Abstract
The purpose of this preliminary study was to (a) examine relationships between the symbolic and language skills of a mixed (developmental language disordered [DLD] and typical language [TL]) Spanish-speaking sample; (b) describe gesture, play, and language skills of DLD and TL groups; (c) compare the development between groups; and (d) explore whether combining symbolic and language measures predict language scores. A major focus of this study was to interpret how this information may be clinically useful, especially for non-Spanish-speaking speech-language pathologists who are responsible for the initial screening, monitoring, and/or assessment of the language development of young Spanish speakers. Analyses revealed that some symbolic skills were significantly associated with language, but group differences were detected only in conventional gesture usage. Conventional gesture and number of different words (NDW) combined significantly predicted language scores. This study showed that play-based approaches provide important developmental information that contributes to screening and assessment decisions.
Keywords
In 2008, 28% of preschool-age children were from households where a language other than English was spoken, with the majority of these children being from Spanish-speaking backgrounds (U.S. Census Bureau, 2008). This growth trend is expected to continue in the coming years (Center for Public Education, 2012), which presents the need for measures that can assist in screening and assessing the language development of young Spanish-speaking children. To complicate this matter, in 2011, only 2.6% of speech-language pathologists (SLPs) in the United States reported that they were Spanish–English bilingual (American Speech-Language-Hearing Association [ASHA], 2012). In two recent surveys from 2012, SLPs reported that they lack confidence when assessing linguistically diverse children (Guiberson & Atkins, 2012; Williams & McLeod, 2012). SLPs who feel unprepared to serve individuals with cultural and linguistic differences should consider the option of referral to a provider who has competence in this area (ASHA, 2013). However, given the extreme shortage of bilingual SLPs, monolingual English-speaking SLPs need to develop strategies to serve the increasingly diverse population in the United States. This may include pairing up with other providers who are bilingual (e.g., special education teachers, English-Language Acquisition or bilingual teachers, or other providers) or developing skills, strategies, and procedures for effectively working with cultural mediators/interpreters (Guiberson & Atkins, 2012; Guiberson & Banerjee, 2012; Williams & McLeod, 2012). While teaming with other bilingual providers or a cultural mediator requires additional time for planning, implementing, and debriefing, it can be an effective way for monolingual English-speaking SLPs to meet the needs of culturally and linguistically diverse populations.
Despite the shortage of bilingual SLPs and the lack of psychometrically sound tools to use with Spanish-speaking children (Dollaghan & Horner, 2011), early intervention programs must screen, monitor, and assess the language skills of Spanish-speaking children and identify when children present with developmental language disorders (DLDs). Primary DLD is a term used to describe children who are not acquiring language as would be expected for their chronological age and for whom language difficulties are not related to neurological, intellectual, social, or sensory deficits (Paul & Norbury, 2012). Behavioral characteristics of DLD change over time as children mature, but generally, DLD includes deficits in morphosyntactic development, vocabulary knowledge, and phonological processing. DLD is presumed to be due to interactions between innate factors and language learning demands, and is not a result of cultural, linguistic, or education experiences (Kohnert, 2008). This last point is especially important when considering DLD in linguistically diverse populations; when assessing these children, development should be compared with behaviors seen in same-age peers with similar linguistic and experiential backgrounds (Guiberson & Banerjee, 2012). For 2-year-old children, DLD is characterized by limited vocabulary and language usage; in addition, family history of DLD, combined with reduced gesture and symbolic play skills, is thought to indicate greater risk of continued and more severe DLD (Kelly, 1998; Kohnert, 2008).
Currently, adequate measures do not exist to screen or assess for DLD in young Spanish-speaking children (Dollaghan & Horner, 2011; U.S. Department of Health and Human Services, 2014). English-speaking SLPs need research-based measures to screen and monitor the language development of Spanish-speaking children to determine when these children need to be referred for a full evaluation by a bilingual assessment team. Measures of gesture and play may be useful because, at least in English-speaking children, these behaviors are associated with linguistic development and DLD (Bates, Bretherton, Snyder, Shore, & Volterra, 1980; Kelly, 1998). Gesture, play, and language are thought to share an underlying symbolic ability, and researchers have hypothesized that children with DLD have a general symbolic deficit (Miller, 1991). This deficit results in slower development of symbolic abilities, which may include deficits in gesture, play, and language skills. It is possible that general symbolic skills observed through gesture and play may provide a window into Spanish-speaking children’s language skills. This would be useful for professionals who do not speak Spanish but are nonetheless responsible for the initial screening or monitoring of the language development of Spanish-speaking children.
