Early diagnostic indicators of childhood apraxia of speech in young children with 7q11.23 duplication syndrome: Preliminary findings

Abstract

Limited evidence for early indicators of childhood apraxia of speech (CAS) precludes reliable diagnosis before 36 months, although a few prior studies have identified several potential early indicators. We examined these possible early indicators in 10 toddlers aged 14 – 24 months at risk for CAS due to a genetic condition: 7q11.23 duplication syndrome (Dup7). Phon Vocalization analyses were conducted on phonetic transcriptions of each child’s vocalizations during an audio-video recorded 30-minute play session with a caregiver and/or a trained research assistant. The resulting data were compared to data previously collected by Overby from similar-aged toddlers developing typically (TD), later diagnosed with CAS (LCAS), or later diagnosed with another speech sound disorder (LSSD). The Dup7 group did not differ significantly from the LCAS group on any measure. In contrast, the Dup7 group evidenced significant delays relative to the LSSD group on canonical babble frequency, volubility, consonant place diversity, and consonant manner diversity and relative to the TD group not only on these measures but also on canonical babble ratio, consonant diversity, and vocalization structure diversity. Toddlers with Dup7 also demonstrated expressive vocabulary delay as measured by both number of word types orally produced during the play sessions and primary caregivers’ responses on a standardized parent-report measure of early expressive vocabulary. Examining babble, phonetic, and phonotactic characteristics from the productions of young children may allow for earlier identification of CAS and a better understanding of the nature of CAS.

Introduction

Childhood apraxia of speech (CAS) is a motor speech disorder affecting the precision and consistency of speech movements; it is assumed to be neurological despite the absence of neuromuscular execution deficits (American Speech-Language-Hearing Association [ASHA], 2007). Impairments are present at the level of motor planning (i.e. spatiotemporal parameters of sound and syllable sequencing) and programming (i.e. commands over corresponding articulator muscles, including range and velocity [van der Merwe, 2009]). The suggested core characteristics of CAS include inconsistent consonant and vowel productions across repetitions, difficulty transitioning across coarticulatory configurations, and prosodic errors (ASHA, 2007). Although children at risk for CAS benefit from proactive early intervention (Peter et al., 2020), current assessment procedures and limited evidence for early indicators preclude reliable diagnosis in children younger than 36 months of age.

Early potential indicators of CAS

Motor speech impairments, such as CAS, may limit an infant’s oral exploration (Maassen, 2002), which may further limit the frequency and quality of prelinguistic vocalizations. Because these early productions are also associated with later speech and language skills (McGillion et al., 2017; Oller et al., 1999), it is important to examine potential early indicators of disorder that may impact emerging and later speech and language skills. Davis and Velleman (2000) proposed a list of characteristics that have the potential for differential diagnosis of CAS in young children based on those reported for older children. These include limited vocal output, gaps in or a limited variety of consonant and vowel sounds, marginal babble (i.e., babble without “true” supraglottal non-glide consonants), lack of consonant-vowel babble, incomplete syllable shapes, and few word shapes. Many of these early potential indicators have been examined in small, retrospective studies.

Expressive and receptive language skills have been assessed in preschoolers diagnosed with CAS using standardized speech/language measures or parent-report questionnaires. When prelinguistic communicative development was retrospectively analyzed in five young children with later CAS (LCAS) (Highman et al., 2012), all five evidenced a delay on standardized tests at 9 months but the pattern of delay varied. Two children had expressive language delay only, one had receptive language delay only, and two had both expressive and receptive language delays. A longitudinal study from age 9 – 24 months comparing eight children at familial risk for CAS to eight without risk using standardized measures (Highman et al., 2013) showed that those in the at-risk group exhibited poorer expressive language skills at most time points and weaker expressive vocabulary skills at 24 months. Highman et al. (2008) compared young children with suspected CAS (sCAS) to children with typical development (TD) and children with specific language impairment (SLI) via retrospective parent-report questionnaires. First words were reported to emerge significantly later in the sCAS group than the TD group, and two-word combinations were significantly more delayed in the sCAS group than in either the TD group or the SLI group.

Measures of babble maturity have been examined both indirectly using retrospective parent questionnaires or interviews and directly using spontaneous speech samples from young children. Parents of young children diagnosed with, or suspected to have, CAS report that their children babbled less, had less mature babble, and were less voluble/vocal during infancy compared to those with TD or other later speech/language conditions (Aziz et al., 2010; Highman et al., 2008). In spontaneous speech samples, volubility did not differ significantly in very young children with LCAS relative to same-aged children with TD (Overby et al., 2015). However, toddlers with LCAS produced significantly fewer canonical babbles and syllables per minute than those later diagnosed with a speech-sound disorder (LSSD) or TD and met the canonical babbling ratio (CBR) criterion of 15% (Lynch et al., 1995) later than either group (Overby et al., 2020).

