Exploring Vocabulary Growth in African American English–Speaking Toddlers: Bridging Linguistic and Cultural Perspectives

Abstract

Purpose:

This study examined vocabulary growth in low-income African American English (AAE)–speaking toddlers aged 30 and 42 months. Specifically, we investigated whether traditional child-level language production, discourse measures, or maternal language input at Time 1 predicted receptive vocabulary growth at Time 2.

Method:

Language samples were collected from 101 African American toddlers during home-based play interactions with their caregivers at 30 and 42 months. Child and maternal language measures were derived from transcripts using Systematic Analysis of Language Transcripts, and receptive vocabulary was assessed at both time points using the Peabody Picture Vocabulary Test–Revised. Hierarchical multiple linear regression analyses examined the predictive value of child- and caregiver-level language variables, including interactions with gender.

Results:

Children demonstrated substantial vocabulary growth from 30 to 42 months. Caregivers demonstrated responsive, developmentally appropriate language and frequent spontaneous questioning. However, neither traditional syntax measures (mean length of utterance [MLU]) nor culturally relevant measures of discourse (caregivers' use of spontaneous questions to direct discourse and children's responses to questions) significantly predicted vocabulary growth. A marginal trend suggested that the relation between maternal MLU and child vocabulary growth may differ by gender.

Conclusions:

Although neither syntax nor discourse measures significantly predicted vocabulary, these findings highlight the importance of continuing to examine culturally embedded interaction patterns in AAE-speaking families. The results suggest that broad language measures may obscure variability in the discourse strategies caregivers use to guide and engage children's language and that more fine-grained approaches may be needed to capture how language input supports vocabulary growth across gender and context.

Supplemental Material:

https://doi.org/10.23641/asha.31894576

The acquisition of a varied and extensive vocabulary plays an important role in fostering effective communication, enhancing critical thinking abilities, and promoting learning across diverse subject areas (Storch & Whitehurst, 2002). In early childhood, commonly used language sampling measures such as mean length of utterance (MLU) are highly related to vocabulary development (Bates & Goodman, 1997; Dethorne et al., 2005; Moyle et al., 2007). Research based on these commonly used language sampling measures has suggested that children's vocabulary and syntactic development may support one another in both typically and atypically developing language populations (Bowerman, 1982; Moyle et al., 2007).

However, the pathways to vocabulary development are determined not solely by children's individual cognitive processes but also by cultural factors that shape language use and acquisition. Cultural influences such as language spoken at home, community norms, and societal values can significantly impact the vocabulary acquisition process (Champion et al., 2003; Roberts et al., 1999). For very young children, the language interactions with parents and caregivers present a particularly important context for language development (Brushe et al., 2025).

Caregiver input provides both the quantity and quality of linguistic experience that supports children's emerging lexical and syntactic language development (Fernald & Weisleder, 2015; Huttenlocher et al., 2010). Research with caregiver language interactions indicates that caregiver language practices such as expanding on children's utterances, using diverse vocabulary, and maintaining conversational contingencies can all support children's development of larger vocabularies and more complex syntax (Hoff, 2003; Pace et al., 2022; Pan et al., 2005). However, these caregiver practices may differ as a function of child gender, which in turn may facilitate gender-based differences in young children's vocabulary development (Heath, 1983; Huttenlocher et al., 2010; Rowe et al., 2005; Schieffelin & Ochs, 1986).

Importantly, many of these studies focus on General American English (GAE)–speaking populations. Much of the extant vocabulary development research that includes African American children has not contextualized their cultural and linguistic communication within context, and African American English (AAE)–speaking children have traditionally been framed from a deficit perspective that positions African American children as lagging behind their GAE-speaking peers in vocabulary development (Avineri et al., 2015; Johnson et al., 2020). The extent to which AAE-speaking children's syntax drives their vocabulary development remains unclear. Similarly, there is a critical research gap for understanding the role of caregiver discourse quality in predicting vocabulary growth in AAE-speaking children, and this gap is particularly evident in studies involving young children from low-income AAE-speaking families, whose everyday communicative practices remain understudied (Shimpi et al., 2011; Washington & Craig, 2002).

The current study examined both AAE-speaking child and caregiver language measures associated with vocabulary development during naturalistic caregiver–child interactions in semistructured play sessions conducted in the home. This study extends existing research by examining caregiver–child exchanges (including caregivers' spontaneous questions and child responses to questions [RTQs]) along with caregiver and child syntax (i.e., mean length of utterance in words [MLUw]) in an underrepresented population, thereby providing a more ecologically valid understanding of early vocabulary development within the linguistic and cultural contexts of AAE-speaking households.

Syntax and Vocabulary Development

The investigation of vocabulary growth in young children represents a foundational area of research on developmental language acquisition. Understanding how vocabulary develops over time is critical for identifying the factors that influence language development and for informing interventions aimed at promoting linguistic ability and academic success. Most research on young children's vocabulary development has focused on White, middle-income, GAE-speaking children (Rowe, 2012). Among this population of children, syntactic complexity tends to be highly correlated with vocabulary. In fact, some researchers have suggested that syntactic complexity is not statistically distinct from vocabulary during the early childhood years (Bates & Goodman, 1997; Dethorne et al., 2005; Moyle et al., 2007).

In research with syntactic complexity and vocabulary development, syntactic complexity has most commonly been measured with MLU (Eisenberg et al., 2001). The popularity of MLU as a measure is likely tied to the ease of its calculation with commonly available transcription software and to the fact that other, more complex metrics of syntax (e.g., Index of Productive Syntax) are more difficult to calculate and tend to overlap significantly with MLU in their measurement (Horton-Ikard et al., 2005).