Gesture
Gesture Development in Typically Developing Children
A large body of research has shown that gesture development progresses steadily from infancy through early childhood and that gesture and language development parallel each other with shared, underlying symbolic abilities (Capone & McGregor, 2004; Guidetti & Nicoladis, 2008). A summary of gesture development is presented below. For a more thorough review, however, see Capone and McGregor (2004).
Infants begin simple deictic gestures (pointing, showing, and giving) from 10 to 13 months of age. Representational gesture emerges from 12 to 13 months and continues through 16 months. Representational gestures include iconic gestures (depicting an object) and conventional gestures defined by the child’s culture (e.g., waving good-bye or holding a finger to the mouth to indicate quiet). By 18 to 20 months, a child’s word usage increases and representational gestures decrease. Also, at 18 to 20 months, an increased use of deictic gestures and gesture + word combinations is observed. From 2 to 5 years of age, gesture + word combinations continue, and deictic gestures continue to supplement spoken messages. Older preschool-age and school-age children use beat gestures, which provide emphasis while speaking but convey no additional semantic information (e.g., moving the hands up and down while talking).
Gesture Development Across Cultures
First, it should be noted that there is a need for more studies with larger samples that describe the development and use of gesture in children from culturally and linguistically diverse backgrounds (Capone & McGregor, 2004; Nicoladis, 2002; Nicoladis, Mayberry, & Genesee, 1999). Gesture usage is both universal and somewhat culture-specific (Guidetti & Nicoladis, 2008). Deictic gestures, with minor variations such as pointing with the pinky finger rather than with the index finger, tend to be consistent across cultures. Conventional gestures, however, are defined as those used within a cultural group and thus vary across cultures (e.g., beckoning “come” with the index finger in some European American cultures or with a flat hand waving downward in some Mexican cultures).
Developmental studies of gesture usage in other cultures have yielded interesting findings. A study conducted in Italy revealed that Italian toddlers used more conventional and iconic gestures than those described in studies of American children. However, gesture + word combinations in Italian children paralleled the combinations seen in American children (Capirci, Iverson, Pizzuto, & Volterra, 1996). In a longitudinal study that compared gestures of Spanish 1-year-olds with those of Spanish 2-year-olds, deictic gestures + word combinations were the most frequently occurring gestures and continued to increase as children aged. Documented cultural variations also exist in the frequency and repertoire of gesture usage in children from different cultural backgrounds. In a study that described the gesture usage of monolingual Italian-speaking and monolingual English-speaking American toddlers, Italian children produced conventional and iconic gestures more frequently than the American children. The American children used mostly deictic gestures, which were also found in the Italian children’s gesture repertoires (Iverson, Capirci, Volterra, & Goldin-Meadow, 2008).
Studies of bilingual children have also shed light on the use of gesture by children from culturally and linguistically diverse backgrounds. In one study, young bilingual children (2–4 years of age) from a variety of linguistic backgrounds demonstrated stronger abilities in reading gestures to understand a speaker’s intent than their monolingual peers (Yow & Markman, 2011). A study of French–English bilingual preschool children described the gesture usage of children from 2 to 3 years of age. While deictic gestures were the most frequently observed, iconic gestures were most closely associated with language development. In a second study of French–English bilingual children, Nicoladis (2002) found that children used a higher rate of iconic gestures in their more proficient language. However, the use of other types of gestures did not differ by language proficiency. The author interpreted these results as indicating that children do not use iconic gestures to compensate for lower proficiency in a given language. She also found that iconic gestures tended to be related to normal word finding difficulties that children typically experience.
Gesture Development in Children With DLDs
Gesture and language development parallel each other in both typically developing children and children with DLD (Capone & McGregor, 2004). In a 1-year follow-up study of toddlers with DLD, Thal, Tobias, and Morrison (1991) found that toddlers with DLD used fewer gestures than their peers. A year later, these children also used fewer conventional gestures than their peers. Subsequent studies describing gestures in toddlers and young preschoolers with DLD have yielded mixed results about specific profiles of gesture usage in this population (Ellis Weismer, Murray-Branch, & Miller, 1994; Olswang, Rodriguez, & Timler, 1998). However, in two studies, young children with DLD who began to combine gestures with words were found to use two-word combinations shortly thereafter (Olswang et al., 1998). These results indicate that toddlers who exhibit few representational gestures or few gesture and word combinations appear to be at higher risk for persistent and more significant DLD.
The relationship between frequency and type of gestures used by children with DLD also appears to change as children age. In a study that included both preschool- and school-age children with and without DLD, researchers found that poorer expressive language skills were related to more frequent gesture production (Iverson & Braddock, 2011). This may be because at later stages in development, children with DLD begin to rely on gestures to supplement their incomplete messages. In a study that described the gesture usage of school-age children, those with DLD did not demonstrate differences in overall gesture usage when compared with typical peers, but they did use slightly more iconic gestures (Blake, Myszczyszyn, Jokel, & Bebiroglu, 2008). Again, this may indicate that older children with DLD use gestures to supplement their messages and perhaps to avoid communication breakdowns.
