Vocabulary Growth From 18 to 24 Months of Age in Children With and Without Repaired Cleft Palate

Abstract

Purpose

This study investigated vocabulary growth from 18 to 24 months of age in young children with repaired cleft palate (CP), children with otitis media, and typically developing (TD) children. In addition, the contributions of factors such as hearing level, middle ear status, size of consonant inventory, maternal education level, and gender to the development of expressive vocabulary were explored.

Method

Vocabulary size of 40 children with repaired CP, 29 children with otitis media, and 25 TD children was measured using the parent report on MacArthur–Bates Communicative Development Inventories: Words and Sentences (Fenson et al., 2007) at 18 and 24 months of age. All participants underwent sound field audiometry at 12 months of age and tympanometry at 18 months of age. A multiple linear regression with and without covariates was used to model vocabulary growth from 18 to 24 months of age across the 3 groups.

Results

Children with CP produced a significantly smaller number of words at 24 months of age and showed significantly slower rate of vocabulary growth from 18 to 24 months of age when compared to TD children (p < .05). Although middle ear status was found to predict vocabulary growth from 18 to 24 months of age across the 3 groups (p < .05), the confidence interval was large, suggesting the effect should be interpreted with caution.

Conclusions

Children with CP showed slower expressive vocabulary growth relative to their age-matched TD peers. Middle ear status may be associated with development of vocabulary skills for some children.

Early expressive vocabulary development has been characterized by two idiosyncratic features. First, there is high individual variability in the onset, size, and rate of vocabulary development (Bates, Bretherton, & Snyder, 1988; Bleses et al., 2008; Devescovi et al., 2005; Fenson et al., 2007; Ganger & Brent, 2004; Marjanovič-Umek, Fekonja-Peklaj, & Podlesek, 2013; Marjanovič-Umek, Fekonja-Peklaj, & Sočan, 2017; Rowe, Raudenbush, & Goldin-Meadow, 2012). Second, regardless of the ambient language the child is exposed to, the transition from prelinguistic vocalizations to canonical babbling and subsequent production of first words is gradual with no discrete landmarks (e.g., Ganger & Brent, 2004; Vihman, Macken, Miller, Simmons, & Miller, 1985).

Although identifying a robust profile of expressive vocabulary development is challenging due to the large individual variability, several longitudinal and cross-sectional studies have provided estimations of vocabulary size along the continuum of children's language development. Children with typical development may babble around 6 months of age, and their first words typically appear around 12 months. It has been reported that a typically developing (TD) child usually has a productive vocabulary of about 50 words by 18 months of age (e.g., Anglin, 1989; Bates et al., 1988; Rescorla, 1980) and about 300 words by 24 months of age (Anglin, 1989). Fenson, Dale, Reznick, Hartung, and Burgess (1990), however, reported a higher average of about 110 words at 18 months and 312 words at 24 months in TD toddlers. In addition, using the MacArthur–Bates Communicative Development Inventories (CDI; see Fenson et al., 1993) in a cross-sectional study on 1,803 middle-class children, Fenson et al. (1994) reported that TD children at the median produced fewer than 10 words, about 40 words, and 573 different words at 12, 16, and 30 months of age, respectively. As of September 2018, Wordbank, an open database of children's vocabulary growth measured on CDI, shows that English-speaking TD children at the median produce 90 and 308 words at 18 and 24 months of age, respectively.

The rate of early vocabulary development has been described as slow in the first few months after the emergence of first words followed by a period of accelerated growth (e.g., Fenson et al., 1994; Huttenlocher, Haight, Bryk, Seltzer, & Lyons, 1991). This transition has often been referred to as the “vocabulary spurt” or “vocabulary burst,” which usually occurs between 18 and 24 months of age (Bates & Goodman, 2001; Fenson et al., 1994; Fernald, Pinto, Swingley, Weinberg, & McRoberts, 1998; Goldfield & Reznick, 1990; Kauschke & Hofmeister, 2002). Research has documented enormous cognitive gains, such as comprehension (e.g., Reznick & Goldfield, 1992), naming (e.g., Dore, Franklin, Miller, & Ramer, 1976; Kamhi, 1986), categorization (e.g., Gopnik & Meltzoff, 1987, 1992; Poulin-Dubois, Graham, & Sippola, 1995), and conceptual knowledge of objects (Gopnik & Meltzoff, 1987; Lifter & Bloom, 1989), around the time of vocabulary spurt. In addition, vocabulary spurt coincides with the emergence of two-word combinations (Frankenburg & Dodds, 1967; Nelson, 1973; Stoel-Gammon, 1998a) that are marked as the onset of syntactic development. There is a bidirectional bootstrapping between lexical and syntactic knowledge during the early stage of language acquisition. On the one hand, the child's lexicon needs to reach the critical mass threshold necessary to trigger grammatical development (Bates & Goodman, 1997, 2001; Marchman & Bates, 1994), and on the other hand, word combinations and syntactic relationships between words trigger lexical development as children become more cognizant about concepts and the demand for conveying concepts with greater specificity (Anisfeld, Rosenberg, Hoberman, & Gasparini, 1998).

Despite the abundance of research on the developmental course of expressive vocabulary, our understanding of the patterns of vocabulary growth is not yet complete. Investigations of lexical development have identified a number of cognitive, physiological, prelinguistic, and sociodemographic factors that singly and in combination influence the onset and rate of vocabulary growth in toddlers. For example, normal development of various cognitive skills, including lexical memory (Woodward, Markman, & Fitzsimmons, 1994), categorization and naming (Gopnik & Meltzoff, 1987; Poulin-Dubois et al., 1995), and the ability to associate objects with words (Woodward & Hoyne, 1999), is significant as they have been identified as essential cognitive precursors of lexical development. In addition, physiological factors such as competent anatomical structure and normal hearing are essential for making articulatory movements and oral–aural adjustments to produce sounds that comprise words. Research has shown that children with cognitive disabilities, abnormal anatomical structure, or disrupted hearing are at risk of developing poor vocabulary skills (Bates et al., 1997; Estrem & Broen, 1989; McGregor et al., 2012; Rvachew, Slawinski, Williams, & Green, 1999; Scherer & D'Antonio, 1995; Singer Harris, Bellugi, Bates, Jones, & Rossen, 1997). More specifically, young children with repaired cleft palate (CP) have been observed to have a smaller size of vocabulary (e.g., Estrem & Broen, 1989; Park & Ha, 2016; Willadsen, 2013). The literature on expressive vocabulary skills of children with CP will be discussed further in the next section.

Various developmental studies also suggested continuity from prespeech vocalizations to speech sounds and consequently to first word productions (Locke, 1989; Stoel-Gammon, 1989, 1998b; Vihman & McCune, 1994). Research has demonstrated a positive association between consonant inventories that occur frequently during canonical babbling and those that typically appear in early word productions of young children (e.g., McCune & Vihman, 2001; Oller, Wieman, Doyle, & Ross, 1976; Stoel-Gammon & Cooper, 1984; Vihman et al., 1985). For example, McCune and Vihman (2001) studied the consonant–lexical relationship in 20 children with typical development from 9 to 16 months of age. They reported that the number of specific consonants produced consistently across the months of observation predicted referential lexical use at 16 months of age. They defined referential words as words that a child was observed to produce at least in two different contexts and/or in relation to at least two different objects, where consistent aspects of object or context were identifiable. The reader is referred to Stoel-Gammon (2011) for a detailed explanation regarding early association between phonological and lexical development. Therefore, due to interdependency between early phonological skills and later lexical development, any pathological condition that disrupts normal development of prespeech vocalizations and canonical babble is likely to jeopardize the development of other aspects of language, in general, and expressive vocabulary, in particular. Empirical evidence in support of continuity between phonological and lexical development can also be found in the literature on children with CP. For example, a study conducted by Chapman, Hardin-Jones, and Halter (2003) showed that the size of the true consonant inventory produced by children with cleft lip and palate (CLP) at 13 months of age (postsurgery) was significantly correlated to the number of different words produced in 10 min of caregiver–child interaction at 21 months of age. Additionally, this study showed that children with CLP demonstrated a significant correlation between the percentage of true stop production at 13 months of age and the number of words on the CDI parent report checklist at 21 months of age.