Prior longitudinal work has demonstrated strong bidirectional relationships between MLU and vocabulary measures, such that advances in one domain facilitate growth in the other (Bowerman, 1982; Moyle et al., 2007). Accordingly, MLU provides an empirically validated indicator of children's productive language that is expected to relate to vocabulary growth over time. Although MLU has been widely accepted as an important measure of syntax in relation to vocabulary development, some researchers have raised concerns about MLU as a sensitive metric. In a review of research on MLU as a clinical indicator of children's language development, Eisenberg et al. (2001) noted that MLU's sensitivity may be limited beyond ages 36–54 months, and the extent to which MLU provides reliably meaningful variance largely depends on elicitation contexts (home vs. lab, parent vs. clinician, narrative vs. free play). Other researchers have argued that MLU remains a reliable and sensitive metric of syntax across the early school-aged years (Potratz et al., 2022). The sensitivity of MLU may also depend on the variety of language it is measuring. In fact, Portelance et al. (2023) found that, across several linguistic and dialectal varieties, MLU was not predictive of all vocabulary acquisition but rather acquisition for specific parts of speech—predicates (e.g., verbs and modifiers) and function words (i.e., articles, prepositions, and pronouns).

Discourse and Vocabulary Development

Examining child language alone paints an incomplete picture of the sociolinguistic factors that drive children's vocabulary development (Johnson et al., 2020). Vocabulary development is highly impacted by input and cultural language practices, and prior work highlights the importance of interactive discourse processes (e.g., conversational contingency and child engagement) for vocabulary growth (Rowe, 2012; Weisleder & Fernald, 2013). Building on this work, recent research has operationalized children's responses to caregiver questions as a meaningful indicator of children's participation in discourse, demonstrating that both mother's rate of asking questions and children's responses to those questions predict children's later receptive vocabulary outcomes (Pace et al., 2022).

In particular, mother–child interactions and the quality of these interactions play a critical role in the development of children's language skills and their linguistic growth trajectories (Justice et al., 2019; Rowe, 2012). For vocabulary acquisition specifically, research has indicated that maternal discourse techniques may play an important role in supporting children's acquisition of new vocabulary words (Fernald & Weisleder, 2015; Hoff, 2003; Pace et al., 2022; Pan et al., 2005). For example, when mothers expand and enhance children's utterances (e.g., a child says, “Hot,” and a mother replies, “The stove is hot”), it is believed to encourage children to use more words (Hoff, 2003). When mothers jointly engage in children's attentional interests and maintain conversational contingencies (e.g., a mother questions, “You cooking me breakfast?” to a child who is engaged with a play kitchen), it is believed to capitalize on children's natural interests to model new vocabulary words (Hoff, 2003).

Caregiver input also appears to be a particularly important factor in accounting for gender differences in vocabulary development. Gendered language differences often reflect broader socialization practices in which caregivers engage boys and girls differently (Heath, 1983; Schieffelin & Ochs, 1986). In fact, after accounting for parent speech as a predictor of vocabulary development, Huttenlocher et al. (2010) found that gender was not a significant predictor of vocabulary growth in children aged 26–46 months. Similarly, Rowe et al. (2005) found that gender differences in vocabulary growth among low-income families were partially explained by variation in the quantity and diversity of parent input.

AAE-Speaking Children's Vocabulary Development

While there is extensive research on language development in the early acquisition stage for middle-income, White children, low-income African American children have often been excluded from these investigations (Horton-Ikard & Ellis Weismer, 2007). Studies that do include low-income, African American children frequently compare them to middle-income, White children, perpetuating deficit-based descriptions (Johnson et al., 2020). For example, Hoff's (2003, 2006) comparative analyses exemplify deficit framing by establishing the language practices of White, middle-class families as the developmental ideal and treating departures from this norm as deficits.

AAE-speaking toddlers are immersed in communities where AAE is the primary linguistic variety, influencing their vocabulary acquisition through interactions with caregivers, peers, and exposure to AAE linguistic features. Language socialization highlights how children learn language norms, values, and identities within their sociocultural context (Watson-Gegeo & Bronson, 2013). Ignoring these influences can lead to an incomplete understanding of language use and development. AAE-speaking toddlers undergo language socialization processes that involve acquiring vocabulary specific to AAE, including culturally relevant words and expressions that reflect their community's linguistic norms and cultural heritage.

There is a dearth of studies focused on vocabulary growth in AAE-speaking toddlers. For instance, Arriaga et al. (1998) reported that low-income toddlers scored significantly lower on expressive vocabulary measures than their middle-income peers, as measured by the MacArthur Communicative Development Inventory (CDI), although the study did not disaggregate the results by race. Failure to disaggregate limits understanding of how linguistic and cultural factors uniquely influence language development within specific groups, such as AAE-speaking children, and risks masking within-group variability and culturally grounded strengths in language use. Similarly, Pan et al. (2004) examined parent-reported vocabulary in a racially diverse low-income sample and found that, while all toddlers showed growth between ages 2 and 3 years, African American children had smaller vocabularies on average than their White peers. Notably, the predictive validity of parent-reported vocabulary (via the CDI) varied by context, being stronger in rural, predominantly White families than in urban, predominantly African American families.

The few studies that focus exclusively on African American children's vocabulary development have emphasized that variations in home environments and socioeconomic resources may be important factors for vocabulary acquisition. For example, in an observational study of low-income African American toddlers' grammatical and vocabulary development, Roberts et al. (1999) found a linear growth in vocabulary size, use of irregular forms, and utterance length between 18 and 30 months. Using parent-reported data from the short form of the MacArthur CDI, the study revealed that children from more responsive and stimulating home environments demonstrated significantly greater language gains. The authors also noted substantial individual variability and raised concerns about potential underreporting and the appropriateness of CDI norms for this population. In an experimental study of vocabulary use and acquisition among African American toddlers, Horton-Ikard and Ellis Weismer (2007) examined vocabulary and word learning abilities in a sample of 30 children aged 30–40 months. They found that low-income African American toddlers (n = 15) scored significantly lower than middle-income African American toddlers (n = 15) on receptive and expressive vocabulary tests. However, no significant difference was observed between the two groups in terms of their ability to learn novel words. This finding draws attention to the complexity of language development and the need for further research to understand the underlying mechanisms driving vocabulary acquisition in diverse populations, including low-income African American toddlers.

