an intervention toward a fully powered clinical trial.

Method:

We piloted Babble Boot Camp (BBC), a proactive, parent-led speech and language intervention, with 10 children with DS ages 4–16 months. Each family participated in weekly sessions via telehealth for 10 months. A pediatric speech-language pathologist coached parents to implement daily routines and activities at home, designed to foster child target behaviors and skills (e.g., increased vocalization rates, babble complexity, word productions) toward building resilience against anticipated challenges. Parents provided daylong audio recordings and questionnaire data at regular intervals.

Results:

Parent participation and compliance metrics in the intervention were high. All rated the intervention as acceptable, convenient, and helpful, whereas three sets of parents found aspects of the data collection time consuming. Children's linguistic environments resembled those of typical controls in terms of child utterance rates, adult word counts, and conversational turns. Babble complexity and receptive and expressive vocabularies increased over time. First words emerged earlier than expected.

Conclusion:

High feasibility metrics and suggestive benefits motivate a larger study to determine more specifically how the various BBC components can improve long-term outcomes for children with DS.

Supplemental Material:

https://doi.org/10.23641/asha.28655222

Down syndrome (DS) is the leading genetic cause of intellectual disability in the United States, with approximately 6,000 children born with DS each year (Mai et al., 2019). DS is caused by an extra copy (full or partial) of chromosome 21. Although there is wide variability across individuals, this chromosomal difference generally results in a characteristic profile of strengths and challenges that is present from birth (D'Souza & D'Souza, 2022; Grieco et al., 2015). Consequently, designing the most promising interventions for children with DS requires a comprehensive understanding of their strengths and the nature and causes of their challenges.

One of the most commonly cited challenges for children with DS is difficulty with spoken language, even beyond what would be expected for IQ and mental age (Abbeduto et al., 2007; Chapman & Hesketh, 2000). Children with DS generally progress through many of the same stages of spoken language development as neurotypical children, but at a slower pace (O'Toole & Chiat, 2006). Here, we summarize some of the main speech and language traits, setting the stage for the design of the proactive intervention that we created and piloted.

Relatively few studies have focused on prespeech sound production, and the findings regarding babbling development are mixed (Kent & Vorperian, 2013). Some lines of evidence suggests that young children with DS have delays in the onset of canonical babbling and babble less than both neurotypical peers (Cobo-Lewis et al., 1996; Lynch et al., 1995) and peers of similar intellectual and communication abilities at the age of 2 years (Sokol & Fey, 2013); other research has not reported significant differences in comparison to typical development (Dodd, 1972; Smith & Oller, 1981; Steffens et al., 1992). However, babbling is important for children with DS, as it predicts later speech and language development (Locatelli et al., 2021).

Most individuals with DS have difficulty developing speech that is clearly understood by others, due in part to slow articulatory motor planning, phonological errors, and anatomical differences such as smaller oral cavities with relatively larger tongues (Guimaraes et al., 2008) and high arched palates (Barnes et al., 2006; Buckley & Bird, 2001; Burgoyne et al., 2021; Chapman et al., 1998; Kumin, 1994). While most speech sound errors appear developmental in nature, some atypical acoustic speech traits have been described, including reduced melodic/prosodic range, reduced rhythmic complexity, smaller F1–F2 area, and greater temporal variability (Cleland et al., 2010; Kent & Vorperian, 2013).

Children with DS also experience challenges with their language development from an early age. For instance, whereas the average age of first words for neurotypical children is around 12 months, for children with DS, this milestone is highly variable and can occur later (Galeote et al., 2011). In a cross-sectional study of 330 Swedish children with DS, first words emerged during an age range of 1–5 years, where production of at least one word was observed in 12% of the 1-year-old children, 80% of 2-year-olds, 90% of 3-year-olds, 98% of 4-year-olds, and 94% of 5-year-olds (Berglund et al., 2001). Another source estimated an average age of 21 months at emergence of first words (Stoel-Gammon, 2001). Many children with DS go on to exhibit difficulties with vocabulary, morphosyntax, and weaker expressive than receptive skills (Abbeduto et al., 2007; Martin et al., 2009). Areas of strengths are joint attention and social interaction skills (Hahn et al., 2018; Martin et al., 2009).

Children with various forms of intellectual disability, including DS, may have weaknesses in language development because it is difficult for them to extract the regularities inherent in the language spoken around them. Examples of such regularities include the meanings of words extracted from the contexts in which they were used, the predominant sentence structure of the language (e.g., subject–verb–object for English), and phonotactic patterns. Typically, children automatically extract regularities from daily exposure to language and become competent users of these linguistic components without explicit help. For children with intellectual disability, however, this bootstrapping approach may be more difficult, leading to deficits in language comprehension and formulation (Kover, 2018). Verbal short-term memory and working memory are also areas of difficulty for many individuals with DS (Jarrold et al., 2002, 2009), which may compound the difficulties with extracting linguistic regularities from the ambient language.

Generally, a broad spectrum of health concerns is associated with DS, all with the potential to impact speech and language development in specific ways. Frequent and recurring hearing loss is likely a barrier to the speech and language development of children with DS. Hearing loss is common, particularly transient conductive hearing loss, with estimates as high as 78% of children with DS (McPherson et al., 2007; Meuwese-Jongejeugd et al., 2006; Nightengale et al., 2017). Individuals with DS are also at increased risk for visual impairment (Creavin & Brown, 2009; Harvey et al., 2020), which can negatively influence accessing the visual contributions to speech perception and interfere with children's ability to map the words they hear to the objects, persons, and activities in their environment when they cannot see them clearly. Hypotonia and motor discoordination are commonly seen in children with DS (Alesi et al., 2022; A. F. Morris et al., 1982), which interferes with their ability to manipulate objects and toys, thus diminishing their ability to learn from play.

Approximately 50% of young children with DS have a congenital heart, gastrointestinal, or genitourinary conditions that may require surgical intervention during the first year of life (Freeman et al., 2008; Visootsak et al., 2011). A study of 178 infants/toddlers, preschoolers, and older children and adolescents with DS who did or did not undergo heart surgery in their first year of life showed that children who underwent surgery experienced a temporary setback in language skills but caught up with their peers who did not undergo surgery by school age (Alsaied et al., 2016).