According to parent report, speech sound development is weaker in toddlers at risk for CAS compared to toddlers with TD (Highman et al., 2013). Furthermore, preschoolers with CAS reportedly have limited phonetic diversity in babble relative to preschoolers with TD or multiple phonological disorders (Aziz et al., 2010). Phonetic and phonotactic characteristics analyzed using speech samples have revealed that young children with LCAS produced significantly fewer resonant productions (defined as “vowels, consonant (C) and vowel (V) sequences, canonical babbling, and words” [p. 49] transcribable in the International Phonetic Alphabet, [IPA]), fewer consonant singletons per minute, and fewer different place features and manner classes than toddlers with TD between 7 and 24 months of age (Overby et al., 2015). Furthermore, young children with LCAS have been shown to use fewer resonant consonants compared to same-aged children with TD (Overby & Caspari, 2015; Overby et al., 2019). Neither age of resonant utterance emergence nor the number of different consonants produced per minute differentiated young children with LCAS from young children with TD or non-CAS LSSD. However, significant differences were present in rate of resonant consonant production per minute and number of different place and manner features per minute (Overby et al., 2019).

While phonotactics have been examined less often than other potential early indicators, parent reports indicate that children with sCAS did not use variegated babble during infancy whereas children with TD or SLI did (Highman et al., 2008). Furthermore, speech sample analysis has shown that young children with LCAS were limited in their use of CV and CVC shapes compared to TD and LSSD groups at 13 – 18 months (Overby et al., 2019).

While these measures have been used to evaluate early potential indicators of CAS in very young children, they have yet to be examined in those at risk due to a known genetic condition. In the present study we focused on toddlers with 7q11.23 duplication syndrome (Dup7), a genetic condition associated with a very high prevalence of CAS or its manifestations (Mervis et al., 2021; Velleman & Mervis, 2011).

7q11.23 duplication syndrome

Dup7 is caused by an extra copy of the approximately 26 genes included in the Williams syndrome region at chromosome 7q11.23 (Mervis et al., 2021). The prevalence of Dup7 is estimated to be one in 7,500 – 20,000 live births (Van der Aa et al., 2009; Velleman & Mervis, 2011). Facial characteristics associated with this syndrome include macrocephaly, brachycephaly, broad forehead, straight eyebrows, deep-set eyes, broad nose tip, short philtrum, thin upper lip, high-arched palate, micrognathia (small lower jaw), and ear anomalies (Mervis et al., 2021; Morris et al., 2015). A few cases of cleft lip and palate and submucous cleft palate also have been reported (Morris et al., 2015). Neurological abnormalities include hypotonia, seizures, and developmental coordination disorder (Morris et al., 2015). The most common cardiovascular manifestation is dilation of the ascending aorta, which has been identified in 46% and is thought to be progressive (Morris et al., 2015).

The median level of intellectual ability for school-aged children and adults with Dup7 is in the low average range, although the full range is from severe intellectual disability to superior ability (Mervis et al., 2021). Toddlers with Dup7 most commonly show relative strengths in receptive language and nonverbal reasoning (with performance in the low average to average range), fine motor abilities in the borderline range, and a specific weakness in expressive language, with almost half performing at floor on standardized assessment (Mervis et al., 2015). Anxiety disorders are very common, with more than half of children with Dup7 meeting full DSM criteria for specific phobia, about 50% meeting full criteria for social anxiety disorder, and about 30% meeting full criteria for selective mutism (Mervis et al., 2021). Externalizing disorders also are common, with about 35% meeting full DSM criteria for attention deficit hyperactivity disorder (Mervis et al., 2021), about 40% reported to show elevated levels of oppositional behavior, and almost 30% reported to show elevated levels of aggression (Klein-Tasman et al., in press). About 19% met gold-standard criteria for autism spectrum disorder (ASD; Klein-Tasman & Mervis, 2018). Most children with Dup7 demonstrate repetitive behaviors and sensory modulation difficulties (Velleman & Mervis, 2011). Developmental delay or intellectual disability, ASD or features of ASD, language delay or disorder, and/or severe speech delay have been noted in almost all case reports of individuals with Dup7.