Syntax and Vocabulary Development for AAE-Speaking Children

For AAE-speaking children, MLU may simply be one of many quantitative measures of syntax that are related to age but not necessarily to language quality during the early childhood years of language development (Craig & Washington, 2000; Horton-Ikard et al., 2005). Quantitative metrics of syntax such as MLU may not be able to reliably distinguish between typically developing, AAE-speaking toddlers with lower expressive output and those who have language impairments (Craig & Washington, 2000). Extant research has not examined MLU as a predictor of AAE-speaking children's vocabulary development, and the extent to which AAE-speaking children's syntax functions as a driver of new vocabulary word acquisition is unclear.

Discourse and Vocabulary Development for AAE-Speaking Children

Research with AAE-speaking families has established that caregiver–child discourse quality may reflect important aspects of language use and development and that these might differ by gender. However, it is important to note that much of this research has not examined African American cultural norms for what “supportive” discourse should be. Traditional assumptions about “supportive” language input have largely been shaped by studies of White, middle-class families, which privilege child-centered, elaborative, and low-directive speech as the hallmark of optimal language environments (see, e.g., Hart & Risley, 1995; Rowe, 2008, 2012). When such GAE-based criteria are applied to AAE-speaking families, culturally grounded discourse styles (e.g., directives, topic shifts, call and response) are often misinterpreted as “less supportive and responsive,” potentially overlooking the socially meaningful functions of caregiver speech in AAE-speaking households (Kuchirko, 2017; Shimpi et al., 2011; Washington & Craig, 2002).

When African American parents' discourse quality is considered from a culturally sensitive perspective, an oft-overlooked strengths-based perspective may emerge. For example, Dyer et al. (2014) investigated differences in mother–child interactions among African American and Latin American low-income families, reporting that, for African American mothers, both directive mothering and child-centered mothering were associated with positive child language outcomes. Additional, recent research has echoed the potential strengths of cultural linguistic differences in the direct and safety-oriented communication style of AAE-speaking communities (Iruka et al., 2015; Shimpi et al., 2011; Tamis-LeMonda et al., 2014). For example, Shimpi et al. (2011) demonstrated that, in a sample of low-income African American mother–infant dyads, mothers' maternal directive speech, although often negatively characterized in mainstream discourse, was positively associated with vocabulary development in this population. These findings underscore the importance of rethinking traditional assumptions about what constitutes supportive language input.

For African American caregivers, a growing body of research suggests that directive language practices can be beneficial for child language development, whereas they may not be beneficial for children from other cultural backgrounds. In particular, in early childhood, African American mothers tend to use noninverted questions (i.e., a statement delivered with rising intonation, such as “It's gonna be a surprise?”) as conversational guides to invite children to respond, interject, and adapt behavior as conversational partners (Washington & Craig, 2002). Thus, African American mothers' spontaneous questions (including noninverted questions and traditional interrogatives beginning with question words such as “who” or “what”) and children's responses to those questions may be particularly important features of discourse to consider for AAE-speaking children's language development. For AAE speakers, the use of questions reflects a culturally sanctioned approach to directing children's attention during language interactions (Washington & Craig, 2002).

Although this line of research is still emerging, recent work suggests that examining the nature of caregiver–child language interactions with an asset-based perspective is important for understanding the role that caregivers play in shaping African American children's language trajectories and that this may differ across children's genders. For example, based on an examination using the Early Childhood Longitudinal Study, Birth Cohort, data set, Iruka (2017) stressed the importance of accounting for contextual risk and promotive factors that may be particularly relevant for African American families when interpreting gender differences in early language development. Iruka noted that African American boys may be significantly less likely to receive sensitive parenting practices (i.e., responsiveness to child cues, scaffolded teaching, positive feedback, and validation of child perspectives) than African American girls and that these more directive interactions may in fact be beneficial for later language and cognitive development. However, the way that these parenting practices emerge linguistically during caregiver–child language interactions remains unclear.

Although research has consistently suggested that there may be gender-based differences in African American children's language development (Craig & Washington, 2002; Roberts et al., 1999), the extent to which language measures (i.e., MLU, number of different words, spontaneous questions, RTQs) map onto African American caregivers' language interactions and shape child language development and the extent to which these interactions may be moderated by child gender remain a research area in critical need of additional examination.

The Current Study

While significant strides have been made in clarifying the factors influencing vocabulary development, there remain gaps in research investigating vocabulary growth in African American toddlers. The extent to which measures of syntax predict vocabulary growth is not well understood for this population. The role of caregiver–child interaction influence on vocabulary growth across gender has yet to be fully explored within this population. This study aims to fill those gaps by investigating vocabulary growth in AAE-speaking toddlers.

The current study examined vocabulary use among low-income African American toddlers and their maternal caregivers, who speak AAE, addressing a critical gap in research by focusing specifically on vocabulary growth in low-income AAE-speaking toddlers aged 30–42 months during caregiver–child interactions. By examining caregiver–child interactions in naturalistic settings where AAE is the primary linguistic variety, we aimed to evaluate whether commonly used language measures, such as MLU, which indexes caregivers' syntactic inputs and children's syntactic complexity, and culturally relevant measures of discourse (caregivers' use of spontaneous questions to direct discourse and children's RTQs) effectively capture and predict vocabulary growth in AAE-speaking toddlers. Our goal was to foreground the linguistic strengths within this population and challenge deficit-based narratives that have historically shaped language research on low-income African American families.