In addition to the potentially negative influences of the cognitive and health variables described above, there is also evidence that environmental factors play a role. Neurotypical children's language development is shaped by input, including hearing speech (Kuhl & Meltzoff, 1996; Werker & Hensch, 2015), seeing others' mouths and faces during communication (Teinonen et al., 2008), and parent responses to children's vocalizations and babbling (Fey et al., 2006; Fiani et al., 2021; Woynaroski et al., 2014). However, young children with DS tend to vocalize sparsely, which diminishes opportunities for caregiver responses. This, in turn, adds a suboptimal linguistic environment to the challenges in learning to talk (Parikh & Mastergeorge, 2018; Thiemann-Bourque et al., 2014). One promising avenue toward supporting the speech and language development of individuals with DS is creating rich language environments by responding frequently to child vocalizations. In one study, mothers were coached to imitate the vocalizations of their 4-month-old infants with DS, which led to greater infant vocalization rates, showcasing the effectiveness of this simple tool (Fiani et al., 2021). This observation, combined with the finding that many families with children who have DS experience barriers in accessing services (Soccorso et al., 2024), supports the idea we are proposing here, namely, that home-based, parent-led interventions may hold promise for improving outcomes.

The fact that the speech and language risks of children with DS are known at or even before birth provides the foundation, in theory, for developing and trialing earliest and proactive interventions via research studies and then implementing those with the best efficacy in clinical practice. A recent review of 11 early speech and language intervention studies (Seager et al., 2022) identified four studies focused on communication training and responsive teaching, two on behavior modifications, one on the dialective-didactic method, and four on early stimulation programs. Sample sizes across groups ranged from four to 16 children. The studies varied in participant start age, with 108 of the 188 total participants under 12 months of age at the start and 80 children between 12 and 55 months old. One of the studies reported better outcomes in a group starting at birth, compared to groups starting at ages 3–4 and 6–7 months. Nine of the 11 studies reported positive outcomes for child language and communication skills as measured up to 18 months after the intervention, however, decelerated growth in requesting behaviors was observed in one study and no improvements were noted in two studies. The review authors generally conclude that interventions with high dosage focus on language and communication in a naturalistic setting, and parent involvement may have the potential to improve outcomes for children with DS ages birth to 6 years. However, the authors also observed limitations: Seven of the 11 studies did not meet PEDro-P appraisal criteria, the sample sizes were small, statistical power was not considered in any of the studies, and even the studies with larger numbers of participants were characterized by severe shortcomings such as inadequate information about the intervention, poorly described outcome measures, missing treatment fidelity data, and control groups lacking the same baseline assessments as those in the intervention groups. According to the authors, these shortcomings urgently call for higher quality intervention studies. This opportunity space motivated the present study.

Babble Boot Camp Intervention

Babble Boot Camp (BBC; Peter et al., 2019) is the first proactive, individualized, sustained, parent-led intervention that addresses anticipated speech and language needs by using evidence-based practices to foster precursor and early communication skills. We developed BBC originally for infants with a newborn diagnosis of classic galactosemia (CG), an inborn error of metabolism that places children at a high risk of severe speech and language delays (Berry, 1993; Potter et al., 2008; Rubio-Gozalbo et al., 2019; Waisbren et al., 1983). Because CG is diagnosed via newborn screening in the United States, these children's risks for speech and language difficulties can be anticipated from birth. This is the reason we selected infants with CG for a clinical trial of BBC.

The parent coaching model was chosen because it has the advantage of high dosage levels as parents incorporate therapeutic strategies into their daily home life. The telehealth modality was chosen for the following reasons: First, it made it possible to include families regardless of physical location in the United States or abroad. Second, it was convenient for families because they could meet with the speech-language pathologist (SLP) without the need to commute to a clinic facility. Third, if successful, it allows for upscaling to serve large numbers of families in regions with limited availability of speech-language pathology services. Fourth, the BBC clinical trial of children with CG showed high caregiver satisfaction regarding the telehealth modality (Finestack et al., 2022). The telehealth modality proved to be the right choice for yet another reason: During the COVID-10 pandemic, we experienced no disruptions in the BBC's implementation.

BBC's set of routines and activities are intended to create resilience against the anticipated speech and language challenges. An example of such a routine is responding frequently to the child's vocalizations, which has the intended effect of increasing infant vocalization rates. This responsive interaction style also builds scaffolding for conversational turn-taking as early as the coo and babble stages. Examples of activities, to be done daily for 5 min or longer, if desired, are showing the child videos of babbling children with the goal of eliciting babble and increasing babble quantity and complexity, and joint picture book reading, intended to build vocabulary and conceptual knowledge. Additional details about the intervention procedures are described in recent publications (Finestack et al., 2022; Peter et al., 2019, 2021, 2022, 2023; Potter et al., 2024), ClinicalTrials.gov (NCT03838016), and Open Science Framework at https://osf.io/yzht4/.

In the BBC theoretical and strategic framework (see Figure 1), target caregiver behaviors are designed to elicit target child behaviors, which, in turn, lead to the child mastering target skills. An example of this cause–effect chain is caregivers surrounding their infants with a word-rich environment, which leads to the child's attending to the verbal input by watching the moving face and hearing the speech sounds, which in turn leads to the child's visual–auditory coupling of the speech signal and build-up of phoneme boundaries. Situational feedback loops are a design feature of the framework, where increases in caregiver behaviors lead to increases in child behaviors, which potentiate the opportunities for caregiver responses. Skill-based feed-forward loops are also built into the framework from the earliest stages. For instance, the child's sensorimotor associations of vocal movements during earliest vocalizations become the building blocks for precursor consonants and vowels in babbled utterances, which, in turn, become building blocks for speech productions (Peter et al., 2019, 2021, 2022, 2023). Figure 1 shows selected components of this framework and their relevance for infants with DS. Because child ages at each of the more advanced target behaviors are highly variable and because children progress through the intervention at their individual pace, target behaviors and skills are not mapped to any particular age.