Play
Play can be defined in many ways. Play involves spontaneous, naturally occurring activities with objects that engage the attention and interest of others (Lifter, Foster-Sanda, Arzamarski, Briesch, & McClure, 2011). Play behaviors reflect a child’s cultural background and can serve as an index of a child’s development (Fenson, 1986; Hwa-Froelich, 2004). Play allows children to embody mental representations of the real world, therefore providing a window into what children know (Westby, 2000). A great deal of research has described the development of play in European American children (for a review, see Gitlin-Weiner, Sandgrund, & Schaefer, 2000; Zigler, Singer, & Bishop-Josef, 2004). In the following sections, cultural considerations and play development in populations with DLD are reviewed.
Play Development and Cultural Considerations
Play occurs within a social context. Care providers are thought to use interaction styles that are consistent with their culture. Two interaction styles have been described: an interdependent/collectivist style and an independent style (Greenfield, Keller, Fuligni, & Maynard, 2003). The interdependent style emphasizes the family and group, whereas the independent style emphasizes independence and individual success. Caregiver interaction style influences behaviors related to attention regulation, language usage, object engagement, and play (Vigil & Hwa-Froelich, 2004). In the context of play development, a care provider’s developmental goals for a child are influenced by his or her interaction style. Care providers with an independent interaction style are likely to hold objects for children to see, allow children to explore objects, and allow them to play with toys in a manner different from the toy’s intended use. Care providers with an interdependent interaction style are more likely to explicitly teach children how to use a toy, use hand-over-hand strategies to teach toy use, and emphasize playing with a toy correctly. In this respect, cultural differences in the context of play may influence teaching and learning behaviors, thereby influencing developmental trajectories (Greenfield et al., 2003).
Like many Latino parents, care providers from Mexi-can backgrounds have been described as having an interdependent interaction style, although variability exists within cultural groups (Kayser & Guiberson, 2008). In a study that contrasted the play behaviors of Mexican dyads to those of European American dyads (Farver & Howes, 1993), researchers found that children in the Mexican dyads joined the mothers’ activities, mothers encouraged interdependence and cooperation, and children learned through observing and imitating adults. In the European American dyads, however, mothers followed their child’s lead, encouraged independence and self-confidence, and the children learned through the adults’ efforts to teach play. Symbolic play was observed in both American and Mexican dyads, but Mexican mothers typically did not initiate symbolic or pretend play with their children. Studies across other cultural groups have also revealed that cultural differences emerge in play schemas, and that everyday routines or family themes may be more prevalent in the play of children from diverse backgrounds (Hwa-Froelich, 2004). These results reflect differences seen in the development of play across different cultural contexts. Results may indicate that cultural differences in maternal play behaviors are related to culture-specific child rearing practices that likely serve adaptive functions (Farver & Howes, 1993). However, more research describing play development in populations from culturally diverse backgrounds is needed.
Play Development in Populations With DLD
A large body of literature describes the play skills of children with DLD (for a review, see Casby, 1997). Two dimensions of play in children with language disorders have been studied extensively: symbolic play and complexity of play. The way in which these dimensions of play are operationalized varies somewhat, making cross-study comparisons difficult. Symbolic play progresses from (a) presymbolic schemes (non-pretend), (b) auto-symbolic schemes (pretend activities with self), (c) single symbolic schemes, (d) combinational schemes, to (e) planned scheme combinations that seem to follow a script (Nicolich, 1977). Complexity of play not only includes some of the same elements of symbolic play but also considers other aspects such as functional and creative use of objects, as well as organization and cohesiveness of play (Linder, 2008). Studies describing the symbolic play of children with DLD have yielded mixed results, with some studies indicating deficits in symbolic play and some indicating that children with DLD use fewer play schemas and fewer combinational schemes than typical peers (Casby, 1997; Leonard, 1998; Terrell, Schwartz, Prelock, & Messick, 1984). Regarding complexity of play, children with DLD have been found to demonstrate less complex play than their typical peers (Casby, 1997; Leonard, 1998). Collectively, these findings indicate that play skills are broadly related to language abilities and that children with language disorders are likely to have constrained symbolic play development, with poorer play associated with lower language skills (Leonard, 1998).