The following research questions were posed:

1. Is vocabulary at 42 months old, measured by scores on the Peabody Picture Vocabulary Test–Revised (PPVT-R), predicted by widely used syntax measure (child MLU), by the quality of the child responsiveness during discourse (RTQs), or by both simultaneously?

2. Are there interactions between syntax (child MLU), quality of child responsiveness to discourse (RTQs), and gender that predict vocabulary growth among low-income AAE-speaking toddlers?

Furthermore, we explored whether caregiver-level language measures might offer additional explanatory power for AAE-speaking children's vocabulary growth:

• Are caregiver-level predictors, including maternal/caregiver syntactic complexity (MLU) and quality of discourse (spontaneous questions), associated with vocabulary growth in AAE-speaking toddlers?

• Does child gender interact with either maternal/caregiver syntactic complexity (MLU) or quality of maternal/caregiver discourse (spontaneous questions) to predict vocabulary growth in low-income AAE-speaking toddlers?

Method

Participants

The participants were selected from a larger longitudinal study focused on examining the development of self-regulation and its relationship to subsequent academic achievement among Hispanic and African American children from low-income households living in a large city in the Southwestern United States. Consent for participation was provided by the primary caregivers. Eligibility for inclusion in the original study required that families had income that fell below 200% of the federal poverty level at the time of data collection, at least one parent identified as either Hispanic or African American, the child was not hospitalized for more than 7 days after birth, and the family intended to remain in the area for at least the next year. A total of 407 children (183 African American and 224 Hispanic) participated in eight waves of data collection: aged 2.5 years, age 3.5 years, kindergarten, first grade, and fourth through seventh grade. Time 1 data collection occurred between November 2009 and February 2011, and Time 2 data collection was concluded in June 2012. Details regarding the recruitment process can be found in Caughy et al. (2013).

Ethics Approval and Consent to Participate

This research was reviewed and approved by the University of Georgia Institutional Review Board (IRB; Human Research Protection Program) as continuing review of the parent project The Role of Fathering in the Language Development Among Young, Low-Income African American and Latino Children (IRB ID: PROJECT00010184; expedited categories: 5 and 6; approval: September 9, 2024; progress report due: September 9, 2027). The IRB determined that the present study involves continued analysis of data collected under an earlier approved protocol (STUDY00003868) and presents no more than minimal risk. The IRB granted a waiver of informed consent under 45 CFR 46.116(f)(3) for secondary analysis of identifiable private information.

The current study included only African American mothers or the female caregiver (n = 183) and toddlers, as the female caregivers represented the majority of participants (95%) identified as “primary caregivers” enrolled in the larger longitudinal study. Ten children were excluded from the analyses because their primary caregiver during the interaction was their father (n = 10, which was too small to meaningfully include in the current study; given that prior research has suggested gender may impact parent–child language interactions, we focused on mothers, as they represented the majority of primary caregivers in the sample). Furthermore, four children were excluded because their mothers were White. Finally, we excluded two children who were diagnosed with a significant developmental disability, 23 children who were lost to follow-up at Time 2, four cases with missing video data at Time 2 due to experimenter error, 38 who had missing Peabody Picture Vocabulary Test (PPVT) scores, and one child that did not produce any verbal utterances during the play interaction. Those families who were lost to follow-up did not differ from those who were not in terms of child gender, family poverty status, or maternal education.

The remaining 101 cases comprised the sample for the present analysis. Specifically, there were 101 African American toddlers aged 30 months (M = 29.78 months, SD = 0.74) at the Time 1 and aged approximately 42 months (M = 41.18 months, SD = 1.38) at Time 2, and their female caregivers. The sample consisted of 51 boys and 50 girls. To confirm caregiver dialect, each transcript was coded for AAE features (e.g., zero copula, invariant BE, zero possessive). All participants lived in homes where AAE was spoken by the primary care provider.

Procedure

All language measures used as predictors in the present study (child MLU, child RTQs, and caregiver MLU and spontaneous questions) were derived from the 30-month (Time 1) language sample. The 42-month (Time 2) session was used only for outcome measurement of children's receptive vocabulary (PPVT). Time 1 data collection sessions involved a play interaction between the toddlers and their caregivers lasting approximately 15 min at age 30 months. The play sessions took place in the participants' homes to ensure a familiar and comfortable environment and were modeled after the National Institute of Child Health & Human Development (NICHD) Study of Early Child Care and Youth Development “three bags” protocol (NICHD Early Child Care Research Network, 1999). At Time 1, the caregiver and the child were presented with three numbered bags, each containing a different book or toy. Bag 1 included a picture book called “Good Dog Carl” by Alexandra Day, Bag 2 included a small kitchen with a frying pan and salt-and-pepper shakers, and Bag 3 included a small playhouse with people and a vehicle. The caregiver was instructed to start with the first bag and proceed in numbered order, spending as much time as desired with each activity. Caregivers were not provided with a script or additional language instructions for these interactions. The sessions were video-recorded for subsequent transcription and analysis. At both time points, trained research personnel also administered standardized assessments of receptive vocabulary (PPVT) in the home.

Transcription Process

Transcription was completed by a team of trained graduate students under the supervision of a speech-language pathologist (SLP). Prior to beginning the transcription, all graduate students underwent a training process led by the SLP, which included guided practice and feedback until they reached at least 80% reliability on a sample transcript. Once trained, the graduate students transcribed the video recordings, and each transcript was subsequently reviewed and cleaned by a secondary transcriber on the team to ensure consistency and accuracy.

To establish transcript reliability, an independent SLP-trained observer retranscribed 25% of each transcript in 10% of the total sample (n = 10 transcripts). Agreement was determined by point-to-point comparison of the original transcripts with the independent retranscriptions, and the number of agreements and disagreements was recorded at both the utterance segmentation and word levels. Bound morphemes (e.g., plural, possessive, progressive verb form) were marked and included in the word agreement. Percent agreement was calculated as the number of agreements divided by the total number of agreements plus disagreements. Transcript reliability was found to be 99.63% at the utterance level and 99.44% at the word level, indicating a high degree of transcription accuracy across coders.