Figure 1.

Selected components of the individualized speech and language intervention in terms of target parent behaviors, target child response behaviors, and target child skills. Yellow arrows in the header row: intended effect chains (target parent behavior → target child behavior → target child skill). Curved blue arrows between the target parent and target child behavior columns: situational feedback loops. Curved green arrows next to the target child skills column: feedforward loops of child skills that, when mastered, provide the building blocks for more advanced skills. Figure adapted from Peter et al. (2021), which is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Following a small pilot study (Peter et al., 2019), a fully powered randomized controlled clinical trial was conducted with National Institutes of Health funding (R01HD098253). Findings to date show positive effects on babble complexity, receptive and expressive vocabulary size, general communication abilities, and evidence of postintervention speech and language benefits (Peter et al., 2019, 2021, 2022, 2023). A feasibility study based on established criteria (Bowen et al., 2009) showed that caregivers rated their experience with BBC as highly acceptable (“acceptability”: the extent to which BBC was satisfying to parents), completion and compliance rates were high (“implementation”: the extent to which the intervention could be successfully delivered to families), and the intervention was low in cost in terms of parent and clinician time (“practicality”: the extent to which the intervention could be delivered to families at a low cost; Finestack et al., 2022). The fact that children with DS also receive an early diagnosis, at or even before birth, and have known risks for speech and communicative challenges motivated the pilot study of BBC with infants with DS described in this report.

To summarize, the overall goal of this pilot study, in analogy to our work with children with CG, was to gain insights that can inform a larger, fully powered clinical trial with a broader set of primary and secondary outcome variables, with the ultimate aspiration to improve clinical outcomes by building resilience against known lifelong speech and language risks that children with DS face. Such a clinical trial would consist of two major components: implementation and evaluation via data collected from the participating families. With the present study, we aimed to observe aspects of intervention feasibility (acceptability, implementation, practicality) and feasibility of evaluation (obtaining sufficient progress data consisting of direct measures of communication skills and also broader aspects of health and well-being). We also addressed potential benefits from BBC for children with DS in the areas of linguistic environment, babble complexity, and language development. In this observational study, the following research questions (RQs) were addressed:

RQ 1. Intervention acceptability: How satisfied were caregivers with their BBC intervention experience?

RQ 2. Intervention implementation feasibility: How successfully was the intervention delivered to the families?

RQ 3. Intervention practicality: How costly was the intervention in terms of time and materials?

RQ 4. Feasibility of evaluation: How successful was the data collection?

RQ 5. Potential benefits: What improvements in linguistic environment, babble complexity, and receptive and expressive language skills could be observed?

RQ 6. Medical status: What was the role of the child's health history at the start of the intervention in the overall outcome?

RQ 7. Age at start of intervention: Are there inherent advantages in starting the intervention during early infancy versus later infancy?

Method

Participants

The supplemental TIDieR form shows where the key components of the intervention are described in this article (see Supplemental Material S1). The study was completed with approval from the institutional review board at Arizona State acting on its own behalf and that of Washington State University (STUDY00004969 and STUDY00018407). Caregivers gave written permission for their children to participate and written consent for their own participation. Inclusionary criteria for the children were a full (not partial or mosaic) Trisomy 21 DS diagnosis, birth not prior to 37 weeks gestation, not awaiting heart surgery at the time of the study, and absence of other conditions with the potential of confounding the results (e.g., cerebral palsy, hearing loss due to cytomegalovirus infection, fetal alcohol spectrum disorder). Presence of hearing loss not due to other causes, feeding difficulty, disordered sleep patterns, and history of surgeries are all common among children with DS and, hence, were not grounds for exclusion. English was required to be the main language spoken at home, although speaking other languages was not cause for exclusion. Two target age ranges were 2 through 5 months for a younger group and 11 through 14 months for an older group. The two ranges were chosen to obtain insights into potential advantages regarding feasibility and effectiveness for either of these early start ages. The younger and older groups received the BBC speech/language intervention with the same intervention format, dosage, and duration. Participants were recruited via organizations serving children with DS (e.g., Gigi's Playhouse, Phoenix Children's Hospital) and social media.

Four children with DS ages 4–5 months (all male) and six children ages 10–16 months (three female, three male) were enrolled. In all cases, both father and mother caregivers participated. One additional family enrolled but withdrew soon after the start of the intervention. One family who participated in the weekly meetings until the near completion of their 10 months of intervention withdrew from further intervention. This family's data are included here because the family attended 89% of the sessions and completed two of the questionnaire packages. In total, 11 families were enrolled, and 10 families' data are reported here.

All children were White, and four were Hispanic. The number of sibling children ranged from zero to four. Two families used English and sign language at home, and two families used English, sign language, and Spanish at home, while the remaining six households were monolingual English. Caregiver education levels included “some college” (n = 5), “college” (n = 11), and “graduate degree” (n = 4). For specific metrics for which no typical norms exist, we consulted data from up to 24 families with typically developing children (12 girls, 12 boys) who participated in the CG study. These families identified their child as White (n = 17), more than one race (n = 5), and Asian (n = 2). Caregivers' educational background included high school (n = 3), some college (n = 13), college (n = 12), and graduate degrees (n = 20).

To protect the identities of the 10 participating children, health conditions are described here in aggregate. The most common health condition was hypotonia, reported for eight children. Five children had recurrent ear infections. Four children had a history of difficulty with chewing and swallowing. Three children were born with cardiac anomalies, one of whom had undergone heart surgery, and three other children had undergone at least one major gastrointestinal surgery. The following conditions were mentioned twice: tongue tie with frenulectomy, fluid aspirated at birth, hospitalization due to low oxygen saturation, heart disease not requiring surgery, and vision impairment. Allergies, thyroid disease, and croup were each mentioned once. In terms of severity, the child with mildest health involvement had only hypotonia. The most severe health involvements were seen in one child who had seven of the listed conditions but no history of surgery and one child who had multiple surgeries plus two of the other conditions.