Current Study
While there is a substantial body of research describing play and gesture in children with DLD as well as some research describing these behaviors in Spanish-speaking children, there is very limited research describing Spanish-speaking children with DLD and the association across these developmental areas. The purpose of this preliminary study was to examine associational relationships between gesture, play, and language development in a mixed DLD and typically developing Spanish-speaking sample; describe gesture, play, and language skills of Spanish-speaking children with and without DLD collected through semi-structured play activities; compare the development of DLD with typically developing Spanish-speaking children; and explore whether the combination of symbolic and language measures predict standardized language scores. A key goal is to interpret how this information may be useful, especially for monolingual English-speaking SLPs who may be responsible for initially screening the language of young Spanish-speaking children.
This preliminary study applied a combination of associational, descriptive, and comparative research approaches with a crafted sample of half DLD and half typically developing children (Gliner, Morgan, & Leech, 2009). This purposefully selected sample, with a disproportionate DLD representation, may assist in better understanding the specific behavioral and developmental profiles of children with DLD, which is one of the primary objectives of this study (Creswell, 2003).
The specific research questions for this preliminary study were as follows:
Method
Participants
Family characteristics
Families were recruited from two states in the Western United States and self-identified their ethnicity as Hispanic and nationality as Mexican. Thirty-eight toddler-age children (2;0–2;11 years of age) and their parents participated in this study. Only predominately Spanish-speaking children and families (those who spoke Spanish 80% of the time or more according to parent report) were included. Families spoke Mexican dialects of Spanish, and none of the parents indicated speaking a language other than Spanish or English in the home. Parent occupations varied with the majority of families working in service and blue-collar industries. Eleven of the families were single-mother households. Maternal level of education (N = 38) ranged from third grade to some college (M = 9.11, SD = 2.93). Paternal level of education (N = 27) was slightly lower than maternal level of education, ranging from third grade to some college (M = 8.44, SD = 2.47). Seventy-six percent of the families qualified for Early Head Start programs.
Child characteristics
DLD is thought to have a prevalence rate of approximately 7% (Paul & Norbury, 2012). For the present study, which aimed to better describe symbolic and linguistic behaviors of children with DLD, researchers decided to oversample children with DLD (approximately 50% DLD and 50% TL). This oversampling of DLD was justified, so that the study would provide a more robust description of the symbolic and linguistic development of children with DLD. Three sources were obtained to establish DLD status: (a) identification of DLD by a bilingual SLP, (b) parent report of concern about language development, and (c) expressive language standard score < 77 (1.5 SD below the mean) on the SPLS-4. The group with DLD (n = 18) was comprised of 12 boys and 6 girls with a mean age of 29.28 months (SD = 3.20 months). The group with TL (n = 20) was comprised of 9 boys and 11 girls with a mean age of 29.60 months (SD = 3.97 months). Children were placed in the TL group if (a) they had not been previously diagnosed as having DLD, (b) their parents did not report concerns about their language development, and (c) they had an expressive language SPLS-4 score ≥ 85. Children with DLD were enrolled in Part C early intervention services, and 90% of children in the TL group were enrolled in Early Head Start home visiting program. As a precaution, group comparisons of maternal education, child and maternal percentage Spanish spoken in the home, and qualification for income-based Early Head Start family services were completed; no significant or substantial group differences were detected. Children from both groups had normal hearing and no known neurological impairment.
Procedures
General recruitment efforts included sharing study flyers with a regional health clinic for Spanish-speaking families, an early childhood center, and an early intervention disability center. In addition, the primary investigator collaborated with disability coordinators from the early intervention disability center to recruit families of children with DLD. Interested families who met the criteria were presented with an informed consent form in Spanish that had been approved by a university institutional review board.
Consenting families scheduled a study visit, which occurred within 2 weeks of initial contact. All study visits, including both play and SPLS testing, were completed in either a child assessment room or an unoccupied classroom at the collaborating early childhood centers. During the study visits, parents (either mothers only or mothers and fathers) engaged children in semi-structured play activities. Play sessions lasted approximately 40 min for each child. After the play sessions, a bilingual SLP or a bilingual graduate student administered the SPLS-4 in Spanish in the same room where the parent–child play had occurred. All visits were videotaped.
Instructions to parents
Based on what is known about interaction styles frequently described in Mexican American families, decisions were made to ensure that play activities would be culturally appropriate. First, families were asked to sit at a child-sized table rather than playing on the floor. Second, mothers were asked to teach children how to play with each set of toys and then to allow children to play with the toys. If parents asked for additional guidance, they were instructed to teach their child to play with the toys like they normally would at home.
Stimuli
Parents interacted with children using six standard sets of toys (see Table 1). These toys were selected because they mapped onto the Westby’s Symbolic Play Scale (SPS, Westby, 1980) as well as Linder’s (2008) complexity of play subscale. In addition, these toys were relatively gender neutral and spanned the expected developmental range for the children in the study. Toy sets were presented in clear boxes, and parents were informed that they would engage in play with each set of items for about 6 min. After all of the toy sets were presented, parents were asked to look at a Spanish version of a My First 100 Words board book with their child for 5 to 6 min. All the play and gesture measures were collected during structured play activities with parents.