Measures

Syntax

For the current study, MLU was used as a measure of syntax. MLU is routinely used in child language research to measure children's syntactic complexity, and some researchers have suggested that MLU may be particularly well suited to capture the oral language strengths observed in African American children because rates of MLU produced by young children do not significantly differ between African American and White children in early childhood (Curenton & Justice, 2004). Importantly, while this may be true for MLU calculated based on words, MLU calculated based on morphemes may be unreliable for AAE speakers because morpheme markers are variably included depending on context (Horton-Ikard et al., 2005; Parker & Brorson, 2005). For the current study, MLUw was calculated using Systematic Analysis of Language Transcripts (SALT; Miller & Chapman, 2000). SALT calculates MLUw by counting the number of words in a language sample and dividing by the number of utterances the child produced. Unintelligible utterances are not included in the calculation. The MLUw was calculated for both caregivers and children in the current study as a measure of syntax.

Child Discourse

RTQs are an important discourse measure associated with toddlers' language use and development (Pace et al., 2022). RTQs, calculated as a percentage by SALT, capture the child's verbal RTQs during interactions (SALT; Miller & Chapman, 2000). In SALT, a response is defined as an utterance that immediately followed an utterance from the caregiver that was marked with a question mark, indicating contingent responsiveness within the conversational exchange. Questions were identified during transcription and checked for reliability. Specifically, utterances including requests for clarification; intonation prompts (including noninverted questions); open-ended, yes/no, or other types of questions were all labeled as question prompts. RTQs were calculated by dividing the number of verbal responses (gestures were not counted) by the total number of questions from the communication partner (SALT; Miller & Chapman, 2000). By examining RTQs, we gain valuable insight into how actively children engage in conversations and how the quality of these exchanges may shape language development in AAE-speaking toddlers.

Caregiver Discourse

Spontaneous questions were included as a key measure to capture the interactive quality of the language environment, consistent with prior work linking caregiver questioning to later vocabulary outcomes (Pace et al., 2022). This measure provides insight into the extent to which caregivers encourage verbal engagement, prompt responses, and structure interactions around child participation. Spontaneous questioning is particularly valuable in assessing discourse quality, as it reflects caregivers' efforts to create conversational opportunities, which can support vocabulary learning through responsive and contingent input. Spontaneous questions represent a specific category of spontaneous utterances, which are identified using the question mark in transcription. In accordance with the SALT (Miller & Chapman, 2000) protocol, which automatically identifies spontaneous utterances using punctuation codes (e.g., “.” or “!” or “?”), spontaneous utterances were defined as those that were not RTQs (i.e., not spontaneous), responses to intonation prompts (i.e., questioning intonation), or imitations of a prior utterance. Thus, spontaneous questions are spontaneous utterances that include open-ended prompts, yes/no questions, clarification requests, and other interrogative forms that are not imitations. This operationalization ensures that only naturally produced, nonechoed utterances are included in the spontaneous question count. Given that young children may not frequently initiate questions themselves (as was true in the current study), focusing on caregiver questioning provides a complementary view of the interactional dynamics and the linguistic input available to the child.

Furthermore, in AAE-speaking families, spontaneous questions often include noninverted question forms (e.g., “It's gonna be a surprise?”) that rely on rising intonation rather than syntactic inversion to mark interrogatives. These structures are a natural and culturally grounded way of guiding conversation and inviting child participation (Washington & Craig, 2002). Including spontaneous questions as a discourse measure therefore captures a meaningful feature of caregiver–child interaction in AAE households, reflecting communicative patterns that support engagement and language learning.

Receptive Vocabulary

The PPVT-R (Dunn & Dunn, 1981), a standardized assessment of receptive vocabulary, was administered. Children were asked by the examiner to point to one of four pictures that corresponded with the stimulus word provided. Raw scores are calculated by counting the number of correct responses including the items before the entry point. Raw scores were used to capture actual growth rather than age-normed standing, an approach recommended for children from low-income or nonmainstream language backgrounds (Vagh et al., 2009), and results were consistent with analyses using standard scores (Sullivan et al., 2014; see Supplemental Material S1).

Analysis Methods

To investigate the factors predicting vocabulary growth from 30 to 42 months of age, we conducted a hierarchical linear regression with one data point per child. The outcome for these analyses was PPVT raw vocabulary score at 42 months, and each predictor was entered into the regression in a planned, stepwise order such that the unique contributions of predictors could be examined. Given that the primary outcome was vocabulary growth (change in vocabulary across 30–42 months), each regression controlled for the PPVT raw score at Time 1 (30 months; i.e., these were residualized change regression models; Castro-Schilo & Grimm, 2017).

In Step 1 of the planned hierarchical regression, we entered PPVT at Time 1 (30 months) and child gender (boys = 0) as control variables. Step 2 introduced the primary child-level predictors: syntax, measured by MLU at 30 months, and discourse, measured by RTQs at 30 months. In Step 3, we included two interaction terms, MLU–gender and RTQs–gender, to explore whether these relationships varied by gender. Prior to conducting the regression analysis, collinearity diagnostics were conducted to assess multicollinearity among the predictors. All variables demonstrated acceptable variance inflation factor (VIF) values (less than 10) and tolerance values (greater than 0.1), indicating that multicollinearity was not a concern.

Next, we explored whether maternal language measures might better explain individual differences in children's vocabulary development. In this secondary analysis, Step 1 included PPVT at 30 months old and child gender (boys = 0) as control variables. In Steps 2 and 3, we examined maternal MLU and spontaneous questions as predictors of child vocabulary growth, both independently and in interaction with child gender. As with the child-level models, caregiver variables demonstrated acceptable VIF values (less than 10) and tolerance values (greater than 0.1), indicating that multicollinearity was not a concern.