Intervention Procedure

The SLP who implemented the intervention had over 30 years of experience, and she had implemented the entire BBC intervention in the CG study. When a family had time conflicts for a scheduled telehealth session, the SLP and the family discussed the session topics including the home video clips and the child's present levels via an e-mail exchange but did not change the treatment plan for the subsequent week. Of the 398 sessions that took place, 16% were conducted over e-mail. In total, 398 of 429 possible sessions took place over telehealth or e-mail, while the remaining 31 (7%) were missed or canceled, an attendance rate of 92% when averaging across the individual families' attendance rates (range [77%, 100%]). For the 398 telehealth or e-mail sessions, the SLP also recorded the number of times caregivers provided videos prior to the session (M = 91%, range [80%, 98%]) and the number of sessions with evidence of follow-through with the intervention plan (M = 91%, range [66%, 100%]. The number of questionnaire packets returned during the intervention ranged from 1 to a maximum of 3 or 4, depending on whether the intake fell on or before a regular collection age, for an average completion rate of 76% (range [33%, 100%]). The number of monthly daylong audio recordings, further described below, completed ranged from three to 11, for an average completion rate of 65%.

The SLP met with the caregivers weekly for approximately 10–20 min. She spent an additional 15 min per family preparing for sessions and maintaining clinical notes. All telehealth sessions were recorded for purposes of offline checks of implementation fidelity. The intervention ended after 10 months, in all cases at or before the child's second birthday. All children were recruited at ages that would have allowed completing the full 10 months before their second birthday; however, one child was enrolled at age 14 months but did not start sessions until age 16 months, cutting the intervention duration from 10 to 8 months. As in previous implementations of BBC, a picture book of first words and a light-up toy were the only materials provided to all families to support verbal interaction. No other specialized materials were needed, as BBC is designed to be scalable to all households using common or even no materials.

The BBC intervention includes many design features that are ideal for infants with DS, given their characteristic profile of strengths and challenges (see Figure 1). Most fundamentally, it leverages the strengths in the areas of social interaction and joint attention, as its framework is entirely based on caregiver–child interactions. The focus on responding to infant vocalizations addresses the known cycle of sparse babbling and reduced opportunities for parent responses (Parikh & Mastergeorge, 2018; Thiemann-Bourque et al., 2014) and the evidence that frequent caregiver responses result in higher infant vocalization rates (Fiani et al., 2021). We modified the intervention further to fit the individual profile of strengths and needs of infants with DS, as described above, in the following ways: First, to address hearing loss, the SLP instructed caregivers to observe their children's responses to sound as an informal way to detect hearing loss and to speak at an adequate loudness level and with clarity (“parentese”). Second, to address both hearing and vision issues, the SLP encouraged caregivers to position their faces within full view of the child when speaking. Third, to compensate for difficulties with working memory and extracting linguistic regularities from the ambient language, and also to support vocabulary growth, the SLP encouraged caregivers to provide a language-rich environment, ensuring that the persons, objects, or events they were describing were in full view of the child. Fourth, to compensate for low muscle tone and difficulties with motor coordination, the SLP coached caregivers to assist their child with holding and manipulating objects during their verbal interactions. The SLP did not discourage caregivers who used sign language with their child from continuing to do so; instead, she coached them to respond to their child both orally and with sign, as the ultimate goal of BBC is to support children's spoken language skills. She coached caregivers to watch for and reinforce their child's attempts at oral language, for instance, closing lips in the context of the word “mama.”

Data Sources and Analytic Procedure

At intake, caregivers completed a questionnaire covering the child's health status and the family's demographic information. Caregivers completed one daylong audio recording each month, using the LENA recording system (Language ENvironment Analysis), on a typical day when one or both caregivers were spending the majority of their time with the child. After a minimum of 4 months of participation in the intervention, caregivers completed a satisfaction survey. Families completed a set of questionnaires covering language development, general development, and physical and psychosocial well-being, further described below, starting at intake and then at 3-month intervals (ages 6, 9, 12, 15, 18, 21, 24 months) during the intervention period, analogous to the clinical trial with children with CG (Peter et al., 2019, 2021, 2022). Due to the small sample sizes in the two intervention groups, comprehensive statistical models were not planned. Where metrics of central tendency or growth over time are reported, no claims of statistical significance are made.

Linguistic Environment

From the monthly LENA recordings, proprietary algorithms were used to extract the number of adult word counts (AWCs), child utterances (CUs), and conversational turns (CTs) and converted into rates per hour, addressing the fact that total hours of recording varied from 10 to 16 hr, which is the maximum recording time. Mean recording time was 13.6 hr (SD = 2.03). Scores from the current study were compared to typical norms for 20 groups of children ages 5–24 months (n = 45–88) from a larger study of 329 children ages 2–48 months (Gilkerson et al., 2017). That study showed relatively stable numbers of AWCs across the sampled child ages but substantial increases of CUs and CTs during the first 2 years of life. Because these normative data were reported as raw numbers for 12-hr recordings, we converted these into per-hour rates for purposes of direct comparisons with our data.

Babble Complexity

Beginning at child age 6 months, which is the earliest age at which babble contains sound productions sufficiently resembling speech sounds for purposes of phonetic analysis, we extracted up to eight 5-min LENA segments with the highest number of child vocalizations based on the LENA processing results. A team of trained transcribers listened to these segments with headphones, transcribed the CUs into the International Phonetic Alphabet, and computed mean babbling level (MBL) scores (S. R. Morris, 2010; Stoel-Gammon, 1989). MBL is a measure of babble complexity based on consonant (C) and vowel (V) content of the babbled utterance, averaged for the first 50 babbled utterances in the recording. It does not capture the actual phonetic or phonemic inventory. MBL scores for individual utterance range from 1 to 3. A score of 1 is assigned for utterances composed of one or more vowels, voiced syllabic consonants, or syllables in which the consonant is a glottal stop, glide, or [h], which are not considered true consonants in the MBL framework (e.g., [æ, wɛ, juhi]). A score of 2 is assigned for utterances composed of CV, VC, or CVC syllables with a single true consonantal type where voicing differences are ignored, so that [g, k] in the same utterance count as one true consonant, not two (e.g., [ʃi, giku, na]). A score of 3 is assigned for utterances composed of utterances with two or more true consonants, again disregarding voicing differences (e.g., [dɛg, nɛbop, gim]). The transcribers computed an average MBL score for each child's monthly recording.