The onset of language for children with Dup7 is almost always delayed. Based on the retrospective reports of parents of 48 children, median age at onset of first words was 2.00 years, with a range from 1.00 – 4.75 years (Morris et al., 2015). Velleman and Mervis (2011) reported that toddlers with Dup7 produced limited speech and few identifiable meaningful words, often relying on consonant-vowel (CV) syllables and exhibiting frequent reduplication or harmony. Almost all school-aged children with Dup7 have speech delay or disorder (Mervis et al., 2021). Of the 25 participants aged 4 – 15 years described by Velleman and Mervis (2011), CAS or its manifestations was evidenced by more than 75%. Hypotonic manifestations of dysarthria were very common; a few children were diagnosed with dysarthria. Most participants presented with mixed motor speech disorders. Word and syllable shapes also were simpler than expected for chronological age, indicating phonotactic difficulties. School-aged children who received consistent and early speech-language therapy typically presented with average receptive and expressive vocabulary and receptive grammar skills on omnibus measures of language abilities. However, expressive grammar tests and language samples showed that expressive grammar was considerably more delayed (Velleman & Mervis, 2011).

Aims

The present study had two aims. The first was to compare the speech characteristics of toddlers with Dup7 to those of three groups: toddlers with LCAS, toddlers with LSSD, and toddlers with TD. We hypothesized that the speech characteristics of the Dup7 group would be most similar to those of the LCAS group. We expected to find significant differences between the Dup7 group and the TD and LSSD groups on many of the early speech characteristics previously associated with a later diagnosis of CAS.

The second aim was to examine the orally produced vocabularies of toddlers with Dup7 during a 30-minute child-adult play session and to consider relations between these vocabularies and both measures of babble maturity and caregiver-reported spontaneous orally produced vocabulary. We expected that correlations measuring these relations would be statistically significant.

Method

Participants

The Dup7 group included 10 toddlers with genetically confirmed classic duplications of the Williams syndrome region of chromosome 7q11.23. Children and families were recruited and tested through a university research laboratory. Cascade testing indicated that the 7q11.23 duplication was de novo for nine children and inherited for one child. Descriptive statistics for chronological age (CA), sex, and maternal education levels are reported in Table 1, separately for the full sample (included in the phonetic, phonotactic, and lexical analyses) and the subsample that was included in the babble maturity analyses. All children lived in monolingual English-speaking households in the Northeast, Midwest, or South regions of the USA. Nine participants (90%) were White non-Hispanic and one (10%) was biracial non-Hispanic. Data collection began in 2008 and ended in 2014.

Table 1.

Participant demographics

Measure Type	Group
	Dup7	LCAS	LSSD	TD
Phonetic, Phonotactic, and Lexical
N	10	7	5	5
Sex	5 girls, 5 boys	2 girls, 5 boys	3 girls, 2 boys	(2 girls, 3 boys)
CA mean (SD) [range]	21.40 mo. (3.32) [14.88 – 24.84]	20.71 mo. (3.77) [14–24]	21.40 mo. (2.61) [18–24]	20.60 mo. (4.10) [14–24]
Maternal education	2 HS, 6 Bach., 2 Mast.	3 Bach., 4 Mast.	3 Bach., 2 Mast.	1 HS, 2 Bach., 2 Mast.
Babble Maturity
N	7	10	4	6
Sex	4 girls, 3 boys	2 girls, 8 boys	3 girls, 1 boy	2 girls, 4 boys
CA mean (SD) [range]	23.14 mo. (1.79) [20.80 – 24.84]	22.13 mo. (1.11) [19–24]	21.56 mo. (0.13) [19–24]	21.89 mo. (1.13) [19–24]
Maternal education	1 HS, 6 Bach.	2 HS, 5 Bach., 3 Mast.	3 Bach., 1 Mast.	4 Bach., 2 Mast.

Note. Lexical measures were available only for the Dup7 group. For the LCAS, LSSD, and TD groups, data were previously reported in Overby et al. (2019) for the phonetic and phonotactic measures and in Overby et al. (2020) for the babble maturity measures. Dup7 = 7q11.23 duplication syndrome, LCAS = later diagnosed with childhood apraxia of speech, LSSD = later diagnosed with a speech sound disorder [non-CAS], TD = typically developing, mo. = months, HS = high school diploma, Bach. = bachelor degree, Mast. = master degree

The data for the LCAS, LSSD, and TD groups were collected by Overby et al. (2019) for the phonetics and phonotactic measures and Overby et al. (2020) for the babble maturity measures and were re-analyzed for the present study. Demographic information for each group is provided in Table 1. All participants in the LCAS, LSSD, and TD groups were from monolingual English-speaking households, were White non-Hispanic, and resided in the Northeast or Midwest regions of the USA. Assignment of participants to the LSSD, LCAS, or TD groups was based on speech-language and motor speech testing performed when the children were between 3.00 and 8.99 years old. Children were included if they did not have factors expected to negatively impact speech development (structural, cognitive, hearing, or receptive language deficits; dysarthria; or a known genetic syndrome). Data collection began in 2014 and ended in 2017.