Furthermore, the effect of sociodemographic factors in the development of expressive language in healthy children has been well documented. Various studies have reported evidence for a significant association between family socioeconomic status (SES) and both onset and rate of vocabulary growth (Arriaga, Fenson, Cronan, & Pethick, 1998; Hoff, 2003; Pan, Rowe, Singer, & Snow, 2005). These studies suggest that children from families lower in SES were exposed to significantly fewer words compared to children from families higher in SES and subsequently experienced slower rates of vocabulary growth. A strong association has also been reported between maternal educational levels (MELs) and early language skills, such as expressive vocabulary, comprehension, and grammatical skills (Dollaghan et al., 1999; Magnuson, Sexton, Davis-Kean, & Huston, 2009; Marjanovič-Umek, Fekonja, Kranjc, & Bajc, 2008; Marjanovič-Umek et al., 2016).

Gender has also been shown to be an influential factor in early expressive language development. Research has established that, compared to boys, girls tend to have a progressively more rapid vocabulary growth (Bauer, Goldfield, & Reznick, 2002; Bornstein, Hahn, & Haynes, 2004; Hadley, Rispoli, Fitzgerald, & Bahnsen, 2011; Huttenlocher et al., 1991) and a larger vocabulary size (Eriksson et al., 2012; Marjanovič-Umek et al., 2008; Marjanovič-Umek, Fekonja-Peklaj, & Podlesek, 2012). Contrary to these findings, Marjanovič-Umek et al. (2016) did not find significant differences between boys' and girls' vocabulary size at different ages, but they reported differences between the two genders in terms of the type of vocabulary (i.e., masculine-type words vs. feminine-type words). In addition, they reported that, although toddlers' gender and parental education did not have a significant effect on vocabulary size at different ages, gender had a significant effect on the shape of the vocabulary growth curve, with the average curve being slightly closer to linear for girls than boys.

Vocabulary Growth in Children With CP

Early expressive vocabulary skills in children with CP have been characterized as smaller in size (Eshghi, Dorry, et al., 2017; Estrem & Broen, 1989; Park & Ha, 2016; Willadsen, 2013) and delayed (Scherer & D'Antonio, 1995; Scherer, Williams, & Proctor-Williams, 2008). Hardin-Jones and Chapman (2014) also reported that, while children with CP exhibited similar expressive vocabulary skills to that of TD children at 13 months of age, they showed a significantly reduced size of their lexicons at 21 and 27 months of age. Additionally, in a study conducted by Hariharan, Raghunathan, Sreedevi, and Ramanan (2017), size of expressive vocabulary (obtained using Tamil translation of MacArthur–Bates Communicative Development Inventories) and the expressive language age (obtained using the Three-Dimensional Language Acquisition Test; Herlekar & Karanth, 1993) were explored in Tamil-speaking children with and without CLP around 30 months of age. They reported that, while young children with repaired CLP exhibited a 6-month delay on the expressive language age compared to healthy controls (i.e., median age of 23.50 months for the CLP group and 30 months for the healthy controls), no significant differences were observed between the two groups for the size of expressive vocabulary. It should be noted, however, that the sample sizes of both groups were small (14 children with CLP and seven healthy controls) and the distribution of gender was unbalanced as children with CLP were predominantly boys.

Less advanced expressive vocabulary skills in children with CP may be due to multiple physiological and environmental factors. Prior to surgical repair, children with CP produce a restricted number of stop consonants and fewer multisyllabic constructions (Chapman, 1991). At prelinguistic stage, infants with unrepaired CP were observed to produce smaller canonical babbling ratios compared to infants without CP (Chapman, Hardin-Jones, Schulte, & Halter, 2001). In their study, Chapman et al. (2001) reported that, while 93% of infants without CP reached the canonical babbling stage by 9 months of age, only 57% of infants with unrepaired CP were able to reach that stage. Because adequate velopharyngeal (VP) function may not occur in up to 30% of children after palatal surgery (Inman, Thomas, Hodgkinson, & Reid, 2005; Webb, Watts, Read-Ward, Hodgkins, & Markus, 2001), ongoing VP dysfunction may also lead to speech and language complications. In a preliminary study using nasal ram pressure (NRP), Eshghi, Vallino, Baylis, Preisser, and Zajac (2017) reported that, while TD children exhibited consistent VP closure at 12 months of age, some children with CP did not achieve consistent VP closure until approximately 3–4 months following palate repair at around 14 months of age. Delayed achievement of consistent VP closure in children with CP may negatively impact early development of speech sounds and subsequent vocabulary. To be sure, Broen, Devers, Doyle, Prouty, and Moller (1998) reported that VP status was a significant factor that contributed to a slower rate of word acquisition in young children with repaired CP.

In addition, children with CP are highly likely to experience mild to moderate degrees of fluctuating conductive hearing loss. Occurrence of otitis media (OM) in babies with CP is common due to malformation of the tensor veli palatini muscles, which are responsible for the opening and closing of the eustachian tubes. Because 26%–55% of children with OM have mild to moderate hearing loss in the speech frequency range (Bess, 1983), fluctuating conductive hearing loss may lead to reduced auditory input for speech processing and acoustic–articulatory matching. Indeed, studies have reported that children with recurring OM demonstrated changes in (a) speech perception abilities (Brandes & Ehinger, 1981; Clarkson, Eimas, & Marean, 1989; Gravel & Wallace, 1992), (b) auditory processing tasks such as identification (Sandeep & Jayaram, 2008), (c) discrimination tasks (Clarkson et al., 1989; Gravel & Wallace, 1992; Haapala et al., 2014; Welsh, Welsh, & Healy, 1983), and (d) tasks of central auditory processing (Moore, Hutchings, & Meyer, 1991; Pillsbury, Grose, & Hall, 1991; Welsh et al., 1983). At the language level, significant associations have been reported between early history of OM and later speech and expressive language development (Donahue, 1993; Friel-Patti & Finitzo, 1990; Luloff, Menyuk, & Teele, 1993; Pearce, Saunders, Creighton, & Sauve, 1988; Petinou, Richard, Maria, & Judith, 1999; Rvachew et al., 1999; Shriberg, Flipsen, et al., 2000; Shriberg, Friel-Patti, Flipsen, & Brown, 2000; Silva, Kirkland, Simpson, Stewart, & Williams, 1982; Teele, Klein, & Rosner, 1984). Despite the conceptual logic of the hypothesized relationship between early OM and later language sequelae, the increased risk for speech and language disorders in children with a history of OM has been a matter of debate in the literature. Some studies showed normal auditory processing and perceptual skills in children with early OM. For example, children who had OM in the first 3 years of life were reported to show later normal perception of words (Gravel et al., 2006; Mody, Schwartz, Gravel, & Ruben, 1999) and sentence in noise (Keogh et al., 2005). In addition, similar expressive language (Black et al., 1988; Rovers et al., 2000; Wright et al., 1988) and vocabulary skills (Fischler, Todd, & Feldman, 1985; Lous, Fiellau-Nikolajsen, & Jeppesen, 1988; Wright et al., 1988) were reported in children with OM compared to children without a history of OM.