Toy Used in Study.
Measures
SPLS-4
The SPLS-4 is a language assessment tool that was standardized using 1,188 Spanish-speaking children living in the United States from monolingual Spanish or mostly Spanish bilingual households (Zimmerman, Steiner, & Pond, 2002). In a normative study that included children between the ages of 2;0 (years; months) and 2;11, test–retest reliability of the SPLS-4 ranged from .73 to .87 (n = 30), and split-half internal consistency correlations ranged from .83 to .87 (n = 129). The classification accuracy (i.e., sensitivity and specificity) of the SPLS-4 was not reported for the 2-year-old age group. Data on 3- to 5-year-old children (N = 140) revealed that the Expressive Communication subtest demonstrated stronger classification qualities (sensitivity = .91 and specificity = .69) than did the Receptive subtest (sensitivity = .83 and specificity = .56). The Expressive Language subtest was selected for the present study because it has stronger psychometric qualities.
Gesture
A gesture coding system developed by Nicoladis (Nicoladis, 2002; Nicoladis et al., 1999) was used to code children’s gestures as conventional, deictic, or iconic. For this coding system, deictic gestures included gestures that direct another’s attention (pointing, showing, and giving), iconic gestures depicted an object, while conventional gestures are defined by the child’s culture (such as waving good-bye or holding a finger to the mouth to indicate quiet). The coding also indicated whether gestures were accompanied by spoken language. Two Spanish-speaking graduate students were trained in the gesture coding. After training, each coder viewed and applied the gesture coding systems to five videos of children (typical and DLD) that had been coded by the author (a certified bilingual SLP who was highly experienced in gesture and play development and coding systems). Prior to independently beginning gesture coding, coders achieved 90% or higher point-by-point inter-rater agreement for the presence of gesture and accuracy of gesture code. A second coder coded eight randomly chosen transcripts. The point-by-point inter-rater agreement for gesture coding was 92%.
Symbolic play
Symbolic play was coded using Westby’s original SPS (Westby, 1980). Westby developed the SPS based on observations of play development in typical and disordered populations. The SPS is a detailed checklist that can be used to code presymbolic and symbolic play and provides developmental levels of symbolic play ranging from Stages I to X. For the present study, symbolic play scores were assigned based upon the highest play behavior demonstrated independently without direct parent scaffolding. Two Spanish-speaking graduate students were trained in the SPS coding. As a first step, each item in the SPS checklist was thoroughly described and discussed by the coding team. Next, the team reviewed approximately 12 short video clips of children not included in the study, and coders identified the level of symbolic play that the children demonstrated. Using the SPS, coders mapped play behaviors independently and compared their coding with the author’s. During this time, questions, discussions, and consensus coding occurred as needed. Next, the coders independently viewed videos and applied SPS coding to videos that the author had already coded. The coders achieved 100% inter-rater reliability for the highest stage of play behavior observed for the five videos. Later, a second coder coded eight randomly chosen transcripts, and the inter-rater agreement for the highest level of symbolic play was 100%.
Complexity of play
Complexity of play was coded using the complexity of play subscale from the cognitive age tables developed by Linder (2008). Behaviors that were immediate imitations of a parent behavior were not counted. Complexity of play developmental scores were based upon the highest play behavior demonstrated independently, which is what practitioners in the field most often use (Linder, 2008). Possible scores ranged from 1 month to 72 months. A similar process of review, practice, and discussion occurred prior to coders independently coding the five tapes that were already coded by the first author. For this coding, point-by-point agreement of complexity of play behaviors that children demonstrated independently was calculated. Coders achieved 100% inter-rater reliability (complexity of play in months), prior to beginning coding complexity of play independently. A second coder coded eight randomly chosen transcripts, and the inter-rater agreement (point by point) for complexity of play coding was 88%, with one tape having a complexity of play discrepancy between coders of 2 months.
Language samples measures
Children’s language production occurring during parent–child play sessions was transcribed into the Systematic Analysis of Language Transcription (SALT; Miller, 2008) program by Spanish bilingual graduate students. The conventions described for the transcription of Spanish speakers were followed. The SALT program was used to compute children’s language sample measures, including mean length of utterance in words (MLUw), mean length of the three longest utterances (MLUw-3), and number of different words (NDW) used. All child utterances were transcribed; however, the SALT program included only complete and intelligible utterances in the calculations of MLU measures, and only the intelligible words were used in the calculation of NDW. Also, the NDW measure was based on stems or word roots (e.g., pato and patos were counted as one word root). Four graduate students were trained to use SALT and achieved 90% or higher inter-rater reliability (point-by-point agreement), prior to beginning independent language transcription. In addition, every fifth tape completed by a transcriber was checked for reliability. The inter-rater reliability for language samples was 94%.