Procedures and measures

Research protocols were reviewed and approved by the Institutional Review Board at the University of Louisville (protocol numbers: 97.096 and 97.097), University of Vermont (approval number: B12–074), and Duquesne University (approval number: 13–179). A parent or legal guardian of each participant signed an informed consent document.

Each toddler with Dup7 was recorded during a 30-minute play session in a university laboratory playroom while interacting with a research assistant and/or caregiver using a variety of developmentally appropriate toys and books. Research assistants were instructed to repeat what the toddler said if it was deemed meaningful. Afterward, video recordings were transcribed independently by two trained transcribers using broad IPA. Transcribers were undergraduate and graduate students majoring in Communication Sciences and Disorders who had completed at least one undergraduate-level phonetics course and additional phonetic transcription training provided by the first and second authors. When transcribers could not agree on an English-relevant IPA symbol, non-English symbols (e.g., [β] for a voiced bilabial fricative), notations about in-between sounds (e.g., [d/g] to indicate that the place of articulation was indeterminate), and other notations (e.g., “distorted [s]” or [s̪] for a lateralized [s]) were used. Vocalizations were defined as word-like units, subjectively identified by the lengths of pauses between them. Transcribers also identified whether a vocalization was a word, using criteria established by Vihman and McCune (1994). Afterward, both transcribers met to calculate point-by-point agreement for vowels and consonants and to create a consensus final transcript. Average agreement between the original transcripts was 47.43% for consonants (SD = 23.49, range: 0–66.32%) and 52.15% for vowels (SD = 25.15, range: 0–72.84%). The final consensus transcript was reviewed and corrected by the second author. Average point-by-point agreement for vowels and consonants for consensus transcripts reviewed and corrected by the second author was 93.94% for consonants (SD = 6.67, range: 77.13–100%) and 94.73% for vowels (SD = 5.74, range: 80.89–100%). Out of the 2597 speech sounds included in the final consensus transcripts for the children with Dup7, 29 (1.12%) could not be transcribed (Cs M = 0.7, range: 0–3; Vs M = 1.5, range: 0–8 or were unknown segments that could not be detected as Cs or Vs (M =0.7, range: 0–4). Analyses were based on the corrected consensus transcripts. Neither the authors nor the transcribers were aware which toddlers with Dup7 later received a diagnosis of CAS.

Final transcripts were entered into the Phon 3.2.0-beta.7 Computer Software (Hedlund & Rose, 2019). Phon transcripts were compared to the final handwritten transcripts and corrected as needed. Four consonants were transcribed as non-English consonants because the transcribers could not agree on an English consonant transcription (e.g. voiced bilabial fricative). Non-English consonants were excluded from the analyses because Overby et al. (2019) only transcribed and analyzed “resonant” English consonants. In the present study, we refer to the latter as “English consonant sounds.” We do not know if the toddlers with TD, LSSD, or LCAS produced any non-English consonants. Furthermore, we refer to the syllable structures of vocalizations as “vocalization structures” rather than “syllable shapes” to make clear that we examined how entire vocalizations (prelinguistic word-length units) were structured with respect to syllables (e.g. V, CV, CVCV).

A Phon Vocalization Analysis was performed on the productions of participants and several variables were retrieved for words and/or vocalizations. The frequency and diversity of consonants and vocalization structures produced were derived from the Phon output. Variables are listed and defined in Table 2.

Table 2.

Variables derived from play session recordings

Measure	Type	Definition
Canonical babble ratio (CBR)	Babble maturity	Number of canonical babbles divided by total number of syllables
Canonical babble frequency (CBF)	Babble maturity	Number of canonical babbles per minute
Volubility	Babble maturity	Number of syllables per minute
Consonant frequency	Phonetic	Number of consonants per minute
Consonant diversity	Phonetic	Number of different consonants per minute
Consonant place diversity	Phonetic	Number of different place features per minute
Consonant manner diversity	Phonetic	Number of different manner features per minute
Consonant place feature diversity	Phonetic	Number of each consonant place feature per minute
Consonant manner feature diversity	Phonetic	Number of each consonant manner feature per minute
Vocalization structure frequency	Phonotactic	Number of predetermined structures (C, V, V+, CV, VC, CVC, VCV, CVCV, VCVC) per minute
Vocalization structure diversity	Phonotactic	Number of different predetermined structures produced at least once every two minutes
Vocabulary	Lexical	Number of orally produced word types

The primary caregiver of each toddler with Dup7 completed the Vocabulary Checklist from the MacArthur-Bates Communication Development Inventories: Words & Sentences (CDI; Fenson et al., 2007). Parents were instructed to mark with an S each word their child both understood and orally produced spontaneously (i.e., not imitated or used as part of a song or routine), mark with an M each word the child both understood and manually signed spontaneously (i.e., not imitated or used as part of a song or routine), and mark with a B each word the child understood and both said orally and signed spontaneously. The child’s orally produced expressive vocabulary was the total number of words the parent marked S or B. Parents also indicated whether their child had begun to combine words.