Another group of studies, however, reported a short-term detrimental effect of early-life recurrent OM on speech and language abilities that tend to disappear at school age (Roberts, Burchinal, & Zeisel, 2002; Schilder et al., 1993; Zumach, Gerrits, Chenault, & Anteunis, 2010). For example, Roberts et al. (2002) reported that children with more episodes of OM during the first 2 years of life scored lower in expressive language during the preschool and early elementary school years, but they caught up by second grade. They also reported that, between home environment and OM, the former factor influenced language outcomes to a greater extent compared to the hearing loss associated with OM. Children from homes that were rated as more stimulating and responsive scored higher on measures of expressive language than did children from less responsive homes. Similarly, in a study conducted by Roberts et al. (1998), OM showed a modest correlation to measures of language (including expressive language and vocabulary acquisition) and cognitive development at 2 years of age, but these direct relationships dissipated upon adding the home and child care environments in the analyses. In other words, the quality of home and child care environments was a stronger predictor of children's language and cognitive development than was OM and associated hearing loss.

It should be noted, however, that variables such as hearing level, age at onset of OM, duration, and number of episodes of middle ear effusion play important roles in OM-related speech outcomes. In fact, these factors might have, to some extent, caused the equivocality of the literature in terms of the effect of OM on speech and language development. Shriberg, Friel-Patti, et al. (2000) reported that hearing levels at 12–18 months were significantly associated with speech delay and low language outcomes at 3 years of age. The risk for speech delay at 3 years of age was 2% for children with less than 20 dB of average hearing levels at 12–18 months. However, children were at a substantially higher risk of speech delay (i.e., 33%) if they displayed greater than 20 dB of average hearing levels at 12–18 months. Similar findings can be found in Miccio, Gallagher, Grossman, Yont, and Vernon-Feagans (2001) in which children with OM and hearing levels below 20 dB demonstrated normal patterns of phonological acquisition. Of interest, a child with low occurrence of OM but the greatest hearing loss (26.4 dB) showed atypical phonological development and language delays. In addition, various studies have found that occurrence of OM at an earlier age (Rvachew & Slawinski, 1995; Rvachew, Slawinski, Williams, & Green, 1996; Rvachew et al., 1999) and a greater number of episodes of middle ear effusion increase the risk of delayed speech and language outcomes (Teele et al., 1984). For example, Rvachew and Slawinski (1995) reported that the mean canonical babble ratio was smaller for children who experienced their first episode of OM at or before the age of 6 months in comparison with those who did not experience OM before 9 months of age. Rvachew et al. (1999) also reported that infants with early-onset OM had consistently smaller canonical syllable ratios throughout the period of 6–18 months of age, in comparison with children who had no ear infections before 6 months of age. Perceptual data also indicate that children who had multiple episodes of OM or for a longer duration had significantly lower speech identification scores for spectrally distorted stimuli than children who had a single episode of OM or OM for less than 1 month in duration (Sandeep & Jayaram, 2008). Regarding the duration of OM, Teele et al. (1984) reported that 3-year old children with prolonged periods of middle ear effusion (at least three episodes of OM with effusion) by 2 years of age significantly scored lower on the Peabody Picture Vocabulary Test (Dunn & Dunn, 1981) compared to their peers who had no or short history of OM (no more than one observation of middle ear effusion) by 2 years of age.

Because children with CP are prone to have multiple episodes of OM during the first year of their life, they may be at a higher risk of developing delayed expressive language skills due to OM. Consistent with the group of studies that suggested pathological consequences of OM on speech and language skills, Jones, Chapman, and Hardin-Jones (2003) reported that children with CP who had a higher number of reported episodes of OM and failed tympanometry pre- or postsurgery showed the tendency to make less progress in the size of the consonant inventory after palate repair. Jocelyn, Penko, and Rode (1996) also reported an association between lower scores of expressive language abilities and higher frequency of middle ear disease and ventilation tubes observed in children with CLP compared to children without CLP. Furthermore, in a longitudinal study conducted by Broen et al. (1998), toddlers with CP exhibited a slower rate of vocabulary acquisition compared to age-matched peers. These investigators reported that poorer hearing, in addition to VP inadequacy, contributed to the findings. In contrast to these studies, in a longitudinal study conducted by Chapman et al. (2003), neither abnormal tympanometry nor having a history of OM (at least four episodes of OM) accounted for the differences observed between children with and without CP for measures of speech and language at 21 months of age.

Finally, another source of complexity in the study of expressive language skills in children with CP pertains to cleft type. The literature has shown uncertainty and equivocality whether speech and language are more apt to be delayed in children with CLP or in children with cleft of palate only (CPO). For example, Hardin-Jones, Chapman, and Schulte (2003) did not find any statistically significant association between cleft type and early frequency of vocalization and the size of consonant inventory in infants with CLP and CPO. However, they reported that infants with CPO showed a trend to produce more consonants with anterior place of articulation. Similarly, Lohmander-Agerskov, Söderpalm, Friede, Persson, and Lilja (1994) demonstrated a correlation between cleft type and place of articulation. They reported that sounds with anterior place of articulations occurred predominately among TD children and children with CPO, whereas children with CLP avoided consonants with anterior place of articulation during early speech production, perhaps due to absent or altered articulatory surface. On the contrary, in a study conducted by Scherer et al. (2000), children with isolated CP scored lower than children with CLP on measures of receptive language as well as expressive language and vocabulary. Although the reason for this finding is not clear, it may be related to the fact that isolated CP is more likely to be associated with syndromes and cognitive deficits. However, Broen et al. (1998) reported that cleft type did not account for the lower performance of children with CP on verbal measures of the Bayley Scales of Infant and Toddler Development (Bayley, 1969), specifically the Receptive Language subscale.

Purpose and Significance of the Study

The purpose of this study was to further investigate the development of expressive vocabulary in children with repaired CP, children with OM, and TD children at 18 and 24 months of age. This period was selected because children's lexicon undergoes an exponential increase from 18 to 24 months of age. Children with OM were included given the evidence—admittedly equivocal—that fluctuating conductive hearing loss may contribute to early language delays. In addition, the effect of variables such as hearing level (at 12 months of age), middle ear status (at 18 months of age), size of consonant inventory (at 18 months of age), MEL, and gender on vocabulary growth from 18 to 24 months of age was examined. The following research questions were asked: (a) How does the expressive vocabulary change from 18 to 24 months of age in young children with CP compared to children with OM and TD children? (b) To what extent do variables such as hearing level, middle ear status, size of consonant inventory, MEL, and gender influence the vocabulary growth from 18 to 24 months of age? It was hypothesized (a) that children with CP would be behind the OM and TD groups with regard to vocabulary growth and children with OM would fall in between and (b) that factors such as hearing, middle ear status, size of consonant inventory, MEL, and gender may account for the hypothesized differences among the three groups. In a secondary data analysis, we were further interested in comparing the vocabulary growth of subgroups of children with CP based on cleft type to the performance of children without cleft.

This study has other theoretical and clinical significance. Findings of the study may provide additional information regarding the pattern of vocabulary growth in children with CP while controlling for the effects of middle ear and hearing status. To our knowledge, this is the first prospective study investigating children with CP while including a control group of children with OM. Because OM occurs frequently in children with CP, incorporating a comparison group of children with OM but without CP would not only increase our understanding of the influence of OM on early expressive vocabulary skills in children without CP but also show the extent to which OM affects early speech and language development in children with CP. In addition, this study may further highlight the importance of early speech and language monitoring to identify children with CP who need early speech and language intervention. Given that impaired expressive language has been shown to negatively impact higher level language and literacy attainment in later childhood (e.g., Felsenfeld, Broen, & McGue, 1994; Lewis et al., 2006; Scarborough, 1990; Sices, Taylor, Freebairn, Hansen, & Lewis, 2007), children with CP may be at risk of developing less advanced skills in other language domains. For example, less syntactically complex utterances (Morris, 1962), a higher likelihood of school underachievement (Broder, Richman, & Matheson, 1998; Knight, Cassell, Meyer, & Strauss, 2015; Wehby et al., 2014), and poorer reading skills (Chapman, 2011; Richman, Eliason, & Lindgren, 1988) have been reported in children with CP compared to controls. Therefore, early assessment of expressive language with subsequent management could prevent persistent language delay and later literacy underachievement.