Results
Convergent Validity
The first research question aimed to examine the convergent validity of nonverbal symbolic skills relative to language measures of Spanish-speaking toddlers. Correlations were calculated between nonverbal symbolic measures and language measures (controlling for age) to establish the convergent validity between these measures for the entire group. Table 2 presents these correlations. Of the gesture measures, conventional gesture, total gesture, and gesture + word combinations were most significantly associated with language measures. Iconic gesture and deictic gesture were less associated with language measures. Symbolic play was not associated with language measures, but complexity of play was significantly associated with language measures. Post hoc analyses were completed to see whether there were any notable changes in the patterns of correlations when analyses were completed by group (DLD or TL). Given the reduction in sample size and loss of power, these analyses were exploratory. Visual inspection revealed no major differences in the general trend of these associations.
Partial Correlations (Controlling for Age) Between Gesture, Play, and Language Measures (N = 39).
Note. SPLS-4 = Spanish Preschool Language Scale–Fourth Edition; MLUw = mean length of utterance in words; NDW = number of different words; MLUw-3 = mean length of the three longest utterances.
p < .05. **p < .01. ***p < .001.
Description of Nonverbal Symbolic and Language Skills of Children
Research Question 2 focused on describing the nonverbal symbolic and language skills demonstrated by Spanish-speaking toddlers during structured play sessions with mothers. Table 3 presents descriptive statistics of the nonverbal symbolic and language skills by group. Each set of variables is described in more detail below. Note that because of the small sample size, visual inspection and description was used for descriptive comparison purposes. Preliminary group comparisons of selected variables are reported later.
Means and Standard Deviations for Language, Gesture, and Play Variables by Group.
Note. TL = typical language; DLD = developmental language disorder; SPLS-4 = Spanish Preschool Language Scale–Fourth Edition; MLUw = mean length of utterance in words; NDW = number of different words; MLUw-3 = mean length of the three longest utterances.
Gesture
All children used some gesture, with the number of gestures used ranging from 7 to 149. The TL group had means for the gesture behaviors that were higher across all variables. Conventional gesture was observed in both groups of children; more than 86% of the children used some conventional gesture. Deictic gesture was the most frequent behavior for both groups and was used by all children in the sample. Iconic gesture was the least frequent gesture observed for both groups; 76% of the children used iconic gestures. Most children (92%) also made gesture + word combinations.
Play
Using the SPS framework (Westby, 1980), children’s symbolic play skills ranged from Stage V to Stage VIII. The DLD group had symbolic play skills that were comparable with those of the TL group. Complexity of play was coded using Linder’s (2008) complexity of play subscale; scores varied from 21- to 36-month developmental scores, with the TL group’s mean developmental score being higher than that of the DLD group’s mean developmental score.
Language sample measures
The DLD group’s mean sentence length in words (MLUw) was 1.2 words per sentence, and the TL group’s MLUw was 1.5 words. However, the DLD group’s average length of three longest sentences (MLUw-3) was 2.8, and the TL group’s MLUw-3 was 4.53. The vocabulary used (NDW) differed by groups; the DLD group’s mean NDW was 40.60, and the TL group’s mean NDW score was 80.35.
Group Comparisons of Nonverbal Symbolic and Language Sample Measures
The focus of Research Question 3 was to compare symbolic skills and language sample measures of a DLD group with those of a TL group. Results from the convergent validity analyses were examined to identify which of the variables should be included in group comparison analyses. Of the gesture measures, conventional gesture was most strongly associated with language. These analyses also showed that of the play measures, complexity of play was the most strongly associated with language. Of the language sample measures, NDW was the most strongly associated with other language measures. Given these results, conventional gesture, complexity of play, and NDW were selected for group comparisons.
The groups demonstrated significant differences in conventional gesture usage, t(36) = 2.71, p ≤ .01, and NDW, t(36) = 3.62, p ≤ .001, with a large effect size observed but no significant group differences were detected in complexity of play, t(36) = 1.25, p = .22. The small sample size in the present study may have limited the statistical power and significance of the statistical comparisons conducted. A post hoc power analysis using the program G*Power 3 (Faul, Erdfelder, Lang, & Buchner, 2007) revealed that the power (1 − β error probability) needed to detect a medium-sized effect was 0.32. There was adequate power to detect group differences between groups with the conventional gesture usage (.76) and NDW (.83), but there was not adequate power to detect group differences with the complexity of play variable (.23).