We used data from the typical control children collected during the CG clinical trial (Peter et al., 2019, 2021, 2022) as a basis for comparison, because our method of calculating MBL scores based on selected 5-min segments has not been previously used by other investigators and typical norms do not yet exist outside of that clinical trial. Typical data are reported here for children ages 6–18 months. Beyond that age, MBL scores were available for decreasing numbers of typical controls. This likely is, at least in part, because increased production of meaningful speech coincided with fewer babbled utterances, and as these children got older, the required 50 babbled utterances for computing MBL scores could be obtained in fewer of the audio clips created from the LENA recordings.

Language and General Development

For each child, age at production of the first word was noted, following established guidelines (clear referent, not imitated, similar to adult form; de Boysson-Bardies & Vihman, 1991). Onset of first word was verified by the SLP based on caregivers' reports or directly observed by her. We used the Words and Gestures form from the MacArthur-Bates Communicative Development Inventories–Second Edition (MBCDI-2; Fenson et al., 2007) to obtain information about the children's receptive and expressive vocabulary sizes. Parent report of vocabulary sizes using this tool has been found to be valid and accurate (Miller et al., 1995).

We used the Ages & Stages Questionnaires–Third Edition (ASQ-3; Squires & Bricker, 2009) to evaluate the children's development in the area of communication (babbling, vocalizing, listening, understanding, talking). The ASQ-3 also measures gross motor (arm, leg, body, sitting, crawling, walking, running), fine motor (hand, finger, coordination), problem solving (learning, playing with toys), and personal–social skills (solitary play, play with other children, self-help skills, interaction with others), which are all areas not directly targeted in the intervention. The questionnaire forms are designed for specific age ranges. Raw score sums for each developmental area, and age range fall into one of three standardized categories: above cutoff (not concerning; interpreted as developing appropriately), close to cutoff (may require close attention, specialized activities, and/or repeat screening), or below cutoff (concerning; further diagnostic assessment is recommended). The ASQ-3 has high test–retest reliability, sensitivity, specificity, and validity (Singh et al., 2017). Note that the ASQ-3 Communication domain differs from the MBCDI-2 in that the questions cover more general communicative behaviors (e.g., baby making the same sound again after the parent copies it; baby looking at an object in response to a question like, “Where is the ball?”) whereas the MBCDI-2 provides more aggregate information on aspects such as expressive and receptive vocabulary size and number of phrases understood.

General Well-Being

We used the Pediatric Quality of Life Inventory (PedsQL) questionnaires to assess physical and psychosocial well-being for children from birth to 12 months and from 13 to 24 months (Varni, 1998a, 1998b). The Physical Health Summary score is the average score of all responses in the Physical Function (e.g., low energy level, fatigue, lethargy, pain) and Physical Symptoms (e.g., difficulty swallowing, spitting up, vomiting, having a rash) categories. The Psychosocial Health Summary score is the average of all questions relating to emotional (e.g., afraid, sad, angry), social (e.g., smiling, eye contact), and cognitive (e.g., attention, imitating sounds) functions. The Likert response options are formatted such that problems with a given function (e.g., low energy level) are checked for never, almost never, sometimes, often, and almost always. In the scoring process, these responses are converted to 100, 75, 50, 25, and 0 points, respectively, and averaged per category, so that the Physical Health Summary and Psychosocial Health Summary scores each have a potential range of 0–100, where high numbers reflect strong functioning. The PedsQL scores are not normed. We compared the scores obtained from the pilot families in this study to a set of scores obtained from 18 control families who participated in the CG study (Finestack et al., 2022; Peter et al., 2019, 2021, 2022, 2023; Potter et al., 2024).

Caregiver Satisfaction

After a minimum of 4 months of participating in the intervention, caregivers completed a satisfaction survey. Questions covered satisfaction with and comments about the start age, session frequency and length, required involvement on the part of the family, interaction with the SLP, most important areas of development targeted by the intervention, effectiveness of the telehealth modality, appropriateness of time requirement for participation, helpful aspects, downsides of the intervention, access to other services, and perceived needs for services that were not available.

Data Reliability and Fidelity

The recorded telehealth sessions were checked for implementation fidelity. Research assistants were trained to 90% accuracy and reviewed 10% of the recordings from each participant, chosen at random per participant. They checked the recordings for the presence of the components of the teach–model–coach–review model on which the caregiver training was based (Roberts et al., 2014). Because the SLP was not present when parents implemented activities and routines during the intervening week, the coach component could not be verified. Instead, the videos were scored for presence of a planning component (explicitly formulating a plan for the coming week) and a question component (SLP asking the caregivers if they had any questions). The teach, review, and plan components were each found in 100% of the checked recordings, and the model and question components were present in 94% and 92% of the recordings, respectively, for an overall average fidelity rate across components of 97%. Thus, the intervention was implemented with high fidelity.

Research assistants scored all questionnaires and entered scores into a Research Electronic Data Capture (REDCap) database. A second assistant confirmed all scores.

The LENA data extraction for AWC, CU, and CT was based on automatized algorithms, as were the extractions of 5-min segments with the highest numbers of child vocalizations that became the basis for the MBL calculations. Members of the transcription team checked all MBL scores for interrater reliability, and any miscalculations discovered in the process were corrected. Past experience showed that retranscriptions were not necessary, as agreement rates of MBL scores between first and second transcriptions were 91.5% (Peter et al., 2019). Members of the questionnaires and transcription teams, who routinely worked on files from children who have CG, DS, other conditions, or typical development, were blinded to the children's diagnoses.

Results

Linguistic Environment

AWC per hour as a function of child age in months was characterized by high levels of variability (see Figure 2). When the individual children's averages were averaged per group, AWC rates for the younger (M = 1,247, range [560, 1,790]) and older (M = 1,221, range [947, 1,773]) groups were similar. Most AWC fell within 1 SD or above the normative data in a larger study of typical children (Gilkerson et al., 2017), mirroring their relatively stable rates across child ages.