For the LCAS, LSSD, and TD groups, analyses were based on acoustically clear home video clips submitted by the participants’ parents to the third author. Individual video clips ranged from several seconds to several minutes long. For calculating babble maturity variables, the mean number of minutes of video (summed across clips) for participants was 15.81 minutes for the TD group, 35.66 minutes for the LSSD group, and 27.51 minutes for the LCAS group (Overby et al., 2020). For calculating phonetic and phonotactic variables, the mean number of minutes of video for participants was 4.82 minutes for the TD group, 6.19 minutes for the LSSD group, and 6.47 minutes for the LCAS group (Overby et al., 2019). Transcription, reliability, and coding procedures are described in Overby et al. (2019, 2020). Data were available for each of the speech characteristics indicated in Table 2. Data from the CDI were not available for the LCAS, LSSD, and TD groups.

Statistical analyses

Data were analyzed using IBM SPSS v. 28. To compare the performance of the Dup7 group to that of the LCAS, LSSD, and TD groups, Mann-Whitney U tests were computed for each babble maturity, phonetic, and phonotactic dependent variable. To control for Type 1 error, the alpha level for determining statistical significance was adjusted to conform to α = .05 two-tailed for each dependent variable type. This yielded an adjusted p-value for statistical significance of .017 (.05/3) for the three babble measures, .008 (.05/6) for the six phonetic measures, .025 (.05/2) for the two phonotactic measures, and .05 for the lexical measure. Cohen’s d (0.2 = small effect, 0.5 = medium effect, 0.8 = large effect; Cohen, 1988) was computed for each analysis using the online Pychometrica software (option 11 in Computation of Effect Sizes; Lenhard & Lenhard, 2016).

To determine the relation between the number of orally produced word types during the 30-minute play session and the number of words in the child’s spontaneous orally produced expressive vocabulary as measured by the primary caregiver’s CDI report, a bootstrapped Spearman correlation (5000 iterations with seed 2000000) was computed. Bootstrapped Spearman correlations also were used to examine relations between the number of orally produced word types during the play session and three of the play-session speech variables (volubility, canonical babble frequency [CBF], consonant diversity) (two-tailed α = .05/3 = .017).

Results

Babble Maturity

Descriptive statistics for the three babble maturity measures for each group of participants are presented in Table 3, along with the results of the Mann-Whitney U tests comparing the performance of the Dup7 group to each of the other groups and the effect size for each analysis. As indicated in the table, the distributions of volubility scores and CBF scores for the Dup7 group were significantly lower (indicating significantly fewer vocalizations and significantly fewer canonical babbles per minute) than for either the TD group or the LSSD group but did not differ significantly from those of the LCAS group. In addition, the distribution of CBR scores for the Dup7 group was significantly lower than the distribution for the TD group.

Table 3.

Descriptive statistics for babble maturity measures and Mann-Whitney U results for comparisons of the Dup7 group to the LCAS, LSSD, and TD groups

Measure	Group	Mean	SD	Median	IQR	Range	z	p	Cohen’s d
Canonical Babble Ratio	Dup7	29.92	19.33	32.81	6.67–47.76	.00–48.75
	LCAS	36.70	13.67	33.48	26.29–50.31	20.07–58.75	−0.29	.813	0.13
	LSSD	51.66	12.33	56.09	38.58–60.30	34.07–60.38	−2.08	.042	2.21
	TD	61.07	13.50	63.95	49.31–70.03	39.95–79.19	−2.57	.008	3.00
Canonical Babble Frequency	Dup7	2.05	2.25	1.40	.20–3.03	.00–6.50
	LCAS	4.22	5.22	2.16	1.34–5.13	.89–16.41	−0.98	.364	0.49
	LSSD	17.68	17.13	11.51	6.43–35.11	4.78–42.93	−2.46	.012	1.61
	TD	11.25	6.25	8.60	7.37–15.14	7.01–23.40	−3.00	.001	2.04
Volubility	Dup7	3.83	3.14	3.97	1.00–5.73	.13–9.37
	LCAS	9.46	8.28	6.57	2.86–15.21	2.14–26.78	−1.66	.109	0.88
	LSSD	27.21	23.14	16.69	14.34–51.21	14.29–61.98	−2.65	.006	2.65
	TD	17.32	9.65	14.10	12.41–20.75	10.57–36.73	−3.00	.001	3.00

Note. Bolded p-values highlight significant differences in comparison to the Dup7 group. Dup7 = 7q11.23 duplication syndrome; LCAS = later diagnosed with childhood apraxia of speech; LSSD = later diagnosed with speech sound disorder other than childhood apraxia of speech; TD = typically developing; canonical babble ratio = number of canonical babbles divided by total number of syllables; canonical babble frequency = number of canonical babbles per minute; volubility = number of syllables per minute. n = 7 for Dup7, 10 for LCAS, 4 for LSSD, 6 for TD.