Method

Participants

Participants of this study were a subset of children recruited for an ongoing multisite longitudinal study that aimed to investigate the development of stop consonants in young children with repaired CP, children with OM, and TD children between 12 and 24 months of age. Inclusion criteria for all participants (CP, OM, and TD) were (a) at least 36 weeks of gestation, (b) from monolingual American English–speaking families, (c) good general health, (d) no documented sensorineural hearing loss, and (e) no documented global developmental delays. All children exhibited normal symbolic language skills (i.e., language comprehension and object use) when enrolled in the study as determined by the Communication and Symbolic Behavior Scales Developmental Profile (CSBS DP; Wetherby & Prizant, 2002; described below). Inclusion criteria for participants with CP were (a) repaired CP with or without cleft lip by 11 months of age, (b) no oronasal fistulae, and (c) no known syndromes including Pierre Robin sequence. The inclusion criterion for participants with OM was multiple (three or more) episodes of OM with effusion during the first year of life. All participants with OM had undergone bilateral myringotomies with the insertion of pressure equalization (PE) tubes prior to enrollment in the project by 12 months of age. The inclusion criteria for TD participants were normal speech and language, as determined by the CSBS DP at enrollment in the study. For the larger project, 104 children with CP were eligible for the study, and 63 (61%) decided to participate. There were 81 children with OM who were eligible, and 34 (42%) decided to participate. Lack of time, distance, and number of study visits were the main reasons not to join the study in the order presented.

At the time of the current study, data for expressive vocabulary skills at 18 and 24 months of age were available for 94 children: 40 children with repaired CP (19 boys, 21 girls), 29 children with OM (20 boys, nine girls), and 25 TD children (13 boys, 12 girls). Twenty-seven of the children (12 CP, six OM, and nine TD) were recruited at the Craniofacial Center, University of North Carolina at Chapel Hill; 36 children (11 CP, 15 OM, and 10 TD) were recruited at Alfred I. duPont Hospital for Children, Wilmington, Delaware; and 31 children (17 CP, eight OM, and six TD) were recruited at the Nationwide Children's Hospital, Columbus, Ohio.

Information about birth and family history, demographics, medical history, surgical history, and cleft classification was obtained from the primary caregiver of the child at a screening visit when the child was initially enrolled in the study. Cleft type and date of palate surgery were verified from medical records. Of the 40 children with repaired CP, 37 children were non-Hispanic/Latino and three children were Hispanic/Latino. In addition, 31 children with CP were Caucasian, two were African American, and seven reported more than one race. Of the 40 children with repaired CP, 15 had CLP (13 boys, two girls) and 25 had cleft of the hard and/or soft palate only (CPO; six boys, 19 girls). All children with CP, either CLP or CPO, had undergone a single surgery to repair the palate by at least 11 months of age (M = 10.3 months, SD = 1.4). Objective assessment of VP status in children with CP was performed at 18 months of age using NRP monitoring (see Eshghi, Vallino, et al., 2017, for details of the procedure). In the current study, NRP data were obtained on 24 of the 40 children with CP (60%). The reason for missing data was due to either a child not tolerating the nasal cannula or a child tolerating the cannula but not producing any stops. Sixteen of the 24 children (67%) exhibited consistent VP closure, while eight (33%) exhibited inconsistent VP closure. Consistent VP closure was defined as achieving complete closure on at least 85% of stops produced by a child during NRP monitoring. This cutoff was selected as all of the noncleft children investigated by Eshghi, Vallino, et al. (2017) exhibited VP closure on at least 85% of stops produced. Oral examinations were performed at both 18 and 24 months to rule out the presence of oronasal fistulae. All children in the CP group had also undergone bilateral myringotomies with insertion of PE tubes at the time of their palate surgery.

All children in the OM group had experienced multiple (three or more) episodes of OM with effusion during their first year of life. All had undergone bilateral myringotomies with insertion of PE tubes prior to enrollment in the project by 12 months of age. None of these children had any type of clefts. Twenty-eight children with OM were non-Hispanic/Latino, and one child was Hispanic/Latino. In addition, 27 children with OM were Caucasian, one was African American, and one reported more than one race.

Finally, none of the children in the TD group had any type of cleft or reported history of OM. All TD children showed normal language development at 12 months of age as part of the larger study, as determined by the CSBS DP. In addition, none received any early speech-language or developmental intervention services. Twenty-three TD children were non-Hispanic/Latino, and two were Hispanic/Latino. In addition, 20 TD children were Caucasian, two were African American, and three reported more than one race. The study was approved by the institutional review boards of the respective sites (University of North Carolina at Chapel Hill, Alfred I. duPont Hospital for Children, Wilmington, Delaware, and Nationwide Children's Hospital, Columbus, Ohio).

Evaluation of Middle Ear Function

As part of the larger study, bilateral tympanograms were obtained at 12, 14, 18, and 24 months of age for all children to assess the status of the middle ear. Tympanograms were obtained by either speech-language pathologists (SLPs) and/or certified audiologists at each of the sites. Normal middle ear status was defined as either a Type A tympanogram (normal compliance and pressure) or a flat Type B tympanogram with a large volume (greater than 1.0 ml when a PE tube was present) for both ears. Abnormal middle ear status was defined as either a Type A_s tympanogram with reduced compliance (peak less than 0.2 ml), a flat Type B tympanogram with a small volume (less than 1.0 ml), or a Type C tympanogram with negative pressure (less than −250 daPa) for either ear. Middle ear status, as determined at 18 months of age, was used as a covariate.

Evaluation of Hearing

Sound field hearing screenings were obtained at 12 and 24 months of age for all children in sound-attenuated booths by licensed audiologists using standard pediatric assessment protocols (i.e., warble tones and visual reinforcement). The majority of the screenings (approximately 82%) were done at 20 dB HL at 500, 1000, 2000, and 4000 Hz; in the remaining cases, screenings were done at 15 dB HL. This occurred because some participants were evaluated outside the larger study as part of their routine standard of care. If a child did not respond at the screening level, then hearing thresholds were obtained. A hearing level for each child was then calculated by averaging across the four frequencies tested. All four frequencies were obtained for approximately 76% of the children. Mean hearing levels were based on three, two, and one frequency for approximately 7%, 7%, and 10% of the children, respectively. Mean hearing levels obtained at 12 months of age were used as a covariate.