Exploratory analyses: Predicting language scores
Through exploratory analyses, Research Question 4 explored the extent to which the combination of nonverbal symbolic and naturalistic language sample measures predicted standardized language scores. Two logistic regression models were completed to establish if combinations of nonverbal symbolic measures or a nonverbal symbolic measure and a language sample measure predicted SPLS-4 scores. Given the small sample, these analyses are exploratory and will help inform next steps in this line of research.
For the first model, the convergent validity results were used to select the nonverbal symbolic measures (one gesture variable and one play variable) that were most closely associated with SPLS-4 scores. Coefficients for the first model are presented in Table 4. The combination of conventional gesture and complexity of play significantly predicted child language scores, F(2, 35) = 6.86, p < .01, with only conventional gesture significantly contributing to the prediction. An adjusted R2 value indicated that the model accounted for 25% of the variance in child language scores.
Summary of Logistic Regression Analysis Models Predicting Language Scores (N = 39).
Note. NDW = number of different words.
p < .01. ***p < .001.
For the second model, the convergent validity results were again used to select the nonverbal symbolic measure (conventional gesture) and language sample measure (NDW) that were most closely associated with SPLS-4 scores. Coefficients for the second model are presented in Table 4. Conventional gesture and NDW were significantly correlated, but collinearity tolerance levels were acceptable (>1 − R2). The combination of conventional gesture and NDW significantly predicted child language scores, F(2, 35) = 14.82, p < .001, with only NDW significantly contributing to the prediction. An adjusted R2 value for conventional gesture combined with NDW indicated that the model accounted for 43% of the variance in child language scores.
Discussion
Convergent Validity Between Symbolic and Language Measures
While studies of English-speaking children and the general symbolic deficit hypothesis suggest that language ability and nonverbal symbolic skills are associated, this has not been established in Spanish-speaking children. In this study, of the symbolic measures, conventional gesture, total gesture, gesture + word combinations, and complexity of play were most significantly associated with language measures. These findings are consistent with earlier studies that have shown that gesture and complexity of play are associated with language ability (Capone & McGregor, 2004; Leonard, 1998). These results also indicate that there is evidence of convergent validity between these symbolic variables and language variables.
Nonverbal Symbolic and Language Skills of Children
There is limited research describing early symbolic and linguistic development in children from culturally and linguistically diverse backgrounds. This study makes a unique contribution in that it describes the nonverbal symbolic and language skills of a growing segment of the U.S. population, young Spanish-speaking children. Most of the children who participated in the present study demonstrated all three types of gestures (conventional, deictic, and iconic), as well as gesture + word combinations. Children also demonstrated symbolic play skills and a range of complexity of play skills when engaged in semi-structured parent-facilitated play activities. More variability was seen in the language sample measures collected, especially in the MLUw-3 and NDW measures. It is not surprising that more variability was seen in language skills, given that the sample included both DLD and TL children. If children with suspected or confirmed developmental disabilities were included, it is likely that more variability would have been observed in symbolic play and complexity of play.
Group Comparisons of Nonverbal and Language Skills
DLD and typical group differences were detected in conventional gesture usage and NDW, but not in complexity of play. These results are consistent with earlier studies that have revealed that young children with DLD may demonstrate fewer conventional gestures than their typical peers (Thal et al., 1991). In addition, significant group differences were seen in the vocabulary used by children with DLD when compared with typical peers. These results combined with results from earlier studies indicate that language sample measures can provide useful information about language development in Spanish-speaking children (Gutiérrez-Clellen, Restrepo, Bedore, Peña, & Anderson, 2000). No significant group differences in play skills were seen between the typically developing group and the group with DLD; this finding does not correspond with earlier studies that have found that children with DLD have significantly weaker play skills than their typical peers’ play (Casby, 1997; Desmarais, Sylvestre, Meyer, Bairati, & Rouleau, 2008; Leonard, 1998; Paul & Norbury, 2012). Both symbolic play and complexity of play measures had rather narrow bands of variability and the group means were rather comparable. In fact, there was not adequate power to detect complexity of play group differences. This suggests that a larger sample and/or play measures that capture more variability are needed. The measures used in this preliminary study were broadly associated with language development in Spanish-speaking children (as the convergent validity analyses indicated), but these measures alone are not strong diagnostic markers of DLD.
Predicting Language Skills From Symbolic Skills and Language Sample Measures
Two exploratory regression models were completed to establish if combinations of nonverbal symbolic measures or a nonverbal symbolic measure and a language sample measure predicted SPLS-4 scores. In the first model, nonverbal symbolic measures (conventional gesture and complexity of play) significantly predicted child language scores and accounted for 25% of the variance in child language scores. In the second model, conventional gesture and NDW significantly predicted child language scores and accounted for 43% of the variance in child language scores. These results indicate that conventional gestures combined with NDW were most strongly associated with language scores.