Phonetics

Descriptive statistics, Mann-Whitney U test results, and effect sizes for the four phonetic measures are presented in Table 4, separately for each participant group. As indicated in the table, the distribution of consonant frequency for the Dup7 group did not differ significantly from that of any of the comparison groups. However, the distribution of consonant diversity for the Dup7 group differed significantly from that of the TD group, and the distributions of both consonant place diversity and consonant manner diversity differed significantly from those of both the TD group and the LSSD group, with the diversities of the Dup7 group being more limited.

Table 4.

Descriptive statistics for phonetic measures and Mann-Whitney U results for comparisons of the Dup7 group to the LCAS, LSSD, and TD groups

Measure	Group	Mean	SD	Median	IQR	Range	z	p	Cohen’s d
Consonant frequency	Dup7	3.52	3.35	2.72	.54–6.20	.03–10.60
	LCAS	1.65	1.60	1.09	.27–2.42	.00–4.70	−1.17	.270	0.59
	LSSD	20.13	30.56	7.73	3.49–42.97	.49–74.33	−1.72	.099	0.99
	TD	11.13	10.26	5.49	3.22–21.87	2.59–25.89	−1.72	.099	0.99
Consonant diversity	Dup7	.36	.19	.39	.19-.47	.03-.67
	LCAS	.36	.30	.23	.14-.55	.00-.87	−0.29	.813	0.14
	LSSD	3.22	4.34	1.25	.54–6.89	.24–10.75	−2.21	.028	1.39
	TD	1.88	1.48	1.42	.76–3.24	.71–4.32	−3.07	<.001	2.58
Consonant place diversity	Dup7	.15	.06	.15	.10-.20	.03-.20
	LCAS	.25	.20	.19	.14-.52	.00-.52	−1.18	.270	0.59
	LSSD	1.71	2.17	.62	.28–3.70	.24–5.37	−3.10	<.001	2.58
	TD	.96	.69	.71	.56–1.49	.41–2.16	−3.10	<.001	2.58
Consonant manner diversity	Dup7	.12	.05	.13	.09-.13	.03-.20
	LCAS	.27	.20	.19	.14-.52	.00-.52	−2.20	.033	1.22
	LSSD	1.54	1.80	.62	.28–3.25	.24–4.48	−3.16	<.001	2.58
	TD	1.07	.93	.71	.56–1.76	.41–2.70	−3.17	<.001	2.58

Note. Bolded p-values highlight significant differences in comparison to the Dup7 group. Dup7 = 7q11.23 duplication syndrome; LCAS = later diagnosed with childhood apraxia of speech; LSSD = later diagnosed with speech sound disorder other than childhood apraxia of speech; TD = typically developing; consonant frequency = number of consonants per minute; consonant diversity = number of different consonants per minute; consonant place diversity = number of different place features per minute; consonant manner diversity = number of different manner features per minute. n = 10 for Dup7, 7 for LCAS, 5 for LSSD, 5 for TD.

Descriptive statistics, Mann-Whitney U test results, and effect sizes are reported in Table S1 for the seven consonant place features and in Table S2 for the seven manner features. As indicated in the tables, there were no significant differences in distributions between the Dup7 group and any of the comparison groups for any of the consonant place or manner features.

Phonotactics

Descriptive statistics, Mann-Whitney U test results, and effect sizes for phonotactic diversity (number of different vocalization structures produced at least once every 2 seconds) are presented in Table 5. As indicated in the table, the phonotactic diversity distribution for the Dup7 group was significantly lower than the distribution for the TD group but did not differ significantly from the distributions for the LSSD or LCAS groups.

Table 5.

Descriptive statistics for phonotactic diversity (number of different vocalization structures produced at least once every two minutes) and Mann-Whitney U results for comparisons of the Dup7 group to the LCAS, LSSD, and TD groups

Group	Mean	SD	Median	IQR	Range	z	p	Cohen’s d
Dup7	1.57	1.40	2.00	0.00–2.00	0.00–4.00
LCAS	0.86	0.90	1.00	0.00–2.00	0.00–2.00	−1.01	.313	0.53
LSSD	2.60	1.14	3.00	1.50–3.50	1.00–4.00	−1.33	.183	0.81
TD	4.20	0.84	4.00	3.50–5.00	3.00–5.00	−2.56	.010	2.12

Note. Bolded p-value highlights a significant difference in comparison to the Dup7 group. Dup7 = 7q11.23 duplication syndrome; LCAS = later diagnosed with childhood apraxia of speech; LSSD = later diagnosed with speech sound disorder other than childhood apraxia of speech; TD = typically developing; n = 10 for Dup7, 7 for LCAS, 5 for LSSD, 5 for TD.