Estimation of the Size of Consonant Inventory

The size of consonant inventories of all children was estimated by direct observation through administration of the CSBS DP (Wetherby & Prizant, 2002) at 18 months of age. The CSBS DP consists of six semistructured communication and play opportunities to sample various speech and language skills. Personnel at each site were trained in administration of the CSBS DP. The participant, caregiver, and examiner were video-recorded during administration of the CSBS DP. The videos were scored at the University of North Carolina at Chapel Hill by three trained SLPs who demonstrated good reliability (described below). Consonants including the stops /p/, /b/, /t/, /d/, /k/, and /g/; the fricatives /s/ and /ʃ/; the nasals /m/ and /n/; the glides /w/ and /j/; and the liquid /l/ were identified per instructions in the CSBS DP scoring manual. It should be noted that the CSBS DP does not distinguish between voiced/voiceless stop cognates. This may be due to the fact that the voice onset time dichotomy of short timing lag for voiced stops and long timing lag for voiceless stops are not adultlike at 2 years of age (Zlatin & Koenigsknecht, 1976). Per the CSBS DP manual, a consonant was counted if the child used it with communicative intent at least once during the assessment procedure. The CSBS DP determines the sound weighted raw score as the number of sampling opportunities with a consonant plus the number of different consonants times two. The maximum number of sampling opportunities is 6, and the maximum number of consonants is 10 (the cognate stops are not distinguished). The Constant 2 is a weighing factor used to make point values across subscales of the CSBS DP comparable. The sound weighed raw score was used as an index of the consonant inventory.

Assessment of Expressive Vocabulary

The MacArthur–Bates Communicative Development Inventories: Words and Sentences form (CDI:WS; Fenson et al., 2007) was used to measure the expressive vocabulary skills of children in the three groups. The fidelity of the CDI:WS form has been verified by various studies (e.g., Dale, Bates, Reznick, & Morisset, 1989; Fenson et al., 1994; Heilmann, Weismer, Evans, & Hollar, 2005). It has also been shown to be a valid assessment tool to measure expressive language development in children with CP as compared to comprehensive speech and language evaluations (Scherer & D'Antonio, 1995). Parents completed the CDI:WS at home when their child was 18 months of age (M = 17.8 months, SD = 0.5, range: 17–19 months) and 24 months of age (M = 23.7 months, SD = 0.6, range: 23–26 months). Parents were given written instructions to complete the CDI:WS. They were instructed to mark words only if they had heard the child produce them spontaneously and not as a result of prompting the child to say a word. The caregivers who completed the CDI:WS form were primarily mothers (i.e., 97% for 18 months and 94% for 24 months).

The CDI:WS consists of two parts. The first part is a vocabulary checklist of 680 words, and the second part samples sentences and grammar. For this study, the first part of the CDI:WS inventory was used to measure the lexical (i.e., expressive vocabulary) development of the participants. The number of words was derived from a checklist of 680 commonly used vocabulary items in 22 categories: 12 words for onomatopoeic words (i.e., sound effects and animal sounds), 43 words for animals, 14 words for vehicles, 18 words for toys, 68 words for food and drink, 28 words for clothing, 27 words for body parts, 50 words for small household items, 33 words for furniture and rooms, 31 words for outside things, 22 words for places, 29 words for people, 25 words for games and routines, 103 action words, 63 descriptive words, 12 time expressions, 25 pronouns, seven question words, 26 prepositions and locations, 17 quantifiers and articles, 21 helping verbs, and six conjunctions.

Reliability of Measurements

Inter- and intrarater reliabilities of the sound weighted raw scores determined by the three SLPs were assessed by randomly selecting 20% of participants and repeating the scoring. Percentages of inter- and intrarater reliabilities were calculated by dividing the smaller score by the larger score obtained by the raters and multiplying by 100. All three scorers achieved at least 83% or higher inter- and intrareliability. In addition, exact agreements were determined relative to the specific consonants used by a participant during the CSBS DP. Inter- and intrarater exact agreements were at least 80% for the three SLPs. Relative to the number of words reported by caregivers on the CDI:WS vocabulary checklist, the number of words was counted by two different scorers, and discrepancies (less than 5%) were resolved.

Statistical Analysis

Descriptive analyses were performed to obtain mean estimates and corresponding standard errors for the number of words across the three groups. The absolute change in the number of words produced by children from 18 to 24 months of age was calculated to answer the research questions. One child with CP was observed to produce fewer words at 24 months than 18 months of age. In this case, the negative change in the number of words within the 6-month period was set to zero. The multiple linear regression based on complete cases was used to model the change in the number of words produced by children from 18 to 24 months of age across the three groups (CP, OM, and TD). After fitting a preliminary model without covariates, mean hearing level (mean-centered at 21.5 dB with linear and quadratic terms to allow for a nonlinear relationship with the number of words), middle ear status (abnormal vs. normal), sound weighted raw score (mean-centered at 13.8), MEL (holding education below bachelor's degree vs. bachelor's degree and above), and gender (male vs. female) were added to the statistical models. All statistical analyses were run in SAS Version 9.4 (TS1M1, SAS Institute), and p values were compared against the .05 significance level selected a priori.

Results

Descriptive statistics including mean estimates for the number of words at 18 and 24 months of age, mean estimates of vocabulary growth from 18 to 24 months of age, and the corresponding standard errors are presented in Table 1. On average, children with CP produced 22 words less than TD children at 18 months of age. However, they lagged behind TD children by approximately 133 words at 24 months of age. Examination of the data revealed that all children with CP produced less than 425 words at 24 months of age except for one child who was reported to produce 653 words at that age. The mean absolute changes from 18 to 24 months of age were 158 and 267 words in children with CP and TD children, respectively. Compared to TD children, children with OM were observed to produce a similar number of words at 18 months of age but, on average, 63 fewer words at 24 months of age.

Table 1.

Means and standard errors for the number of words produced by children with cleft palate (CP), children with otitis media (OM), and typically developing (TD) children at 18 and 24 months of age as well as the 6-month vocabulary growth from 18 to 24 months of age.

Cohort	n	18 months		24 months		Change in number of words
		M (SE)	Mdn (NIQR)	M (SE)	Mdn (NIQR)	M (SE)	Mdn (NIQR)
CP	40	38.4 (5.6)	26.0 (20.3)	195.1 (23.0)	196.5 (159.0)	157.5 (20.4)	147.0 (139.1)
OM	29	57.4 (7.9)	48.0 (21.8)	261.6 (27.7)	242.0 (169.5)	204.2 (21.6)	203.0 (151.5)
TD	25	60.8 (9.8)	44.0 (36.8)	327.6 (33.2)	352.0 (182.3)	266.8 (29.0)	271.0 (146.3)

Note. NIQR = normalized interquartile range = 0.75 × (Q3 − Q1), where Q1 and Q3 are the first and third quartiles, respectively.

Table 2 presents descriptive statistics for the number of words produced by children in the CP, OM, and TD groups at 18 and 24 months of age and the absolute change in the number of words from 18 to 24 months of age while taking the categorical covariates of middle ear status, MEL, and gender into account. The proportions of children in the CP, OM, and TD groups with abnormal middle ear status were 28%, 14%, and 28%, respectively. The relatively large proportion of TD children with abnormal tympanograms (i.e., Type C) was due to testing during the winter months when colds were present. In addition, 48% of children with CP, 69% of children with OM, and 52% of TD children were male. Finally, while only 8% of TD children had mothers with education below bachelor's degree, 53% and 21% of children with CP and OM, respectively, had mothers with education below that level. As seen in the table, for every level of tympanogram, gender, and MEL, respectively, the mean change in number of words was highest for the TD group and lowest for the CP group, with the OM group having mean number of words between that of the TD and CP groups.

Table 2.

Descriptive statistics for the number of words produced by children with cleft palate (CP), children with otitis media (OM), and typically developing (TD) children while categorizing children based on middle ear status (tympanogram), gender, and maternal education level (MEL).