Figure 2.

Adult word counts (AWCs) per hour. Green lines = children with DS in the younger group; blue lines = children with DS in the older group; thick black line = means from 45 to 88 typical children (Gilkerson et al., 2017); DS = Down syndrome. Error bars: ±1 SD.

CU rates per hour were variable across time and overall lower in the younger group (M = 84, range [61,106]), compared to the older group (M = 114, range [59, 166]; see Figure 3). In both groups, most CU rates fell within 1 SD of the typical data, generally mirroring their steady increase across child ages.

Figure 3.

Child utterance rates per hour. Green lines = children with DS in the younger group; blue lines = children with DS in the older group; thick black line = means from 45 to 88 typical children (Gilkerson et al., 2017); DS = Down syndrome. Error bars: ±1 SD.

As with AWC and CU, CT rates per hour were variable. Mean CU rates were lower in the younger group (M = 25, range [16, 36]), compared to the older group (M = 32, range [17, 48]; see Figure 4). Most CU rates in both groups were within or above the typical data reported by Gilkerson et al. (2017), again mirroring their steady increase over time. For linear Pearson regression equations for each participant across age, see Supplemental Tables S1, S2 and S3 for AWC, CU, and CT, respectively.

Figure 4.

Conversational turn rates per hour. Green lines = children with DS in the younger group; blue lines = children with DS in the older group; thick black line = means from 45 to 88 typical children (Gilkerson et al., 2017); DS = Down syndrome. Error bars: ±1 SD.

Babble Complexity

In the younger group, sufficient numbers of canonical babble utterances for calculating MBL scores were obtained for two children starting at 6 months, one child at 7 months, and one child at 8 months. MBL scores in the children with DS in the early start group resembled those obtained from the typical controls and fell nearly completely within ±1 SD of their scores (see Figure 5). All but two children showed growth over time. For linear Pearson regression equations, see Supplemental Table S4.

Figure 5.

Mean babbling level scores as a function of child age in months. Black line = averages from up to 24 typical controls; green lines = children with DS in the younger group; blue lines = children with DS in the older group; DS = Down syndrome. Error bars for typical controls: ±1 SD.

Language and General Development

Nine of the 10 children started to produce words at ages ranging from 16 to 22 months (M = 19). One child had not yet started producing words at 21 months, the last available age point for this child (see Table 1). Average ages for the younger and older groups, respectively, were 18.0 and 19.4 months.

Table 1.

Age at first words as determined by the speech-language pathologist.

Code	Sex	Age at first session (months)	Age at first word (months)
1_1	M	5	20
1_2	M	4	19
1_3	M	4	17
1_4	M	5	16
2_1	M	10	Not yet at 21
2_2	F	12	16
2_3	F	13	20
2_4	M	14	20
2_5	M	14	22
2_6	F	16	19

Note. M = male; F = female.

The receptive and expressive word counts from the MBCDI-2 show increases over time for all children except one for whom only a single data point was available (see Supplemental Figures S1 and S2).

Of the ASQ-3 scores for the five domains (Communication, Gross Motor, Fine Motor, Problem Solving, Personal/Social), the highest scores were seen in the Communication domain (Above Cutoff: 5, Close to Cutoff: 5, Below Cutoff: 15). The respective counts for Gross Motor were 3, 1, 21; for Fine Motor 2, 6, 17; for Problem Solving 3, 7, 15; and for Personal/Social 1, 5, 19. Note that all Communication scores of Above Cutoff occurred at or before age 12 months. Supplemental Table S5 summarizes the scores in all five domains.

General Well-Being

Many children's Physical Health Summary scores fell below −1 SD of scores obtained from 18 controls (see Supplemental Figure S3). For two children in the younger group and three children in the older group, all data points were below this threshold. Average scores in the younger and older groups were 78.0 and 74.6, respectively.

Children's Psychosocial Health Summary scores fell largely below those obtained from typical peers (see Supplemental Figure S4). Averages for the younger and older groups were 76.4 and 67.9, respectively. Separate averages across children for the categories contributing to this score (social, emotional, and cognitive functions) showed that areas of strength were social function (87.9 and 79.6 in the younger and older groups, respectively) and emotional function (75.6 and 78.2 in the younger and older groups, respectively), while some of the cognitive function scores were lower (68.7 and 48.4 in the younger and older groups, respectively; see Supplemental Figure S5).

General Summary

For a simplified overview of potential intervention effects, six areas of communication skill progress were estimated for each family. Progress was noted for having at least half of the AWC, CU, and CT metrics within or above 1 SD in a typical sample (Gilkerson et al., 2017), evidence of growth in babble complexity; first words at or before age 21 months; and at least one above Cutoff or Close to Cutoff score in ASQ-3 Communication. Five participants showed progress in all six areas. These included the child whose health concerns consisted of only hypotonia, but also four children with additional health concerns, in two cases including histories of abdominal or heart surgery. Two participants showed progress in five areas, one in three areas, and two in two areas, where one had a history of surgery.

Caregiver Intervention Satisfaction

The average amount of time since starting the intervention when completing the satisfaction surveys was 6.2 months (range [4, 13]). In the younger group, three of four sets of caregivers completed the satisfaction survey, whereas in the older group, all six sets of caregivers completed it. Supplemental Table S6 lists the responses.

All respondents rated the videoconference mode as effective. Eight of the nine families selected the “Just right” option for session duration; one family would have preferred longer sessions, 30 instead of 10–20 min. All but two families selected the “Just right” option for the start age; in the older group, two sets of caregivers would have preferred an earlier start date, 6 or 6–9 months (their children had started at 14 and 12 months, respectively). Of three families who said that the time requirement was too great, one cited the questionnaires, one cited the brief home videos, and one cited both the videos and the LENA recordings as being time consuming. One set of caregivers mentioned that they would prefer to focus more on signing than spoken language because their child seemed to respond better with sign than with speech. Importantly, none of the 10 families currently had access to other speech/language interventions. Only two families listed any additional services, and these included occupational and physical therapy.