Descriptive statistics, Mann-Whitney U test results, and effect sizes for each of the seven vocalization structures are reported in Table S3. As indicated in that table, there was only one significant difference in distributions between the Dup7 group and any of the comparison groups: For the CVCV structure, the distribution was significantly lower for the Dup7 group than for the TD group.

Lexicon

Lexical data were available only for the Dup7 group. Descriptive statistics for number of word types produced orally during the 30-minute play session and number of words from the CDI vocabulary checklist that the primary caregiver indicated the child spontaneously produced orally are reported in Table 6. The Spearman correlation between these two measures was rho = .88 (p = .001). Four toddlers with Dup7 did not produce any oral words during the play session. These findings are consistent with their caregivers’ responses on the CDI: One (aged 21 months) was reported not to produce any manual signs or oral words spontaneously; one (aged 14 months) was reported to produce three manual signs but no oral words spontaneously; the other two (aged 21 and 24 months) were reported to produce a small number of oral words spontaneously.

Table 6.

Oral expressive vocabulary of participants with 7q11.23 duplication syndrome as measured by play session transcripts and parent report (n = 10)

Measure	Mean	SD	Median	IQR	Range
Number of word types produced orally during 30-minute play session	3.30	4.06	3.00	0.00–4.50	0–13
Number of spontaneous word types produced orally based on parent report a	33.90	50.08	17.50	3.75–44.25	0–167

^aBased on primary caregivers’ responses on the MacArthur-Bates Communicative Development Inventory: Words and Sentences vocabulary checklist (Fenson et al., 2007).

To determine if the expressive vocabularies of the participants with Dup7 were delayed relative to CA expectations, the age-based percentile norms for the CDI (Fenson et al., 2007) were used. Two participants (aged 18 and 24 months) scored at the 25^th percentile, indicating that their expressive vocabulary was not delayed relative to normative expectations. The expressive vocabularies of the remaining participants were delayed: Four participants scored at the 5^th percentile and the remaining four participants scored below the 5^th percentile (the lowest percentile included in the CDI norms).

Spearman correlations between number of word types produced orally during the play session and three measures of babble maturity also were computed. All three correlations were statistically significant: for volubility, rho = .82 (p = .004); for CBF, rho = .82 (p = .003), and for consonant diversity, rho = .79 (p = .007).

Based on primary-caregiver report on the CDI, only one of the four 24-month-olds with Dup7 (25%) had begun to combine words spontaneously. In contrast, 86% of the 24-month-olds in the CDI norming sample produced word combinations. One 18-month-old with Dup7 also had begun to combine words. The two participants with Dup7 who were reported to be combining words both scored at the 25^th percentile for expressive vocabulary size on the age-based CDI norms.

Discussion

This study contributes to the sparse literature about early indicators of CAS, specifically focusing on toddlers with an increased risk for CAS due to a neurodevelopmental genetic syndrome. Our results show that measures of babble maturity and some measures of phonetics and phonotactics differentiated toddlers with Dup7 from toddlers with TD, and, to a lesser extent, from toddlers later diagnosed with a speech sound disorder other than CAS. Toddlers with Dup7 did not differ significantly from toddlers later diagnosed with CAS with respect to any of the early CAS indicators examined. Statistically significant relations were also shown between the orally produced vocabularies of toddlers with Dup7 during play sessions and measures of babble maturity and caregiver-reported orally produced vocabulary as measured by the CDI. In the remainder of the Discussion, we consider these findings in the context of prior research, address the theoretical and clinical implications of these findings, and consider the most important limitations of the study.

Our findings that toddlers with Dup7 have less mature babble, reduced volubility, and delayed expressive vocabulary compared to toddlers with TD are consistent with the literature on children later diagnosed with CAS. Studies that examined parent recollections of babbling in infants and toddlers later diagnosed with, or suspected to have, CAS report that these children babbled infrequently as infants and toddlers compared to toddlers with TD and sometimes to toddlers with other speech/language conditions (Aziz et al., 2010; Highman et al., 2008, 2013). Parents recalled that, as infants, their children with CAS had less varied phonetic repertoires (Aziz et al., 2010) and were less vocal and had delayed expressive language emergence (Highman et al. 2008, 2013). Based on direct examination, infants with later CAS also babbled and vocalized less relative to infants with TD or later SSD (Overby et al., 2020). Restricted or lack of babbling and vocalizations are suggested early features of CAS (Davis & Velleman, 2000; Maassen, 2002). Canonical babbling serves as a motoric precursor to language acquisition (McGillion et al., 2017; Oller et al., 1999). The presence of lower CBF and CBR, and their relations with delayed expressive vocabularies, in toddlers with Dup7 suggests they may play a role in expressive language acquisition. Furthermore, the strong positive correlation between play-session orally produced vocabulary size and caregiver-reported orally produced expressive vocabulary on the CDI vocabulary checklist validates the use of both strategies for relating babble features to orally produced expressive single-word vocabulary.