Variables		n	18 months		24 months		Change in number of words
			M (SE)	Mdn	M (SE)	Mdn	M (SE)	Mdn
Tymp a
Normal	CP	28	37.1 (6.6)	25.5	195.2 (28.7)	200.0	159.3 (26.2)	146.5
	OM	22	62.3 (9.3)	49.5	287.0 (31.2)	270.5	224.6 (23.9)	223.0
	TD	13	53.9 (9.8)	44.0	346.8 (36.5)	353.0	292.9 (29.8)	286.0
Abnormal	CP	11	41.7 (11.9)	26.0	179.7 (39.9)	157.0	138.0 (30.1)	138.0
	OM	4	35.0 (9.3)	35.5	196.0 (70.4)	171.0	161.0 (65.3)	126.0
	TD	7	88.0 (27.6)	81.0	262.9 (76.8)	155.0	174.9 (55.2)	125.0
Gender
Male	CP	19	32.1 (7.3)	22.0	174.4 (37.3)	114.0	142.3 (34.3)	92.0
	OM	20	55.1 (10.1)	44.0	260.5 (35.9)	251.0	205.4 (27.7)	207.0
	TD	13	67.2 (16.1)	48.0	343.5 (50.0)	359.0	276.3 (43.0)	271.0
Female	CP	21	44.1 (8.3)	37.0	213.8 (28.2)	227.0	171.2 (23.9)	174.0
	OM	9	62.7 (12.3)	57.0	264.1 (43.1)	242.0	201.4 (35.1)	183.0
	TD	12	53.9 (10.9)	40.5	310.4 (44.7)	317.0	256.5 (40.4)	274.5
MEL
Low	CP	21	32.3 (7.6)	21.0	174.7 (35.8)	114.0	143.9 (32.7)	92.0
	OM	6	76.3 (23.7)	54.0	278.3 (90.4)	194.5	202.0 (69.3)	124.5
	TD	2	77.0 (45.0)	77.0	390.0 (235.0)	390.0	313.0 (280.0)	313.0
High	CP	19	45.2 (8.2)	38.0	217.6 (28.0)	237.0	172.4 (23.7)	194.0
	OM	23	52.5 (7.8)	46.0	257.2 (27.3)	260.0	204.7 (21.6)	211.0
	TD	23	59.4 (10.2)	44.0	322.2 (32.7)	352.0	262.8 (26.1)	271.0

^a Nine participants (one CP, three OM, and five TD) had missing tympanograms (Tymp).

Audiogram data were available for 77 children at 12 months of age. The range for average hearing level was from 15.0 to 40.0 dB. Ninety-two percent of children (30 children with CP, 21 children with OM, 20 TD children) had normal hearing, as indicated by levels of 20.0 dB or less. The grand means of hearing level were 21.02 dB (SD = 3.60), 20.83 dB (SD = 3.72), and 22.31dB (SD = 7.81) for the CP, OM, and TD groups, respectively. No significant differences in hearing level were observed among groups. It should be noted that all children with OM had tubes and were categorized as having normal tympanograms if the tubes were present (i.e., large-volume Type B). In addition, the slightly larger grand mean of hearing level for the TD children was due to three children who were tested during the winter months (cold season) and obtained average hearing levels of 40 dB. Mean values for sound weighted raw score at 18 months of age were 11.8 (SD = 6.0), 14.4 (SD = 6.2), and 15.6 (SD = 6.0) for the CP, OM, and TD groups, respectively. The mean sound weighted raw score for children with CP was significantly lower than that for TD children (p < .05). Relative to specific consonants, chi-square analyses indicated that fewer children with CP produced the bilabial stop /b/ or /p/ (p = .024), the glide /w/ (p = .026), and the fricative /s/ (p = .020) than children in either the OM or TD groups. There were no group differences for other sounds.

Unadjusted and adjusted model estimates for the absolute change in the number of words observed from 18 to 24 months of age are presented in Table 3 and Figure 1. The intercept estimates the change in the mean number of words observed for a female TD child with a mean hearing level of 21.5 (the sample mean), a sound weighted raw score of 13.8 (the sample mean), and normal middle ear status whose mother had at least a bachelor's degree. Negative estimates indicate a decrease in the effect (i.e., number of estimated words).

Table 3.

Unadjusted and adjusted linear regression model coefficient estimates and standard errors for change in the mean number of words across the three groups of children with cleft palate (CP), children with otitis media (OM), and typically developing (TD) children.

Parameter	Estimate (SE)	p	95% CI
Unadjusted model
Intercept	266.8 (26.0)	< .001	[215.2, 318.4]
CP	−109.4 (33.1)	.001	[−175.1, −43.6]
OM	−62.6 (35.5)	.081	[−133.1, 7.8]
Adjusted model
Intercept	339.0 (40.7)	< .001	[257.8, 420.2]
Cohort (TD is reference)
CP	−116.6 (41.7)	.007	[−199.7, −33.5]
OM	−65.7 (41.3)	.116	[−148.3, 16.6]
Centered mean hearing a (linear effect)	8.8 (5.9)	.141	[−3.0, 20.7]
Centered squared mean hearing a (quadratic effect)	−1.0 (0.4)	.027	[−1.8, −0.1]
Abnormal tympanogram	−69.7 (34.5)	.048	[−138.6, −0.8]
Centered sound weighted raw score b	1.8 (2.4)	.469	[−3.1, 6.6]
Lower maternal education	−17.7 (34.1)	.606	[−85.7, 50.4]
Male	−25.3 (29.8)	.399	[−84.9, 34.2]

Note. Unadjusted model is based on n = 94 participants; the adjusted model is based on n = 77 participants, owing to 17 participants having missing covariates.

^a Mean hearing level is centered at 21.5.

^b Mean hearing level is centered at 13.8.

Figure 1.

Mean vocabulary growth and its 95% confidence interval based on the adjusted multiple linear regression model comparing children with cleft palate (CP), children with otitis media (OM), and typically developing (TD) children from 18 to 24 months of age.

The unadjusted multiple linear regression analysis suggested that children with CP showed a statistically significant smaller absolute change of vocabulary (mean number of words) from 18 to 24 months of age in comparison with children in the TD group (95% CI [−175.1, −43.6]). The observed difference between children with CP and TD children remained significant when the model was adjusted for the effect of covariates (i.e., linear and quadratic terms for mean hearing level, middle ear status, sound weighted raw score, MEL, and gender) on vocabulary growth (95% CI [−199.7, −33.5]). Among these variables, only middle ear status (i.e., type of tympanogram; 95% CI [−138.6, −0.8]) and mean hearing level (p = .037; 2-df joint F test testing the null hypothesis that the linear and quadratic effects are zero) were observed to significantly predict vocabulary growth from 18 to 24 months of age. Although hearing level was statistically significant in the regression model, the partial eta-squared from the joint F test indicated a small effect size (i.e., .098). In addition, a scatter plot did not suggest a strong relationship of hearing level and vocabulary growth (see Figure 2). Note that Figure 2 shows part of the curve from 20 dB HL because 82% of screenings were performed at this level rather than 15 dB.

Figure 2.

Predicted change in mean number of words from 18 to 24 months of age in children with cleft palate (CP), children with otitis media (OM), and typically developing (TD) children as a function of mean hearing level.

Secondary Data Analysis

In separate statistical analyses, children with CP were stratified into two groups of children with CLP and CPO, and their vocabulary skills were compared to those of children without CP (i.e., OM and TD). Demographic information and descriptive analyses of the number of words produced by children with CLP, children with CPO, and children without cleft (OM and TD groups) are shown in Table 4. A Kruskal–Wallis test revealed significant differences among children with CLP, children with CPO, and children without cleft in terms of the number of words produced at 18 months (p = .01), the number of words produced at 24 months (p = .002), and vocabulary growth from 18 to 24 months of age (p = .002). Post hoc comparisons revealed significant group differences between the CLP and CPO groups for vocabulary growth from 18 to 24 months of age (p = .04). In addition, children with CLP showed a significantly smaller number of words at 18 months (p = .01) and 24 months (p = .001) as well as a slower rate of vocabulary growth from 18 to 24 months of age (p = .001) compared to children without cleft. However, children with CPO were observed to produce a statistically smaller number of vocabularies only at 18 months of age compared to children without cleft (p = .04).