Results

Descriptive Statistics

Descriptive statistics for child language measures are presented in Table 1. Toddlers with typical language development aged 27–30 months should have, on average, a mean length of utterance in morphemes (MLUm; calculated based on morphemes) between 2.0 and 2.5 with an MLUm of 2.25 (Brown, 1973). Similar MLU rates have been reported in other research with African American toddlers aged approximately 30 months (MLUm: M = 2.70, SD = 0.34 for age 30 months and M = 3.08, SD = 0.48 for age 42 months; Horton-Ikard et al., 2005), African American children aged 36–51 months (MLUm: M = 3.38, SD = 0.74; Qi et al., 2011), and African American preschoolers and kindergarteners aged 59–65 months (MLU in C-units based on the first 50 words uttered in the language sample: M = 3.15, SD = 0.72; Craig & Washington, 2002). Thus, much of what is developmentally expected for AAE-speaking children reported in the literature varies as a result of participants' socioeconomic status, gender, language sampling technique, and MLU calculation technique. The average MLUw for the children in the current study was M = 1.91 (SD = 0.43) at age 30 months and M = 2.50 (SD = 0.58) at age 42 months (see Table 1). Given that parents/caregivers tend to adjust their MLUs to match the outputs of their children, it is not surprising that the average MLUw for caregivers was similar to that of our child participants (M = 3.63, SD = .61 at age 30 months and M = 3.52, SD = 0.54 at 42 months; see Table 1).

Table 1.

Descriptive statistics of child and caregiver language measure variables by age and gender.

Variables	30 months 29.8 (0.74)	30-month boys 29.7 (0.83)	30-month girls 29.9 (0.63)	42 months 41.2 (1.39)	42-month boys 41.27 (1.27)	42-month girls 41.08 (1.51)
Child-level variables
MLU	1.91 (0.43)	1.86 (0.38)	1.96 (0.47)	2.50 (0.58)	2.47 (.50)	2.54 (0.65)
Responses to questions	0.56 (0.16)	0.56 (0.15)	0.55 (0.17)	0.57 (0.24)	0.63 (0.14)	0.52 (0.29)
Raw PPVT scores	9.43 (6.40)	9.51 (6.59)	9.34 (6.27)	24.70 (12.68)	24.45 (12.97)	24.94 (12.53)
Caregiver-level variables
Maternal MLU	3.63 (0.61)	3.59 (0.57)	3.68 (0.65)	3.52 (0.54)	3.58 (0.47)	3.45 (0.60)
Maternal spontaneous questions	63.03 (31.74)	60.84 (30.34)	65.26 (33.36)	34.65 (21.38)	36.00 (20.48)	33.28 (22.39)

Note. MLU = mean length of utterance; PPVT = Peabody Picture Vocabulary Test.

As expected, children showed clear developmental gains between 30 and 42 months. Children's RTQs remained consistent across age and gender groups. Raw PPVT scores showed a marked increase from 30 to 42 months, reflecting growth in receptive vocabulary during this developmental period. None of the differences between boys and girls were statistically significant.

Descriptive statistics for maternal language measures are presented in Table 1. Maternal MLU remained stable across child age groups, while the frequency of spontaneous questions was highest at 30 months and decreased by 42 months. The decline in spontaneous questions may reflect task differences across sessions or developmental shifts in how mothers scaffold language as their children's verbal abilities increase. Although the analytic models used only 30-month (Time 1) language measures as predictors of children's vocabulary growth, descriptive statistics for both time points are presented to provide developmental context.

Regression Analyses

Table 2 presents model summary results for all child-level regression models. These analyses used a residualized change approach, predicting children's PPVT raw scores at 42 months old while controlling for PPVT raw scores at 30 months old (Castro-Schilo & Grimm, 2017). Model 1, which included PPVT at 30 months and gender, accounted for a significant portion of variance in PPVT at 42 months, F(2, 98) = 23.43, p < .001, R² = .32. In Model 2, the addition of child-level language predictors (MLU and RTQs) did not improve model fit (ΔF = 0.21, p = .81, ΔR² = .003). Neither MLU (b = 1.13, SE = 2.61, t = 0.43, p = .67) nor RTQs (b = 2.71, SE = 6.95, t = 0.39, p = .70) were significant predictors (see Figure 1). Model 3 introduced two interaction terms, gender by MLU and gender by RTQs, which again did not significantly improve the model (ΔF = 0.28, p = .75, ΔR² = .004). Overall, none of the child-level language measures or their interactions with gender significantly predicted vocabulary growth after controlling for earlier vocabulary scores. ¹

Table 2.

Linear regression results for child-level research questions.

Model	Variable	b	SE	t	p value	F test	R 2	ΔR2
1	Intercept	13.73	2.15	6.38	< .001***	23.43	.32
	PPVT_T1	1.13	0.17	6.84	< .001***
	Gender	0.68	2.10	0.32	.75
2	Intercept	14.07	2.24	6.29	< .001***	11.63	.33	.003
	PPVT_T1	1.10	0.17	6.36	< .001***
	Gender	0.60	2.14	0.28	.78
	Child MLU	1.13	2.61	0.43	.67
	Child RTQs	2.71	6.95	0.39	.70
3	Intercept	13.82	2.29	6.05	< .001***	7.73	.33	.004
	PPVT_T1	1.11	0.18	6.33	< .001***
	Gender	0.63	2.15	0.29	.77
	Child MLU	−1.15	4.14	−0.28	.78
	Child RTQs	4.91	10.15	0.48	.63
	Gender* MLU	3.86	5.31	0.73	.47
	Gender* RTQs	−4.77	13.86	−0.34	.73

Note. PPVT_T1 = Peabody Picture Vocabulary Test at 30 months; MLU = mean length of utterance; RTQs = responses to questions.