In concordance with Overby et al. (2019), we also found differences with respect to vocalization structure usage. In particular, toddlers with Dup7 (who are at considerable risk for CAS) produced less varied vocalization structures than either toddlers with TD or LSSD but did not differ significantly from toddlers with LCAS. At the same time, in contrast to Overby et al.’s (2019) finding that children with LCAS differed significantly from children with TD and children with LSSD on consonant frequency or frequency of certain features, we did not find significant differences between the toddlers with Dup7 and any of the other groups for these measures. These contrasting findings were likely due in part to our use of an adjusted p level. Nevertheless, our findings for consonant diversity and place and manner feature diversity were similar. While differing slightly, the collective findings from our study and Overby et al. (2019) highlight gaps in the consonant repertoires of toddlers with Dup7 or LCAS.

The findings of the present study suggest that the core deficit in CAS may be an inability to combine motor units (e.g., oral postures and articulatory gestures) and linguistic elements (e.g., phones, syllables, words) into larger wholes, as proposed by Velleman and Strand (1994). This hypothesis parallels a view of motor planning/programming and of speech and language as hierarchical processes of systematically embedding smaller components into larger ones that was emerging at the time that chapter was published and has since gained acceptance in the vast majority of both motor and linguistic theories. The fact that, from prelinguistic stages onward, toddlers with CAS demonstrate difficulties with producing syllable and larger structures (as demonstrated by deficits in canonical babbling and vocalization structures) as well as with producing a variety of consonants within those structures supports the perspective that CAS is a deficit of hierarchical speech planning/programming.

The present study has three important limitations. First, the speech sampling procedure for the Dup7 group (a single 30-minute adult-child play session) was quite different from the procedure for the other three groups (multiple relatively short home videos for each participant, recorded over a period of time). As a result, several variables differ with respect to the video recordings from the Dup7 group and TD, LSSD, and LCAS groups. These include the lengths of the recordings, how many videos were recorded and at what time intervals for toddlers with multiple video recordings, and the contexts and communication partners interacting with each toddler in each recording. On the one hand, the multiple short recordings in the Overby studies may be more robust to daily fluctuations in a child. Nevertheless, the conditions of the Dup7 recordings were far more consistent across children and less impacted by external factors such as varying numbers and types of interlocutors and varying amounts and types of environmental distraction. Given these differences, it is impressive that the present findings for the Dup7 group relative to the TD and LSSD groups are for the most part similar to those reported by Overby et al. (2019, 2020) for comparisons between the LCAS group and the TD and LSSD groups. It also is noteworthy that across the many analyses conducted as part of the present study, there was not even one significant difference between the Dup7 group and the LCAS group. This pattern is consistent with the strong association previously reported between Dup7 and CAS for school-age children. Second, although the small sample sizes in the present study are typical for studies involving phonetic transcription, especially studies of populations with rare conditions, they nevertheless resulted in reduced power to identify significant differences between groups. With larger sample sizes, additional significant differences may have been found, especially between the Dup7 group and the LSSD group, given that the latter group was smaller and had much higher within-group variability than the other three groups. This increased variability was likely due to the LSSD group being defined by exclusion (presence of a later diagnosis of any type of speech sound disorder other than CAS). Third, we currently do not know which of the participants with Dup7 eventually was diagnosed with CAS.

Despite these limitations, our preliminary findings highlight the importance of examining babble maturity, speech, phonology, and orally produced expressive vocabulary to identify potential risk for CAS at younger ages. Given the heterogeneity in CAS profiles, it will be important for clinicians and researchers to use these indicators to identify where breakdowns occur with respect to hierarchical speech planning/programming processes. Findings from continued research will inform the development and implementation of targeted early intervention approaches.

In conclusion, this study is the first to examine potential early indicators of CAS in young children with an increased risk for CAS due to a specific genetic syndrome. Measures of babble maturity, consonant diversity, and vocalization structure diversity were associated with genetic risk for CAS; they differentiated this small group of toddlers with Dup7 from toddlers with TD or LSSD, but not toddlers with LCAS. These early indicators are not only potentially useful for diagnosing CAS prior to 36 months of age, but also for understanding which symptoms are inherent versus less central to CAS.