Table 4.

Means and standard errors for the number of words produced at 18 and 24 months of age as well the vocabulary growth from 18 to 24 months for children with cleft lip and palate (CLP), children with cleft palate only (CPO), and children without cleft.

Cleft group	n	Gender		MEL		Tymp a		Average HL	18 months		24 months		Vocabulary growth
		Male	Female	Low	High	Normal	Abnormal		M (SE)	Mdn (NIQR)	M (SE)	Mdn (NIQR)	M (SE)	Mdn (NIQR)
CLP	15	13	2	6	9	9	6	20.83	36.0 (10.9)	20.0 (24)	139.4 (29.4)	84.0 (164.3)	103.4 (22.6)	71.0 (108.0)
CPO	25	6	19	15	10	19	5	21.09	39.8 (6.3)	35.0 (20.3)	228.4 (30.8)	227.0 (150.8)	189.9 (28.1)	188.0 (114.0)
None	54	33	21	8	46	35	11	21.57	59.0 (6.1)	47.0 (26.3)	292.2 (21.7)	284.0 (173.3)	233.2 (18.1)	236.0 (142.5)

Note. MEL = maternal educational level; HL = hearing level; NIQR = normalized interquartile range = 0.75 × (Q3 − Q1), where Q1 and Q3 are the first and third quartiles, respectively.

^a Nine participants (one CPO and eight children without cleft) had missing tympanograms (Tymp).

Discussion

The aims of this study were to investigate expressive vocabulary skills of children with CP, children with OM, and TD children at 18 and 24 months of age and identify the potential factors that could account for vocabulary growth. Considering the size of vocabulary from a developmental perspective, TD children were observed to produce, on average, approximately 61 and 328 words at 18 and 24 months of age, respectively. These numbers are similar to what has been suggested before by various studies. For example, Bates et al. (1988), Rescorla (1980), and Anglin (1989) reported a productive vocabulary of approximately 50 words by 18 months and 300 words by 24 months for children with typical development. Compared to Wordbank archive normative CDI data obtained from English-speaking children, the average number of words produced by TD children in this study was at the 25th percentile at 18 months of age and at the 50th percentile at 24 months of age. Similar to TD children, children with OM and CP were also at the 25th percentile of vocabulary norms at 18 months of age, but both groups remained at the first quantile (and below median) for the number of words at 24 months of age. Children with OM were observed to be in proximity to TD children in terms of the number of words produced at 18 months of age. Although the difference between these groups increased at 24 months of age, it did not reach statistical significance.

Children with CP displayed a slightly smaller number of words at 18 months of age compared to TD children. Significant differences between the two groups, however, emerged at 24 months of age. These findings are similar to those obtained by Hardin-Jones and Chapman (2014), as they also reported that, while size of the expressive vocabulary between children with and without CP was comparable at 13 months (earlier age), children with CP showed a significantly smaller number of words around the age of 21–27 months compared to their age-matched peers without CP. In addition, findings of this study characterized the vocabulary growth of children with CP as occurring at a slower rate (i.e., a smaller absolute vocabulary change from 18 to 24 months of age) compared to TD children. In this study, the mean percent changes for the increase in the number of words from 18 to 24 months of age were 529%, 440%, and 717% in the CP, OM, and TD groups, respectively. Therefore, on average, while TD children showed about a sevenfold increase in productive vocabulary over the 6-month period (i.e., from 18 to 24 months), children with CP and children with OM showed only a five fold and fourfold increase, respectively.

Additionally, a closer look at the data obtained from children in the CP group revealed that seven children with CP produced less than 50 words (mean number of words = 26) at 24 months of age compared to the rest of the children in this group who produced over 50 words at 24 months of age (mean number of words = 231). Excluding these seven children, the difference between children with CP who produced greater than 50 words at 24 months of age and TD children remained significant (p < .05). In addition, of the 40 children with CP, 25 (63%) and 21 (53%) were at or below the 16th percentile for the number of words on the CDI:WS at 18 and 24 months, respectively. In comparison, of the TD children, only seven (28%) and three (12%) were below the 16th percentile at 18 and 24 months, respectively. This reveals that at least half of the children with CP were delayed (i.e., below normal levels) in vocabulary at 24 months of age. Although more research is warranted about higher level linguistic skills and later language measures of these subgroups of children with CP, it is possible that children with CP who are below the 16th percentile for the number of words on the CDI:WS are at a higher risk for developing later impaired language skills such as reading and narratives.

Numerous explanations have been proposed as to why expressive language development is slower among children with CP. The widely accepted assumption is that delayed or deficient speech sound development is one of the main factors that lead to reduced size of vocabulary in children with CP. This argument is well grounded in previous correlational studies that show a strong tie between early speech production and later language performance of children with CP (e.g., Chapman et al., 2003; Scherer et al., 2008). In a longitudinal study conducted by Chapman et al. (2003), the percentage of true stops produced at both presurgery (about 9 months) and postsurgery (about 13 months) was positively correlated with measures of speech production (including size of consonant inventory and lexical items) at 21 months. Scherer et al. (2008) also reported a significant correlation between the frequency of babbling at 6 months and the size of consonant inventory and vocabulary at 30 months in children with clefts. In addition, Eshghi, Dorry, et al. (2017) reported that children with CP and adequate VP function exhibited a significantly reduced composite speech standard score (rate of sound and word acquisition) on the CSBS DP compared to TD children at both 18 and 24 months of age. They reported, however, that, despite the observed lower speech scores, the speech performance of most children with CP was still within normal limits. Therefore, because children with CP lag behind their age-matched noncleft peers in the onset of babbling (Chapman et al., 2001), the size of true consonant inventory (Chapman et al., 2003), and the composition of babbling (Lohmander, Lillvik, & Friede, 2004), it is conceptually plausible to link poorer expressive vocabulary in these children to delayed development of sounds at earlier stages of language acquisition.

In this study, children with CP exhibited significantly lower mean sound weighted raw scores compared to TD children at 18 months of age with fewer consonants, especially bilabial stops. NRP evaluation at 18 months of age identified at least eight children with CP who had inconsistent VP closure for stop consonants, suggesting the possibility of VP inadequacy. In a small preliminary study, Krochmal, Zajac, Alhudaib, Emodi, and van Aalst (2013) reported high sensitivity and specificity of early NRP measures in predicting the need for later secondary palatal surgery when less than 75% of stops were produced with VP closure. At least six of the children with NRP data in this study were below this cutoff. Despite the potential influence of VP status on speech sound development, sound weighted raw score did not significantly predict vocabulary growth.

Rather, results of this study suggest that middle ear status might influence the development of vocabulary skills. Middle ear status significantly predicted the development of vocabulary skills across all children. Abnormal middle ear status was associated with a 70-word decrease (21% reduction) in vocabulary. This finding is consistent with a study conducted by Jones et al. (2003) in which children with CP who had a higher number of reported episodes of OM and failed tympanometry pre- or postsurgery exhibited less progress in the size of the consonant inventory after palate repair. It must be noted, however, that the confidence interval for the effect of middle ear status in this study was large, suggesting uncertainty of the estimate and the need for caution relative to interpretation of findings.