^* p < .05.

^*** p < .001.

Figure 1.

Relations between child language predictors and vocabulary growth by gender. Scatter plots depicting the relations between children's mean length of utterance and growth in receptive vocabulary (Peabody Picture Vocabulary Test [PPVT]) scores and between children's responses to caregiver questions and growth in receptive vocabulary (PPVT) scores from 30 to 42 months.

Table 3 presents the regression results for caregiver-level language measures. Model 1 included PPVT at 30 months and gender and explained a significant proportion of variance in PPVT at 42 months, F(2, 98) = 23.43, p < .001, R² = .32. Model 2 added maternal MLU and spontaneous questions but did not significantly improve the model (ΔF = 0.06, p = .95, ΔR² = .001). Neither maternal MLU (b = 0.32, SE = 1.81, t = 0.18, p = .86) nor spontaneous questions (b = 0.01, SE = 0.04, t = 0.24, p = .81) were significant predictors (see Figure 2). In Model 3, interaction terms were introduced, and while the overall model remained nonsignificant, F(5, 95) = 0.89, p = .49, R² = .04, the interaction between gender and maternal MLU approached significance (b = −6.99, SE = 3.66, t = −1.91, p = .06). This suggests a possible trend indicating that the relation between caregiver syntactic complexity and children's vocabulary growth may differ by gender. ²

Table 3.

Linear regression results for caregiver-level research questions.

Model	Variable	b	SE	t	p value	F test	R 2	ΔR2
1	Intercept	13.73	2.15	6.38	< .001***	23.43	.32
	PPVT_T1	1.13	0.17	6.84	< .001***
	Gender	0.68	2.10	0.32	.75
2	Intercept	13.98	2.31	6.05	< .001***	11.52	.32	.001
	PPVT_T1	1.12	0.18	6.10	< .001***
	Gender	0.61	2.13	0.28	.78
	MLU	0.32	1.81	0.18	.86
	Spontaneous questions	0.01	0.04	0.24	.81
3	Intercept	14.68	2.33	6.30	< .001***	8.42	.35	.03
	PPVT_T1	1.05	0.18	5.76	< .001***
	Gender	0.49	2.11	0.02	.82
	MLU	4.10	2.68	1.53	.99
	Spontaneous questions	0.01	0.05	0.13	.81
	Gender* MLU	−6.99	3.66	−1.91	.41
	Gender* SponQT	0.03	0.07	0.39	.99

Note. PPVT_T1 = Peabody Picture Vocabulary Test at 30 months; MLU = mean length of utterance; SponQT = spontaneous questions.

^* p < .05.

^*** p < .001.

Figure 2.

Relations between caregiver language predictors and vocabulary growth by gender. Scatter plots depicting the relations between caregiver's mean length of utterance and growth in receptive vocabulary (Peabody Picture Vocabulary Test [PPVT]) scores and between caregiver's spontaneous and growth in receptive vocabulary (PPVT) scores from 30 to 42 months.

We ran a post hoc power analysis using G*Power 3.1 (Faul et al., 2009). Given that model main effects for the current study are tendered by interaction effects, we focused on power analysis for the interactions. For the caregiver-level full model (n = 101; six predictors) a post hoc test of the overall R² indicated power = .35 to detect the small effect size of the parent/caregiver MLU–child gender interaction effect (Cohen ⨍² = .025). Thus, a sample size of n = 316 participants would have been needed to detect this effect at power = .80. Note that the additional interaction effect for parent/caregiver spontaneous questions–child gender was extremely small (Cohen ⨍² < .001) and likely not a practically significant effect. However, for the sake of transparency, we report that this effect was powered = .06 and would require a minimum of n = 7,851 participants to detect this effect at power = .80.

Discussion

This study examined whether traditional measures of language production (MLU), child RTQs, and caregiver discourse predicted vocabulary growth in low-income AAE-speaking toddlers between the ages of 30 and 42 months. It also explored how these relationships might differ by gender and whether caregiver language, measured by maternal MLU and spontaneous questions, provided additional insight into children's vocabulary development.

Despite children in the sample demonstrating significant vocabulary growth over the 1-year period, none of the child-level or caregiver-level predictors significantly explained that growth. Although traditional measures such as syntactic complexity and children's responsiveness during discourse appear to be important predictors of vocabulary acquisition for GAE-speaking children (Moyle et al., 2007; Pace et al., 2022), they were not significant predictors of vocabulary growth for AAE-speaking children, regardless of gender, during ages 30–42 months. Recent research has also emphasized the role of maternal/caregiver input as an important factor shaping the vocabulary development of young AAE-speaking children (Rowe, 2012; Weisleder & Fernald, 2013). However, in the current study, syntactic complexity (maternal caregivers' MLU) and spontaneous questioning during discourse were also not predictors of AAE-speaking children's vocabulary growth, regardless of child gender. Additional exploratory analyses revealed that, across other traditional measures of maternal caregivers' and children's language, we consistently found that these traditionally informative language metrics lacked explanatory power when applied to this population of children and families.

Traditional Syntax Metrics

MLU is so closely related to vocabulary that some researchers have suggested they may not be distinguishable from one another at this age for GAE-speaking children (Bates & Goodman, 1997), and research has suggested that lexical and syntactic development are intimately intertwined (Moyle et al., 2007). For GAE speakers, questioning during discourse offers opportunities to elaborate, expand, and probe, and thus, understanding children's RTQs may be a very important indicator of their later vocabulary development. Questions invite children to produce more language.