Limitations
There are several limitations to this study that should be considered. First, the relatively small sample size resulted in reduced power for statistical analyses. A larger sample might provide clearer evidence for the relative performance of the play and gesture measures. Furthermore, a crafted sample that included half children with DLD and half typically developing children was used for this study, which is quite different than the estimated occurrence of DLD in the population. While a crafted sample is appropriate for a preliminary study, caution needs to be exercised in interpreting results and generalizing findings to the general population. Future studies should include samples that more closely resemble the occurrence of DLD in the population. The nature and size of the sample (Mexican American backgrounds) also limit the extent to which these results can be generalized to children from Latino populations residing in the United States or other countries. A final limitation of this study is that it does not include information about specific developmental goals and interventions being applied through Part C or Early Head Start services. Given the age of the children in this study, it is very likely that both programs included play and may have encouraged gesture, but this information was not gathered. Future studies should collect this information.
Conclusion
Findings from the present study provide mixed results for the general symbolic deficit hypothesis. The significant association between gesture and language scores and the moderate association between the complexity of play and language scores provide some support for the hypothesis that children with DLD may possess a general symbolic deficit. However, when combined, complexity of play and conventional gesture accounted for a significant but small percentage of variability in language skills. It may be that children with DLD do have a general symbolic deficit and that individual symbolic behaviors are influenced in different ways at different times in development. These results are consistent with early studies that have shown that the associations between gesture, play, and language skills change over time, and individual symbolic skills and their associations with language may evolve over time (Capone & McGregor, 2004).
Findings from the present study also indicate that complexity of play is broadly associated with language skills, but complexity of play does not appear to significantly differ in DLD and typical Spanish-speaking toddlers. Play measures may still be useful in developmental screenings or assessments with Spanish-speaking toddlers, especially in cases where cognitive or other developmental disabilities are suspected. While measures of play did not differentiate between DLD and TL groups, a play-based format did lend itself to a naturalistic opportunity to observe communicative development. Language sampling measures collected through semi-structured play schemas yielded very promising results, in that the language skills of children with DLD differed from those of typical peers. Language sampling has been described as another ecologically valid and accurate way to obtain information about the language skills of Spanish-speaking toddlers and preschool-age children (Gutiérrez-Clellen et al., 2000; Gutiérrez-Clellen & Simon-Cereijido, 2009). The results from the present study provide additional support for the inclusion and meaningful engagement of parents during screening and assessment activities, and the use of language sampling as a measure of linguistic development in Spanish-speaking toddlers.
Implications for Practice
Knowledge gained from this study may be useful to practitioners in many ways. Preliminary analyses showed that observing conventional gesture and vocabulary usage may be especially useful measures during play-based assessments with young Spanish-speaking children. The play measures applied in this study were not strongly associated with language development, and the groups did not differ significantly on these measures. This means that relying on these play measures alone may not assist in identifying children with DLD. However, the method of play-based assessment (in this case, parent-facilitated semi-structured play activities) appears to be a useful way to gain information about communication development.
How can nonbilingual SLPs use this information to screen and assess the language of young Spanish-speaking children, given that they do not have the language skills to collect Spanish-language measures? Nonbilingual SLPs can team with a bilingual early childhood professional or a cultural mediator/interpreter who can assist in collecting and recording language data during parent-facilitated play interactions. Working with these other professionals requires careful coordination, planning, and time for preassessment and postassessment meetings but can be very effective (Guiberson & Atkins, 2012). These bilingual early childhood professionals should receive specialized training on play-based assessment and bilingual language development. This investment in developing an early childhood bilingual assessment team will result in improved outcomes for children with DLD who are from linguistically diverse backgrounds.
The use of developmental observation and semi-structured parent-facilitated activities parallels the call for the use of rich behavioral observation, informed opinion, and multiple sources of data when appraising a child’s development (Durán, Cheatham, & Santos, 2011; Guiberson & Banerjee, 2012; Neisworth & Bagnato, 2011). A structured play-based approach provides a meaningful way to engage Spanish-speaking families in the assessment/screening process and provides important developmental information. Results from the present study indicate that a gesture usage rubric combined with language sampling measures (e.g., NDW and word combinations children produced) would yield diagnostically important information.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received the following financial support for the research, authorship, and/or publication of this article: This project was supported by grants from the National Center for Research Resources (5P20RR016474) and the National Institute of General Medical Sciences (8 P20 GM103432) from the National Institutes of Health.