Although a statistically significant relationship was found between hearing level and vocabulary growth, the effect size was small. As shown in Figure 2, the relationship between hearing level and the absolute change in the number of words shows a fairly flat curve in the range of the majority of the data, suggesting a negligible effect of hearing level on vocabulary growth. The statistically significant relationship, therefore, most likely occurred because a small number of outlying values at 40 dB (not shown in Figure 2) exerted undue influence on the data. As further seen in Figure 2, although a majority of participants across the three groups aligned at 20 dB, the absolute change in their number of words ranged from almost zero to above 300. This is an indication that, even in the presence of normal hearing (i.e., 20 dB), expressive vocabulary output varies considerably, probably due to other confounding variables. Given that middle ear effusion is associated with mild to moderate conductive hearing loss (Bess, 1983), the observed discrepancy between the effects of middle ear status and hearing level on vocabulary growth is most likely due to the nonconcurrent nature of the data. As reported in the Method section, although tympanometry and vocabulary data were obtained beginning at 18 months of age, audiometric data were obtained earlier at 12 months of age.

Findings of this study also suggest that gender and MEL may influence development of expressive vocabulary skills. Although not statistically significant, boys tended to have reduced vocabulary growth by approximately 29 words (9%) compared to girls. This finding is similar to previous studies in which girls were reported to outperform boys by producing a larger size of vocabulary (e.g., Eriksson et al., 2012; Marjanovič-Umek et al., 2008, 2012) and showing more rapid vocabulary growth (Bauer et al., 2002; Bornstein et al., 2004; Hadley et al., 2011; Huttenlocher et al., 1991). In addition, lower MEL, although not statistically significant, was associated with an approximately 23-word reduction (7%) in vocabulary. This finding is consistent with previous literature (e.g., Dollaghan et al., 1999; Magnuson et al., 2009; Marjanovič-Umek et al., 2016, 2008).

Based on findings related to the effect of gender and MEL on the development of vocabulary skills, it is plausible that the slight difference between the OM and TD groups in terms of the number of words might be due to two factors: the overpresentation of males (69% vs. 52%) and the higher incidence of mothers with education below bachelor's degree (21% vs. 8%) in the OM group compared to the TD group. As seen in Table 2, the effect of MEL was most pronounced in children with CP. It must be noted, however, that there were relatively few mothers with low education levels in the OM and TD groups. Conversely, approximately half (53%) of children with CP had mothers with education below bachelor's degree.

In our secondary data analysis, children with CLP were observed to have significantly slower vocabulary growth compared to children without cleft. This finding is in contrast to the broadly accepted belief that children with CPO are more vulnerable to develop poorer language and cognitive skills than children with CLP (Eliason & Richman, 1990; Richman, 1980; Scherer & D'Antonio, 1997). For example, Scherer et al. (2000) reported poorer expressive language and vocabulary skills in children with isolated CP compared to children with CLP. The observed discrepancies may be related to differences in the severity of the cleft and/or gender differences. That is, while approximately 87% of children with CLP were boys in this study, children with CPO were predominantly girls (i.e., approximately 76%). In addition, 40% of children with CLP had abnormal tympanograms, whereas 20% of children with CPO had abnormal tympanograms. Therefore, it seems plausible that the known gender distributions and middle ear status in subgroups of children with CP might have contributed to the observed differences between the two groups, especially with the knowledge that middle ear status was found to be a strong predictor of vocabulary growth in this study. Of note, the number of children with CPO who had mothers with education below bachelor's degree was larger than that of children with CLP (60% vs. 40%, respectively). It is possible that the effect of MEL was dissipated by gender and middle ear status in this subgroup.

Clinical Implications

This research suggests that a substantial portion of young children with CP may be at risk for delayed vocabulary growth compared to their age-matched peers without CP. Despite the fact that many children whose language is delayed at the second or third year of life will spontaneously (without intervention) catch up before school starts (e.g., Dale, Price, Bishop, & Plomin, 2003; Whitehurst & Fischel, 1994), research has demonstrated that some children without clefts who showed delayed onset of expressive language by 24 months of age are at considerable risk for continuing language problems (e.g., Fernald & Marchman, 2012; Rescorla & Schwartz, 1990). In addition, persisting speech and expressive language deficits have been reported to account for later reading and academic difficulties in children without CP (Felsenfeld et al., 1994; Lewis et al., 2006; Lewis, Freebairn, & Taylor, 2000, 2002; Nathan, Stackhouse, Goulandris, & Snowling, 2004; Sices et al., 2007).

Although it is likely that some children with CP show improvement in later years before they start school, early speech and language monitoring is needed to identify subgroups of children with persistent expressive language delay. Because reduced size of vocabulary can cause delayed transition to two-word utterances and subsequently impact normal development of syntactic skills and higher level linguistic abilities, receiving appropriate speech interventions is critical for children with persistent expressive language deficits. Evidence for poor grammatical skills in children with CP has indeed been reported in some studies (e.g., Eshghi et al., 2018; Morris, 1962). In addition, persisting speech and expressive language deficits have been reported to account for later reading and academic difficulties in children without CP (Felsenfeld et al., 1994; Lewis et al., 2006, 2000, 2002; Nathan et al., 2004; Sices et al., 2007). Similarly, poor academic outcomes (Broder et al., 1998; Knight et al., 2015; Wehby et al., 2014) and reading skills in children with CP were reported by a number of other studies (Chapman, 2011; Richman et al., 1988). Therefore, identifying subgroups of children at risk for expressive language delay and applying appropriate speech interventions are of paramount importance. Finally, and perhaps most important, the results of this study reinforce the current protocol of early and ongoing management of middle ear disease in children with repaired CP. Such care may be critical to optimizing vocabulary growth for some children. Clinicians need to continue to vigilantly monitor young children and encourage caregivers to seek medical treatment when children exhibit signs of middle ear disease, such as fever, irritability, and/or drainage.

Limitations of the Study

Several methodological limitations must be acknowledged. First, children with OM were observed to show slower vocabulary growth compared to TD children, but the small sample size may have contributed to the observed trend that approached but did not reach significance. Future prospective studies with larger sample sizes are needed to substantiate the effects of OM. Second, although data on expressive vocabulary were obtained from children at approximately the same time (i.e., 18 and 24 months of age), matching between the three groups in terms of gender, MEL, and SES was not performed. Although gender and MEL were entered as covariates in the statistical models, a priori matching would have been preferred. Third, as already noted, hearing data were not obtained concurrently with middle ear and vocabulary data. Given these limitations, precise information regarding the actual magnitude of conductive hearing loss at the time of vocabulary assessment could not be determined. In addition, various intervening factors such as age at the onset of OM as well as magnitude and duration of conductive hearing loss, which can influence speech outcomes, were not measured in this study. Finally, research on the development of expressive vocabulary and other language skills has demonstrated that factors such as child–caregiver interactions and amount of language stimulation influence expressive vocabulary and language development (e.g., Huttenlocher et al., 1991; National Institute of Child Health and Human Development Early Child Care Research Network, 2000; Rowe, 2012; Tamis-LeMonda, Bornstein, & Baumwell, 2001). Information on these factors, however, were not obtained in this study.

Conclusions

The current study demonstrated that children with CP had a significantly smaller number of words at 24 months of age and a slower rate of vocabulary growth from 18 to 24 months of age compared to TD children. Among variables including hearing level, tympanogram status, size of consonant inventory, MEL, and gender, only abnormal tympanogram as an indication of middle ear dysfunction was found to be a predictor of expressive vocabulary skills. This variable, however, had a large confidence interval, suggesting uncertainty with the estimate and caution relative to interpretation of the findings. Nevertheless, findings of this study reinforce current management protocols that call for vigilant monitoring and management of middle ear disease in young children with repaired CP.

Funding Statement

Research reported in this publication was supported by National Institute of Dental and Craniofacial Research Award R01DE022566 to David Zajac. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.