However, these traditional measures may not play the same important predictive role for AAE-speaking children's vocabulary development. For AAE-speaking children, MLU tends to increase with age, but lower MLU is not reliably indicative of impairment (Craig & Washington, 2000; Horton-Ikard et al., 2005). Extant literature has not established MLU as a predictor of AAE-speaking children's vocabulary development. Similarly, questioning during discourse looks different and plays a different communicative function for AAE speakers. Adults often use regular interrogatives and noninverted question forms (e.g., that's where it go?) to redirect children's attention, engage, and provide opportunities for scaffolding. Results from the current study suggest that these traditional, quantitative language metrics may indeed have less utility in predicting AAE-speaking children's vocabulary development than they do for GAE speakers. However, question frequency alone did not explain vocabulary outcomes, suggesting that the type and communicative purpose of questions may matter more than their quantity.

Parent/Caregiver Language Input

African American caregivers have been characterized as harsh, critical, and linguistically impoverished in the literature, and this is especially true of lower income African American mothers (Hammer & Weiss, 1999; Iruka et al., 2015). The idea that African American mothers provide their children with limited language input has been normalized as common knowledge, and it has been blamed for poor linguistic and later academic outcomes (Iruka et al., 2015).

In contrast to this characterization of African American mothers, caregivers in the current study demonstrated responsive and developmentally appropriate language interactions with their toddlers. They used shorter utterances aligned with their children's language abilities and sustained back-and-forth in ways that reflect natural scaffolding of language (Tamis-LeMonda et al., 2014). Importantly, mothers made frequent use of spontaneous utterances, which suggests an intentional effort to invite participation and elicit children's language (Washington & Craig, 2002). Caregivers consistently responded when children initiated questions, often elaborating with multiple responses, showing high levels of engagement and conversational responsiveness. Their language use included repetition alongside a reasonably diverse set of words (Weisleder & Fernald, 2013). Together, these patterns illustrate the richness and intentionality of maternal language input in AAE-speaking families. Our results emphasize African American mothers' linguistic competence when interacting with their children, adding to a growing body of literature challenging the way that African American mothers have been characterized and the way that AAE speakers use form and function to convey meaning to their children.

Prior research has also suggested that African American parents may provide differential input to their children as a result of gender (Iruka, 2017), and gender-based differences in children's patterns of language acquisition may be best explained by gendered interactions with caregivers (Huttenlocher et al., 2010). In the current study, however, maternal caregivers' inputs to boys and girls were not significantly different, and gender-based differences in parental input did not predict children's vocabulary growth. This being said, the interaction between maternal/caregiver syntax and child gender was very close to the margin for statistical significance. The current study may have been underpowered to detect the relatively small effect size for this interaction. This small effect size is also echoed in prior research (e.g., Huttenlocher et al., 2010), but gendered language interactions may be an important sociocultural dynamic for future researchers to consider.

Limitations

Although this study provides important insights into vocabulary development among AAE-speaking toddlers, some limitations should be noted. First, the sample was limited to families that experienced economic hardship. As such, the findings may not be generalizable to African American toddlers from middle- or high-income households, whose language environments and developmental experiences may differ in meaningful ways. Prior research has demonstrated socioeconomic differences in vocabulary size and MLU, with children from low-income backgrounds (Fernald et al., 2012; Hirsh-Pasek et al., 2015; Hoff, 2003). In the present study, children's MLUs were lower than those reported in studies of middle-income African American children. However, interpreting these findings through a developmental rather than deficit lens is essential. The goal is not to compare children to a monolithic standard but to understand the diversity of developmental trajectories within and across groups. Further research is needed to explore how economic, cultural, and contextual factors interact to shape language development across a wider range of African American families.

Additionally, this study focused on verbal measures of language input and response. Nonverbal and paralinguistic aspects of caregiver–child interaction, such as gesture, prosody, and affective engagement, were not analyzed, although these may play a key role in language development and should be included in future work. The use of naturalistic home-based interactions is a strength of the study, allowing for authentic observations of language use that better reflect children's real-world communicative environments; however, naturalistic home-based interactions also represent one of many linguistic contexts, which may influence measures such as MLU or spontaneous questions.

Conclusions

The development of a broad and diverse vocabulary is central to children's communication, thinking, and learning across domains (Storch & Whitehurst, 2002). In this study, receptive vocabulary not only increased but appeared to accelerate between 30 and 42 months for this sample of low-income AAE-speaking children, yet traditional predictors such as MLU and RTQs did not explain why. This disconnect signals that traditional language metrics may be capturing facets of language that have a different functionality for AAE speakers. As a result, we lack culturally grounded tools to accurately predict, assess, and support their language development. Thus, we may greatly benefit from reexamining AAE-speaking caregiver–child interactions to understand the true purposes, meanings, and functions of culturally embedded communication techniques that foster vocabulary growth. Sociolinguistic theory supports this critique, emphasizing that language learning is deeply embedded in cultural routines, pragmatic uses of language, and identity building (Green, 2012; Watson-Gegeo & Bronson, 2013). Children may be acquiring vocabulary through culturally meaningful interactions that are not fully visible in surface-level counts of utterances or syntactic complexity.

Future research should continue to explore the roles of input quality, cultural discourse styles, and interactional dynamics important in shaping language development among African American children. In particular, attention to the types of questions asked, the function of directive language, and gender-based differences in interactional styles may offer further insights into the mechanisms of vocabulary growth. Expanding our analytic tools to include culturally relevant measures will not only deepen our theoretical understanding but also support more equitable and effective practices in language assessment and intervention.

Author Contributions

Gennie R. Laramore: Conceptualization, Formal analysis, Data curation, Visualization, Methodology, Writing – original draft. Katherine T. Rhodes: Methodology, Formal analysis, Writing – review & editing, Visualization. Julie A. Washington: Conceptualization, Supervision, Writing – review & editing.

Data Availability Statement

The data sets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Funding Statement

This study was supported by the Eunice Kennedy Shriver National Institute of Child Health and Development (Grants 1R01HD058643 and 1R01HD100557, both awarded to Margaret Caughy).