Caregivers appreciated the professional support to enhance their communication with their child, learning to recognize their child's various attempts at communicating, the focus on their child's current functioning, and awareness that their child was making progress.

Discussion

The pilot study of BBC in a sample of four children ages 4–5 months and six children ages 10–16 months, all with DS, was completed to investigate aspects of acceptability, implementation feasibility, practicality, feasibility of data collection, and potential benefits, with the goal of scaling up to a fully powered clinical trial in this population. Additional areas of interest were child wellness and the role of the child's age and medical status at the start of the intervention. To our knowledge, this was the first pilot implementation of a proactive, sustained, caregiver-led intervention via telehealth aiming to build resilience against known risk of speech and language delays in children with DS, with rigorous implementation and data scoring fidelity.

Acceptability of the intervention (RQ 1) was evident in the caregivers' feedback in the satisfaction survey. The early start age, a key BBC design feature, was unavailable to the families outside of BBC, and it was clearly endorsed by all families. The telehealth modality and session frequency and duration were all rated as highly acceptable. Caregivers appreciated the professional help to better recognize their child's attempts at communication and to build communication skills. All comments on the interactions with the SLP were positive, and all caregivers described ways in which they found the intervention helpful. Overall, parent responses are consistent with high levels of intervention acceptability.

Implementation feasibility of the BBC intervention (RQ 2) was demonstrated in the following ways: Average session attendance rate across families was 92%. Caregivers' participation metrics in terms of providing home videos and following through with the planned routines and activities were also high, both with an average compliance rate across families of 91%. Dropout rates were low, with one family withdrawing at the beginning of their intervention and one family withdrawing close to the end of the intervention. The average implementation fidelity rate for the five tracked components was 97%. Thus, the intervention was delivered to the families with high implementation feasibility. Of note is the fact that high implementation feasibility was evident across all child ages including the very youngest participants who started the intervention as early as age 4 months.

The intervention was provided at low cost (RQ 3). Families participated via telehealth, which eliminated the need to commute to a facility. All families already had access to computers, the internet, and video recording devices such as smartphones, which are required for this implementation format. The home routines and activities did not require any specialized materials. The intervention was also cost-effective in terms of clinician time due to its design feature of parent coaching. As caregivers gained competency and implemented the activities and routines tailored to their child's skill set on a daily basis at home, the intervention dosage was maximized to levels not possible with direct professional treatment. Thus, the BBC implementation was characterized by high levels of practicality.

While all families contributed to the data collection, the participation rates in these research-related activities (RQ 4), with average return rates of questionnaires at 74% and daylong audio recordings at 65%, were lower than the participation rates in the intervention activities, and the survey feedback from two families corroborated this by specifically mentioning the questionnaires and audio recordings as time consuming. Nonetheless, we successfully obtained relevant data characterizing the children's communication skills, health status, general development, and physical and psychosocial well-being. Together, these findings are valuable in that they can inform the design of a larger, fully powered study.

Due to the small sample sizes of four children in the younger group and six children in the older one, no conclusive statistical evidence of benefits can be offered. However, some observations can be made toward a tentative appraisal of benefits regarding child speech and language development (RQ 5):

One of the known areas of concern for infants with DS is their sparse vocalization rate, which limits caregivers' opportunities to respond to their child, leading to overall lower AWC and CT rates (Parikh & Mastergeorge, 2018; Thiemann-Bourque et al., 2014). As demonstrated via an intervention study, caregivers who responded to their 4-month-old children's vocalizations by imitating the vocalizations saw an increase in infant vocalization rates (Fiani et al., 2021). One focus of BBC is caregiver responsiveness to infant vocalizations. Comparing the AWC, CU, and CT rates in the participating families to those in a larger study of families with typical children (Gilkerson et al., 2017) showed no differences for most of the families. This is consistent with successfully changing caregiver behaviors in terms of AWC and CT rates, which may also have positively influenced CU rates.

In the younger group, babble complexity as measured with the MBL increased over time for three of the four children, and this increase was in keeping with what was observed in typical peers. In the older group, sufficient numbers of babbled utterances toward calculating the MBL were collected up to age 24 months, with the MBL scores showing a steady increase in complexity for five of the six children. Although comparison data from typical controls beyond age 18 months were not available, the increase in babble complexity in the older DS group was evident.

Emergence of first words as determined by the interventionist using established procedures showed that eight of the nine children who said their first words during the intervention phase did so by age 21 months, with one child producing first words at age 22 months and one child completing the intervention at age 21 months when words had not yet emerged. The early ages at word emergence may imply an intervention effect, given that only 12% of 330 Swedish children with DS said their first word before age 2 years (Berglund et al., 2001) and that one estimate of age of first words is 21 months (Stoel-Gammon, 2001). Per caregiver report using the MBCDI-2, receptive and expressive vocabulary increased steadily over time. These findings are all suggestive of beneficial BBC effects on child language skills but require validation in larger samples and a control group with DS receiving conventional care. General communication skill as measured with the ASQ-3 was the domain with the overall highest scores. Evidence for intervention benefits was weaker here, as the highest scores were mainly seen in the younger group, and they did not show growth over time. One possible reason for this pattern is that the ASQ-3 scoring system compared our participants to typical children of the same age, whose communication skills at the youngest ages may differ less from those of children with DS during the youngest ages, compared to older ages. Here again, to fully appraise the intervention benefits in the area of communication, a control group with DS receiving conventional care is needed. Also, the ASQ-3 may yield more precise insights if, instead of normed scores, the actual skills that parents indicated as fully or partially mastered are described.

The role of the children's medical history in the overall outcomes (RQ 6) is not conclusive. Among the children with the most areas of growth was the child with the fewest health concerns, namely, only hypotonia, but also four children with more severe health concerns, of whom two had histories of abdominal or heart surgeries. One possible explanation is that while a history of major surgery may disrupt speech and language development temporarily (Alsaied et al., 2016; Snookes et al., 2010), earliest and intense parent-led intervention strategies offset the effects of this disruption. A larger participant sample is needed to investigate the influence of the child's health history including number and nature of health concerns on the amount of benefit from BBC.

Regarding metrics of general well-being, many Physical Health Summary scores from children with DS fell below −1 SD of typical controls, with five children's scores entirely below that threshold. This is consistent with health challenges encountered by many children with DS. In the intake questionnaires, hypotonia and difficulty swallowing were among the most frequently mentioned health concerns, and these areas are reflected in the PedsQL child physical health questions.

The children with DS had emotional and social scores similar to typical controls; it was their cognitive development scores that contributed to an overall lowered Psychosocial Health Summary score. This profile of strengths in the interpersonal domain and delays in the cognitive domain reflects previous findings (D'Souza & D'Souza, 2022; Grieco et al., 2015; Hahn et al., 2018; Martin et al., 2009), and it was as expected, because the intervention did not address children's psychosocial health. The PedsQL questionnaire as well as the ASQ-3 questionnaire, especially when used in qualitative (actual skill set) rather than quantitative ways, may prove to be useful in future studies that aim to address cognition and other areas of development in a proactive and intensive way, similar to how speech and language skills were addressed with BBC.

The findings from this study offer some suggestions about the most effective start age (RQ 7), with the caveat that the sample size was too small for more definitive guidance. First, caregiver responses in the satisfaction survey showed that the start age, which was earlier than what was currently available to them, was viewed as highly desirable, with two of six caregiver sets in the older group wishing they could have started even earlier, whereas all caregiver sets in the younger group were satisfied with their early start age. Thus, the earliest start age was generally preferred by caregivers.

Children in both groups began to produce words at younger ages than expected (Berglund et al., 2001; Stoel-Gammon, 2001). The average age at first word production was slightly lower in the younger group, compared to the older group. A larger participant sample is needed to corroborate this finding.

Children in the younger group obtained higher scores in the ASQ-3 Communication category than children in the older group, but there was no evidence of growth over time. The highest scores were obtained at the youngest ages. As mentioned, a possible interpretation is that children with DS and typical children do not differ substantially in their earliest communicative development and that differences become more evident at older ages.

No clear advantage for either of the two start ages emerged for AWC, CU, and CT rates. In both groups, these metrics resembled those in families with typical children, consistent with similar benefits, given that lower AWC, CU, and CT rates would be expected for children with DS, compared to typical controls (Parikh & Mastergeorge, 2018; Thiemann-Bourque et al., 2014). Thus, caregivers in both age groups appeared to be responsive to the SLP's coaching.

Limitations and Future Studies

This pilot study is characterized by several limitations, strengths, and guidance for future directions. To begin, no DS control group receiving conventional care was included. We used published studies reporting on the emergence of first words as one basis of comparison. Because published data from infants with DS covering linguistic environment, babble complexity, general development, and medical conditions in the context of communication skills are scarce, we compared the findings to typically developing children, either by consulting data that we had collected during the CG clinical trial or typical norms in the assessment tools. To fully appraise BBC benefits against conventional care, future studies should include a control group with DS receiving conventional care.

Some of the data collected for this study came from parent report, for instance, the ASQ-3, and PedsQL responses, which carry the potential of subjectivity or bias. Validity for the MBCDI has been shown previously (Miller et al., 1995). The measures of the linguistic environment (AWC, CU, and CT) and the MBL data were based on automatized data extraction and objective quantification.

LENA metrics can be quite variable, due to various reasons, such as changes in the child's daily social environment, the child's health, and, importantly, caregiver behaviors as a result of the intervention. Collecting a series of multiple monthly recordings, as we have done in the present study, is one way to address variability while also showing longitudinal trends in caregiver behaviors and child responses.

Although nine of the 10 children had begun to say words, we did not complete measures of articulation accuracy and speech sound inventories. Future studies should investigate correspondences between babble complexity and speech sound accuracy in children with DS who receive earliest and proactive interventions. Future studies should also investigate the phonetic and phonemic inventories of children with DS. The MBL does not capture phonetic inventory, only utterance complexity based on the number, not the nature, of consonants.

The design of this study does not allow us to make any conclusions of causality. For instance, to what extent an increase in child vocalization rates influences babble complexity and, from there, emergence of words cannot be answered directly with this pilot study. Although the hypotheses underlying the study design include feed-forward loops where one skill mastered by the child becomes the foundation for an advanced skill, children received continuous BBC support at each skill level. For instance, to tease apart whether support during the babble stage contributes substantially to speech and language development later on, a research design is needed that allows comparing long-term outcomes of children with DS who did and did not receive babble support.

Importantly, this study was designed as a pilot study with a small sample size, no control group of infants with DS receiving conventional care, and a limited duration—10 months compared to 18–22 months in our previous CG clinical trial. Overall, the acceptability, feasibility, and practicality that it demonstrates, along with metrics consistent with growth in child skills and successful data collection, encourage follow-up with a fully powered clinical trial of longer duration that includes a control group of infants with DS receiving conventional care. Such a larger and more diverse participant samples will further afford more sophisticated statistical and analytic techniques to be employed. Those larger, planned studies will constitute stronger evidence and bring insights from this work closer to clinical implementation for the benefit of children with DS.

Future studies should also trial BBC in additional groups of children whose risk for speech and language challenges is predictable in early infancy, for instance, children born with hearing impairment, children with high familial risk for conditions like autism spectrum disorder, and children born preterm. For each of these populations, a thorough understanding of their profiles of strengths and challenges is needed to adapt the BBC strategies for maximal benefit.

Data Availability Statement

Anonymized data can be requested from the corresponding author.

Funding Statement

The original Babble Boot Camp clinical trial was funded with a local grant from the Arizona State University Institute for Social Science Research (B. Peter) and a National Institute of Child Health and Human Development grant (R01HD098253) to B. Peter, M. VanDam, and N. Potter. The present project was funded with Research Incentive Distribution funding from Arizona State University (B. Peter) and private donations to the Arizona State University Foundation Gift Account “Babble Boot Camp” (B. Peter). It was also supported by Down Syndrome Education International.